ref: 40486d3641407466dfc08c7cddbee4a2f230758a
parent: 835d20a095ef12d9c6054d752241ed8b7d3cd20c
author: Ori Bernstein <ori@eigenstate.org>
date: Sat Mar 4 15:03:54 EST 2023
diff, merge3: refactor diff, implement merge3 Refactor diff internals to allow multiple diffs to be done in the same process. This allows a merge3 to be implemented off the guts of diff. Tests are added, files with no end of line terminator are currently broken.
--- /dev/null
+++ b/sys/src/cmd/diff/diff.c
@@ -1,0 +1,84 @@
+#include <u.h>
+#include <libc.h>
+#include <bio.h>
+#include "diff.h"
+
+void
+done(int status)
+{
+ switch(status)
+ {
+ case 0:
+ exits("");
+ case 1:
+ exits("some");
+ default:
+ exits("error");
+ }
+}
+
+void
+usage(void)
+{
+ fprint(2, "usage: %s [-abcefmnrw] file1 ... file2\n", argv0);
+ exits("usage");
+}
+
+void
+main(int argc, char *argv[])
+{
+ int i;
+ Dir *fsb, *tsb;
+
+ Binit(&stdout, 1, OWRITE);
+ ARGBEGIN{
+ case 'e':
+ case 'f':
+ case 'n':
+ case 'c':
+ case 'a':
+ case 'u':
+ mode = ARGC();
+ break;
+ case 'w':
+ bflag = 2;
+ break;
+
+ case 'b':
+ bflag = 1;
+ break;
+
+ case 'r':
+ rflag = 1;
+ break;
+
+ case 'm':
+ mflag = 1;
+ break;
+
+ case 'h':
+ default:
+ usage();
+ }ARGEND;
+ if (argc < 2)
+ usage();
+ if ((tsb = dirstat(argv[argc-1])) == nil)
+ sysfatal("can't stat %s", argv[argc-1]);
+ if (argc > 2) {
+ if (!DIRECTORY(tsb))
+ sysfatal("not directory: %s", argv[argc-1]);
+ mflag = 1;
+ } else {
+ if ((fsb = dirstat(argv[0])) == nil)
+ sysfatal("can't stat %s", argv[0]);
+ if (DIRECTORY(fsb) && DIRECTORY(tsb))
+ mflag = 1;
+ free(fsb);
+ }
+ free(tsb);
+ for (i = 0; i < argc-1; i++)
+ diff(argv[i], argv[argc-1], 0);
+
+ done(anychange);
+ /*NOTREACHED*/
+}
--- a/sys/src/cmd/diff/diff.h
+++ b/sys/src/cmd/diff/diff.h
@@ -1,13 +1,53 @@
-typedef struct Line Line;
+typedef struct Line Line;
+typedef struct Cand Cand;
+typedef struct Diff Diff;
+typedef struct Change Change;
struct Line {
int serial;
int value;
};
-extern Line *file[2];
-extern int len[2];
-extern long *ixold, *ixnew;
-extern int *J;
+
+struct Cand {
+ int x;
+ int y;
+ int pred;
+};
+
+struct Change
+{
+ int a;
+ int b;
+ int c;
+ int d;
+};
+
+struct Diff {
+ Cand cand;
+ Line *file[2], line;
+ int len[2];
+ int binary;
+ Line *sfile[2]; /*shortened by pruning common prefix and suffix*/
+ int slen[2];
+ int pref, suff; /*length of prefix and suffix*/
+ int *class; /*will be overlaid on file[0]*/
+ int *member; /*will be overlaid on file[1]*/
+ int *klist; /*will be overlaid on file[0] after class*/
+ Cand *clist; /* merely a free storage pot for candidates */
+ int clen;
+ int *J; /*will be overlaid on class*/
+ long *ixold; /*will be overlaid on klist*/
+ long *ixnew; /*will be overlaid on file[1]*/
+ char *file1;
+ char *file2;
+ Biobuf *input[2];
+ Biobuf *b0;
+ Biobuf *b1;
+ int firstchange;
+ Change *changes;
+ int nchanges;
+};
+
extern char mode;
extern char bflag;
extern char rflag;
@@ -14,19 +54,24 @@
extern char mflag;
extern int anychange;
extern Biobuf stdout;
-extern int binary;
#define MAXPATHLEN 1024
+#define DIRECTORY(s) ((s)->qid.type&QTDIR)
+#define REGULAR_FILE(s) ((s)->type == 'M' && !DIRECTORY(s))
+
int mkpathname(char *, char *, char *);
+char *mktmpfile(int, Dir **);
+char *statfile(char *, Dir **);
void *emalloc(unsigned);
void *erealloc(void *, unsigned);
void diff(char *, char *, int);
+void diffreg(char*, char*, char*, char*);
void diffdir(char *, char *, int);
-void diffreg(char *, char *, char *, char *);
-Biobuf *prepare(int, char *, char *);
-void panic(int, char *, ...);
-void check(Biobuf *, Biobuf *);
-void change(int, int, int, int);
-void flushchanges(void);
-
+void calcdiff(Diff *, char *, char *, char *, char *);
+Biobuf *prepare(Diff*, int, char *, char *);
+void check(Diff *, Biobuf *, Biobuf *);
+void change(Diff *, int, int, int, int);
+void freediff(Diff *);
+void flushchanges(Diff *);
+void fetch(Diff *d, long *f, int a, int b, Biobuf *bp, char *s);
--- a/sys/src/cmd/diff/diffdir.c
+++ b/sys/src/cmd/diff/diffdir.c
@@ -111,3 +111,45 @@
free(dirf);
free(dirt);
}
+
+void
+diff(char *f, char *t, int level)
+{
+ char *fp, *tp, *p, fb[MAXPATHLEN+1], tb[MAXPATHLEN+1];
+ Dir *fsb, *tsb;
+
+ fsb = nil;
+ tsb = nil;
+ if ((fp = statfile(f, &fsb)) == 0)
+ goto Return;
+ if ((tp = statfile(t, &tsb)) == 0)
+ goto Return;
+ if (DIRECTORY(fsb) && DIRECTORY(tsb)) {
+ if (rflag || level == 0)
+ diffdir(fp, tp, level);
+ else
+ Bprint(&stdout, "Common subdirectories: %s and %s\n", fp, tp);
+ }
+ else if (REGULAR_FILE(fsb) && REGULAR_FILE(tsb))
+ diffreg(fp, f, tp, t);
+ else {
+ if (REGULAR_FILE(fsb)) {
+ if ((p = utfrrune(f, '/')) == 0)
+ p = f;
+ else
+ p++;
+ if (mkpathname(tb, tp, p) == 0)
+ diffreg(fp, f, tb, t);
+ } else {
+ if ((p = utfrrune(t, '/')) == 0)
+ p = t;
+ else
+ p++;
+ if (mkpathname(fb, fp, p) == 0)
+ diffreg(fb, f, tp, t);
+ }
+ }
+Return:
+ free(fsb);
+ free(tsb);
+}
--- a/sys/src/cmd/diff/diffio.c
+++ b/sys/src/cmd/diff/diffio.c
@@ -4,10 +4,6 @@
#include <ctype.h>
#include "diff.h"
-static Biobuf *input[2];
-static char *file1, *file2;
-static int firstchange;
-
#define MAXLINELEN 4096
#define MIN(x, y) ((x) < (y) ? (x): (y))
@@ -104,7 +100,7 @@
}
Biobuf *
-prepare(int i, char *arg, char *orig)
+prepare(Diff *d, int i, char *arg, char *orig)
{
Line *p;
int j, h;
@@ -115,10 +111,10 @@
bp = Bopen(arg, OREAD);
if (!bp) {
- panic(mflag ? 0: 2, "cannot open %s: %r\n", arg);
+ sysfatal("cannot open %s: %r", arg);
return 0;
}
- if (binary)
+ if (d->binary)
return bp;
nbytes = Bread(bp, buf, MIN(1024, MAXLINELEN));
if (nbytes > 0) {
@@ -130,7 +126,7 @@
*/
cp += chartorune(&r, cp);
if (r == 0 || (r > 0x7f && r <= 0xa0)) {
- binary++;
+ d->binary++;
return bp;
}
}
@@ -139,14 +135,14 @@
p = emalloc(3*sizeof(Line));
for (j = 0; h = readhash(bp, buf); p[j].value = h)
p = erealloc(p, (++j+3)*sizeof(Line));
- len[i] = j;
- file[i] = p;
- input[i] = bp;
+ d->len[i] = j;
+ d->file[i] = p;
+ d->input[i] = bp;
if (i == 0) {
- file1 = orig;
- firstchange = 0;
+ d->file1 = orig;
+ d->firstchange = 0;
} else
- file2 = orig;
+ d->file2 = orig;
return bp;
}
@@ -175,31 +171,32 @@
* need to fix up for unexpected EOF's
*/
void
-check(Biobuf *bf, Biobuf *bt)
+check(Diff *d, Biobuf *bf, Biobuf *bt)
{
int f, t, flen, tlen;
char fbuf[MAXLINELEN], tbuf[MAXLINELEN];
- ixold[0] = ixnew[0] = 0;
- for (f = t = 1; f < len[0]; f++) {
+ d->ixold[0] = 0;
+ d->ixnew[0] = 0;
+ for (f = t = 1; f < d->len[0]; f++) {
flen = readline(bf, fbuf);
- ixold[f] = ixold[f-1] + flen + 1; /* ftell(bf) */
- if (J[f] == 0)
+ d->ixold[f] = d->ixold[f-1] + flen + 1; /* ftell(bf) */
+ if (d->J[f] == 0)
continue;
do {
tlen = readline(bt, tbuf);
- ixnew[t] = ixnew[t-1] + tlen + 1; /* ftell(bt) */
- } while (t++ < J[f]);
+ d->ixnew[t] = d->ixnew[t-1] + tlen + 1; /* ftell(bt) */
+ } while (t++ < d->J[f]);
if (bflag) {
flen = squishspace(fbuf);
tlen = squishspace(tbuf);
}
if (flen != tlen || strcmp(fbuf, tbuf))
- J[f] = 0;
+ d->J[f] = 0;
}
- while (t < len[1]) {
+ while (t < d->len[1]) {
tlen = readline(bt, tbuf);
- ixnew[t] = ixnew[t-1] + tlen + 1; /* fseek(bt) */
+ d->ixnew[t] = d->ixnew[t-1] + tlen + 1; /* fseek(bt) */
t++;
}
}
@@ -212,8 +209,8 @@
Bprint(&stdout, "%s%d", separator, b);
}
-static void
-fetch(long *f, int a, int b, Biobuf *bp, char *s)
+void
+fetch(Diff *d, long *f, int a, int b, Biobuf *bp, char *s)
{
char buf[MAXLINELEN];
int maxb;
@@ -220,10 +217,10 @@
if(a <= 1)
a = 1;
- if(bp == input[0])
- maxb = len[0];
+ if(bp == d->input[0])
+ maxb = d->len[0];
else
- maxb = len[1];
+ maxb = d->len[1];
if(b > maxb)
b = maxb;
if(a > maxb)
@@ -232,23 +229,12 @@
while (a++ <= b) {
readline(bp, buf);
Bprint(&stdout, "%s%s\n", s, buf);
+ Bflush(&stdout);
}
}
-typedef struct Change Change;
-struct Change
-{
- int a;
- int b;
- int c;
- int d;
-};
-
-Change *changes;
-int nchanges;
-
void
-change(int a, int b, int c, int d)
+change(Diff *df, int a, int b, int c, int d)
{
char verb;
char buf[4];
@@ -257,7 +243,7 @@
if (a > b && c > d)
return;
anychange = 1;
- if (mflag && firstchange == 0) {
+ if (mflag && df->firstchange == 0) {
if(mode) {
buf[0] = '-';
buf[1] = mode;
@@ -266,8 +252,8 @@
} else {
buf[0] = '\0';
}
- Bprint(&stdout, "diff %s%s %s\n", buf, file1, file2);
- firstchange = 1;
+ Bprint(&stdout, "diff %s%s %s\n", buf, df->file1, df->file2);
+ df->firstchange = 1;
}
verb = a > b ? 'a': c > d ? 'd': 'c';
switch(mode) {
@@ -281,10 +267,10 @@
range(c, d, ",");
break;
case 'n':
- Bprint(&stdout, "%s:", file1);
+ Bprint(&stdout, "%s:", df->file1);
range(a, b, ",");
Bprint(&stdout, " %c ", verb);
- Bprint(&stdout, "%s:", file2);
+ Bprint(&stdout, "%s:", df->file2);
range(c, d, ",");
break;
case 'f':
@@ -294,9 +280,9 @@
case 'c':
case 'a':
case 'u':
- if(nchanges%1024 == 0)
- changes = erealloc(changes, (nchanges+1024)*sizeof(changes[0]));
- ch = &changes[nchanges++];
+ if(df->nchanges%1024 == 0)
+ df->changes = erealloc(df->changes, (df->nchanges+1024)*sizeof(df->changes[0]));
+ ch = &df->changes[df->nchanges++];
ch->a = a;
ch->b = b;
ch->c = c;
@@ -305,11 +291,11 @@
}
Bputc(&stdout, '\n');
if (mode == 0 || mode == 'n') {
- fetch(ixold, a, b, input[0], "< ");
+ fetch(df, df->ixold, a, b, df->input[0], "< ");
if (a <= b && c <= d)
Bprint(&stdout, "---\n");
}
- fetch(ixnew, c, d, input[1], mode == 0 || mode == 'n' ? "> ": "");
+ fetch(df, df->ixnew, c, d, df->input[1], mode == 0 || mode == 'n' ? "> ": "");
if (mode != 0 && mode != 'n' && c <= d)
Bprint(&stdout, ".\n");
}
@@ -320,69 +306,69 @@
};
int
-changeset(int i)
+changeset(Diff *d, int i)
{
- while(i<nchanges && changes[i].b+1+2*Lines > changes[i+1].a)
+ while(i < d->nchanges && d->changes[i].b + 1 + 2*Lines > d->changes[i+1].a)
i++;
- if(i<nchanges)
+ if(i < d->nchanges)
return i+1;
- return nchanges;
+ return d->nchanges;
}
void
-flushchanges(void)
+flushchanges(Diff *df)
{
- int a, b, c, d, at, hdr;
- int i, j;
+ vlong a, b, c, d, at, hdr;
+ vlong i, j;
- if(nchanges == 0)
+ if(df->nchanges == 0)
return;
hdr = 0;
- for(i=0; i<nchanges; ){
- j = changeset(i);
- a = changes[i].a-Lines;
- b = changes[j-1].b+Lines;
- c = changes[i].c-Lines;
- d = changes[j-1].d+Lines;
+ for(i=0; i < df->nchanges; ){
+ j = changeset(df, i);
+ a = df->changes[i].a - Lines;
+ b = df->changes[j-1].b + Lines;
+ c = df->changes[i].c - Lines;
+ d = df->changes[j-1].d + Lines;
if(a < 1)
a = 1;
if(c < 1)
c = 1;
- if(b > len[0])
- b = len[0];
- if(d > len[1])
- d = len[1];
+ if(b > df->len[0])
+ b = df->len[0];
+ if(d > df->len[1])
+ d = df->len[1];
if(mode == 'a'){
a = 1;
- b = len[0];
+ b = df->len[0];
c = 1;
- d = len[1];
- j = nchanges;
+ d = df->len[1];
+ j = df->nchanges;
}
if(mode == 'u'){
if(!hdr){
- Bprint(&stdout, "--- %s\n", file1);
- Bprint(&stdout, "+++ %s\n", file2);
+ Bprint(&stdout, "--- %s\n", df->file1);
+ Bprint(&stdout, "+++ %s\n", df->file2);
hdr = 1;
}
- Bprint(&stdout, "@@ -%d,%d +%d,%d @@\n", a, b-a+1, c, d-c+1);
+ Bprint(&stdout, "@@ -%lld,%lld +%lld,%lld @@\n", a, b-a+1, c, d-c+1);
}else{
- Bprint(&stdout, "%s:", file1);
+ Bprint(&stdout, "%s:", df->file1);
range(a, b, ",");
Bprint(&stdout, " - ");
- Bprint(&stdout, "%s:", file2);
+ Bprint(&stdout, "%s:", df->file2);
range(c, d, ",");
Bputc(&stdout, '\n');
}
at = a;
for(; i<j; i++){
- fetch(ixold, at, changes[i].a-1, input[0], mode == 'u' ? " " : " ");
- fetch(ixold, changes[i].a, changes[i].b, input[0], mode == 'u' ? "-" : "- ");
- fetch(ixnew, changes[i].c, changes[i].d, input[1], mode == 'u' ? "+" : "+ ");
- at = changes[i].b+1;
+ fetch(df, df->ixold, at, df->changes[i].a-1, df->input[0], mode == 'u' ? " " : " ");
+ fetch(df, df->ixold, df->changes[i].a, df->changes[i].b, df->input[0], mode == 'u' ? "-" : "- ");
+ fetch(df, df->ixnew, df->changes[i].c, df->changes[i].d, df->input[1], mode == 'u' ? "+" : "+ ");
+ at = df->changes[i].b+1;
}
- fetch(ixold, at, b, input[0], mode == 'u' ? " " : " ");
+ fetch(df, df->ixold, at, b, df->input[0], mode == 'u' ? " " : " ");
}
- nchanges = 0;
+ df->nchanges = 0;
}
--- a/sys/src/cmd/diff/diffreg.c
+++ b/sys/src/cmd/diff/diffreg.c
@@ -66,30 +66,7 @@
* 3*(number of k-candidates installed), typically about
* 6n words for files of length n.
*/
-typedef struct Cand Cand;
-struct Cand {
- int x;
- int y;
- int pred;
-};
-
-Cand cand;
-Line *file[2], line;
-int len[2];
-int binary;
-Line *sfile[2]; /*shortened by pruning common prefix and suffix*/
-int slen[2];
-int pref, suff; /*length of prefix and suffix*/
-int *class; /*will be overlaid on file[0]*/
-int *member; /*will be overlaid on file[1]*/
-int *klist; /*will be overlaid on file[0] after class*/
-Cand *clist; /* merely a free storage pot for candidates */
-int clen;
-int *J; /*will be overlaid on class*/
-long *ixold; /*will be overlaid on klist*/
-long *ixnew; /*will be overlaid on file[1]*/
-
static void
sort(Line *a, int n) /*shellsort CACM #201*/
{
@@ -137,21 +114,21 @@
}
static void
-prune(void)
+prune(Diff *d)
{
int i,j;
- for(pref=0;pref<len[0]&&pref<len[1]&&
- file[0][pref+1].value==file[1][pref+1].value;
- pref++ ) ;
- for(suff=0;suff<len[0]-pref&&suff<len[1]-pref&&
- file[0][len[0]-suff].value==file[1][len[1]-suff].value;
- suff++) ;
+ for(d->pref = 0; d->pref < d->len[0] && d->pref < d->len[1] &&
+ d->file[0][d->pref+1].value == d->file[1][d->pref+1].value;
+ d->pref++) ;
+ for(d->suff=0; d->suff < d->len[0] - d->pref && d->suff < d->len[1] - d->pref &&
+ d->file[0][d->len[0] - d->suff].value == d->file[1][d->len[1] - d->suff].value;
+ d->suff++) ;
for(j=0;j<2;j++) {
- sfile[j] = file[j]+pref;
- slen[j] = len[j]-pref-suff;
- for(i=0;i<=slen[j];i++)
- sfile[j][i].serial = i;
+ d->sfile[j] = d->file[j]+d->pref;
+ d->slen[j] = d->len[j]-d->pref-d->suff;
+ for(i=0;i<=d->slen[j];i++)
+ d->sfile[j][i].serial = i;
}
}
@@ -184,30 +161,30 @@
}
static int
-newcand(int x, int y, int pred)
+newcand(Diff *d, int x, int y, int pred)
{
Cand *q;
- clist = erealloc(clist, (clen+1)*sizeof(Cand));
- q = clist + clen;
+ d->clist = erealloc(d->clist, (d->clen+1)*sizeof(Cand));
+ q = d->clist + d->clen;
q->x = x;
q->y = y;
q->pred = pred;
- return clen++;
+ return d->clen++;
}
static int
-search(int *c, int k, int y)
+search(Diff *d, int *c, int k, int y)
{
int i, j, l;
int t;
- if(clist[c[k]].y < y) /*quick look for typical case*/
+ if(d->clist[c[k]].y < y) /*quick look for typical case*/
return k+1;
i = 0;
j = k+1;
while((l=(i+j)/2) > i) {
- t = clist[c[l]].y;
+ t = d->clist[c[l]].y;
if(t > y)
j = l;
else if(t < y)
@@ -219,7 +196,7 @@
}
static int
-stone(int *a, int n, int *b, int *c)
+stone(Diff *d, int *a, int n, int *b, int *c)
{
int i, k,y;
int j, l;
@@ -227,7 +204,7 @@
int oldl;
k = 0;
- c[0] = newcand(0,0,0);
+ c[0] = newcand(d, 0, 0, 0);
for(i=1; i<=n; i++) {
j = a[i];
if(j==0)
@@ -236,20 +213,20 @@
oldl = 0;
oldc = c[0];
do {
- if(y <= clist[oldc].y)
+ if(y <= d->clist[oldc].y)
continue;
- l = search(c, k, y);
+ l = search(d, c, k, y);
if(l!=oldl+1)
oldc = c[l-1];
if(l<=k) {
- if(clist[c[l]].y <= y)
+ if(d->clist[c[l]].y <= y)
continue;
tc = c[l];
- c[l] = newcand(i,y,oldc);
+ c[l] = newcand(d, i, y, oldc);
oldc = tc;
oldl = l;
} else {
- c[l] = newcand(i,y,oldc);
+ c[l] = newcand(d, i,y,oldc);
k++;
break;
}
@@ -259,61 +236,23 @@
}
static void
-unravel(int p)
+unravel(Diff *d, int p)
{
int i;
Cand *q;
- for(i=0; i<=len[0]; i++) {
- if (i <= pref)
- J[i] = i;
- else if (i > len[0]-suff)
- J[i] = i+len[1]-len[0];
+ for(i=0; i<=d->len[0]; i++) {
+ if (i <= d->pref)
+ d->J[i] = i;
+ else if (i > d->len[0]-d->suff)
+ d->J[i] = i+d->len[1] - d->len[0];
else
- J[i] = 0;
+ d->J[i] = 0;
}
- for(q=clist+p;q->y!=0;q=clist+q->pred)
- J[q->x+pref] = q->y+pref;
+ for(q=d->clist+p; q->y != 0; q= d->clist + q->pred)
+ d->J[q->x+d->pref] = q->y+d->pref;
}
-static void
-output(void)
-{
- int m, i0, i1, j0, j1;
-
- m = len[0];
- J[0] = 0;
- J[m+1] = len[1]+1;
- if (mode != 'e') {
- for (i0 = 1; i0 <= m; i0 = i1+1) {
- while (i0 <= m && J[i0] == J[i0-1]+1)
- i0++;
- j0 = J[i0-1]+1;
- i1 = i0-1;
- while (i1 < m && J[i1+1] == 0)
- i1++;
- j1 = J[i1+1]-1;
- J[i1] = j1;
- change(i0, i1, j0, j1);
- }
- } else {
- for (i0 = m; i0 >= 1; i0 = i1-1) {
- while (i0 >= 1 && J[i0] == J[i0+1]-1 && J[i0])
- i0--;
- j0 = J[i0+1]-1;
- i1 = i0+1;
- while (i1 > 1 && J[i1-1] == 0)
- i1--;
- j1 = J[i1-1]+1;
- J[i1] = j1;
- change(i1 , i0, j1, j0);
- }
- }
- if (m == 0)
- change(1, 0, 1, len[1]);
- flushchanges();
-}
-
#define BUF 4096
static int
cmp(Biobuf* b1, Biobuf* b2)
@@ -361,21 +300,20 @@
}
void
-diffreg(char *f, char *fo, char *t, char *to)
+calcdiff(Diff *d, char *f, char *fo, char *t, char *to)
{
Biobuf *b0, *b1;
int k;
- binary = 0;
- b0 = prepare(0, f, fo);
+ b0 = prepare(d, 0, f, fo);
if (!b0)
return;
- b1 = prepare(1, t, to);
+ b1 = prepare(d, 1, t, to);
if (!b1) {
Bterm(b0);
return;
}
- if (binary){
+ if (d->binary){
// could use b0 and b1 but this is simpler.
if (cmp(b0, b1))
print("binary files %s %s differ\n", f, t);
@@ -383,38 +321,91 @@
Bterm(b1);
return;
}
- clen = 0;
- prune();
- sort(sfile[0], slen[0]);
- sort(sfile[1], slen[1]);
+ d->clen = 0;
+ prune(d);
+ sort(d->sfile[0], d->slen[0]);
+ sort(d->sfile[1], d->slen[1]);
- member = (int *)file[1];
- equiv(sfile[0], slen[0], sfile[1], slen[1], member);
- member = erealloc(member, (slen[1]+2)*sizeof(int));
+ d->member = (int *)d->file[1];
+ equiv(d->sfile[0], d->slen[0], d->sfile[1], d->slen[1], d->member);
+ d->member = erealloc(d->member, (d->slen[1]+2)*sizeof(int));
- class = (int *)file[0];
- unsort(sfile[0], slen[0], class);
- class = erealloc(class, (slen[0]+2)*sizeof(int));
+ d->class = (int *)d->file[0];
+ unsort(d->sfile[0], d->slen[0], d->class);
+ d->class = erealloc(d->class, (d->slen[0]+2)*sizeof(int));
- klist = emalloc((slen[0]+2)*sizeof(int));
- clist = emalloc(sizeof(Cand));
- k = stone(class, slen[0], member, klist);
- free(member);
- free(class);
+ d->klist = emalloc((d->slen[0]+2)*sizeof(int));
+ d->clist = emalloc(sizeof(Cand));
+ k = stone(d, d->class, d->slen[0], d->member, d->klist);
+ free(d->member);
+ free(d->class);
- J = emalloc((len[0]+2)*sizeof(int));
- unravel(klist[k]);
- free(clist);
- free(klist);
+ d->J = emalloc((d->len[0]+2)*sizeof(int));
+ unravel(d, d->klist[k]);
+ free(d->clist);
+ free(d->klist);
- ixold = emalloc((len[0]+2)*sizeof(long));
- ixnew = emalloc((len[1]+2)*sizeof(long));
+ d->ixold = emalloc((d->len[0]+2)*sizeof(long));
+ d->ixnew = emalloc((d->len[1]+2)*sizeof(long));
Bseek(b0, 0, 0); Bseek(b1, 0, 0);
- check(b0, b1);
- output();
- free(J);
- free(ixold);
- free(ixnew);
- Bterm(b0);
- Bterm(b1);
+ check(d, b0, b1);
+}
+
+static void
+output(Diff *d)
+{
+ int m, i0, i1, j0, j1;
+
+ m = d->len[0];
+ d->J[0] = 0;
+ d->J[m+1] = d->len[1]+1;
+ if (mode != 'e') {
+ for (i0 = 1; i0 <= m; i0 = i1+1) {
+ while (i0 <= m && d->J[i0] == d->J[i0-1]+1)
+ i0++;
+ j0 = d->J[i0-1]+1;
+ i1 = i0-1;
+ while (i1 < m && d->J[i1+1] == 0)
+ i1++;
+ j1 = d->J[i1+1]-1;
+ d->J[i1] = j1;
+ change(d, i0, i1, j0, j1);
+ }
+ } else {
+ for (i0 = m; i0 >= 1; i0 = i1-1) {
+ while (i0 >= 1 && d->J[i0] == d->J[i0+1]-1 && d->J[i0])
+ i0--;
+ j0 = d->J[i0+1]-1;
+ i1 = i0+1;
+ while (i1 > 1 && d->J[i1-1] == 0)
+ i1--;
+ j1 = d->J[i1-1]+1;
+ d->J[i1] = j1;
+ change(d, i1 , i0, j1, j0);
+ }
+ }
+ if (m == 0)
+ change(d, 1, 0, 1, d->len[1]);
+ flushchanges(d);
+}
+
+void
+diffreg(char *f, char *fo, char *t, char *to)
+{
+ Diff d;
+
+ memset(&d, 0, sizeof(d));
+ calcdiff(&d, f, fo, t, to);
+ output(&d);
+ freediff(&d);
+}
+
+void
+freediff(Diff *d)
+{
+ Bterm(d->input[0]);
+ Bterm(d->input[1]);
+ free(d->J);
+ free(d->ixold);
+ free(d->ixnew);
}
--- a/sys/src/cmd/diff/main.c
+++ /dev/null
@@ -1,261 +1,0 @@
-#include <u.h>
-#include <libc.h>
-#include <bio.h>
-#include "diff.h"
-
-#define DIRECTORY(s) ((s)->qid.type&QTDIR)
-#define REGULAR_FILE(s) ((s)->type == 'M' && !DIRECTORY(s))
-
-Biobuf stdout;
-char mode; /* '\0', 'e', 'f', 'h' */
-char bflag; /* ignore multiple and trailing blanks */
-char rflag; /* recurse down directory trees */
-char mflag; /* pseudo flag: doing multiple files, one dir */
-int anychange;
-
-static char *tmp[] = {"/tmp/diff1XXXXXXXXXXX", "/tmp/diff2XXXXXXXXXXX"};
-static int whichtmp;
-
-static void
-rmtmpfiles(void)
-{
- while (whichtmp > 0) {
- whichtmp--;
- remove(tmp[whichtmp]);
- }
-}
-
-void
-done(int status)
-{
- rmtmpfiles();
- switch(status)
- {
- case 0:
- exits("");
- case 1:
- exits("some");
- default:
- exits("error");
- }
- /*NOTREACHED*/
-}
-
-void
-panic(int status, char *fmt, ...)
-{
- va_list arg;
-
- Bflush(&stdout);
-
- fprint(2, "%s: ", argv0);
- va_start(arg, fmt);
- vfprint(2, fmt, arg);
- va_end(arg);
- if (status)
- done(status);
- /*NOTREACHED*/
-}
-
-static int
-catch(void *a, char *msg)
-{
- USED(a);
- panic(2, msg);
- return 1;
-}
-
-int
-mkpathname(char *pathname, char *path, char *name)
-{
- if (strlen(path) + strlen(name) > MAXPATHLEN) {
- panic(0, "pathname %s/%s too long\n", path, name);
- return 1;
- }
- sprint(pathname, "%s/%s", path, name);
- return 0;
-}
-
-static char *
-mktmpfile(int input, Dir **sb)
-{
- int fd, i;
- char *p;
- char buf[8192];
-
- atnotify(catch, 1);
- p = mktemp(tmp[whichtmp++]);
- fd = create(p, OWRITE, 0600);
- if (fd < 0) {
- panic(mflag ? 0: 2, "cannot create %s: %r\n", p);
- return 0;
- }
- while ((i = read(input, buf, sizeof(buf))) > 0) {
- if ((i = write(fd, buf, i)) < 0)
- break;
- }
- *sb = dirfstat(fd);
- close(fd);
- if (i < 0) {
- panic(mflag ? 0: 2, "cannot read/write %s: %r\n", p);
- return 0;
- }
- return p;
-}
-
-static char *
-statfile(char *file, Dir **sb)
-{
- Dir *dir;
- int input;
-
- dir = dirstat(file);
- if(dir == nil) {
- if (strcmp(file, "-") || (dir = dirfstat(0)) == nil) {
- panic(mflag ? 0: 2, "cannot stat %s: %r\n", file);
- return 0;
- }
- free(dir);
- return mktmpfile(0, sb);
- } else if (!REGULAR_FILE(dir) && !DIRECTORY(dir)) {
- free(dir);
- if ((input = open(file, OREAD)) == -1) {
- panic(mflag ? 0: 2, "cannot open %s: %r\n", file);
- return 0;
- }
- file = mktmpfile(input, sb);
- close(input);
- } else
- *sb = dir;
- return file;
-}
-
-void
-diff(char *f, char *t, int level)
-{
- char *fp, *tp, *p, fb[MAXPATHLEN+1], tb[MAXPATHLEN+1];
- Dir *fsb, *tsb;
-
- if ((fp = statfile(f, &fsb)) == 0)
- goto Return;
- if ((tp = statfile(t, &tsb)) == 0){
- free(fsb);
- goto Return;
- }
- if (DIRECTORY(fsb) && DIRECTORY(tsb)) {
- if (rflag || level == 0)
- diffdir(fp, tp, level);
- else
- Bprint(&stdout, "Common subdirectories: %s and %s\n", fp, tp);
- }
- else if (REGULAR_FILE(fsb) && REGULAR_FILE(tsb))
- diffreg(fp, f, tp, t);
- else {
- if (REGULAR_FILE(fsb)) {
- if ((p = utfrrune(f, '/')) == 0)
- p = f;
- else
- p++;
- if (mkpathname(tb, tp, p) == 0)
- diffreg(fp, f, tb, t);
- } else {
- if ((p = utfrrune(t, '/')) == 0)
- p = t;
- else
- p++;
- if (mkpathname(fb, fp, p) == 0)
- diffreg(fb, f, tp, t);
- }
- }
- free(fsb);
- free(tsb);
-Return:
- rmtmpfiles();
-}
-
-void
-usage(void)
-{
- fprint(2, "usage: %s [-abcefmnrw] file1 ... file2\n", argv0);
- exits("usage");
-}
-
-void
-main(int argc, char *argv[])
-{
- int i;
- Dir *fsb, *tsb;
-
- Binit(&stdout, 1, OWRITE);
- ARGBEGIN{
- case 'e':
- case 'f':
- case 'n':
- case 'c':
- case 'a':
- case 'u':
- mode = ARGC();
- break;
- case 'w':
- bflag = 2;
- break;
-
- case 'b':
- bflag = 1;
- break;
-
- case 'r':
- rflag = 1;
- break;
-
- case 'm':
- mflag = 1;
- break;
-
- case 'h':
- default:
- usage();
- }ARGEND;
- if (argc < 2)
- usage();
- if ((tsb = dirstat(argv[argc-1])) == nil)
- panic(2, "can't stat %s\n", argv[argc-1]);
- if (argc > 2) {
- if (!DIRECTORY(tsb))
- panic(2, "not directory: %s", argv[argc-1]);
- mflag = 1;
- } else {
- if ((fsb = dirstat(argv[0])) == nil)
- panic(2, "can't stat %s\n", argv[0]);
- if (DIRECTORY(fsb) && DIRECTORY(tsb))
- mflag = 1;
- free(fsb);
- }
- free(tsb);
- for (i = 0; i < argc-1; i++)
- diff(argv[i], argv[argc-1], 0);
- done(anychange);
- /*NOTREACHED*/
-}
-
-static char noroom[] = "out of memory - try diff -h\n";
-
-void *
-emalloc(unsigned n)
-{
- register void *p;
-
- if ((p = malloc(n)) == 0)
- panic(2, noroom);
- return p;
-}
-
-void *
-erealloc(void *p, unsigned n)
-{
- void *rp;
-
- if ((rp = realloc(p, n)) == 0)
- panic(2, noroom);
- return rp;
-}
--- /dev/null
+++ b/sys/src/cmd/diff/merge3.c
@@ -1,0 +1,169 @@
+#include <u.h>
+#include <libc.h>
+#include <bio.h>
+#include "diff.h"
+
+static int
+changecmp(void *a, void *b)
+{
+ return ((Change*)a)->a - ((Change*)b)->a;
+}
+
+static void
+addchange(Diff *df, int a, int b, int c, int d)
+{
+ Change *ch;
+
+ if (a > b && c > d)
+ return;
+ if(df->nchanges%1024 == 0)
+ df->changes = erealloc(df->changes, (df->nchanges+1024)*sizeof(df->changes[0]));
+ ch = &df->changes[df->nchanges++];
+ ch->a = a;
+ ch->b = b;
+ ch->c = c;
+ ch->d = d;
+}
+
+static void
+collect(Diff *d)
+{
+ int m, i0, i1, j0, j1;
+
+ m = d->len[0];
+ d->J[0] = 0;
+ d->J[m+1] = d->len[1]+1;
+ for (i0 = m; i0 >= 1; i0 = i1-1) {
+ while (i0 >= 1 && d->J[i0] == d->J[i0+1]-1 && d->J[i0])
+ i0--;
+ j0 = d->J[i0+1]-1;
+ i1 = i0+1;
+ while (i1 > 1 && d->J[i1-1] == 0)
+ i1--;
+ j1 = d->J[i1-1]+1;
+ d->J[i1] = j1;
+ addchange(d, i1 , i0, j1, j0);
+ }
+ if (m == 0)
+ change(d, 1, 0, 1, d->len[1]);
+ qsort(d->changes, d->nchanges, sizeof(Change), changecmp);
+}
+
+static int
+overlaps(Change *l, Change *r)
+{
+ if(l == nil || r == nil)
+ return 0;
+ if(l->a <= r->a)
+ return l->b >= r->a;
+ else
+ return r->b >= l->a;
+}
+
+char*
+merge(Diff *l, Diff *r)
+{
+ int il, ir, x, y, δ;
+ Change *lc, *rc;
+ char *status;
+ vlong ln;
+
+ il = 0;
+ ir = 0;
+ ln = 0;
+ status = nil;
+ collect(l);
+ collect(r);
+ while(il < l->nchanges || ir < r->nchanges){
+ lc = nil;
+ rc = nil;
+ if(il < l->nchanges)
+ lc = &l->changes[il];
+ if(ir < r->nchanges)
+ rc = &r->changes[ir];
+ if(overlaps(lc, rc)){
+ /*
+ * align the edges of the chunks
+ */
+ if(lc->a < rc->a){
+ x = lc->c;
+ δ = rc->a - lc->a;
+ rc->a -= δ;
+ rc->c -= δ;
+ }else{
+ x = rc->c;
+ δ = lc->a - rc->a;
+ lc->a -= δ;
+ lc->c -= δ;
+ }
+ if(lc->b > rc->b){
+ y = lc->d;
+ δ = lc->b - rc->b;
+ rc->b += δ;
+ rc->d += δ;
+ }else{
+ y = rc->d;
+ δ = rc->b - lc->b;
+ lc->b += δ;
+ lc->d += δ;
+ }
+ fetch(l, l->ixold, ln, x-1, l->input[0], "");
+ Bprint(&stdout, "<<<<<<<<<< %s\n", l->file2);
+ fetch(l, l->ixnew, lc->c, lc->d, l->input[1], "");
+ Bprint(&stdout, "========== original\n");
+ fetch(l, l->ixold, x, y, l->input[0], "");
+ Bprint(&stdout, "========== %s\n", r->file2);
+ fetch(r, r->ixnew, rc->c, rc->d, r->input[1], "");
+ Bprint(&stdout, ">>>>>>>>>>\n");
+ ln = y+1;
+ il++;
+ ir++;
+ status = "conflict";
+ }else if(rc == nil || (lc != nil && lc->a < rc->a)){
+ fetch(l, l->ixold, ln, lc->a-1, l->input[0], "");
+ fetch(l, l->ixnew, lc->c, lc->d, l->input[1], "");
+ ln = lc->b+1;
+ il++;
+ }else if(lc == nil || (rc != nil && rc->a < lc->a)){
+ fetch(l, l->ixold, ln, rc->a-1, l->input[0], "");
+ fetch(r, r->ixnew, rc->c, rc->d, r->input[1], "");
+ ln = rc->b+1;
+ ir++;
+ }else
+ abort();
+ }
+ if(ln < l->len[0])
+ fetch(l, l->ixold, ln, l->len[0], l->input[0], "");
+ return status;
+}
+
+void
+usage(void)
+{
+ fprint(2, "usage: %s theirs base ours\n", argv0);
+ exits("usage");
+}
+
+void
+main(int argc, char **argv)
+{
+ Diff l, r;
+ char *x;
+
+ ARGBEGIN{
+ default:
+ usage();
+ }ARGEND;
+
+ if(argc != 3)
+ usage();
+ Binit(&stdout, 1, OWRITE);
+ memset(&l, 0, sizeof(l));
+ memset(&r, 0, sizeof(r));
+ calcdiff(&l, argv[1], argv[1], argv[0], argv[0]);
+ calcdiff(&r, argv[1], argv[1], argv[2], argv[2]);
+ x = merge(&l, &r);
+ freediff(&l);
+ freediff(&r);
+ exits(x);
+}
--- a/sys/src/cmd/diff/mkfile
+++ b/sys/src/cmd/diff/mkfile
@@ -1,13 +1,13 @@
< /$objtype/mkfile
-TARG=diff
+TARG=diff merge3
OFILES=\
diffdir.$O\
diffio.$O\
diffreg.$O\
- main.$O\
+ util.$O
HFILES=diff.h
BIN=/$objtype/bin
-</sys/src/cmd/mkone
+</sys/src/cmd/mkmany
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t10.1
@@ -1,0 +1,1 @@
+a line of text
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t10.2
@@ -1,0 +1,1 @@
+Another line of text
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t10.expected
@@ -1,0 +1,5 @@
+--- diff-t10.1
++++ diff-t10.2
+@@ -1 +1 @@
+-a line of text
++Another line of text
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t11.1
@@ -1,0 +1,1003 @@
+.\" $OpenBSD: t11.1,v 1.2 2007/11/27 16:22:12 martynas Exp $
+.\" $NetBSD: ed.1,v 1.13 1995/03/21 09:04:38 cgd Exp $
+.\"
+.TH ED 1 "21 May 1993"
+.SH NAME
+.\" ed, red \- text editor
+ed \- text editor
+.SH SYNOPSIS
+ed [-] [-sx] [-p \fIstring\fR] [\fIfile\fR]
+.\" .LP
+.\" red [-] [-sx] [-p \fIstring\fR] [\fIfile\fR]
+.SH DESCRIPTION
+.B ed
+is a line-oriented text editor.
+It is used to create, display, modify and otherwise manipulate text
+files.
+.\" .B red
+.\" is a restricted
+.\" .BR ed :
+.\" it can only edit files in the current
+.\" directory and cannot execute shell commands.
+
+If invoked with a
+.I file
+argument, then a copy of
+.I file
+is read into the editor's buffer.
+Changes are made to this copy and not directly to
+.I file
+itself.
+Upon quitting
+.BR ed ,
+any changes not explicitly saved with a
+.I `w'
+command are lost.
+
+Editing is done in two distinct modes:
+.I command
+and
+.IR input .
+When first invoked,
+.B ed
+is in command mode.
+In this mode commands are read from the standard input and
+executed to manipulate the contents of the editor buffer.
+A typical command might look like:
+.sp
+.RS
+,s/\fIold\fR/\fInew\fR/g
+.RE
+.sp
+which replaces all occurrences of the string
+.I old
+with
+.IR new .
+
+When an input command, such as
+.I `a'
+(append),
+.I `i'
+(insert) or
+.I `c'
+(change), is given,
+.B ed
+enters input mode. This is the primary means
+of adding text to a file.
+In this mode, no commands are available;
+instead, the standard input is written
+directly to the editor buffer. Lines consist of text up to and
+including a
+.IR newline
+character.
+Input mode is terminated by
+entering a single period (\fI.\fR) on a line.
+
+All
+.B ed
+commands operate on whole lines or ranges of lines; e.g.,
+the
+.I `d'
+command deletes lines; the
+.I `m'
+command moves lines, and so on.
+It is possible to modify only a portion of a line by means of replacement,
+as in the example above. However even here, the
+.I `s'
+command is applied to whole lines at a time.
+
+In general,
+.B ed
+commands consist of zero or more line addresses, followed by a single
+character command and possibly additional parameters; i.e.,
+commands have the structure:
+.sp
+.RS
+.I [address [,address]]command[parameters]
+.RE
+.sp
+The address(es) indicate the line or range of lines to be affected by the
+command. If fewer addresses are given than the command accepts, then
+default addresses are supplied.
+
+.SS OPTIONS
+.TP 8
+-s
+Suppresses diagnostics. This should be used if
+.BR ed 's
+standard input is from a script.
+
+.TP 8
+-x
+Prompts for an encryption key to be used in subsequent reads and writes
+(see the
+.I `x'
+command).
+
+.TP 8
+.RI \-p \ string
+Specifies a command prompt. This may be toggled on and off with the
+.I `P'
+command.
+
+.TP 8
+.I file
+Specifies the name of a file to read. If
+.I file
+is prefixed with a
+bang (!), then it is interpreted as a shell command. In this case,
+what is read is
+the standard output of
+.I file
+executed via
+.IR sh (1).
+To read a file whose name begins with a bang, prefix the
+name with a backslash (\\).
+The default filename is set to
+.I file
+only if it is not prefixed with a bang.
+
+.SS LINE ADDRESSING
+An address represents the number of a line in the buffer.
+.B ed
+maintains a
+.I current address
+which is
+typically supplied to commands as the default address when none is specified.
+When a file is first read, the current address is set to the last line
+of the file. In general, the current address is set to the last line
+affected by a command.
+
+A line address is
+constructed from one of the bases in the list below, optionally followed
+by a numeric offset. The offset may include any combination
+of digits, operators (i.e.,
+.IR + ,
+.I -
+and
+.IR ^ )
+and whitespace.
+Addresses are read from left to right, and their values are computed
+relative to the current address.
+
+One exception to the rule that addresses represent line numbers is the
+address
+.I 0
+(zero).
+This means "before the first line,"
+and is legal wherever it makes sense.
+
+An address range is two addresses separated either by a comma or
+semi-colon. The value of the first address in a range cannot exceed the
+value of the second. If only one address is given in a range, then
+the second address is set to the given address. If an
+.IR n- tuple
+of addresses is given where
+.I n > 2,
+then the corresponding range is determined by the last two addresses in
+the
+.IR n- tuple.
+If only one address is expected, then the last address is used.
+
+Each address in a comma-delimited range is interpreted relative to the
+current address. In a semi-colon-delimited range, the first address is
+used to set the current address, and the second address is interpreted
+relative to the first.
+
+
+The following address symbols are recognized.
+
+.TP 8
+\&.
+The current line (address) in the buffer.
+
+.TP 8
+$
+The last line in the buffer.
+
+.TP 8
+n
+The
+.IR n th,
+line in the buffer
+where
+.I n
+is a number in the range
+.I [0,$].
+
+.TP 8
+- or ^
+The previous line.
+This is equivalent to
+.I -1
+and may be repeated with cumulative effect.
+
+.TP 8
+-\fIn\fR or ^\fIn\fR
+The
+.IR n th
+previous line, where
+.I n
+is a non-negative number.
+
+.TP 8
++
+The
+next line.
+This is equivalent to
+.I +1
+and may be repeated with cumulative effect.
+
+.TP 8
++\fIn\fR or whitespace\fIn\fR
+The
+.IR n th
+next line, where
+.I n
+is a non-negative number.
+.I whitespace
+followed by a number
+.I n
+is interpreted as
+.IR +n .
+
+.TP 8
+, \fRor\fB %
+The first through last lines in the buffer. This is equivalent to
+the address range
+.I 1,$.
+
+.TP 8
+;
+The
+current through last lines in the buffer. This is equivalent to
+the address range
+.I .,$.
+
+.TP 8
+.RI / re/
+The
+next line containing the regular expression
+.IR re .
+The search wraps to the beginning of the buffer and continues down to the
+current line, if necessary.
+// repeats the last search.
+
+.TP 8
+.RI ? re?
+The
+previous line containing the regular expression
+.IR re .
+The search wraps to the end of the buffer and continues up to the
+current line, if necessary.
+?? repeats the last search.
+
+.TP 8
+.RI \' lc
+The
+line previously marked by a
+.I `k'
+(mark) command, where
+.I lc
+is a lower case letter.
+
+.SS REGULAR EXPRESSIONS
+Regular expressions are patterns used in selecting text.
+For example, the
+.B ed
+command
+.sp
+.RS
+g/\fIstring\fR/
+.RE
+.sp
+prints all lines containing
+.IR string .
+Regular expressions are also
+used by the
+.I `s'
+command for selecting old text to be replaced with new.
+
+In addition to a specifying string literals, regular expressions can
+represent
+classes of strings. Strings thus represented are said to be matched
+by the corresponding regular expression.
+If it is possible for a regular expression
+to match several strings in a line, then the left-most longest match is
+the one selected.
+
+The following symbols are used in constructing regular expressions:
+
+.TP 8
+c
+Any character
+.I c
+not listed below, including `{', '}', `(', `)', `<' and `>',
+matches itself.
+
+.TP 8
+\fR\e\fIc\fR
+Any backslash-escaped character
+.IR c ,
+except for `{', '}', `(', `)', `<' and `>',
+matches itself.
+
+.TP 8
+\fR.\fR
+Matches any single character.
+
+.TP 8
+.I [char-class]
+Matches any single character in
+.IR char-class .
+To include a `]'
+in
+.IR char-class ,
+it must be the first character.
+A range of characters may be specified by separating the end characters
+of the range with a `-', e.g., `a-z' specifies the lower case characters.
+The following literal expressions can also be used in
+.I char-class
+to specify sets of characters:
+.sp
+\ \ [:alnum:]\ \ [:cntrl:]\ \ [:lower:]\ \ [:space:]
+.PD 0
+\ \ [:alpha:]\ \ [:digit:]\ \ [:print:]\ \ [:upper:]
+.PD 0
+\ \ [:blank:]\ \ [:graph:]\ \ [:punct:]\ \ [:xdigit:]
+.sp
+If `-' appears as the first or last
+character of
+.IR char-class ,
+then it matches itself.
+All other characters in
+.I char-class
+match themselves.
+.sp
+Patterns in
+.I char-class
+of the form:
+.sp
+\ \ [.\fIcol-elm\fR.] or,
+.PD 0
+\ \ [=\fIcol-elm\fR=]
+.sp
+where
+.I col-elm
+is a
+.I collating element
+are interpreted according to
+.IR locale (5)
+(not currently supported).
+See
+.IR regex (3)
+for an explanation of these constructs.
+
+.TP 8
+[^\fIchar-class\fR]
+Matches any single character, other than newline, not in
+.IR char-class .
+.IR char-class
+is defined
+as above.
+
+.TP 8
+^
+If `^' is the first character of a regular expression, then it
+anchors the regular expression to the beginning of a line.
+Otherwise, it matches itself.
+
+.TP 8
+$
+If `$' is the last character of a regular expression, it
+anchors the regular expression to the end of a line.
+Otherwise, it matches itself.
+
+.TP 8
+\fR\e<\fR
+Anchors the single character regular expression or subexpression
+immediately following it to the beginning of a word.
+(This may not be available)
+
+.TP 8
+\fR\e>\fR
+Anchors the single character regular expression or subexpression
+immediately following it to the end of a word.
+(This may not be available)
+
+.TP 8
+\fR\e(\fIre\fR\e)\fR
+Defines a subexpression
+.IR re .
+Subexpressions may be nested.
+A subsequent backreference of the form \fI`\en'\fR, where
+.I n
+is a number in the range [1,9], expands to the text matched by the
+.IR n th
+subexpression.
+For example, the regular expression `\e(.*\e)\e1' matches any string
+consisting of identical adjacent substrings.
+Subexpressions are ordered relative to
+their left delimiter.
+
+.TP 8
+*
+Matches the single character regular expression or subexpression
+immediately preceding it zero or more times. If '*' is the first
+character of a regular expression or subexpression, then it matches
+itself. The `*' operator sometimes yields unexpected results.
+For example, the regular expression `b*' matches the beginning of
+the string `abbb' (as opposed to the substring `bbb'), since a null match
+is the only left-most match.
+
+.TP 8
+\fR\e{\fIn,m\fR\e}\fR or \fR\e{\fIn,\fR\e}\fR or \fR\e{\fIn\fR\e}\fR
+Matches the single character regular expression or subexpression
+immediately preceding it at least
+.I n
+and at most
+.I m
+times.
+If
+.I m
+is omitted, then it matches at least
+.I n
+times.
+If the comma is also omitted, then it matches exactly
+.I n
+times.
+
+.LP
+Additional regular expression operators may be defined depending on the
+particular
+.IR regex (3)
+implementation.
+
+.SS COMMANDS
+All
+.B ed
+commands are single characters, though some require additonal parameters.
+If a command's parameters extend over several lines, then
+each line except for the last
+must be terminated with a backslash (\\).
+
+In general, at most one command is allowed per line.
+However, most commands accept a print suffix, which is any of
+.I `p'
+(print),
+.I `l'
+(list) ,
+or
+.I `n'
+(enumerate),
+to print the last line affected by the command.
+
+An interrupt (typically ^C) has the effect of aborting the current command
+and returning the editor to command mode.
+
+.B ed
+recognizes the following commands. The commands are shown together with
+the default address or address range supplied if none is
+specified (in parenthesis).
+
+.TP 8
+(.)a
+Appends text to the buffer after the addressed line.
+Text is entered in input mode.
+The current address is set to last line entered.
+
+.TP 8
+(.,.)c
+Changes lines in the buffer. The addressed lines are deleted
+from the buffer, and text is appended in their place.
+Text is entered in input mode.
+The current address is set to last line entered.
+
+.TP 8
+(.,.)d
+Deletes the addressed lines from the buffer.
+If there is a line after the deleted range, then the current address is set
+to this line. Otherwise the current address is set to the line
+before the deleted range.
+
+.TP 8
+.RI e \ file
+Edits
+.IR file ,
+and sets the default filename.
+If
+.I file
+is not specified, then the default filename is used.
+Any lines in the buffer are deleted before
+the new file is read.
+The current address is set to the last line read.
+
+.TP 8
+.RI e \ !command
+Edits the standard output of
+.IR `!command' ,
+(see
+.RI ! command
+below).
+The default filename is unchanged.
+Any lines in the buffer are deleted before the output of
+.I command
+is read.
+The current address is set to the last line read.
+
+.TP 8
+.RI E \ file
+Edits
+.I file
+unconditionally.
+This is similar to the
+.I e
+command,
+except that unwritten changes are discarded without warning.
+The current address is set to the last line read.
+
+.TP 8
+.RI f \ file
+Sets the default filename to
+.IR file .
+If
+.I file
+is not specified, then the default unescaped filename is printed.
+
+.TP 8
+.RI (1,$)g /re/command-list
+Applies
+.I command-list
+to each of the addressed lines matching a regular expression
+.IR re .
+The current address is set to the
+line currently matched before
+.I command-list
+is executed.
+At the end of the
+.I `g'
+command, the current address is set to the last line affected by
+.IR command-list .
+
+Each command in
+.I command-list
+must be on a separate line,
+and every line except for the last must be terminated by a backslash
+(\\).
+Any commands are allowed, except for
+.IR `g' ,
+.IR `G' ,
+.IR `v' ,
+and
+.IR `V' .
+A newline alone in
+.I command-list
+is equivalent to a
+.I `p'
+command.
+
+.TP 8
+.RI (1,$)G /re/
+Interactively edits the addressed lines matching a regular expression
+.IR re.
+For each matching line,
+the line is printed,
+the current address is set,
+and the user is prompted to enter a
+.IR command-list .
+At the end of the
+.I `G'
+command, the current address
+is set to the last line affected by (the last)
+.IR command-list .
+
+The format of
+.I command-list
+is the same as that of the
+.I `g'
+command. A newline alone acts as a null command list.
+A single `&' repeats the last non-null command list.
+
+.TP 8
+H
+Toggles the printing of error explanations.
+By default, explanations are not printed.
+It is recommended that ed scripts begin with this command to
+aid in debugging.
+
+.TP 8
+h
+Prints an explanation of the last error.
+
+.TP 8
+(.)i
+Inserts text in the buffer before the current line.
+Text is entered in input mode.
+The current address is set to the last line entered.
+
+.TP 8
+(.,.+1)j
+Joins the addressed lines. The addressed lines are
+deleted from the buffer and replaced by a single
+line containing their joined text.
+The current address is set to the resultant line.
+
+.TP 8
+.RI (.)k lc
+Marks a line with a lower case letter
+.IR lc .
+The line can then be addressed as
+.I 'lc
+(i.e., a single quote followed by
+.I lc
+) in subsequent commands. The mark is not cleared until the line is
+deleted or otherwise modified.
+
+.TP 8
+(.,.)l
+Prints the addressed lines unambiguously.
+If a single line fills for than one screen (as might be the case
+when viewing a binary file, for instance), a `--More--'
+prompt is printed on the last line.
+.B ed
+waits until the RETURN key is pressed
+before displaying the next screen.
+The current address is set to the last line
+printed.
+
+.TP 8
+(.,.)m(.)
+Moves lines in the buffer. The addressed lines are moved to after the
+right-hand destination address, which may be the address
+.IR 0
+(zero).
+The current address is set to the
+last line moved.
+
+.TP 8
+(.,.)n
+Prints the addressed lines along with
+their line numbers. The current address is set to the last line
+printed.
+
+.TP 8
+(.,.)p
+Prints the addressed lines. The current address is set to the last line
+printed.
+
+.TP 8
+P
+Toggles the command prompt on and off.
+Unless a prompt was specified by with command-line option
+\fI-p string\fR, the command prompt is by default turned off.
+
+.TP 8
+q
+Quits ed.
+
+.TP 8
+Q
+Quits ed unconditionally.
+This is similar to the
+.I q
+command,
+except that unwritten changes are discarded without warning.
+
+.TP 8
+.RI ($)r \ file
+Reads
+.I file
+to after the addressed line. If
+.I file
+is not specified, then the default
+filename is used. If there was no default filename prior to the command,
+then the default filename is set to
+.IR file .
+Otherwise, the default filename is unchanged.
+The current address is set to the last line read.
+
+.TP 8
+.RI ($)r \ !command
+Reads
+to after the addressed line
+the standard output of
+.IR `!command' ,
+(see the
+.RI ! command
+below).
+The default filename is unchanged.
+The current address is set to the last line read.
+
+.HP
+.RI (.,.)s /re/replacement/
+.PD 0
+.HP
+.RI (.,.)s /re/replacement/\fRg\fR
+.HP
+.RI (.,.)s /re/replacement/n
+.br
+Replaces text in the addressed lines
+matching a regular expression
+.I re
+with
+.IR replacement .
+By default, only the first match in each line is replaced.
+If the
+.I `g'
+(global) suffix is given, then every match to be replaced.
+The
+.I `n'
+suffix, where
+.I n
+is a postive number, causes only the
+.IR n th
+match to be replaced.
+It is an error if no substitutions are performed on any of the addressed
+lines.
+The current address is set the last line affected.
+
+.I re
+and
+.I replacement
+may be delimited by any character other than space and newline
+(see the
+.I `s'
+command below).
+If one or two of the last delimiters is omitted, then the last line
+affected is printed as though the print suffix
+.I `p'
+were specified.
+
+
+An unescaped `&' in
+.I replacement
+is replaced by the currently matched text.
+The character sequence
+\fI`\em'\fR,
+where
+.I m
+is a number in the range [1,9], is replaced by the
+.IR m th
+backreference expression of the matched text.
+If
+.I replacement
+consists of a single `%', then
+.I replacement
+from the last substitution is used.
+Newlines may be embedded in
+.I replacement
+if they are escaped with a backslash (\\).
+
+.TP 8
+(.,.)s
+Repeats the last substitution.
+This form of the
+.I `s'
+command accepts a count suffix
+.IR `n' ,
+or any combination of the characters
+.IR `r' ,
+.IR `g' ,
+and
+.IR `p' .
+If a count suffix
+.I `n'
+is given, then only the
+.IR n th
+match is replaced.
+The
+.I `r'
+suffix causes
+the regular expression of the last search to be used instead of the
+that of the last substitution.
+The
+.I `g'
+suffix toggles the global suffix of the last substitution.
+The
+.I `p'
+suffix toggles the print suffix of the last substitution
+The current address is set to the last line affected.
+
+.TP 8
+(.,.)t(.)
+Copies (i.e., transfers) the addressed lines to after the right-hand
+destination address, which may be the address
+.IR 0
+(zero).
+The current address is set to the last line
+copied.
+
+.TP 8
+u
+Undoes the last command and restores the current address
+to what it was before the command.
+The global commands
+.IR `g' ,
+.IR `G' ,
+.IR `v' ,
+and
+.IR `V' .
+are treated as a single command by undo.
+.I `u'
+is its own inverse.
+
+.TP 8
+.RI (1,$)v /pat/command-list
+Applies
+.I command-list
+to each of the addressed lines not matching a regular expression
+.IR re .
+This is similar to the
+.I `g'
+command.
+
+.TP 8
+.RI (1,$)V /re/
+Interactively edits the addressed lines not matching a regular expression
+.IR re.
+This is similar to the
+.I `G'
+command.
+
+.TP 8
+.RI (1,$)w \ file
+Writes the addressed lines to
+.IR file .
+Any previous contents of
+.I file
+is lost without warning.
+If there is no default filename, then the default filename is set to
+.IR file,
+otherwise it is unchanged. If no filename is specified, then the default
+filename is used.
+The current address is unchanged.
+
+.TP 8
+.RI (1,$)wq \ file
+Writes the addressed lines to
+.IR file ,
+and then executes a
+.I `q'
+command.
+
+.TP 8
+.RI (1,$)w \ !command
+Writes the addressed lines to the standard input of
+.IR `!command' ,
+(see the
+.RI ! command
+below).
+The default filename and current address are unchanged.
+
+.TP 8
+.RI (1,$)W \ file
+Appends the addressed lines to the end of
+.IR file .
+This is similar to the
+.I `w'
+command, expect that the previous contents of file is not clobbered.
+The current address is unchanged.
+
+.TP 8
+x
+Prompts for an encryption key which is used in subsequent reads and
+writes. If a newline alone is entered as the key, then encryption is
+turned off. Otherwise, echoing is disabled while a key is read.
+Encryption/decryption is done using the bdes(1) algorithm.
+
+.TP 8
+.RI (.+1)z n
+Scrolls
+.I n
+lines at a time starting at addressed line. If
+.I n
+is not specified, then the current window size is used.
+The current address is set to the last line printed.
+
+.TP 8
+.RI ! command
+Executes
+.I command
+via
+.IR sh (1).
+If the first character of
+.I command
+is `!', then it is replaced by text of the
+previous
+.IR `!command' .
+.B ed
+does not process
+.I command
+for backslash (\\) escapes.
+However, an unescaped
+.I `%'
+is replaced by the default filename.
+When the shell returns from execution, a `!'
+is printed to the standard output.
+The current line is unchanged.
+
+.TP 8
+($)=
+Prints the line number of the addressed line.
+
+.TP 8
+(.+1)newline
+Prints the addressed line, and sets the current address to
+that line.
+
+.SH FILES
+.TP 20
+/tmp/ed.*
+Buffer file
+.PD 0
+.TP 20
+ed.hup
+The file to which
+.B ed
+attempts to write the buffer if the terminal hangs up.
+
+.SH SEE ALSO
+
+.IR vi (1),
+.IR sed (1),
+.IR regex (3),
+.IR bdes (1),
+.IR sh (1).
+
+USD:12-13
+
+B. W. Kernighan and P. J. Plauger,
+.I Software Tools in Pascal ,
+Addison-Wesley, 1981.
+
+.SH LIMITATIONS
+.B ed
+processes
+.I file
+arguments for backslash escapes, i.e., in a filename,
+any characters preceded by a backslash (\\) are
+interpreted literally.
+
+If a text (non-binary) file is not terminated by a newline character,
+then
+.B ed
+appends one on reading/writing it. In the case of a binary file,
+.B ed
+does not append a newline on reading/writing.
+
+per line overhead: 4 ints
+
+.SH DIAGNOSTICS
+When an error occurs,
+.B ed
+prints a `?' and either returns to command mode
+or exits if its input is from a script.
+An explanation of the last error can be
+printed with the
+.I `h'
+(help) command.
+
+Since the
+.I `g'
+(global) command masks any errors from failed searches and substitutions,
+it can be used to perform conditional operations in scripts; e.g.,
+.sp
+.RS
+g/\fIold\fR/s//\fInew\fR/
+.RE
+.sp
+replaces any occurrences of
+.I old
+with
+.IR new .
+If the
+.I `u'
+(undo) command occurs in a global command list, then
+the command list is executed only once.
+
+If diagnostics are not disabled, attempting to quit
+.B ed
+or edit another file before writing a modified buffer
+results in an error.
+If the command is entered a second time, it succeeds,
+but any changes to the buffer are lost.
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t11.2
@@ -1,0 +1,908 @@
+.\" $OpenBSD: t11.2,v 1.1 2003/07/21 20:16:21 otto Exp $
+.\"
+.Dd May 2, 1993
+.Dt ED 1
+.Os
+.Sh NAME
+.Nm ed
+.Nd text editor
+.Sh SYNOPSIS
+.Nm ed
+.Op Fl
+.Op Fl sx
+.Op Fl p Ar string
+.Op Ar file
+.Sh DESCRIPTION
+.Nm
+is a line-oriented text editor.
+It is used to create, display, modify, and otherwise manipulate text files.
+If invoked with a
+.Ar file
+argument, then a copy of
+.Ar file
+is read into the editor's buffer.
+Changes are made to this copy and not directly to
+.Ar file
+itself.
+Upon quitting
+.Nm ed ,
+any changes not explicitly saved with a
+.Em w
+command are lost.
+.Pp
+Editing is done in two distinct modes:
+.Em command
+and
+.Em input .
+When first invoked,
+.Nm
+is in command mode.
+In this mode, commands are read from the standard input and
+executed to manipulate the contents of the editor buffer.
+.Pp
+A typical command might look like:
+.Bd -literal -offset indent
+,s/old/new/g
+.Ed
+.Pp
+which replaces all occurrences of the string
+.Pa old
+with
+.Pa new .
+.Pp
+When an input command, such as
+.Em a
+(append),
+.Em i
+(insert),
+or
+.Em c
+(change) is given,
+.Nm
+enters input mode.
+This is the primary means of adding text to a file.
+In this mode, no commands are available;
+instead, the standard input is written directory to the editor buffer.
+Lines consist of text up to and including a newline character.
+Input mode is terminated by entering a single period
+.Pq Ql \&.
+on a line.
+.Pp
+All
+.Nm
+commands operate on whole lines or ranges of lines; e.g.,
+the
+.Em d
+command deletes lines; the
+.Em m
+command moves lines, and so on.
+It is possible to modify only a portion of a line by means of replacement,
+as in the example above.
+However, even here, the
+.Em s
+command is applied to whole lines at a time.
+.Pp
+In general,
+.Nm
+commands consist of zero or more line addresses, followed by a single
+character command and possibly additional parameters; i.e.,
+commands have the structure:
+.Bd -literal -offset indent
+[address [,address]]command[parameters]
+.Ed
+.Pp
+The address(es) indicate the line or range of lines to be affected by the
+command.
+If fewer addresses are given than the command accepts, then
+default addresses are supplied.
+.Pp
+The options are as follows:
+.Bl -tag -width Ds
+.It Fl
+Same as the
+.Fl s
+option (deprecated).
+.It Fl s
+Suppress diagnostics.
+This should be used if
+.Nm
+standard input is from a script.
+.Fl s
+flag.
+.It Fl x
+Prompt for an encryption key to be used in subsequent reads and writes
+(see the
+.Em x
+command).
+.It Fl p Ar string
+Specifies a command prompt.
+This may be toggled on and off with the
+.Em P
+command.
+.It Ar file
+Specifies the name of a file to read.
+If
+.Ar file
+is prefixed with a
+bang
+.Pq Ql \&! ,
+then it is interpreted as a shell command.
+In this case, what is read is the standard output of
+.Ar file
+executed via
+.Xr sh 1 .
+To read a file whose name begins with a bang, prefix the
+name with a backslash
+.Pq Ql \e .
+The default filename is set to
+.Ar file
+only if it is not prefixed with a bang.
+.El
+.Ss LINE ADDRESSING
+An address represents the number of a line in the buffer.
+.Nm
+maintains a
+.Em current address
+which is typically supplied to commands as the default address
+when none is specified.
+When a file is first read, the current address is set to the last line
+of the file.
+In general, the current address is set to the last line affected by a command.
+.Pp
+A line address is
+constructed from one of the bases in the list below, optionally followed
+by a numeric offset.
+The offset may include any combination of digits, operators (i.e.,
+.Em + ,
+.Em - ,
+and
+.Em ^ ) ,
+and whitespace.
+Addresses are read from left to right, and their values are computed
+relative to the current address.
+.Pp
+One exception to the rule that addresses represent line numbers is the
+address
+.Em 0
+(zero).
+This means
+.Dq before the first line ,
+and is legal wherever it makes sense.
+.Pp
+An address range is two addresses separated either by a comma or semi-colon.
+The value of the first address in a range cannot exceed the
+value of the second.
+If only one address is given in a range,
+then the second address is set to the given address.
+If an
+.Em n Ns No -tuple
+of addresses is given where
+.Em n > 2 ,
+then the corresponding range is determined by the last two addresses in the
+.Em n Ns No -tuple.
+If only one address is expected, then the last address is used.
+.Pp
+Each address in a comma-delimited range is interpreted relative to the
+current address.
+In a semi-colon-delimited range, the first address is
+used to set the current address, and the second address is interpreted
+relative to the first.
+.Pp
+The following address symbols are recognized:
+.Bl -tag -width Ds
+.It Em \&.
+The current line (address) in the buffer.
+.It Em $
+The last line in the buffer.
+.It Em n
+The
+.Em n Ns No th
+line in the buffer where
+.Em n
+is a number in the range
+.Em [0,$] .
+.It Em - No or Em ^
+The previous line.
+This is equivalent to
+.Em -1
+and may be repeated with cumulative effect.
+.It Em -n No or Em ^n
+The
+.Em n Ns No th
+previous line, where
+.Em n
+is a non-negative number.
+.It Em +
+The next line.
+This is equivalent to
+.Em +1
+and may be repeated with cumulative effect.
+.It Em +n
+The
+.Em n Ns No th
+next line, where
+.Em n
+is a non-negative number.
+.It Em \&, No or Em %
+The first through last lines in the buffer.
+This is equivalent to the address range
+.Em 1,$ .
+.It Em \&;
+The current through last lines in the buffer.
+This is equivalent to the address range
+.Em .,$ .
+.It Em / Ns No re Ns Em /
+The next line containing the regular expression
+.Em re .
+The search wraps to the beginning of the buffer and continues down to the
+current line, if necessary.
+.Em //
+repeats the last search.
+.It Em ? Ns No re Ns Em ?
+The previous line containing the regular expression
+.Em re .
+The search wraps to the end of the buffer and continues up to the
+current line, if necessary.
+.Em ??
+repeats the last search.
+.It Em \&\' Ns No lc
+The line previously marked by a
+.Em k
+(mark) command, where
+.Em lc
+is a lower case letter.
+.El
+.Ss REGULAR EXPRESSIONS
+Regular expressions are patterns used in selecting text.
+For example, the
+.Nm
+command
+.Bd -literal -offset indent
+g/string/
+.Ed
+.Pp
+prints all lines containing
+.Em string .
+Regular expressions are also used by the
+.Em s
+command for selecting old text to be replaced with new.
+.Pp
+In addition to a specifying string literals, regular expressions can
+represent classes of strings.
+Strings thus represented are said to be matched by the
+corresponding regular expression.
+If it is possible for a regular expression to match several strings in
+a line, then the leftmost longest match is the one selected.
+.Pp
+The following symbols are used in constructing regular expressions:
+.Bl -tag -width Dsasdfsd
+.It Em c
+Any character
+.Em c
+not listed below, including
+.Em { Ns No ,
+.Em } Ns No ,
+.Em \&( Ns No ,
+.Em \&) Ns No ,
+.Em < Ns No ,
+and
+.Em >
+matches itself.
+.It Em \ec
+Any backslash-escaped character
+.Em c Ns No ,
+except for
+.Em { Ns No ,
+.Em } Ns No ,
+.Em \&( Ns No ,
+.Em \&) Ns No ,
+.Em < Ns No , and
+.Em >
+matches itself.
+.It Em \&.
+Matches any single character.
+.It Em [char-class]
+Matches any single character in
+.Em char-class .
+To include a
+.Ql \&]
+in
+.Em char-class Ns No ,
+it must be the first character.
+A range of characters may be specified by separating the end characters
+of the range with a
+.Ql - ;
+e.g.,
+.Em a-z
+specifies the lower case characters.
+The following literal expressions can also be used in
+.Em char-class
+to specify sets of characters:
+.Pp
+.Em \ \ [:alnum:]\ \ [:cntrl:]\ \ [:lower:]\ \ [:space:]
+.Em \ \ [:alpha:]\ \ [:digit:]\ \ [:print:]\ \ [:upper:]
+.Em \ \ [:blank:]\ \ [:graph:]\ \ [:punct:]\ \ [:xdigit:]
+.Pp
+If
+.Ql -
+appears as the first or last character of
+.Em char-class Ns No ,
+then it matches itself.
+All other characters in
+.Em char-class
+match themselves.
+.Pp
+Patterns in
+.Em char-class
+of the form
+.Em [.col-elm.] No or Em [=col-elm=]
+where
+.Em col-elm
+is a collating element are interpreted according to
+.Xr locale 5
+(not currently supported).
+See
+.Xr regex 3
+for an explanation of these constructs.
+.It Em [^char-class]
+Matches any single character, other than newline, not in
+.Em char-class Ns No .
+.Em char-class
+is defined as above.
+.It Em ^
+If
+.Em ^
+is the first character of a regular expression, then it
+anchors the regular expression to the beginning of a line.
+Otherwise, it matches itself.
+.It Em $
+If
+.Em $
+is the last character of a regular expression,
+it anchors the regular expression to the end of a line.
+Otherwise, it matches itself.
+.It Em \e<
+Anchors the single character regular expression or subexpression
+immediately following it to the beginning of a word.
+(This may not be available.)
+.It Em \e>
+Anchors the single character regular expression or subexpression
+immediately following it to the end of a word.
+(This may not be available.)
+.It Em \e( Ns No re Ns Em \e)
+Defines a subexpression
+.Em re .
+Subexpressions may be nested.
+A subsequent backreference of the form
+.Em \en Ns No ,
+where
+.Em n
+is a number in the range [1,9], expands to the text matched by the
+.Em n Ns No th
+subexpression.
+For example, the regular expression
+.Em \e(.*\e)\e1
+matches any string consisting of identical adjacent substrings.
+Subexpressions are ordered relative to their left delimiter.
+.It Em *
+Matches the single character regular expression or subexpression
+immediately preceding it zero or more times.
+If
+.Em *
+is the first character of a regular expression or subexpression,
+then it matches itself.
+The
+.Em *
+operator sometimes yields unexpected results.
+For example, the regular expression
+.Em b*
+matches the beginning of the string
+.Em abbb
+(as opposed to the substring
+.Em bbb Ns No ),
+since a null match is the only leftmost match.
+.Sm off
+.It Xo Em \e{ No n,m
+.Em \e}\ \e{ No n, Em \e}\
+.Em \e{ No n Em \e}
+.Xc
+.Sm on
+Matches the single character regular expression or subexpression
+immediately preceding it at least
+.Em n
+and at most
+.Em m
+times.
+If
+.Em m
+is omitted, then it matches at least
+.Em n
+times.
+If the comma is also omitted, then it matches exactly
+.Em n
+times.
+.El
+.Pp
+Additional regular expression operators may be defined depending on the
+particular
+.Xr regex 3
+implementation.
+.Ss COMMANDS
+All
+.Nm
+commands are single characters, though some require additional parameters.
+If a command's parameters extend over several lines, then
+each line except for the last must be terminated with a backslash
+.Pq Ql \e .
+.Pp
+In general, at most one command is allowed per line.
+However, most commands accept a print suffix, which is any of
+.Em p No (print),
+.Em l No (list),
+or
+.Em n No (enumerate),
+to print the last line affected by the command.
+.Pp
+An interrupt (typically ^C) has the effect of aborting the current command
+and returning the editor to command mode.
+.Pp
+.Nm
+recognizes the following commands.
+The commands are shown together with
+the default address or address range supplied if none is
+specified (in parentheses), and other possible arguments on the right.
+.Bl -tag -width Dxxs
+.It (.) Ns Em a
+Appends text to the buffer after the addressed line.
+Text is entered in input mode.
+The current address is set to last line entered.
+.It (.,.) Ns Em c
+Changes lines in the buffer.
+The addressed lines are deleted from the buffer,
+and text is appended in their place.
+Text is entered in input mode.
+The current address is set to last line entered.
+.It (.,.) Ns Em d
+Deletes the addressed lines from the buffer.
+If there is a line after the deleted range, then the current address is set
+to this line.
+Otherwise the current address is set to the line before the deleted range.
+.It Em e No file
+Edits
+.Em file Ns No ,
+and sets the default filename.
+If
+.Em file
+is not specified, then the default filename is used.
+Any lines in the buffer are deleted before the new file is read.
+The current address is set to the last line read.
+.It Em e No !command
+Edits the standard output of
+.Em !command Ns No ,
+(see
+.Em ! No command
+below).
+The default filename is unchanged.
+Any lines in the buffer are deleted before the output of
+.Em command
+is read.
+The current address is set to the last line read.
+.It Em E No file
+Edits
+.Em file
+unconditionally.
+This is similar to the
+.Em e
+command, except that unwritten changes are discarded without warning.
+The current address is set to the last line read.
+.It Em f No file
+Sets the default filename to
+.Em file Ns No .
+If
+.Em file
+is not specified, then the default unescaped filename is printed.
+.It (1,$) Ns Em g Ns No /re/command-list
+Applies
+.Em command-list
+to each of the addressed lines matching a regular expression
+.Em re Ns No .
+The current address is set to the line currently matched before
+.Em command-list
+is executed.
+At the end of the
+.Em g
+command, the current address is set to the last line affected by
+.Em command-list Ns No .
+.Pp
+Each command in
+.Em command-list
+must be on a separate line,
+and every line except for the last must be terminated by
+.Em \e No (backslash).
+Any commands are allowed, except for
+.Em g Ns No ,
+.Em G Ns No ,
+.Em v Ns No ,
+and
+.Em V Ns No .
+A newline alone in
+.Em command-list
+is equivalent to a
+.Em p
+command.
+.It (1,$) Ns Em G Ns No /re/
+Interactively edits the addressed lines matching a regular expression
+.Em re Ns No .
+For each matching line, the line is printed, the current address is set,
+and the user is prompted to enter a
+.Em command-list Ns No .
+At the end of the
+.Em g
+command, the current address is set to the last line affected by (the last)
+.Em command-list Ns No .
+.Pp
+The format of
+.Em command-list
+is the same as that of the
+.Em g
+command.
+A newline alone acts as a null command list.
+A single
+.Em &
+repeats the last non-null command list.
+.It Em H
+Toggles the printing of error explanations.
+By default, explanations are not printed.
+It is recommended that
+.Nm
+scripts begin with this command to aid in debugging.
+.It Em h
+Prints an explanation of the last error.
+.It (.) Ns Em i
+Inserts text in the buffer before the current line.
+Text is entered in input mode.
+The current address is set to the last line entered.
+.It (.,.+1) Ns Em j
+Joins the addressed lines.
+The addressed lines are deleted from the buffer and replaced by a single
+line containing their joined text.
+The current address is set to the resultant line.
+.It (.) Ns Em klc
+Marks a line with a lower case letter
+.Em lc Ns No \&.
+The line can then be addressed as
+.Em \&'lc
+(i.e., a single quote followed by
+.Em lc Ns No )
+in subsequent commands.
+The mark is not cleared until the line is deleted or otherwise modified.
+.It (.,.) Ns Em l
+Prints the addressed lines unambiguously.
+If a single line fills more than one screen (as might be the case
+when viewing a binary file, for instance), a
+.Dq --More--
+prompt is printed on the last line.
+.Nm
+waits until the RETURN key is pressed before displaying the next screen.
+The current address is set to the last line printed.
+.It (.,.) Ns Em m Ns No (.)
+Moves lines in the buffer.
+The addressed lines are moved to after the
+right-hand destination address, which may be the address
+.Em 0
+(zero).
+The current address is set to the last line moved.
+.It (.,.) Ns Em n
+Prints the addressed lines along with their line numbers.
+The current address is set to the last line printed.
+.It (.,.) Ns Em p
+Prints the addressed lines.
+The current address is set to the last line printed.
+.It Em P
+Toggles the command prompt on and off.
+Unless a prompt was specified by with command-line option
+.Fl p Ar string Ns No ,
+the command prompt is by default turned off.
+.It Em q
+Quits
+.Nm ed .
+.It Em Q
+Quits
+.Nm
+unconditionally.
+This is similar to the
+.Em q
+command, except that unwritten changes are discarded without warning.
+.It ($) Ns Em r No file
+Reads
+.Em file
+to after the addressed line.
+If
+.Em file
+is not specified, then the default filename is used.
+If there was no default filename prior to the command,
+then the default filename is set to
+.Em file Ns No .
+Otherwise, the default filename is unchanged.
+The current address is set to the last line read.
+.It ($) Ns Em r No !command
+Reads to after the addressed line the standard output of
+.Em !command Ns No ,
+(see the
+.Em !
+command below).
+The default filename is unchanged.
+The current address is set to the last line read.
+.Sm off
+.It Xo (.,.) Em s No /re/replacement/ , \ (.,.)
+.Em s No /re/replacement/ Em g , No \ (.,.)
+.Em s No /re/replacement/ Em n
+.Xc
+.Sm on
+Replaces text in the addressed lines matching a regular expression
+.Em re
+with
+.Em replacement Ns No .
+By default, only the first match in each line is replaced.
+If the
+.Em g
+(global) suffix is given, then every match to be replaced.
+The
+.Em n
+suffix, where
+.Em n
+is a positive number, causes only the
+.Em n Ns No th
+match to be replaced.
+It is an error if no substitutions are performed on any of the addressed
+lines.
+The current address is set the last line affected.
+.Pp
+.Em re
+and
+.Em replacement
+may be delimited by any character other than space and newline
+(see the
+.Em s
+command below).
+If one or two of the last delimiters is omitted, then the last line
+affected is printed as though the print suffix
+.Em p
+were specified.
+.Pp
+An unescaped
+.Ql \e
+in
+.Em replacement
+is replaced by the currently matched text.
+The character sequence
+.Em \em Ns No ,
+where
+.Em m
+is a number in the range [1,9], is replaced by the
+.Em m Ns No th
+backreference expression of the matched text.
+If
+.Em replacement
+consists of a single
+.Ql % ,
+then
+.Em replacement
+from the last substitution is used.
+Newlines may be embedded in
+.Em replacement
+if they are escaped with a backslash
+.Pq Ql \e .
+.It (.,.) Ns Em s
+Repeats the last substitution.
+This form of the
+.Em s
+command accepts a count suffix
+.Em n Ns No ,
+or any combination of the characters
+.Em r Ns No ,
+.Em g Ns No ,
+and
+.Em p Ns No .
+If a count suffix
+.Em n
+is given, then only the
+.Em n Ns No th
+match is replaced.
+The
+.Em r
+suffix causes
+the regular expression of the last search to be used instead of the
+that of the last substitution.
+The
+.Em g
+suffix toggles the global suffix of the last substitution.
+The
+.Em p
+suffix toggles the print suffix of the last substitution
+The current address is set to the last line affected.
+.It (.,.) Ns Em t Ns No (.)
+Copies (i.e., transfers) the addressed lines to after the right-hand
+destination address, which may be the address
+.Em 0
+(zero).
+The current address is set to the last line copied.
+.It Em u
+Undoes the last command and restores the current address
+to what it was before the command.
+The global commands
+.Em g Ns No ,
+.Em G Ns No ,
+.Em v Ns No ,
+and
+.Em V Ns No .
+are treated as a single command by undo.
+.Em u
+is its own inverse.
+.It (1,$) Ns Em v Ns No /re/command-list
+Applies
+.Em command-list
+to each of the addressed lines not matching a regular expression
+.Em re Ns No .
+This is similar to the
+.Em g
+command.
+.It (1,$) Ns Em V Ns No /re/
+Interactively edits the addressed lines not matching a regular expression
+.Em re Ns No .
+This is similar to the
+.Em G
+command.
+.It (1,$) Ns Em w No file
+Writes the addressed lines to
+.Em file Ns No .
+Any previous contents of
+.Em file
+is lost without warning.
+If there is no default filename, then the default filename is set to
+.Em file Ns No ,
+otherwise it is unchanged.
+If no filename is specified, then the default filename is used.
+The current address is unchanged.
+.It (1,$) Ns Em wq No file
+Writes the addressed lines to
+.Em file Ns No ,
+and then executes a
+.Em q
+command.
+.It (1,$) Ns Em w No !command
+Writes the addressed lines to the standard input of
+.Em !command Ns No ,
+(see the
+.Em !
+command below).
+The default filename and current address are unchanged.
+.It (1,$) Ns Em W No file
+Appends the addressed lines to the end of
+.Em file Ns No .
+This is similar to the
+.Em w
+command, expect that the previous contents of file is not clobbered.
+The current address is unchanged.
+.It Em x
+Prompts for an encryption key which is used in subsequent reads and writes.
+If a newline alone is entered as the key, then encryption is turned off.
+Otherwise, echoing is disabled while a key is read.
+Encryption/decryption is done using the
+.Xr bdes 1
+algorithm.
+.It (.+1) Ns Em z Ns No n
+Scrolls
+.Em n
+lines at a time starting at addressed line.
+If
+.Em n
+is not specified, then the current window size is used.
+The current address is set to the last line printed.
+.It ($) Ns Em =
+Prints the line number of the addressed line.
+.It (.+1) Ns Em newline
+Prints the addressed line, and sets the current address to that line.
+.It Em ! Ns No command
+Executes
+.Em command
+via
+.Xr sh 1 .
+If the first character of
+.Em command
+is
+.Em ! Ns No ,
+then it is replaced by text of the previous
+.Em !command Ns No .
+.Nm
+does not process
+.Em command
+for
+.Em \e
+(backslash) escapes.
+However, an unescaped
+.Em %
+is replaced by the default filename.
+When the shell returns from execution, a
+.Em !
+is printed to the standard output.
+The current line is unchanged.
+.El
+.Sh LIMITATIONS
+.Nm
+processes
+.Em file
+arguments for backslash escapes, i.e., in a filename,
+any characters preceded by a backslash
+.Pq Ql \e
+are interpreted literally.
+.Pp
+If a text (non-binary) file is not terminated by a newline character,
+then
+.Nm
+appends one on reading/writing it.
+In the case of a binary file,
+.Nm
+does not append a newline on reading/writing.
+.Sh DIAGNOSTICS
+When an error occurs,
+.Nm
+prints a
+.Dq ?
+and either returns to command mode or exits if its input is from a script.
+An explanation of the last error can be printed with the
+.Em h
+(help) command.
+.Pp
+Since the
+.Em g
+(global) command masks any errors from failed searches and substitutions,
+it can be used to perform conditional operations in scripts; e.g.,
+.Bd -literal -offset indent
+g/old/s//new/
+.Ed
+.Pp
+replaces any occurrences of
+.Em old
+with
+.Em new Ns No .
+.Pp
+If the
+.Em u
+(undo) command occurs in a global command list, then
+the command list is executed only once.
+.Pp
+If diagnostics are not disabled, attempting to quit
+.Nm
+or edit another file before writing a modified buffer results in an error.
+If the command is entered a second time, it succeeds,
+but any changes to the buffer are lost.
+.Sh FILES
+.Bl -tag -width /tmp/ed.* -compact
+.It Pa /tmp/ed.*
+buffer file
+.It Pa ed.hup
+where
+.Nm
+attempts to write the buffer if the terminal hangs up
+.El
+.Sh SEE ALSO
+.Xr bdes 1 ,
+.Xr sed 1 ,
+.Xr sh 1 ,
+.Xr vi 1 ,
+.Xr regex 3
+.Pp
+USD:12-13
+.Rs
+.%A B. W. Kernighan
+.%A P. J. Plauger
+.%B Software Tools in Pascal
+.%O Addison-Wesley
+.%D 1981
+.Re
+.Sh HISTORY
+An
+.Nm
+command appeared in
+.At v1 .
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t11.expected
@@ -1,0 +1,1612 @@
+--- diff-t11.1
++++ diff-t11.2
+@@ -1,1003 +1,908 @@
+-.\" $OpenBSD: t11.1,v 1.2 2007/11/27 16:22:12 martynas Exp $
+-.\" $NetBSD: ed.1,v 1.13 1995/03/21 09:04:38 cgd Exp $
++.\" $OpenBSD: t11.2,v 1.1 2003/07/21 20:16:21 otto Exp $
+ .\"
+-.TH ED 1 "21 May 1993"
+-.SH NAME
+-.\" ed, red \- text editor
+-ed \- text editor
+-.SH SYNOPSIS
+-ed [-] [-sx] [-p \fIstring\fR] [\fIfile\fR]
+-.\" .LP
+-.\" red [-] [-sx] [-p \fIstring\fR] [\fIfile\fR]
+-.SH DESCRIPTION
+-.B ed
++.Dd May 2, 1993
++.Dt ED 1
++.Os
++.Sh NAME
++.Nm ed
++.Nd text editor
++.Sh SYNOPSIS
++.Nm ed
++.Op Fl
++.Op Fl sx
++.Op Fl p Ar string
++.Op Ar file
++.Sh DESCRIPTION
++.Nm
+ is a line-oriented text editor.
+-It is used to create, display, modify and otherwise manipulate text
+-files.
+-.\" .B red
+-.\" is a restricted
+-.\" .BR ed :
+-.\" it can only edit files in the current
+-.\" directory and cannot execute shell commands.
+-
++It is used to create, display, modify, and otherwise manipulate text files.
+ If invoked with a
+-.I file
++.Ar file
+ argument, then a copy of
+-.I file
++.Ar file
+ is read into the editor's buffer.
+ Changes are made to this copy and not directly to
+-.I file
++.Ar file
+ itself.
+ Upon quitting
+-.BR ed ,
+-any changes not explicitly saved with a
+-.I `w'
++.Nm ed ,
++any changes not explicitly saved with a
++.Em w
+ command are lost.
+-
++.Pp
+ Editing is done in two distinct modes:
+-.I command
++.Em command
+ and
+-.IR input .
++.Em input .
+ When first invoked,
+-.B ed
++.Nm
+ is in command mode.
+-In this mode commands are read from the standard input and
++In this mode, commands are read from the standard input and
+ executed to manipulate the contents of the editor buffer.
++.Pp
+ A typical command might look like:
+-.sp
+-.RS
+-,s/\fIold\fR/\fInew\fR/g
+-.RE
+-.sp
++.Bd -literal -offset indent
++,s/old/new/g
++.Ed
++.Pp
+ which replaces all occurrences of the string
+-.I old
++.Pa old
+ with
+-.IR new .
+-
++.Pa new .
++.Pp
+ When an input command, such as
+-.I `a'
++.Em a
+ (append),
+-.I `i'
+-(insert) or
+-.I `c'
+-(change), is given,
+-.B ed
+-enters input mode. This is the primary means
+-of adding text to a file.
++.Em i
++(insert),
++or
++.Em c
++(change) is given,
++.Nm
++enters input mode.
++This is the primary means of adding text to a file.
+ In this mode, no commands are available;
+-instead, the standard input is written
+-directly to the editor buffer. Lines consist of text up to and
+-including a
+-.IR newline
+-character.
+-Input mode is terminated by
+-entering a single period (\fI.\fR) on a line.
+-
++instead, the standard input is written directory to the editor buffer.
++Lines consist of text up to and including a newline character.
++Input mode is terminated by entering a single period
++.Pq Ql \&.
++on a line.
++.Pp
+ All
+-.B ed
++.Nm
+ commands operate on whole lines or ranges of lines; e.g.,
+ the
+-.I `d'
++.Em d
+ command deletes lines; the
+-.I `m'
++.Em m
+ command moves lines, and so on.
+ It is possible to modify only a portion of a line by means of replacement,
+-as in the example above. However even here, the
+-.I `s'
++as in the example above.
++However, even here, the
++.Em s
+ command is applied to whole lines at a time.
+-
++.Pp
+ In general,
+-.B ed
++.Nm
+ commands consist of zero or more line addresses, followed by a single
+ character command and possibly additional parameters; i.e.,
+ commands have the structure:
+-.sp
+-.RS
+-.I [address [,address]]command[parameters]
+-.RE
+-.sp
++.Bd -literal -offset indent
++[address [,address]]command[parameters]
++.Ed
++.Pp
+ The address(es) indicate the line or range of lines to be affected by the
+-command. If fewer addresses are given than the command accepts, then
++command.
++If fewer addresses are given than the command accepts, then
+ default addresses are supplied.
+-
+-.SS OPTIONS
+-.TP 8
+--s
+-Suppresses diagnostics. This should be used if
+-.BR ed 's
++.Pp
++The options are as follows:
++.Bl -tag -width Ds
++.It Fl
++Same as the
++.Fl s
++option (deprecated).
++.It Fl s
++Suppress diagnostics.
++This should be used if
++.Nm
+ standard input is from a script.
+-
+-.TP 8
+--x
+-Prompts for an encryption key to be used in subsequent reads and writes
++.Fl s
++flag.
++.It Fl x
++Prompt for an encryption key to be used in subsequent reads and writes
+ (see the
+-.I `x'
++.Em x
+ command).
+-
+-.TP 8
+-.RI \-p \ string
+-Specifies a command prompt. This may be toggled on and off with the
+-.I `P'
++.It Fl p Ar string
++Specifies a command prompt.
++This may be toggled on and off with the
++.Em P
+ command.
+-
+-.TP 8
+-.I file
+-Specifies the name of a file to read. If
+-.I file
++.It Ar file
++Specifies the name of a file to read.
++If
++.Ar file
+ is prefixed with a
+-bang (!), then it is interpreted as a shell command. In this case,
+-what is read is
+-the standard output of
+-.I file
++bang
++.Pq Ql \&! ,
++then it is interpreted as a shell command.
++In this case, what is read is the standard output of
++.Ar file
+ executed via
+-.IR sh (1).
++.Xr sh 1 .
+ To read a file whose name begins with a bang, prefix the
+-name with a backslash (\\).
++name with a backslash
++.Pq Ql \e .
+ The default filename is set to
+-.I file
++.Ar file
+ only if it is not prefixed with a bang.
+-
+-.SS LINE ADDRESSING
++.El
++.Ss LINE ADDRESSING
+ An address represents the number of a line in the buffer.
+-.B ed
++.Nm
+ maintains a
+-.I current address
+-which is
+-typically supplied to commands as the default address when none is specified.
+-When a file is first read, the current address is set to the last line
+-of the file. In general, the current address is set to the last line
+-affected by a command.
+-
++.Em current address
++which is typically supplied to commands as the default address
++when none is specified.
++When a file is first read, the current address is set to the last line
++of the file.
++In general, the current address is set to the last line affected by a command.
++.Pp
+ A line address is
+ constructed from one of the bases in the list below, optionally followed
+-by a numeric offset. The offset may include any combination
+-of digits, operators (i.e.,
+-.IR + ,
+-.I -
++by a numeric offset.
++The offset may include any combination of digits, operators (i.e.,
++.Em + ,
++.Em - ,
+ and
+-.IR ^ )
++.Em ^ ) ,
+ and whitespace.
+ Addresses are read from left to right, and their values are computed
+ relative to the current address.
+-
++.Pp
+ One exception to the rule that addresses represent line numbers is the
+ address
+-.I 0
++.Em 0
+ (zero).
+-This means "before the first line,"
++This means
++.Dq before the first line ,
+ and is legal wherever it makes sense.
+-
+-An address range is two addresses separated either by a comma or
+-semi-colon. The value of the first address in a range cannot exceed the
+-value of the second. If only one address is given in a range, then
+-the second address is set to the given address. If an
+-.IR n- tuple
++.Pp
++An address range is two addresses separated either by a comma or semi-colon.
++The value of the first address in a range cannot exceed the
++value of the second.
++If only one address is given in a range,
++then the second address is set to the given address.
++If an
++.Em n Ns No -tuple
+ of addresses is given where
+-.I n > 2,
+-then the corresponding range is determined by the last two addresses in
+-the
+-.IR n- tuple.
++.Em n > 2 ,
++then the corresponding range is determined by the last two addresses in the
++.Em n Ns No -tuple.
+ If only one address is expected, then the last address is used.
+-
++.Pp
+ Each address in a comma-delimited range is interpreted relative to the
+-current address. In a semi-colon-delimited range, the first address is
++current address.
++In a semi-colon-delimited range, the first address is
+ used to set the current address, and the second address is interpreted
+ relative to the first.
+-
+-
+-The following address symbols are recognized.
+-
+-.TP 8
+-\&.
++.Pp
++The following address symbols are recognized:
++.Bl -tag -width Ds
++.It Em \&.
+ The current line (address) in the buffer.
+-
+-.TP 8
+-$
++.It Em $
+ The last line in the buffer.
+-
+-.TP 8
+-n
++.It Em n
+ The
+-.IR n th,
+-line in the buffer
+-where
+-.I n
++.Em n Ns No th
++line in the buffer where
++.Em n
+ is a number in the range
+-.I [0,$].
+-
+-.TP 8
+-- or ^
++.Em [0,$] .
++.It Em - No or Em ^
+ The previous line.
+ This is equivalent to
+-.I -1
++.Em -1
+ and may be repeated with cumulative effect.
+-
+-.TP 8
+--\fIn\fR or ^\fIn\fR
++.It Em -n No or Em ^n
+ The
+-.IR n th
++.Em n Ns No th
+ previous line, where
+-.I n
++.Em n
+ is a non-negative number.
+-
+-.TP 8
+-+
+-The
+-next line.
++.It Em +
++The next line.
+ This is equivalent to
+-.I +1
++.Em +1
+ and may be repeated with cumulative effect.
+-
+-.TP 8
+-+\fIn\fR or whitespace\fIn\fR
++.It Em +n
+ The
+-.IR n th
++.Em n Ns No th
+ next line, where
+-.I n
++.Em n
+ is a non-negative number.
+-.I whitespace
+-followed by a number
+-.I n
+-is interpreted as
+-.IR +n .
+-
+-.TP 8
+-, \fRor\fB %
+-The first through last lines in the buffer. This is equivalent to
+-the address range
+-.I 1,$.
+-
+-.TP 8
+-;
+-The
+-current through last lines in the buffer. This is equivalent to
+-the address range
+-.I .,$.
+-
+-.TP 8
+-.RI / re/
+-The
+-next line containing the regular expression
+-.IR re .
++.It Em \&, No or Em %
++The first through last lines in the buffer.
++This is equivalent to the address range
++.Em 1,$ .
++.It Em \&;
++The current through last lines in the buffer.
++This is equivalent to the address range
++.Em .,$ .
++.It Em / Ns No re Ns Em /
++The next line containing the regular expression
++.Em re .
+ The search wraps to the beginning of the buffer and continues down to the
+ current line, if necessary.
+-// repeats the last search.
+-
+-.TP 8
+-.RI ? re?
+-The
+-previous line containing the regular expression
+-.IR re .
++.Em //
++repeats the last search.
++.It Em ? Ns No re Ns Em ?
++The previous line containing the regular expression
++.Em re .
+ The search wraps to the end of the buffer and continues up to the
+ current line, if necessary.
+-?? repeats the last search.
+-
+-.TP 8
+-.RI \' lc
+-The
+-line previously marked by a
+-.I `k'
++.Em ??
++repeats the last search.
++.It Em \&\' Ns No lc
++The line previously marked by a
++.Em k
+ (mark) command, where
+-.I lc
++.Em lc
+ is a lower case letter.
+-
+-.SS REGULAR EXPRESSIONS
++.El
++.Ss REGULAR EXPRESSIONS
+ Regular expressions are patterns used in selecting text.
+ For example, the
+-.B ed
++.Nm
+ command
+-.sp
+-.RS
+-g/\fIstring\fR/
+-.RE
+-.sp
++.Bd -literal -offset indent
++g/string/
++.Ed
++.Pp
+ prints all lines containing
+-.IR string .
+-Regular expressions are also
+-used by the
+-.I `s'
++.Em string .
++Regular expressions are also used by the
++.Em s
+ command for selecting old text to be replaced with new.
+-
++.Pp
+ In addition to a specifying string literals, regular expressions can
+-represent
+-classes of strings. Strings thus represented are said to be matched
+-by the corresponding regular expression.
+-If it is possible for a regular expression
+-to match several strings in a line, then the left-most longest match is
+-the one selected.
+-
++represent classes of strings.
++Strings thus represented are said to be matched by the
++corresponding regular expression.
++If it is possible for a regular expression to match several strings in
++a line, then the leftmost longest match is the one selected.
++.Pp
+ The following symbols are used in constructing regular expressions:
+-
+-.TP 8
+-c
++.Bl -tag -width Dsasdfsd
++.It Em c
+ Any character
+-.I c
+-not listed below, including `{', '}', `(', `)', `<' and `>',
++.Em c
++not listed below, including
++.Em { Ns No ,
++.Em } Ns No ,
++.Em \&( Ns No ,
++.Em \&) Ns No ,
++.Em < Ns No ,
++and
++.Em >
+ matches itself.
+-
+-.TP 8
+-\fR\e\fIc\fR
++.It Em \ec
+ Any backslash-escaped character
+-.IR c ,
+-except for `{', '}', `(', `)', `<' and `>',
++.Em c Ns No ,
++except for
++.Em { Ns No ,
++.Em } Ns No ,
++.Em \&( Ns No ,
++.Em \&) Ns No ,
++.Em < Ns No , and
++.Em >
+ matches itself.
+-
+-.TP 8
+-\fR.\fR
++.It Em \&.
+ Matches any single character.
+-
+-.TP 8
+-.I [char-class]
++.It Em [char-class]
+ Matches any single character in
+-.IR char-class .
+-To include a `]'
++.Em char-class .
++To include a
++.Ql \&]
+ in
+-.IR char-class ,
++.Em char-class Ns No ,
+ it must be the first character.
+ A range of characters may be specified by separating the end characters
+-of the range with a `-', e.g., `a-z' specifies the lower case characters.
++of the range with a
++.Ql - ;
++e.g.,
++.Em a-z
++specifies the lower case characters.
+ The following literal expressions can also be used in
+-.I char-class
++.Em char-class
+ to specify sets of characters:
+-.sp
+-\ \ [:alnum:]\ \ [:cntrl:]\ \ [:lower:]\ \ [:space:]
+-.PD 0
+-\ \ [:alpha:]\ \ [:digit:]\ \ [:print:]\ \ [:upper:]
+-.PD 0
+-\ \ [:blank:]\ \ [:graph:]\ \ [:punct:]\ \ [:xdigit:]
+-.sp
+-If `-' appears as the first or last
+-character of
+-.IR char-class ,
++.Pp
++.Em \ \ [:alnum:]\ \ [:cntrl:]\ \ [:lower:]\ \ [:space:]
++.Em \ \ [:alpha:]\ \ [:digit:]\ \ [:print:]\ \ [:upper:]
++.Em \ \ [:blank:]\ \ [:graph:]\ \ [:punct:]\ \ [:xdigit:]
++.Pp
++If
++.Ql -
++appears as the first or last character of
++.Em char-class Ns No ,
+ then it matches itself.
+ All other characters in
+-.I char-class
++.Em char-class
+ match themselves.
+-.sp
++.Pp
+ Patterns in
+-.I char-class
+-of the form:
+-.sp
+-\ \ [.\fIcol-elm\fR.] or,
+-.PD 0
+-\ \ [=\fIcol-elm\fR=]
+-.sp
++.Em char-class
++of the form
++.Em [.col-elm.] No or Em [=col-elm=]
+ where
+-.I col-elm
+-is a
+-.I collating element
+-are interpreted according to
+-.IR locale (5)
++.Em col-elm
++is a collating element are interpreted according to
++.Xr locale 5
+ (not currently supported).
+ See
+-.IR regex (3)
++.Xr regex 3
+ for an explanation of these constructs.
+-
+-.TP 8
+-[^\fIchar-class\fR]
++.It Em [^char-class]
+ Matches any single character, other than newline, not in
+-.IR char-class .
+-.IR char-class
+-is defined
+-as above.
+-
+-.TP 8
+-^
+-If `^' is the first character of a regular expression, then it
++.Em char-class Ns No .
++.Em char-class
++is defined as above.
++.It Em ^
++If
++.Em ^
++is the first character of a regular expression, then it
+ anchors the regular expression to the beginning of a line.
+ Otherwise, it matches itself.
+-
+-.TP 8
+-$
+-If `$' is the last character of a regular expression, it
+-anchors the regular expression to the end of a line.
++.It Em $
++If
++.Em $
++is the last character of a regular expression,
++it anchors the regular expression to the end of a line.
+ Otherwise, it matches itself.
+-
+-.TP 8
+-\fR\e<\fR
++.It Em \e<
+ Anchors the single character regular expression or subexpression
+ immediately following it to the beginning of a word.
+-(This may not be available)
+-
+-.TP 8
+-\fR\e>\fR
++(This may not be available.)
++.It Em \e>
+ Anchors the single character regular expression or subexpression
+ immediately following it to the end of a word.
+-(This may not be available)
+-
+-.TP 8
+-\fR\e(\fIre\fR\e)\fR
++(This may not be available.)
++.It Em \e( Ns No re Ns Em \e)
+ Defines a subexpression
+-.IR re .
++.Em re .
+ Subexpressions may be nested.
+-A subsequent backreference of the form \fI`\en'\fR, where
+-.I n
++A subsequent backreference of the form
++.Em \en Ns No ,
++where
++.Em n
+ is a number in the range [1,9], expands to the text matched by the
+-.IR n th
++.Em n Ns No th
+ subexpression.
+-For example, the regular expression `\e(.*\e)\e1' matches any string
+-consisting of identical adjacent substrings.
+-Subexpressions are ordered relative to
+-their left delimiter.
+-
+-.TP 8
+-*
++For example, the regular expression
++.Em \e(.*\e)\e1
++matches any string consisting of identical adjacent substrings.
++Subexpressions are ordered relative to their left delimiter.
++.It Em *
+ Matches the single character regular expression or subexpression
+-immediately preceding it zero or more times. If '*' is the first
+-character of a regular expression or subexpression, then it matches
+-itself. The `*' operator sometimes yields unexpected results.
+-For example, the regular expression `b*' matches the beginning of
+-the string `abbb' (as opposed to the substring `bbb'), since a null match
+-is the only left-most match.
+-
+-.TP 8
+-\fR\e{\fIn,m\fR\e}\fR or \fR\e{\fIn,\fR\e}\fR or \fR\e{\fIn\fR\e}\fR
++immediately preceding it zero or more times.
++If
++.Em *
++is the first character of a regular expression or subexpression,
++then it matches itself.
++The
++.Em *
++operator sometimes yields unexpected results.
++For example, the regular expression
++.Em b*
++matches the beginning of the string
++.Em abbb
++(as opposed to the substring
++.Em bbb Ns No ),
++since a null match is the only leftmost match.
++.Sm off
++.It Xo Em \e{ No n,m
++.Em \e}\ \e{ No n, Em \e}\
++.Em \e{ No n Em \e}
++.Xc
++.Sm on
+ Matches the single character regular expression or subexpression
+ immediately preceding it at least
+-.I n
++.Em n
+ and at most
+-.I m
++.Em m
+ times.
+ If
+-.I m
++.Em m
+ is omitted, then it matches at least
+-.I n
++.Em n
+ times.
+ If the comma is also omitted, then it matches exactly
+-.I n
++.Em n
+ times.
+-
+-.LP
++.El
++.Pp
+ Additional regular expression operators may be defined depending on the
+ particular
+-.IR regex (3)
++.Xr regex 3
+ implementation.
+-
+-.SS COMMANDS
++.Ss COMMANDS
+ All
+-.B ed
+-commands are single characters, though some require additonal parameters.
++.Nm
++commands are single characters, though some require additional parameters.
+ If a command's parameters extend over several lines, then
+-each line except for the last
+-must be terminated with a backslash (\\).
+-
++each line except for the last must be terminated with a backslash
++.Pq Ql \e .
++.Pp
+ In general, at most one command is allowed per line.
+ However, most commands accept a print suffix, which is any of
+-.I `p'
+-(print),
+-.I `l'
+-(list) ,
++.Em p No (print),
++.Em l No (list),
+ or
+-.I `n'
+-(enumerate),
++.Em n No (enumerate),
+ to print the last line affected by the command.
+-
++.Pp
+ An interrupt (typically ^C) has the effect of aborting the current command
+ and returning the editor to command mode.
+-
+-.B ed
+-recognizes the following commands. The commands are shown together with
++.Pp
++.Nm
++recognizes the following commands.
++The commands are shown together with
+ the default address or address range supplied if none is
+-specified (in parenthesis).
+-
+-.TP 8
+-(.)a
++specified (in parentheses), and other possible arguments on the right.
++.Bl -tag -width Dxxs
++.It (.) Ns Em a
+ Appends text to the buffer after the addressed line.
+ Text is entered in input mode.
+ The current address is set to last line entered.
+-
+-.TP 8
+-(.,.)c
+-Changes lines in the buffer. The addressed lines are deleted
+-from the buffer, and text is appended in their place.
++.It (.,.) Ns Em c
++Changes lines in the buffer.
++The addressed lines are deleted from the buffer,
++and text is appended in their place.
+ Text is entered in input mode.
+ The current address is set to last line entered.
+-
+-.TP 8
+-(.,.)d
++.It (.,.) Ns Em d
+ Deletes the addressed lines from the buffer.
+ If there is a line after the deleted range, then the current address is set
+-to this line. Otherwise the current address is set to the line
+-before the deleted range.
+-
+-.TP 8
+-.RI e \ file
++to this line.
++Otherwise the current address is set to the line before the deleted range.
++.It Em e No file
+ Edits
+-.IR file ,
++.Em file Ns No ,
+ and sets the default filename.
+ If
+-.I file
+-is not specified, then the default filename is used.
+-Any lines in the buffer are deleted before
+-the new file is read.
++.Em file
++is not specified, then the default filename is used.
++Any lines in the buffer are deleted before the new file is read.
+ The current address is set to the last line read.
+-
+-.TP 8
+-.RI e \ !command
++.It Em e No !command
+ Edits the standard output of
+-.IR `!command' ,
++.Em !command Ns No ,
+ (see
+-.RI ! command
++.Em ! No command
+ below).
+ The default filename is unchanged.
+ Any lines in the buffer are deleted before the output of
+-.I command
++.Em command
+ is read.
+ The current address is set to the last line read.
+-
+-.TP 8
+-.RI E \ file
++.It Em E No file
+ Edits
+-.I file
++.Em file
+ unconditionally.
+ This is similar to the
+-.I e
+-command,
+-except that unwritten changes are discarded without warning.
++.Em e
++command, except that unwritten changes are discarded without warning.
+ The current address is set to the last line read.
+-
+-.TP 8
+-.RI f \ file
++.It Em f No file
+ Sets the default filename to
+-.IR file .
++.Em file Ns No .
+ If
+-.I file
++.Em file
+ is not specified, then the default unescaped filename is printed.
+-
+-.TP 8
+-.RI (1,$)g /re/command-list
++.It (1,$) Ns Em g Ns No /re/command-list
+ Applies
+-.I command-list
++.Em command-list
+ to each of the addressed lines matching a regular expression
+-.IR re .
+-The current address is set to the
+-line currently matched before
+-.I command-list
++.Em re Ns No .
++The current address is set to the line currently matched before
++.Em command-list
+ is executed.
+ At the end of the
+-.I `g'
++.Em g
+ command, the current address is set to the last line affected by
+-.IR command-list .
+-
++.Em command-list Ns No .
++.Pp
+ Each command in
+-.I command-list
++.Em command-list
+ must be on a separate line,
+-and every line except for the last must be terminated by a backslash
+-(\\).
++and every line except for the last must be terminated by
++.Em \e No (backslash).
+ Any commands are allowed, except for
+-.IR `g' ,
+-.IR `G' ,
+-.IR `v' ,
++.Em g Ns No ,
++.Em G Ns No ,
++.Em v Ns No ,
+ and
+-.IR `V' .
++.Em V Ns No .
+ A newline alone in
+-.I command-list
+-is equivalent to a
+-.I `p'
++.Em command-list
++is equivalent to a
++.Em p
+ command.
+-
+-.TP 8
+-.RI (1,$)G /re/
++.It (1,$) Ns Em G Ns No /re/
+ Interactively edits the addressed lines matching a regular expression
+-.IR re.
+-For each matching line,
+-the line is printed,
+-the current address is set,
+-and the user is prompted to enter a
+-.IR command-list .
++.Em re Ns No .
++For each matching line, the line is printed, the current address is set,
++and the user is prompted to enter a
++.Em command-list Ns No .
+ At the end of the
+-.I `G'
+-command, the current address
+-is set to the last line affected by (the last)
+-.IR command-list .
+-
++.Em g
++command, the current address is set to the last line affected by (the last)
++.Em command-list Ns No .
++.Pp
+ The format of
+-.I command-list
++.Em command-list
+ is the same as that of the
+-.I `g'
+-command. A newline alone acts as a null command list.
+-A single `&' repeats the last non-null command list.
+-
+-.TP 8
+-H
++.Em g
++command.
++A newline alone acts as a null command list.
++A single
++.Em &
++repeats the last non-null command list.
++.It Em H
+ Toggles the printing of error explanations.
+ By default, explanations are not printed.
+-It is recommended that ed scripts begin with this command to
+-aid in debugging.
+-
+-.TP 8
+-h
++It is recommended that
++.Nm
++scripts begin with this command to aid in debugging.
++.It Em h
+ Prints an explanation of the last error.
+-
+-.TP 8
+-(.)i
++.It (.) Ns Em i
+ Inserts text in the buffer before the current line.
+ Text is entered in input mode.
+ The current address is set to the last line entered.
+-
+-.TP 8
+-(.,.+1)j
+-Joins the addressed lines. The addressed lines are
+-deleted from the buffer and replaced by a single
++.It (.,.+1) Ns Em j
++Joins the addressed lines.
++The addressed lines are deleted from the buffer and replaced by a single
+ line containing their joined text.
+ The current address is set to the resultant line.
+-
+-.TP 8
+-.RI (.)k lc
++.It (.) Ns Em klc
+ Marks a line with a lower case letter
+-.IR lc .
+-The line can then be addressed as
+-.I 'lc
++.Em lc Ns No \&.
++The line can then be addressed as
++.Em \&'lc
+ (i.e., a single quote followed by
+-.I lc
+-) in subsequent commands. The mark is not cleared until the line is
+-deleted or otherwise modified.
+-
+-.TP 8
+-(.,.)l
++.Em lc Ns No )
++in subsequent commands.
++The mark is not cleared until the line is deleted or otherwise modified.
++.It (.,.) Ns Em l
+ Prints the addressed lines unambiguously.
+-If a single line fills for than one screen (as might be the case
+-when viewing a binary file, for instance), a `--More--'
+-prompt is printed on the last line.
+-.B ed
+-waits until the RETURN key is pressed
+-before displaying the next screen.
+-The current address is set to the last line
+-printed.
+-
+-.TP 8
+-(.,.)m(.)
+-Moves lines in the buffer. The addressed lines are moved to after the
++If a single line fills more than one screen (as might be the case
++when viewing a binary file, for instance), a
++.Dq --More--
++prompt is printed on the last line.
++.Nm
++waits until the RETURN key is pressed before displaying the next screen.
++The current address is set to the last line printed.
++.It (.,.) Ns Em m Ns No (.)
++Moves lines in the buffer.
++The addressed lines are moved to after the
+ right-hand destination address, which may be the address
+-.IR 0
++.Em 0
+ (zero).
+-The current address is set to the
+-last line moved.
+-
+-.TP 8
+-(.,.)n
+-Prints the addressed lines along with
+-their line numbers. The current address is set to the last line
+-printed.
+-
+-.TP 8
+-(.,.)p
+-Prints the addressed lines. The current address is set to the last line
+-printed.
+-
+-.TP 8
+-P
++The current address is set to the last line moved.
++.It (.,.) Ns Em n
++Prints the addressed lines along with their line numbers.
++The current address is set to the last line printed.
++.It (.,.) Ns Em p
++Prints the addressed lines.
++The current address is set to the last line printed.
++.It Em P
+ Toggles the command prompt on and off.
+ Unless a prompt was specified by with command-line option
+-\fI-p string\fR, the command prompt is by default turned off.
+-
+-.TP 8
+-q
+-Quits ed.
+-
+-.TP 8
+-Q
+-Quits ed unconditionally.
++.Fl p Ar string Ns No ,
++the command prompt is by default turned off.
++.It Em q
++Quits
++.Nm ed .
++.It Em Q
++Quits
++.Nm
++unconditionally.
+ This is similar to the
+-.I q
+-command,
+-except that unwritten changes are discarded without warning.
+-
+-.TP 8
+-.RI ($)r \ file
++.Em q
++command, except that unwritten changes are discarded without warning.
++.It ($) Ns Em r No file
+ Reads
+-.I file
+-to after the addressed line. If
+-.I file
+-is not specified, then the default
+-filename is used. If there was no default filename prior to the command,
++.Em file
++to after the addressed line.
++If
++.Em file
++is not specified, then the default filename is used.
++If there was no default filename prior to the command,
+ then the default filename is set to
+-.IR file .
++.Em file Ns No .
+ Otherwise, the default filename is unchanged.
+ The current address is set to the last line read.
+-
+-.TP 8
+-.RI ($)r \ !command
+-Reads
+-to after the addressed line
+-the standard output of
+-.IR `!command' ,
++.It ($) Ns Em r No !command
++Reads to after the addressed line the standard output of
++.Em !command Ns No ,
+ (see the
+-.RI ! command
+-below).
++.Em !
++command below).
+ The default filename is unchanged.
+ The current address is set to the last line read.
+-
+-.HP
+-.RI (.,.)s /re/replacement/
+-.PD 0
+-.HP
+-.RI (.,.)s /re/replacement/\fRg\fR
+-.HP
+-.RI (.,.)s /re/replacement/n
+-.br
+-Replaces text in the addressed lines
+-matching a regular expression
+-.I re
++.Sm off
++.It Xo (.,.) Em s No /re/replacement/ , \ (.,.)
++.Em s No /re/replacement/ Em g , No \ (.,.)
++.Em s No /re/replacement/ Em n
++.Xc
++.Sm on
++Replaces text in the addressed lines matching a regular expression
++.Em re
+ with
+-.IR replacement .
++.Em replacement Ns No .
+ By default, only the first match in each line is replaced.
+ If the
+-.I `g'
++.Em g
+ (global) suffix is given, then every match to be replaced.
+ The
+-.I `n'
++.Em n
+ suffix, where
+-.I n
+-is a postive number, causes only the
+-.IR n th
++.Em n
++is a positive number, causes only the
++.Em n Ns No th
+ match to be replaced.
+ It is an error if no substitutions are performed on any of the addressed
+ lines.
+ The current address is set the last line affected.
+-
+-.I re
++.Pp
++.Em re
+ and
+-.I replacement
++.Em replacement
+ may be delimited by any character other than space and newline
+ (see the
+-.I `s'
++.Em s
+ command below).
+ If one or two of the last delimiters is omitted, then the last line
+ affected is printed as though the print suffix
+-.I `p'
++.Em p
+ were specified.
+-
+-
+-An unescaped `&' in
+-.I replacement
++.Pp
++An unescaped
++.Ql \e
++in
++.Em replacement
+ is replaced by the currently matched text.
+ The character sequence
+-\fI`\em'\fR,
++.Em \em Ns No ,
+ where
+-.I m
++.Em m
+ is a number in the range [1,9], is replaced by the
+-.IR m th
++.Em m Ns No th
+ backreference expression of the matched text.
+ If
+-.I replacement
+-consists of a single `%', then
+-.I replacement
++.Em replacement
++consists of a single
++.Ql % ,
++then
++.Em replacement
+ from the last substitution is used.
+ Newlines may be embedded in
+-.I replacement
+-if they are escaped with a backslash (\\).
+-
+-.TP 8
+-(.,.)s
++.Em replacement
++if they are escaped with a backslash
++.Pq Ql \e .
++.It (.,.) Ns Em s
+ Repeats the last substitution.
+ This form of the
+-.I `s'
++.Em s
+ command accepts a count suffix
+-.IR `n' ,
++.Em n Ns No ,
+ or any combination of the characters
+-.IR `r' ,
+-.IR `g' ,
++.Em r Ns No ,
++.Em g Ns No ,
+ and
+-.IR `p' .
++.Em p Ns No .
+ If a count suffix
+-.I `n'
++.Em n
+ is given, then only the
+-.IR n th
++.Em n Ns No th
+ match is replaced.
+ The
+-.I `r'
++.Em r
+ suffix causes
+ the regular expression of the last search to be used instead of the
+ that of the last substitution.
+ The
+-.I `g'
++.Em g
+ suffix toggles the global suffix of the last substitution.
+ The
+-.I `p'
++.Em p
+ suffix toggles the print suffix of the last substitution
+ The current address is set to the last line affected.
+-
+-.TP 8
+-(.,.)t(.)
++.It (.,.) Ns Em t Ns No (.)
+ Copies (i.e., transfers) the addressed lines to after the right-hand
+ destination address, which may be the address
+-.IR 0
++.Em 0
+ (zero).
+-The current address is set to the last line
+-copied.
+-
+-.TP 8
+-u
++The current address is set to the last line copied.
++.It Em u
+ Undoes the last command and restores the current address
+ to what it was before the command.
+ The global commands
+-.IR `g' ,
+-.IR `G' ,
+-.IR `v' ,
++.Em g Ns No ,
++.Em G Ns No ,
++.Em v Ns No ,
+ and
+-.IR `V' .
++.Em V Ns No .
+ are treated as a single command by undo.
+-.I `u'
++.Em u
+ is its own inverse.
+-
+-.TP 8
+-.RI (1,$)v /pat/command-list
++.It (1,$) Ns Em v Ns No /re/command-list
+ Applies
+-.I command-list
++.Em command-list
+ to each of the addressed lines not matching a regular expression
+-.IR re .
++.Em re Ns No .
+ This is similar to the
+-.I `g'
++.Em g
+ command.
+-
+-.TP 8
+-.RI (1,$)V /re/
++.It (1,$) Ns Em V Ns No /re/
+ Interactively edits the addressed lines not matching a regular expression
+-.IR re.
++.Em re Ns No .
+ This is similar to the
+-.I `G'
++.Em G
+ command.
+-
+-.TP 8
+-.RI (1,$)w \ file
++.It (1,$) Ns Em w No file
+ Writes the addressed lines to
+-.IR file .
++.Em file Ns No .
+ Any previous contents of
+-.I file
++.Em file
+ is lost without warning.
+ If there is no default filename, then the default filename is set to
+-.IR file,
+-otherwise it is unchanged. If no filename is specified, then the default
+-filename is used.
++.Em file Ns No ,
++otherwise it is unchanged.
++If no filename is specified, then the default filename is used.
+ The current address is unchanged.
+-
+-.TP 8
+-.RI (1,$)wq \ file
++.It (1,$) Ns Em wq No file
+ Writes the addressed lines to
+-.IR file ,
++.Em file Ns No ,
+ and then executes a
+-.I `q'
++.Em q
+ command.
+-
+-.TP 8
+-.RI (1,$)w \ !command
++.It (1,$) Ns Em w No !command
+ Writes the addressed lines to the standard input of
+-.IR `!command' ,
++.Em !command Ns No ,
+ (see the
+-.RI ! command
+-below).
++.Em !
++command below).
+ The default filename and current address are unchanged.
+-
+-.TP 8
+-.RI (1,$)W \ file
++.It (1,$) Ns Em W No file
+ Appends the addressed lines to the end of
+-.IR file .
++.Em file Ns No .
+ This is similar to the
+-.I `w'
++.Em w
+ command, expect that the previous contents of file is not clobbered.
+ The current address is unchanged.
+-
+-.TP 8
+-x
+-Prompts for an encryption key which is used in subsequent reads and
+-writes. If a newline alone is entered as the key, then encryption is
+-turned off. Otherwise, echoing is disabled while a key is read.
+-Encryption/decryption is done using the bdes(1) algorithm.
+-
+-.TP 8
+-.RI (.+1)z n
++.It Em x
++Prompts for an encryption key which is used in subsequent reads and writes.
++If a newline alone is entered as the key, then encryption is turned off.
++Otherwise, echoing is disabled while a key is read.
++Encryption/decryption is done using the
++.Xr bdes 1
++algorithm.
++.It (.+1) Ns Em z Ns No n
+ Scrolls
+-.I n
+-lines at a time starting at addressed line. If
+-.I n
++.Em n
++lines at a time starting at addressed line.
++If
++.Em n
+ is not specified, then the current window size is used.
+ The current address is set to the last line printed.
+-
+-.TP 8
+-.RI ! command
++.It ($) Ns Em =
++Prints the line number of the addressed line.
++.It (.+1) Ns Em newline
++Prints the addressed line, and sets the current address to that line.
++.It Em ! Ns No command
+ Executes
+-.I command
++.Em command
+ via
+-.IR sh (1).
++.Xr sh 1 .
+ If the first character of
+-.I command
+-is `!', then it is replaced by text of the
+-previous
+-.IR `!command' .
+-.B ed
++.Em command
++is
++.Em ! Ns No ,
++then it is replaced by text of the previous
++.Em !command Ns No .
++.Nm
+ does not process
+-.I command
+-for backslash (\\) escapes.
++.Em command
++for
++.Em \e
++(backslash) escapes.
+ However, an unescaped
+-.I `%'
++.Em %
+ is replaced by the default filename.
+-When the shell returns from execution, a `!'
++When the shell returns from execution, a
++.Em !
+ is printed to the standard output.
+ The current line is unchanged.
+-
+-.TP 8
+-($)=
+-Prints the line number of the addressed line.
+-
+-.TP 8
+-(.+1)newline
+-Prints the addressed line, and sets the current address to
+-that line.
+-
+-.SH FILES
+-.TP 20
+-/tmp/ed.*
+-Buffer file
+-.PD 0
+-.TP 20
+-ed.hup
+-The file to which
+-.B ed
+-attempts to write the buffer if the terminal hangs up.
+-
+-.SH SEE ALSO
+-
+-.IR vi (1),
+-.IR sed (1),
+-.IR regex (3),
+-.IR bdes (1),
+-.IR sh (1).
+-
+-USD:12-13
+-
+-B. W. Kernighan and P. J. Plauger,
+-.I Software Tools in Pascal ,
+-Addison-Wesley, 1981.
+-
+-.SH LIMITATIONS
+-.B ed
++.El
++.Sh LIMITATIONS
++.Nm
+ processes
+-.I file
+-arguments for backslash escapes, i.e., in a filename,
+-any characters preceded by a backslash (\\) are
+-interpreted literally.
+-
++.Em file
++arguments for backslash escapes, i.e., in a filename,
++any characters preceded by a backslash
++.Pq Ql \e
++are interpreted literally.
++.Pp
+ If a text (non-binary) file is not terminated by a newline character,
+ then
+-.B ed
+-appends one on reading/writing it. In the case of a binary file,
+-.B ed
++.Nm
++appends one on reading/writing it.
++In the case of a binary file,
++.Nm
+ does not append a newline on reading/writing.
+-
+-per line overhead: 4 ints
+-
+-.SH DIAGNOSTICS
++.Sh DIAGNOSTICS
+ When an error occurs,
+-.B ed
+-prints a `?' and either returns to command mode
+-or exits if its input is from a script.
+-An explanation of the last error can be
+-printed with the
+-.I `h'
++.Nm
++prints a
++.Dq ?
++and either returns to command mode or exits if its input is from a script.
++An explanation of the last error can be printed with the
++.Em h
+ (help) command.
+-
+-Since the
+-.I `g'
+-(global) command masks any errors from failed searches and substitutions,
++.Pp
++Since the
++.Em g
++(global) command masks any errors from failed searches and substitutions,
+ it can be used to perform conditional operations in scripts; e.g.,
+-.sp
+-.RS
+-g/\fIold\fR/s//\fInew\fR/
+-.RE
+-.sp
++.Bd -literal -offset indent
++g/old/s//new/
++.Ed
++.Pp
+ replaces any occurrences of
+-.I old
++.Em old
+ with
+-.IR new .
++.Em new Ns No .
++.Pp
+ If the
+-.I `u'
++.Em u
+ (undo) command occurs in a global command list, then
+ the command list is executed only once.
+-
++.Pp
+ If diagnostics are not disabled, attempting to quit
+-.B ed
+-or edit another file before writing a modified buffer
+-results in an error.
++.Nm
++or edit another file before writing a modified buffer results in an error.
+ If the command is entered a second time, it succeeds,
+ but any changes to the buffer are lost.
++.Sh FILES
++.Bl -tag -width /tmp/ed.* -compact
++.It Pa /tmp/ed.*
++buffer file
++.It Pa ed.hup
++where
++.Nm
++attempts to write the buffer if the terminal hangs up
++.El
++.Sh SEE ALSO
++.Xr bdes 1 ,
++.Xr sed 1 ,
++.Xr sh 1 ,
++.Xr vi 1 ,
++.Xr regex 3
++.Pp
++USD:12-13
++.Rs
++.%A B. W. Kernighan
++.%A P. J. Plauger
++.%B Software Tools in Pascal
++.%O Addison-Wesley
++.%D 1981
++.Re
++.Sh HISTORY
++An
++.Nm
++command appeared in
++.At v1 .
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t12.1
@@ -1,0 +1,10 @@
+# Test HMAC:
+
+PROG= hmactest
+SRCS= hash.c hmactest.c
+.PATH: ${.CURDIR}/../../
+NOMAN=
+CFLAGS+= -I${.CURDIR}/../../ -Wall
+DEBUG= -g
+
+.include <bsd.prog.mk>
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t12.2
@@ -1,0 +1,12 @@
+# $OpenBSD: t12.2,v 1.1 2003/07/21 20:16:21 otto Exp $
+
+# Test HMAC:
+
+PROG= hmactest
+SRCS= hash.c hmactest.c
+.PATH: ${.CURDIR}/../../
+NOMAN=
+CFLAGS+= -I${.CURDIR}/../../ -Wall
+DEBUG= -g
+
+.include <bsd.prog.mk>
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t12.expected
@@ -1,0 +1,8 @@
+--- diff-t12.1
++++ diff-t12.2
+@@ -1,3 +1,5 @@
++# $OpenBSD: t12.2,v 1.1 2003/07/21 20:16:21 otto Exp $
++
+ # Test HMAC:
+
+ PROG= hmactest
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t13.1
@@ -1,0 +1,11 @@
+A line of text
+.
+Another line of text
+..
+A third line
+...
+A fourth line
+.
+We keep counting
+.
+.
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t13.2
@@ -1,0 +1,9 @@
+A line of text
+Another line of text
+..
+.
+A third line
+...
+..
+We keep counting
+.
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t13.expected
@@ -1,0 +1,16 @@
+--- diff-t13.1
++++ diff-t13.2
+@@ -1,11 +1,9 @@
+ A line of text
+-.
+ Another line of text
+ ..
++.
+ A third line
+ ...
+-A fourth line
+-.
++..
+ We keep counting
+-.
+ .
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t14.1
@@ -1,0 +1,2 @@
+1
+2
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t14.2
@@ -1,0 +1,3 @@
+1
+2
+3
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t14.expected
@@ -1,0 +1,7 @@
+--- diff-t14.1
++++ diff-t14.2
+@@ -1,2 +1,3 @@
+ 1
+-2
++2
++3
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t15.1
@@ -1,0 +1,3 @@
+1
+2
+3
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t15.2
@@ -1,0 +1,2 @@
+1
+2
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t15.expected
@@ -1,0 +1,6 @@
+--- diff-t15.1
++++ diff-t15.2
+@@ -1,3 +1,2 @@
+ 1
+ 2
+-3
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t2.1
@@ -1,0 +1,25 @@
+Below is an example license to be used for new code in OpenBSD,
+modeled after the ISC license.
+
+It is important to specify the year of the copyright. Additional years
+should be separated by a comma, e.g.
+ Copyright (c) 2003, 2004
+
+If you add extra text to the body of the license, be careful not to
+add further restrictions.
+
+/*
+ * Copyright (c) CCYY YOUR NAME HERE <user@your.dom.ain>
+ *
+ * Permission to use, copy, modify, and distribute this software for any
+ * purpose with or without fee is hereby granted, provided that the above
+ * copyright notice and this permission notice appear in all copies.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+ * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
+ * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
+ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
+ * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+ */
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t2.2
@@ -1,0 +1,25 @@
+Below is an example license to be used for new code in OpenBSD,
+modeled after the ISC license.
+
+It is important to specify the year of the copyright. Additional years
+should be separated by a comma, e.g.
+ Copyright (c) 2003, 2004, 2005
+
+If you add extra text to the body of the license, be careful not to
+add further restrictions.
+
+/*
+ * Copyright (c) CCYY YOUR NAME HERE <user@your.dom.ain>
+ *
+ * purpose with or without fee is hereby granted, provided that the above
+ * copyright notice and this permission notice appear in all copies.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+ * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
+ * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
+ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
+ * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+ */
+An extra line
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t2.expected
@@ -1,0 +1,24 @@
+--- diff-t2.1
++++ diff-t2.2
+@@ -3,7 +3,7 @@
+
+ It is important to specify the year of the copyright. Additional years
+ should be separated by a comma, e.g.
+- Copyright (c) 2003, 2004
++ Copyright (c) 2003, 2004, 2005
+
+ If you add extra text to the body of the license, be careful not to
+ add further restrictions.
+@@ -11,7 +11,6 @@
+ /*
+ * Copyright (c) CCYY YOUR NAME HERE <user@your.dom.ain>
+ *
+- * Permission to use, copy, modify, and distribute this software for any
+ * purpose with or without fee is hereby granted, provided that the above
+ * copyright notice and this permission notice appear in all copies.
+ *
+@@ -23,3 +22,4 @@
+ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
+ * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+ */
++An extra line
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t3.2
@@ -1,0 +1,25 @@
+Below is an example license to be used for new code in OpenBSD,
+modeled after the ISC license.
+
+It is important to specify the year of the copyright. Additional years
+should be separated by a comma, e.g.
+ Copyright (c) 2003, 2004
+
+If you add extra text to the body of the license, be careful not to
+add further restrictions.
+
+/*
+ * Copyright (c) CCYY YOUR NAME HERE <user@your.dom.ain>
+ *
+ * Permission to use, copy, modify, and distribute this software for any
+ * purpose with or without fee is hereby granted, provided that the above
+ * copyright notice and this permission notice appear in all copies.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+ * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
+ * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
+ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
+ * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+ */
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t3.expected
@@ -1,0 +1,28 @@
+--- diff-t3.1
++++ diff-t3.2
+@@ -1,0 +1,25 @@
++Below is an example license to be used for new code in OpenBSD,
++modeled after the ISC license.
++
++It is important to specify the year of the copyright. Additional years
++should be separated by a comma, e.g.
++ Copyright (c) 2003, 2004
++
++If you add extra text to the body of the license, be careful not to
++add further restrictions.
++
++/*
++ * Copyright (c) CCYY YOUR NAME HERE <user@your.dom.ain>
++ *
++ * Permission to use, copy, modify, and distribute this software for any
++ * purpose with or without fee is hereby granted, provided that the above
++ * copyright notice and this permission notice appear in all copies.
++ *
++ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
++ * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
++ * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
++ * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
++ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
++ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
++ * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
++ */
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t4.1
@@ -1,0 +1,25 @@
+Below is an example license to be used for new code in OpenBSD,
+modeled after the ISC license.
+
+It is important to specify the year of the copyright. Additional years
+should be separated by a comma, e.g.
+ Copyright (c) 2003, 2004
+
+If you add extra text to the body of the license, be careful not to
+add further restrictions.
+
+/*
+ * Copyright (c) CCYY YOUR NAME HERE <user@your.dom.ain>
+ *
+ * Permission to use, copy, modify, and distribute this software for any
+ * purpose with or without fee is hereby granted, provided that the above
+ * copyright notice and this permission notice appear in all copies.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+ * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
+ * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
+ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
+ * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+ */
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t4.expected
@@ -1,0 +1,28 @@
+--- diff-t4.1
++++ diff-t4.2
+@@ -1,25 +1,0 @@
+-Below is an example license to be used for new code in OpenBSD,
+-modeled after the ISC license.
+-
+-It is important to specify the year of the copyright. Additional years
+-should be separated by a comma, e.g.
+- Copyright (c) 2003, 2004
+-
+-If you add extra text to the body of the license, be careful not to
+-add further restrictions.
+-
+-/*
+- * Copyright (c) CCYY YOUR NAME HERE <user@your.dom.ain>
+- *
+- * Permission to use, copy, modify, and distribute this software for any
+- * purpose with or without fee is hereby granted, provided that the above
+- * copyright notice and this permission notice appear in all copies.
+- *
+- * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+- * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+- * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
+- * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+- * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
+- * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
+- * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+- */
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t5.1
@@ -1,0 +1,10 @@
+OpenBSD 3.3-current (GENERIC) #47: Mon Jun 30 11:19:56 CEST 2003
+
+Welcome to OpenBSD: The proactively secure Unix-like operating system.
+
+Please use the sendbug(1) utility to report bugs in the system.
+Before reporting a bug, please try to reproduce it with the latest
+version of the code. With bug reports, please try to ensure that
+enough information to reproduce the problem is enclosed, and if a
+known fix for it exists, include that as well.
+
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t5.2
@@ -1,0 +1,10 @@
+OpenBSD 3.3-current (GENERIC) #47: Mon Jun 30 11:19:56 CEST 2003
+
+Welcome to OpenBSD: The proactively secure Unix-like operating system.
+
+Please use the sendbug(1) utility to report bugs in the system.
+Before reporting a bug, please try to reproduce it with the latest
+version of the code. With bug reports, please try to ensure that
+enough information to reproduce the problem is enclosed, and if a
+known fix for it exists, include that as well.
+
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t6.1
@@ -1,0 +1,9 @@
+OpenBSD 3.3-current (GENERIC) #47: Mon Jun 30 11:19:56 CEST 2003
+
+Welcome to OpenBSD: The proactively secure Unix-like operating system.
+
+Please use the sendbug(1) utility to report bugs in the system.
+Before reporting a bug, please try to reproduce it with the latest
+version of the code. With bug reports, please try to ensure that
+enough information to reproduce the problem is enclosed, and if a
+known fix for it exists, include that as well.
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t6.2
@@ -1,0 +1,9 @@
+OpenBSD 3.3-current (GENERIC) #47: Mon Jun 30 11:19:56 CEST 2003
+
+Welcome to OpenBSD: The proactively secure Unix-like operating system.
+
+Please use the sendbug(1) utility to report bugs in the system.
+Before reporting a bug, please try to reproduce it with the latest
+version of the code. With bug reports, please try to ensure that
+enough information to reproduce the problem is enclosed, and if a
+known fix for it exists, include that as well.
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t6.expected
@@ -1,0 +1,8 @@
+--- diff-t6.1
++++ diff-t6.2
+@@ -6,4 +6,4 @@
+ Before reporting a bug, please try to reproduce it with the latest
+ version of the code. With bug reports, please try to ensure that
+ enough information to reproduce the problem is enclosed, and if a
+-known fix for it exists, include that as well.
++known fix for it exists, include that as well.
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t7.1
@@ -1,0 +1,9 @@
+OpenBSD 3.3-current (GENERIC) #47: Mon Jun 30 11:19:56 CEST 2003
+
+Welcome to OpenBSD: The proactively secure Unix-like operating system.
+
+Please use the sendbug(1) utility to report bugs in the system.
+Before reporting a bug, please try to reproduce it with the latest
+version of the code. With bug reports, please try to ensure that
+enough information to reproduce the problem is enclosed, and if a
+known fix for it exists, include that as well.
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t7.2
@@ -1,0 +1,9 @@
+OpenBSD 3.3-current (GENERIC) #47: Mon Jun 30 11:19:56 CEST 2003
+
+Welcome to OpenBSD: The proactively secure Unix-like operating system.
+
+Please use the sendbug(1) utility to report bugs in the system.
+Before reporting a bug, please try to reproduce it with the latest
+version of the code. With bug reports, please try to ensure that
+enough information to reproduce the problem is enclosed, and if a
+known fix for it exists, include that as well.
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t7.expected
@@ -1,0 +1,8 @@
+--- diff-t7.1
++++ diff-t7.2
+@@ -6,4 +6,4 @@
+ Before reporting a bug, please try to reproduce it with the latest
+ version of the code. With bug reports, please try to ensure that
+ enough information to reproduce the problem is enclosed, and if a
+-known fix for it exists, include that as well.
++known fix for it exists, include that as well.
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t8.1
@@ -1,0 +1,392 @@
+/* $NetBSD: kern_malloc.c,v 1.11 1995/05/01 22:39:11 cgd Exp $ */
+
+/*
+ * Copyright (c) 1987, 1991, 1993
+ * The Regents of the University of California. All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * 3. Neither the name of the University nor the names of its contributors
+ * may be used to endorse or promote products derived from this software
+ * without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * @(#)kern_malloc.c 8.3 (Berkeley) 1/4/94
+ */
+
+#include <sys/param.h>
+#include <sys/proc.h>
+#include <sys/map.h>
+#include <sys/kernel.h>
+#include <sys/malloc.h>
+
+#include <vm/vm.h>
+#include <vm/vm_kern.h>
+
+struct kmembuckets bucket[MINBUCKET + 16];
+struct kmemstats kmemstats[M_LAST];
+struct kmemusage *kmemusage;
+char *kmembase, *kmemlimit;
+char *memname[] = INITKMEMNAMES;
+
+#ifdef DIAGNOSTIC
+/*
+ * This structure provides a set of masks to catch unaligned frees.
+ */
+long addrmask[] = { 0,
+ 0x00000001, 0x00000003, 0x00000007, 0x0000000f,
+ 0x0000001f, 0x0000003f, 0x0000007f, 0x000000ff,
+ 0x000001ff, 0x000003ff, 0x000007ff, 0x00000fff,
+ 0x00001fff, 0x00003fff, 0x00007fff, 0x0000ffff,
+};
+
+/*
+ * The WEIRD_ADDR is used as known text to copy into free objects so
+ * that modifications after frees can be detected.
+ */
+#define WEIRD_ADDR 0xdeadbeef
+#define MAX_COPY 32
+
+/*
+ * Normally the freelist structure is used only to hold the list pointer
+ * for free objects. However, when running with diagnostics, the first
+ * 8 bytes of the structure is unused except for diagnostic information,
+ * and the free list pointer is at offst 8 in the structure. Since the
+ * first 8 bytes is the portion of the structure most often modified, this
+ * helps to detect memory reuse problems and avoid free list corruption.
+ */
+struct freelist {
+ int32_t spare0;
+ int16_t type;
+ int16_t spare1;
+ caddr_t next;
+};
+#else /* !DIAGNOSTIC */
+struct freelist {
+ caddr_t next;
+};
+#endif /* DIAGNOSTIC */
+
+/*
+ * Allocate a block of memory
+ */
+void *
+malloc(size, type, flags)
+ unsigned long size;
+ int type, flags;
+{
+ register struct kmembuckets *kbp;
+ register struct kmemusage *kup;
+ register struct freelist *freep;
+ long indx, npg, allocsize;
+ int s;
+ caddr_t va, cp, savedlist;
+#ifdef DIAGNOSTIC
+ int32_t *end, *lp;
+ int copysize;
+ char *savedtype;
+#endif
+#ifdef KMEMSTATS
+ register struct kmemstats *ksp = &kmemstats[type];
+
+ if (((unsigned long)type) > M_LAST)
+ panic("malloc - bogus type");
+#endif
+ indx = BUCKETINDX(size);
+ kbp = &bucket[indx];
+ s = splimp();
+#ifdef KMEMSTATS
+ while (ksp->ks_memuse >= ksp->ks_limit) {
+ if (flags & M_NOWAIT) {
+ splx(s);
+ return ((void *) NULL);
+ }
+ if (ksp->ks_limblocks < 65535)
+ ksp->ks_limblocks++;
+ tsleep((caddr_t)ksp, PSWP+2, memname[type], 0);
+ }
+ ksp->ks_size |= 1 << indx;
+#endif
+#ifdef DIAGNOSTIC
+ copysize = 1 << indx < MAX_COPY ? 1 << indx : MAX_COPY;
+#endif
+ if (kbp->kb_next == NULL) {
+ kbp->kb_last = NULL;
+ if (size > MAXALLOCSAVE)
+ allocsize = roundup(size, CLBYTES);
+ else
+ allocsize = 1 << indx;
+ npg = clrnd(btoc(allocsize));
+ va = (caddr_t) kmem_malloc(kmem_map, (vm_size_t)ctob(npg),
+ !(flags & M_NOWAIT));
+ if (va == NULL) {
+ splx(s);
+ return ((void *) NULL);
+ }
+#ifdef KMEMSTATS
+ kbp->kb_total += kbp->kb_elmpercl;
+#endif
+ kup = btokup(va);
+ kup->ku_indx = indx;
+ if (allocsize > MAXALLOCSAVE) {
+ if (npg > 65535)
+ panic("malloc: allocation too large");
+ kup->ku_pagecnt = npg;
+#ifdef KMEMSTATS
+ ksp->ks_memuse += allocsize;
+#endif
+ goto out;
+ }
+#ifdef KMEMSTATS
+ kup->ku_freecnt = kbp->kb_elmpercl;
+ kbp->kb_totalfree += kbp->kb_elmpercl;
+#endif
+ /*
+ * Just in case we blocked while allocating memory,
+ * and someone else also allocated memory for this
+ * bucket, don't assume the list is still empty.
+ */
+ savedlist = kbp->kb_next;
+ kbp->kb_next = cp = va + (npg * NBPG) - allocsize;
+ for (;;) {
+ freep = (struct freelist *)cp;
+#ifdef DIAGNOSTIC
+ /*
+ * Copy in known text to detect modification
+ * after freeing.
+ */
+ end = (int32_t *)&cp[copysize];
+ for (lp = (int32_t *)cp; lp < end; lp++)
+ *lp = WEIRD_ADDR;
+ freep->type = M_FREE;
+#endif /* DIAGNOSTIC */
+ if (cp <= va)
+ break;
+ cp -= allocsize;
+ freep->next = cp;
+ }
+ freep->next = savedlist;
+ if (kbp->kb_last == NULL)
+ kbp->kb_last = (caddr_t)freep;
+ }
+ va = kbp->kb_next;
+ kbp->kb_next = ((struct freelist *)va)->next;
+#ifdef DIAGNOSTIC
+ freep = (struct freelist *)va;
+ savedtype = (unsigned)freep->type < M_LAST ?
+ memname[freep->type] : "???";
+ if (kbp->kb_next &&
+ !kernacc(kbp->kb_next, sizeof(struct freelist), 0)) {
+ printf("%s %d of object %p size %d %s %s (invalid addr %p)\n",
+ "Data modified on freelist: word",
+ (int32_t *)&kbp->kb_next - (int32_t *)kbp, va, size,
+ "previous type", savedtype, kbp->kb_next);
+ kbp->kb_next = NULL;
+ }
+
+ /* Fill the fields that we've used with WEIRD_ADDR */
+#if BYTE_ORDER == BIG_ENDIAN
+ freep->type = WEIRD_ADDR >> 16;
+#endif
+#if BYTE_ORDER == LITTLE_ENDIAN
+ freep->type = (short)WEIRD_ADDR;
+#endif
+ end = (int32_t *)&freep->next +
+ (sizeof(freep->next) / sizeof(int32_t));
+ for (lp = (int32_t *)&freep->next; lp < end; lp++)
+ *lp = WEIRD_ADDR;
+
+ /* and check that the data hasn't been modified. */
+ end = (int32_t *)&va[copysize];
+ for (lp = (int32_t *)va; lp < end; lp++) {
+ if (*lp == WEIRD_ADDR)
+ continue;
+ printf("%s %d of object %p size %d %s %s (%p != %p)\n",
+ "Data modified on freelist: word", lp - (int32_t *)va,
+ va, size, "previous type", savedtype, *lp, WEIRD_ADDR);
+ break;
+ }
+
+ freep->spare0 = 0;
+#endif /* DIAGNOSTIC */
+#ifdef KMEMSTATS
+ kup = btokup(va);
+ if (kup->ku_indx != indx)
+ panic("malloc: wrong bucket");
+ if (kup->ku_freecnt == 0)
+ panic("malloc: lost data");
+ kup->ku_freecnt--;
+ kbp->kb_totalfree--;
+ ksp->ks_memuse += 1 << indx;
+out:
+ kbp->kb_calls++;
+ ksp->ks_inuse++;
+ ksp->ks_calls++;
+ if (ksp->ks_memuse > ksp->ks_maxused)
+ ksp->ks_maxused = ksp->ks_memuse;
+#else
+out:
+#endif
+ splx(s);
+ return ((void *) va);
+}
+
+/*
+ * Free a block of memory allocated by malloc.
+ */
+void
+free(addr, type)
+ void *addr;
+ int type;
+{
+ register struct kmembuckets *kbp;
+ register struct kmemusage *kup;
+ register struct freelist *freep;
+ long size;
+ int s;
+#ifdef DIAGNOSTIC
+ caddr_t cp;
+ int32_t *end, *lp;
+ long alloc, copysize;
+#endif
+#ifdef KMEMSTATS
+ register struct kmemstats *ksp = &kmemstats[type];
+#endif
+
+ kup = btokup(addr);
+ size = 1 << kup->ku_indx;
+ kbp = &bucket[kup->ku_indx];
+ s = splimp();
+#ifdef DIAGNOSTIC
+ /*
+ * Check for returns of data that do not point to the
+ * beginning of the allocation.
+ */
+ if (size > NBPG * CLSIZE)
+ alloc = addrmask[BUCKETINDX(NBPG * CLSIZE)];
+ else
+ alloc = addrmask[kup->ku_indx];
+ if (((u_long)addr & alloc) != 0)
+ panic("free: unaligned addr 0x%x, size %d, type %s, mask %d\n",
+ addr, size, memname[type], alloc);
+#endif /* DIAGNOSTIC */
+ if (size > MAXALLOCSAVE) {
+ kmem_free(kmem_map, (vm_offset_t)addr, ctob(kup->ku_pagecnt));
+#ifdef KMEMSTATS
+ size = kup->ku_pagecnt << PGSHIFT;
+ ksp->ks_memuse -= size;
+ kup->ku_indx = 0;
+ kup->ku_pagecnt = 0;
+ if (ksp->ks_memuse + size >= ksp->ks_limit &&
+ ksp->ks_memuse < ksp->ks_limit)
+ wakeup((caddr_t)ksp);
+ ksp->ks_inuse--;
+ kbp->kb_total -= 1;
+#endif
+ splx(s);
+ return;
+ }
+ freep = (struct freelist *)addr;
+#ifdef DIAGNOSTIC
+ /*
+ * Check for multiple frees. Use a quick check to see if
+ * it looks free before laboriously searching the freelist.
+ */
+ if (freep->spare0 == WEIRD_ADDR) {
+ for (cp = kbp->kb_next; cp; cp = *(caddr_t *)cp) {
+ if (addr != cp)
+ continue;
+ printf("multiply freed item %p\n", addr);
+ panic("free: duplicated free");
+ }
+ }
+ /*
+ * Copy in known text to detect modification after freeing
+ * and to make it look free. Also, save the type being freed
+ * so we can list likely culprit if modification is detected
+ * when the object is reallocated.
+ */
+ copysize = size < MAX_COPY ? size : MAX_COPY;
+ end = (int32_t *)&((caddr_t)addr)[copysize];
+ for (lp = (int32_t *)addr; lp < end; lp++)
+ *lp = WEIRD_ADDR;
+ freep->type = type;
+#endif /* DIAGNOSTIC */
+#ifdef KMEMSTATS
+ kup->ku_freecnt++;
+ if (kup->ku_freecnt >= kbp->kb_elmpercl)
+ if (kup->ku_freecnt > kbp->kb_elmpercl)
+ panic("free: multiple frees");
+ else if (kbp->kb_totalfree > kbp->kb_highwat)
+ kbp->kb_couldfree++;
+ kbp->kb_totalfree++;
+ ksp->ks_memuse -= size;
+ if (ksp->ks_memuse + size >= ksp->ks_limit &&
+ ksp->ks_memuse < ksp->ks_limit)
+ wakeup((caddr_t)ksp);
+ ksp->ks_inuse--;
+#endif
+ if (kbp->kb_next == NULL)
+ kbp->kb_next = addr;
+ else
+ ((struct freelist *)kbp->kb_last)->next = addr;
+ freep->next = NULL;
+ kbp->kb_last = addr;
+ splx(s);
+}
+
+/*
+ * Initialize the kernel memory allocator
+ */
+kmeminit()
+{
+ register long indx;
+ int npg;
+
+#if ((MAXALLOCSAVE & (MAXALLOCSAVE - 1)) != 0)
+ ERROR!_kmeminit:_MAXALLOCSAVE_not_power_of_2
+#endif
+#if (MAXALLOCSAVE > MINALLOCSIZE * 32768)
+ ERROR!_kmeminit:_MAXALLOCSAVE_too_big
+#endif
+#if (MAXALLOCSAVE < CLBYTES)
+ ERROR!_kmeminit:_MAXALLOCSAVE_too_small
+#endif
+
+ if (sizeof(struct freelist) > (1 << MINBUCKET))
+ panic("minbucket too small/struct freelist too big");
+
+ npg = VM_KMEM_SIZE/ NBPG;
+ kmemusage = (struct kmemusage *) kmem_alloc(kernel_map,
+ (vm_size_t)(npg * sizeof(struct kmemusage)));
+ kmem_map = kmem_suballoc(kernel_map, (vm_offset_t *)&kmembase,
+ (vm_offset_t *)&kmemlimit, (vm_size_t)(npg * NBPG), FALSE);
+#ifdef KMEMSTATS
+ for (indx = 0; indx < MINBUCKET + 16; indx++) {
+ if (1 << indx >= CLBYTES)
+ bucket[indx].kb_elmpercl = 1;
+ else
+ bucket[indx].kb_elmpercl = CLBYTES / (1 << indx);
+ bucket[indx].kb_highwat = 5 * bucket[indx].kb_elmpercl;
+ }
+ for (indx = 0; indx < M_LAST; indx++)
+ kmemstats[indx].ks_limit = npg * NBPG * 6 / 10;
+#endif
+}
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t8.2
@@ -1,0 +1,616 @@
+/* $OpenBSD: t8.2,v 1.1 2003/07/17 21:04:04 otto Exp $ */
+/* $NetBSD: kern_malloc.c,v 1.15.4.2 1996/06/13 17:10:56 cgd Exp $ */
+
+/*
+ * Copyright (c) 1987, 1991, 1993
+ * The Regents of the University of California. All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * 3. Neither the name of the University nor the names of its contributors
+ * may be used to endorse or promote products derived from this software
+ * without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * @(#)kern_malloc.c 8.3 (Berkeley) 1/4/94
+ */
+
+#include <sys/param.h>
+#include <sys/proc.h>
+#include <sys/kernel.h>
+#include <sys/malloc.h>
+#include <sys/systm.h>
+#include <sys/sysctl.h>
+
+#include <uvm/uvm_extern.h>
+
+static struct vm_map_intrsafe kmem_map_store;
+struct vm_map *kmem_map = NULL;
+
+#ifdef NKMEMCLUSTERS
+#error NKMEMCLUSTERS is obsolete; remove it from your kernel config file and use NKMEMPAGES instead or let the kernel auto-size
+#endif
+
+/*
+ * Default number of pages in kmem_map. We attempt to calculate this
+ * at run-time, but allow it to be either patched or set in the kernel
+ * config file.
+ */
+#ifndef NKMEMPAGES
+#define NKMEMPAGES 0
+#endif
+int nkmempages = NKMEMPAGES;
+
+/*
+ * Defaults for lower- and upper-bounds for the kmem_map page count.
+ * Can be overridden by kernel config options.
+ */
+#ifndef NKMEMPAGES_MIN
+#define NKMEMPAGES_MIN NKMEMPAGES_MIN_DEFAULT
+#endif
+
+#ifndef NKMEMPAGES_MAX
+#define NKMEMPAGES_MAX NKMEMPAGES_MAX_DEFAULT
+#endif
+
+struct kmembuckets bucket[MINBUCKET + 16];
+struct kmemstats kmemstats[M_LAST];
+struct kmemusage *kmemusage;
+char *kmembase, *kmemlimit;
+char buckstring[16 * sizeof("123456,")];
+int buckstring_init = 0;
+#if defined(KMEMSTATS) || defined(DIAGNOSTIC) || defined(FFS_SOFTUPDATES)
+char *memname[] = INITKMEMNAMES;
+char *memall = NULL;
+extern struct lock sysctl_kmemlock;
+#endif
+
+#ifdef DIAGNOSTIC
+/*
+ * This structure provides a set of masks to catch unaligned frees.
+ */
+const long addrmask[] = { 0,
+ 0x00000001, 0x00000003, 0x00000007, 0x0000000f,
+ 0x0000001f, 0x0000003f, 0x0000007f, 0x000000ff,
+ 0x000001ff, 0x000003ff, 0x000007ff, 0x00000fff,
+ 0x00001fff, 0x00003fff, 0x00007fff, 0x0000ffff,
+};
+
+/*
+ * The WEIRD_ADDR is used as known text to copy into free objects so
+ * that modifications after frees can be detected.
+ */
+#define WEIRD_ADDR ((unsigned) 0xdeadbeef)
+#define MAX_COPY 32
+
+/*
+ * Normally the freelist structure is used only to hold the list pointer
+ * for free objects. However, when running with diagnostics, the first
+ * 8 bytes of the structure is unused except for diagnostic information,
+ * and the free list pointer is at offset 8 in the structure. Since the
+ * first 8 bytes is the portion of the structure most often modified, this
+ * helps to detect memory reuse problems and avoid free list corruption.
+ */
+struct freelist {
+ int32_t spare0;
+ int16_t type;
+ int16_t spare1;
+ caddr_t next;
+};
+#else /* !DIAGNOSTIC */
+struct freelist {
+ caddr_t next;
+};
+#endif /* DIAGNOSTIC */
+
+/*
+ * Allocate a block of memory
+ */
+void *
+malloc(size, type, flags)
+ unsigned long size;
+ int type, flags;
+{
+ register struct kmembuckets *kbp;
+ register struct kmemusage *kup;
+ register struct freelist *freep;
+ long indx, npg, allocsize;
+ int s;
+ caddr_t va, cp, savedlist;
+#ifdef DIAGNOSTIC
+ int32_t *end, *lp;
+ int copysize;
+ char *savedtype;
+#endif
+#ifdef KMEMSTATS
+ register struct kmemstats *ksp = &kmemstats[type];
+
+ if (((unsigned long)type) >= M_LAST)
+ panic("malloc - bogus type");
+#endif
+
+#ifdef MALLOC_DEBUG
+ if (debug_malloc(size, type, flags, (void **)&va))
+ return ((void *) va);
+#endif
+
+ indx = BUCKETINDX(size);
+ kbp = &bucket[indx];
+ s = splvm();
+#ifdef KMEMSTATS
+ while (ksp->ks_memuse >= ksp->ks_limit) {
+ if (flags & M_NOWAIT) {
+ splx(s);
+ return ((void *) NULL);
+ }
+ if (ksp->ks_limblocks < 65535)
+ ksp->ks_limblocks++;
+ tsleep((caddr_t)ksp, PSWP+2, memname[type], 0);
+ }
+ ksp->ks_size |= 1 << indx;
+#endif
+#ifdef DIAGNOSTIC
+ copysize = 1 << indx < MAX_COPY ? 1 << indx : MAX_COPY;
+#endif
+ if (kbp->kb_next == NULL) {
+ kbp->kb_last = NULL;
+ if (size > MAXALLOCSAVE)
+ allocsize = round_page(size);
+ else
+ allocsize = 1 << indx;
+ npg = btoc(allocsize);
+ va = (caddr_t) uvm_km_kmemalloc(kmem_map, uvmexp.kmem_object,
+ (vsize_t)ctob(npg),
+ (flags & M_NOWAIT) ? UVM_KMF_NOWAIT : 0);
+ if (va == NULL) {
+ /*
+ * Kmem_malloc() can return NULL, even if it can
+ * wait, if there is no map space available, because
+ * it can't fix that problem. Neither can we,
+ * right now. (We should release pages which
+ * are completely free and which are in buckets
+ * with too many free elements.)
+ */
+ if ((flags & M_NOWAIT) == 0)
+ panic("malloc: out of space in kmem_map");
+ splx(s);
+ return ((void *) NULL);
+ }
+#ifdef KMEMSTATS
+ kbp->kb_total += kbp->kb_elmpercl;
+#endif
+ kup = btokup(va);
+ kup->ku_indx = indx;
+ if (allocsize > MAXALLOCSAVE) {
+ if (npg > 65535)
+ panic("malloc: allocation too large");
+ kup->ku_pagecnt = npg;
+#ifdef KMEMSTATS
+ ksp->ks_memuse += allocsize;
+#endif
+ goto out;
+ }
+#ifdef KMEMSTATS
+ kup->ku_freecnt = kbp->kb_elmpercl;
+ kbp->kb_totalfree += kbp->kb_elmpercl;
+#endif
+ /*
+ * Just in case we blocked while allocating memory,
+ * and someone else also allocated memory for this
+ * bucket, don't assume the list is still empty.
+ */
+ savedlist = kbp->kb_next;
+ kbp->kb_next = cp = va + (npg * PAGE_SIZE) - allocsize;
+ for (;;) {
+ freep = (struct freelist *)cp;
+#ifdef DIAGNOSTIC
+ /*
+ * Copy in known text to detect modification
+ * after freeing.
+ */
+ end = (int32_t *)&cp[copysize];
+ for (lp = (int32_t *)cp; lp < end; lp++)
+ *lp = WEIRD_ADDR;
+ freep->type = M_FREE;
+#endif /* DIAGNOSTIC */
+ if (cp <= va)
+ break;
+ cp -= allocsize;
+ freep->next = cp;
+ }
+ freep->next = savedlist;
+ if (kbp->kb_last == NULL)
+ kbp->kb_last = (caddr_t)freep;
+ }
+ va = kbp->kb_next;
+ kbp->kb_next = ((struct freelist *)va)->next;
+#ifdef DIAGNOSTIC
+ freep = (struct freelist *)va;
+ savedtype = (unsigned)freep->type < M_LAST ?
+ memname[freep->type] : "???";
+ if (kbp->kb_next) {
+ int rv;
+ vaddr_t addr = (vaddr_t)kbp->kb_next;
+
+ vm_map_lock(kmem_map);
+ rv = uvm_map_checkprot(kmem_map, addr,
+ addr + sizeof(struct freelist), VM_PROT_WRITE);
+ vm_map_unlock(kmem_map);
+
+ if (!rv) {
+ printf("%s %d of object %p size 0x%lx %s %s (invalid addr %p)\n",
+ "Data modified on freelist: word",
+ (int32_t *)&kbp->kb_next - (int32_t *)kbp, va, size,
+ "previous type", savedtype, kbp->kb_next);
+ kbp->kb_next = NULL;
+ }
+ }
+
+ /* Fill the fields that we've used with WEIRD_ADDR */
+#if BYTE_ORDER == BIG_ENDIAN
+ freep->type = WEIRD_ADDR >> 16;
+#endif
+#if BYTE_ORDER == LITTLE_ENDIAN
+ freep->type = (short)WEIRD_ADDR;
+#endif
+ end = (int32_t *)&freep->next +
+ (sizeof(freep->next) / sizeof(int32_t));
+ for (lp = (int32_t *)&freep->next; lp < end; lp++)
+ *lp = WEIRD_ADDR;
+
+ /* and check that the data hasn't been modified. */
+ end = (int32_t *)&va[copysize];
+ for (lp = (int32_t *)va; lp < end; lp++) {
+ if (*lp == WEIRD_ADDR)
+ continue;
+ printf("%s %d of object %p size 0x%lx %s %s (0x%x != 0x%x)\n",
+ "Data modified on freelist: word", lp - (int32_t *)va,
+ va, size, "previous type", savedtype, *lp, WEIRD_ADDR);
+ break;
+ }
+
+ freep->spare0 = 0;
+#endif /* DIAGNOSTIC */
+#ifdef KMEMSTATS
+ kup = btokup(va);
+ if (kup->ku_indx != indx)
+ panic("malloc: wrong bucket");
+ if (kup->ku_freecnt == 0)
+ panic("malloc: lost data");
+ kup->ku_freecnt--;
+ kbp->kb_totalfree--;
+ ksp->ks_memuse += 1 << indx;
+out:
+ kbp->kb_calls++;
+ ksp->ks_inuse++;
+ ksp->ks_calls++;
+ if (ksp->ks_memuse > ksp->ks_maxused)
+ ksp->ks_maxused = ksp->ks_memuse;
+#else
+out:
+#endif
+ splx(s);
+ return ((void *) va);
+}
+
+/*
+ * Free a block of memory allocated by malloc.
+ */
+void
+free(addr, type)
+ void *addr;
+ int type;
+{
+ register struct kmembuckets *kbp;
+ register struct kmemusage *kup;
+ register struct freelist *freep;
+ long size;
+ int s;
+#ifdef DIAGNOSTIC
+ caddr_t cp;
+ int32_t *end, *lp;
+ long alloc, copysize;
+#endif
+#ifdef KMEMSTATS
+ register struct kmemstats *ksp = &kmemstats[type];
+#endif
+
+#ifdef MALLOC_DEBUG
+ if (debug_free(addr, type))
+ return;
+#endif
+
+#ifdef DIAGNOSTIC
+ if (addr < (void *)kmembase || addr >= (void *)kmemlimit)
+ panic("free: non-malloced addr %p type %s", addr,
+ memname[type]);
+#endif
+
+ kup = btokup(addr);
+ size = 1 << kup->ku_indx;
+ kbp = &bucket[kup->ku_indx];
+ s = splvm();
+#ifdef DIAGNOSTIC
+ /*
+ * Check for returns of data that do not point to the
+ * beginning of the allocation.
+ */
+ if (size > PAGE_SIZE)
+ alloc = addrmask[BUCKETINDX(PAGE_SIZE)];
+ else
+ alloc = addrmask[kup->ku_indx];
+ if (((u_long)addr & alloc) != 0)
+ panic("free: unaligned addr %p, size %ld, type %s, mask %ld",
+ addr, size, memname[type], alloc);
+#endif /* DIAGNOSTIC */
+ if (size > MAXALLOCSAVE) {
+ uvm_km_free(kmem_map, (vaddr_t)addr, ctob(kup->ku_pagecnt));
+#ifdef KMEMSTATS
+ size = kup->ku_pagecnt << PGSHIFT;
+ ksp->ks_memuse -= size;
+ kup->ku_indx = 0;
+ kup->ku_pagecnt = 0;
+ if (ksp->ks_memuse + size >= ksp->ks_limit &&
+ ksp->ks_memuse < ksp->ks_limit)
+ wakeup((caddr_t)ksp);
+ ksp->ks_inuse--;
+ kbp->kb_total -= 1;
+#endif
+ splx(s);
+ return;
+ }
+ freep = (struct freelist *)addr;
+#ifdef DIAGNOSTIC
+ /*
+ * Check for multiple frees. Use a quick check to see if
+ * it looks free before laboriously searching the freelist.
+ */
+ if (freep->spare0 == WEIRD_ADDR) {
+ for (cp = kbp->kb_next; cp;
+ cp = ((struct freelist *)cp)->next) {
+ if (addr != cp)
+ continue;
+ printf("multiply freed item %p\n", addr);
+ panic("free: duplicated free");
+ }
+ }
+ /*
+ * Copy in known text to detect modification after freeing
+ * and to make it look free. Also, save the type being freed
+ * so we can list likely culprit if modification is detected
+ * when the object is reallocated.
+ */
+ copysize = size < MAX_COPY ? size : MAX_COPY;
+ end = (int32_t *)&((caddr_t)addr)[copysize];
+ for (lp = (int32_t *)addr; lp < end; lp++)
+ *lp = WEIRD_ADDR;
+ freep->type = type;
+#endif /* DIAGNOSTIC */
+#ifdef KMEMSTATS
+ kup->ku_freecnt++;
+ if (kup->ku_freecnt >= kbp->kb_elmpercl) {
+ if (kup->ku_freecnt > kbp->kb_elmpercl)
+ panic("free: multiple frees");
+ else if (kbp->kb_totalfree > kbp->kb_highwat)
+ kbp->kb_couldfree++;
+ }
+ kbp->kb_totalfree++;
+ ksp->ks_memuse -= size;
+ if (ksp->ks_memuse + size >= ksp->ks_limit &&
+ ksp->ks_memuse < ksp->ks_limit)
+ wakeup((caddr_t)ksp);
+ ksp->ks_inuse--;
+#endif
+ if (kbp->kb_next == NULL)
+ kbp->kb_next = addr;
+ else
+ ((struct freelist *)kbp->kb_last)->next = addr;
+ freep->next = NULL;
+ kbp->kb_last = addr;
+ splx(s);
+}
+
+/*
+ * Compute the number of pages that kmem_map will map, that is,
+ * the size of the kernel malloc arena.
+ */
+void
+kmeminit_nkmempages()
+{
+ int npages;
+
+ if (nkmempages != 0) {
+ /*
+ * It's already been set (by us being here before, or
+ * by patching or kernel config options), bail out now.
+ */
+ return;
+ }
+
+ /*
+ * We use the following (simple) formula:
+ *
+ * - Starting point is physical memory / 4.
+ *
+ * - Clamp it down to NKMEMPAGES_MAX.
+ *
+ * - Round it up to NKMEMPAGES_MIN.
+ */
+ npages = physmem / 4;
+
+ if (npages > NKMEMPAGES_MAX)
+ npages = NKMEMPAGES_MAX;
+
+ if (npages < NKMEMPAGES_MIN)
+ npages = NKMEMPAGES_MIN;
+
+ nkmempages = npages;
+}
+
+/*
+ * Initialize the kernel memory allocator
+ */
+void
+kmeminit()
+{
+ vaddr_t base, limit;
+#ifdef KMEMSTATS
+ long indx;
+#endif
+
+#ifdef DIAGNOSTIC
+ if (sizeof(struct freelist) > (1 << MINBUCKET))
+ panic("kmeminit: minbucket too small/struct freelist too big");
+#endif
+
+ /*
+ * Compute the number of kmem_map pages, if we have not
+ * done so already.
+ */
+ kmeminit_nkmempages();
+ base = vm_map_min(kernel_map);
+ kmem_map = uvm_km_suballoc(kernel_map, &base, &limit,
+ (vsize_t)(nkmempages * PAGE_SIZE), VM_MAP_INTRSAFE, FALSE,
+ &kmem_map_store.vmi_map);
+ kmembase = (char *)base;
+ kmemlimit = (char *)limit;
+ kmemusage = (struct kmemusage *) uvm_km_zalloc(kernel_map,
+ (vsize_t)(nkmempages * sizeof(struct kmemusage)));
+#ifdef KMEMSTATS
+ for (indx = 0; indx < MINBUCKET + 16; indx++) {
+ if (1 << indx >= PAGE_SIZE)
+ bucket[indx].kb_elmpercl = 1;
+ else
+ bucket[indx].kb_elmpercl = PAGE_SIZE / (1 << indx);
+ bucket[indx].kb_highwat = 5 * bucket[indx].kb_elmpercl;
+ }
+ for (indx = 0; indx < M_LAST; indx++)
+ kmemstats[indx].ks_limit = nkmempages * PAGE_SIZE * 6 / 10;
+#endif
+#ifdef MALLOC_DEBUG
+ debug_malloc_init();
+#endif
+}
+
+/*
+ * Return kernel malloc statistics information.
+ */
+int
+sysctl_malloc(name, namelen, oldp, oldlenp, newp, newlen, p)
+ int *name;
+ u_int namelen;
+ void *oldp;
+ size_t *oldlenp;
+ void *newp;
+ size_t newlen;
+ struct proc *p;
+{
+ struct kmembuckets kb;
+ int i, siz;
+
+ if (namelen != 2 && name[0] != KERN_MALLOC_BUCKETS &&
+ name[0] != KERN_MALLOC_KMEMNAMES)
+ return (ENOTDIR); /* overloaded */
+
+ switch (name[0]) {
+ case KERN_MALLOC_BUCKETS:
+ /* Initialize the first time */
+ if (buckstring_init == 0) {
+ buckstring_init = 1;
+ bzero(buckstring, sizeof(buckstring));
+ for (siz = 0, i = MINBUCKET; i < MINBUCKET + 16; i++) {
+ snprintf(buckstring + siz,
+ sizeof buckstring - siz,
+ "%d,", (u_int)(1<<i));
+ siz += strlen(buckstring + siz);
+ }
+ /* Remove trailing comma */
+ if (siz)
+ buckstring[siz - 1] = '\0';
+ }
+ return (sysctl_rdstring(oldp, oldlenp, newp, buckstring));
+
+ case KERN_MALLOC_BUCKET:
+ bcopy(&bucket[BUCKETINDX(name[1])], &kb, sizeof(kb));
+ kb.kb_next = kb.kb_last = 0;
+ return (sysctl_rdstruct(oldp, oldlenp, newp, &kb, sizeof(kb)));
+ case KERN_MALLOC_KMEMSTATS:
+#ifdef KMEMSTATS
+ if ((name[1] < 0) || (name[1] >= M_LAST))
+ return (EINVAL);
+ return (sysctl_rdstruct(oldp, oldlenp, newp,
+ &kmemstats[name[1]], sizeof(struct kmemstats)));
+#else
+ return (EOPNOTSUPP);
+#endif
+ case KERN_MALLOC_KMEMNAMES:
+#if defined(KMEMSTATS) || defined(DIAGNOSTIC) || defined(FFS_SOFTUPDATES)
+ if (memall == NULL) {
+ int totlen;
+
+ i = lockmgr(&sysctl_kmemlock, LK_EXCLUSIVE, NULL, p);
+ if (i)
+ return (i);
+
+ /* Figure out how large a buffer we need */
+ for (totlen = 0, i = 0; i < M_LAST; i++) {
+ if (memname[i])
+ totlen += strlen(memname[i]);
+ totlen++;
+ }
+ memall = malloc(totlen + M_LAST, M_SYSCTL, M_WAITOK);
+ bzero(memall, totlen + M_LAST);
+ for (siz = 0, i = 0; i < M_LAST; i++) {
+ snprintf(memall + siz,
+ totlen + M_LAST - siz,
+ "%s,", memname[i] ? memname[i] : "");
+ siz += strlen(memall + siz);
+ }
+ /* Remove trailing comma */
+ if (siz)
+ memall[siz - 1] = '\0';
+
+ /* Now, convert all spaces to underscores */
+ for (i = 0; i < totlen; i++)
+ if (memall[i] == ' ')
+ memall[i] = '_';
+ lockmgr(&sysctl_kmemlock, LK_RELEASE, NULL, p);
+ }
+ return (sysctl_rdstring(oldp, oldlenp, newp, memall));
+#else
+ return (EOPNOTSUPP);
+#endif
+ default:
+ return (EOPNOTSUPP);
+ }
+ /* NOTREACHED */
+}
+
+/*
+ * Round up a size to how much malloc would actually allocate.
+ */
+size_t
+malloc_roundup(size_t sz)
+{
+ if (sz > MAXALLOCSAVE)
+ return round_page(sz);
+
+ return (1 << BUCKETINDX(sz));
+}
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t8.expected
@@ -1,0 +1,460 @@
+--- diff-t8.1
++++ diff-t8.2
+@@ -1,4 +1,5 @@
+-/* $NetBSD: kern_malloc.c,v 1.11 1995/05/01 22:39:11 cgd Exp $ */
++/* $OpenBSD: t8.2,v 1.1 2003/07/17 21:04:04 otto Exp $ */
++/* $NetBSD: kern_malloc.c,v 1.15.4.2 1996/06/13 17:10:56 cgd Exp $ */
+
+ /*
+ * Copyright (c) 1987, 1991, 1993
+@@ -33,24 +34,59 @@
+
+ #include <sys/param.h>
+ #include <sys/proc.h>
+-#include <sys/map.h>
+ #include <sys/kernel.h>
+ #include <sys/malloc.h>
++#include <sys/systm.h>
++#include <sys/sysctl.h>
+
+-#include <vm/vm.h>
+-#include <vm/vm_kern.h>
++#include <uvm/uvm_extern.h>
++
++static struct vm_map_intrsafe kmem_map_store;
++struct vm_map *kmem_map = NULL;
++
++#ifdef NKMEMCLUSTERS
++#error NKMEMCLUSTERS is obsolete; remove it from your kernel config file and use NKMEMPAGES instead or let the kernel auto-size
++#endif
++
++/*
++ * Default number of pages in kmem_map. We attempt to calculate this
++ * at run-time, but allow it to be either patched or set in the kernel
++ * config file.
++ */
++#ifndef NKMEMPAGES
++#define NKMEMPAGES 0
++#endif
++int nkmempages = NKMEMPAGES;
++
++/*
++ * Defaults for lower- and upper-bounds for the kmem_map page count.
++ * Can be overridden by kernel config options.
++ */
++#ifndef NKMEMPAGES_MIN
++#define NKMEMPAGES_MIN NKMEMPAGES_MIN_DEFAULT
++#endif
++
++#ifndef NKMEMPAGES_MAX
++#define NKMEMPAGES_MAX NKMEMPAGES_MAX_DEFAULT
++#endif
+
+ struct kmembuckets bucket[MINBUCKET + 16];
+ struct kmemstats kmemstats[M_LAST];
+ struct kmemusage *kmemusage;
+ char *kmembase, *kmemlimit;
++char buckstring[16 * sizeof("123456,")];
++int buckstring_init = 0;
++#if defined(KMEMSTATS) || defined(DIAGNOSTIC) || defined(FFS_SOFTUPDATES)
+ char *memname[] = INITKMEMNAMES;
++char *memall = NULL;
++extern struct lock sysctl_kmemlock;
++#endif
+
+ #ifdef DIAGNOSTIC
+ /*
+ * This structure provides a set of masks to catch unaligned frees.
+ */
+-long addrmask[] = { 0,
++const long addrmask[] = { 0,
+ 0x00000001, 0x00000003, 0x00000007, 0x0000000f,
+ 0x0000001f, 0x0000003f, 0x0000007f, 0x000000ff,
+ 0x000001ff, 0x000003ff, 0x000007ff, 0x00000fff,
+@@ -61,14 +97,14 @@
+ * The WEIRD_ADDR is used as known text to copy into free objects so
+ * that modifications after frees can be detected.
+ */
+-#define WEIRD_ADDR 0xdeadbeef
++#define WEIRD_ADDR ((unsigned) 0xdeadbeef)
+ #define MAX_COPY 32
+
+ /*
+ * Normally the freelist structure is used only to hold the list pointer
+ * for free objects. However, when running with diagnostics, the first
+ * 8 bytes of the structure is unused except for diagnostic information,
+- * and the free list pointer is at offst 8 in the structure. Since the
++ * and the free list pointer is at offset 8 in the structure. Since the
+ * first 8 bytes is the portion of the structure most often modified, this
+ * helps to detect memory reuse problems and avoid free list corruption.
+ */
+@@ -106,12 +142,18 @@
+ #ifdef KMEMSTATS
+ register struct kmemstats *ksp = &kmemstats[type];
+
+- if (((unsigned long)type) > M_LAST)
++ if (((unsigned long)type) >= M_LAST)
+ panic("malloc - bogus type");
+ #endif
++
++#ifdef MALLOC_DEBUG
++ if (debug_malloc(size, type, flags, (void **)&va))
++ return ((void *) va);
++#endif
++
+ indx = BUCKETINDX(size);
+ kbp = &bucket[indx];
+- s = splimp();
++ s = splvm();
+ #ifdef KMEMSTATS
+ while (ksp->ks_memuse >= ksp->ks_limit) {
+ if (flags & M_NOWAIT) {
+@@ -130,13 +172,24 @@
+ if (kbp->kb_next == NULL) {
+ kbp->kb_last = NULL;
+ if (size > MAXALLOCSAVE)
+- allocsize = roundup(size, CLBYTES);
++ allocsize = round_page(size);
+ else
+ allocsize = 1 << indx;
+- npg = clrnd(btoc(allocsize));
+- va = (caddr_t) kmem_malloc(kmem_map, (vm_size_t)ctob(npg),
+- !(flags & M_NOWAIT));
++ npg = btoc(allocsize);
++ va = (caddr_t) uvm_km_kmemalloc(kmem_map, uvmexp.kmem_object,
++ (vsize_t)ctob(npg),
++ (flags & M_NOWAIT) ? UVM_KMF_NOWAIT : 0);
+ if (va == NULL) {
++ /*
++ * Kmem_malloc() can return NULL, even if it can
++ * wait, if there is no map space available, because
++ * it can't fix that problem. Neither can we,
++ * right now. (We should release pages which
++ * are completely free and which are in buckets
++ * with too many free elements.)
++ */
++ if ((flags & M_NOWAIT) == 0)
++ panic("malloc: out of space in kmem_map");
+ splx(s);
+ return ((void *) NULL);
+ }
+@@ -164,7 +217,7 @@
+ * bucket, don't assume the list is still empty.
+ */
+ savedlist = kbp->kb_next;
+- kbp->kb_next = cp = va + (npg * NBPG) - allocsize;
++ kbp->kb_next = cp = va + (npg * PAGE_SIZE) - allocsize;
+ for (;;) {
+ freep = (struct freelist *)cp;
+ #ifdef DIAGNOSTIC
+@@ -192,13 +245,22 @@
+ freep = (struct freelist *)va;
+ savedtype = (unsigned)freep->type < M_LAST ?
+ memname[freep->type] : "???";
+- if (kbp->kb_next &&
+- !kernacc(kbp->kb_next, sizeof(struct freelist), 0)) {
+- printf("%s %d of object %p size %d %s %s (invalid addr %p)\n",
++ if (kbp->kb_next) {
++ int rv;
++ vaddr_t addr = (vaddr_t)kbp->kb_next;
++
++ vm_map_lock(kmem_map);
++ rv = uvm_map_checkprot(kmem_map, addr,
++ addr + sizeof(struct freelist), VM_PROT_WRITE);
++ vm_map_unlock(kmem_map);
++
++ if (!rv) {
++ printf("%s %d of object %p size 0x%lx %s %s (invalid addr %p)\n",
+ "Data modified on freelist: word",
+ (int32_t *)&kbp->kb_next - (int32_t *)kbp, va, size,
+ "previous type", savedtype, kbp->kb_next);
+ kbp->kb_next = NULL;
++ }
+ }
+
+ /* Fill the fields that we've used with WEIRD_ADDR */
+@@ -218,7 +280,7 @@
+ for (lp = (int32_t *)va; lp < end; lp++) {
+ if (*lp == WEIRD_ADDR)
+ continue;
+- printf("%s %d of object %p size %d %s %s (%p != %p)\n",
++ printf("%s %d of object %p size 0x%lx %s %s (0x%x != 0x%x)\n",
+ "Data modified on freelist: word", lp - (int32_t *)va,
+ va, size, "previous type", savedtype, *lp, WEIRD_ADDR);
+ break;
+@@ -270,25 +332,36 @@
+ register struct kmemstats *ksp = &kmemstats[type];
+ #endif
+
++#ifdef MALLOC_DEBUG
++ if (debug_free(addr, type))
++ return;
++#endif
++
++#ifdef DIAGNOSTIC
++ if (addr < (void *)kmembase || addr >= (void *)kmemlimit)
++ panic("free: non-malloced addr %p type %s", addr,
++ memname[type]);
++#endif
++
+ kup = btokup(addr);
+ size = 1 << kup->ku_indx;
+ kbp = &bucket[kup->ku_indx];
+- s = splimp();
++ s = splvm();
+ #ifdef DIAGNOSTIC
+ /*
+ * Check for returns of data that do not point to the
+ * beginning of the allocation.
+ */
+- if (size > NBPG * CLSIZE)
+- alloc = addrmask[BUCKETINDX(NBPG * CLSIZE)];
++ if (size > PAGE_SIZE)
++ alloc = addrmask[BUCKETINDX(PAGE_SIZE)];
+ else
+ alloc = addrmask[kup->ku_indx];
+ if (((u_long)addr & alloc) != 0)
+- panic("free: unaligned addr 0x%x, size %d, type %s, mask %d\n",
++ panic("free: unaligned addr %p, size %ld, type %s, mask %ld",
+ addr, size, memname[type], alloc);
+ #endif /* DIAGNOSTIC */
+ if (size > MAXALLOCSAVE) {
+- kmem_free(kmem_map, (vm_offset_t)addr, ctob(kup->ku_pagecnt));
++ uvm_km_free(kmem_map, (vaddr_t)addr, ctob(kup->ku_pagecnt));
+ #ifdef KMEMSTATS
+ size = kup->ku_pagecnt << PGSHIFT;
+ ksp->ks_memuse -= size;
+@@ -310,7 +383,8 @@
+ * it looks free before laboriously searching the freelist.
+ */
+ if (freep->spare0 == WEIRD_ADDR) {
+- for (cp = kbp->kb_next; cp; cp = *(caddr_t *)cp) {
++ for (cp = kbp->kb_next; cp;
++ cp = ((struct freelist *)cp)->next) {
+ if (addr != cp)
+ continue;
+ printf("multiply freed item %p\n", addr);
+@@ -331,11 +405,12 @@
+ #endif /* DIAGNOSTIC */
+ #ifdef KMEMSTATS
+ kup->ku_freecnt++;
+- if (kup->ku_freecnt >= kbp->kb_elmpercl)
++ if (kup->ku_freecnt >= kbp->kb_elmpercl) {
+ if (kup->ku_freecnt > kbp->kb_elmpercl)
+ panic("free: multiple frees");
+ else if (kbp->kb_totalfree > kbp->kb_highwat)
+ kbp->kb_couldfree++;
++ }
+ kbp->kb_totalfree++;
+ ksp->ks_memuse -= size;
+ if (ksp->ks_memuse + size >= ksp->ks_limit &&
+@@ -353,40 +428,189 @@
+ }
+
+ /*
++ * Compute the number of pages that kmem_map will map, that is,
++ * the size of the kernel malloc arena.
++ */
++void
++kmeminit_nkmempages()
++{
++ int npages;
++
++ if (nkmempages != 0) {
++ /*
++ * It's already been set (by us being here before, or
++ * by patching or kernel config options), bail out now.
++ */
++ return;
++ }
++
++ /*
++ * We use the following (simple) formula:
++ *
++ * - Starting point is physical memory / 4.
++ *
++ * - Clamp it down to NKMEMPAGES_MAX.
++ *
++ * - Round it up to NKMEMPAGES_MIN.
++ */
++ npages = physmem / 4;
++
++ if (npages > NKMEMPAGES_MAX)
++ npages = NKMEMPAGES_MAX;
++
++ if (npages < NKMEMPAGES_MIN)
++ npages = NKMEMPAGES_MIN;
++
++ nkmempages = npages;
++}
++
++/*
+ * Initialize the kernel memory allocator
+ */
++void
+ kmeminit()
+ {
+- register long indx;
+- int npg;
+-
+-#if ((MAXALLOCSAVE & (MAXALLOCSAVE - 1)) != 0)
+- ERROR!_kmeminit:_MAXALLOCSAVE_not_power_of_2
+-#endif
+-#if (MAXALLOCSAVE > MINALLOCSIZE * 32768)
+- ERROR!_kmeminit:_MAXALLOCSAVE_too_big
+-#endif
+-#if (MAXALLOCSAVE < CLBYTES)
+- ERROR!_kmeminit:_MAXALLOCSAVE_too_small
++ vaddr_t base, limit;
++#ifdef KMEMSTATS
++ long indx;
+ #endif
+
++#ifdef DIAGNOSTIC
+ if (sizeof(struct freelist) > (1 << MINBUCKET))
+- panic("minbucket too small/struct freelist too big");
++ panic("kmeminit: minbucket too small/struct freelist too big");
++#endif
+
+- npg = VM_KMEM_SIZE/ NBPG;
+- kmemusage = (struct kmemusage *) kmem_alloc(kernel_map,
+- (vm_size_t)(npg * sizeof(struct kmemusage)));
+- kmem_map = kmem_suballoc(kernel_map, (vm_offset_t *)&kmembase,
+- (vm_offset_t *)&kmemlimit, (vm_size_t)(npg * NBPG), FALSE);
++ /*
++ * Compute the number of kmem_map pages, if we have not
++ * done so already.
++ */
++ kmeminit_nkmempages();
++ base = vm_map_min(kernel_map);
++ kmem_map = uvm_km_suballoc(kernel_map, &base, &limit,
++ (vsize_t)(nkmempages * PAGE_SIZE), VM_MAP_INTRSAFE, FALSE,
++ &kmem_map_store.vmi_map);
++ kmembase = (char *)base;
++ kmemlimit = (char *)limit;
++ kmemusage = (struct kmemusage *) uvm_km_zalloc(kernel_map,
++ (vsize_t)(nkmempages * sizeof(struct kmemusage)));
+ #ifdef KMEMSTATS
+ for (indx = 0; indx < MINBUCKET + 16; indx++) {
+- if (1 << indx >= CLBYTES)
++ if (1 << indx >= PAGE_SIZE)
+ bucket[indx].kb_elmpercl = 1;
+ else
+- bucket[indx].kb_elmpercl = CLBYTES / (1 << indx);
++ bucket[indx].kb_elmpercl = PAGE_SIZE / (1 << indx);
+ bucket[indx].kb_highwat = 5 * bucket[indx].kb_elmpercl;
+ }
+ for (indx = 0; indx < M_LAST; indx++)
+- kmemstats[indx].ks_limit = npg * NBPG * 6 / 10;
++ kmemstats[indx].ks_limit = nkmempages * PAGE_SIZE * 6 / 10;
++#endif
++#ifdef MALLOC_DEBUG
++ debug_malloc_init();
++#endif
++}
++
++/*
++ * Return kernel malloc statistics information.
++ */
++int
++sysctl_malloc(name, namelen, oldp, oldlenp, newp, newlen, p)
++ int *name;
++ u_int namelen;
++ void *oldp;
++ size_t *oldlenp;
++ void *newp;
++ size_t newlen;
++ struct proc *p;
++{
++ struct kmembuckets kb;
++ int i, siz;
++
++ if (namelen != 2 && name[0] != KERN_MALLOC_BUCKETS &&
++ name[0] != KERN_MALLOC_KMEMNAMES)
++ return (ENOTDIR); /* overloaded */
++
++ switch (name[0]) {
++ case KERN_MALLOC_BUCKETS:
++ /* Initialize the first time */
++ if (buckstring_init == 0) {
++ buckstring_init = 1;
++ bzero(buckstring, sizeof(buckstring));
++ for (siz = 0, i = MINBUCKET; i < MINBUCKET + 16; i++) {
++ snprintf(buckstring + siz,
++ sizeof buckstring - siz,
++ "%d,", (u_int)(1<<i));
++ siz += strlen(buckstring + siz);
++ }
++ /* Remove trailing comma */
++ if (siz)
++ buckstring[siz - 1] = '\0';
++ }
++ return (sysctl_rdstring(oldp, oldlenp, newp, buckstring));
++
++ case KERN_MALLOC_BUCKET:
++ bcopy(&bucket[BUCKETINDX(name[1])], &kb, sizeof(kb));
++ kb.kb_next = kb.kb_last = 0;
++ return (sysctl_rdstruct(oldp, oldlenp, newp, &kb, sizeof(kb)));
++ case KERN_MALLOC_KMEMSTATS:
++#ifdef KMEMSTATS
++ if ((name[1] < 0) || (name[1] >= M_LAST))
++ return (EINVAL);
++ return (sysctl_rdstruct(oldp, oldlenp, newp,
++ &kmemstats[name[1]], sizeof(struct kmemstats)));
++#else
++ return (EOPNOTSUPP);
+ #endif
++ case KERN_MALLOC_KMEMNAMES:
++#if defined(KMEMSTATS) || defined(DIAGNOSTIC) || defined(FFS_SOFTUPDATES)
++ if (memall == NULL) {
++ int totlen;
++
++ i = lockmgr(&sysctl_kmemlock, LK_EXCLUSIVE, NULL, p);
++ if (i)
++ return (i);
++
++ /* Figure out how large a buffer we need */
++ for (totlen = 0, i = 0; i < M_LAST; i++) {
++ if (memname[i])
++ totlen += strlen(memname[i]);
++ totlen++;
++ }
++ memall = malloc(totlen + M_LAST, M_SYSCTL, M_WAITOK);
++ bzero(memall, totlen + M_LAST);
++ for (siz = 0, i = 0; i < M_LAST; i++) {
++ snprintf(memall + siz,
++ totlen + M_LAST - siz,
++ "%s,", memname[i] ? memname[i] : "");
++ siz += strlen(memall + siz);
++ }
++ /* Remove trailing comma */
++ if (siz)
++ memall[siz - 1] = '\0';
++
++ /* Now, convert all spaces to underscores */
++ for (i = 0; i < totlen; i++)
++ if (memall[i] == ' ')
++ memall[i] = '_';
++ lockmgr(&sysctl_kmemlock, LK_RELEASE, NULL, p);
++ }
++ return (sysctl_rdstring(oldp, oldlenp, newp, memall));
++#else
++ return (EOPNOTSUPP);
++#endif
++ default:
++ return (EOPNOTSUPP);
++ }
++ /* NOTREACHED */
++}
++
++/*
++ * Round up a size to how much malloc would actually allocate.
++ */
++size_t
++malloc_roundup(size_t sz)
++{
++ if (sz > MAXALLOCSAVE)
++ return round_page(sz);
++
++ return (1 << BUCKETINDX(sz));
+ }
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t9.1
@@ -1,0 +1,2045 @@
+/* $NetBSD: vfs_syscalls.c,v 1.57 1995/10/07 06:28:51 mycroft Exp $ */
+
+/*
+ * Copyright (c) 1989, 1993
+ * The Regents of the University of California. All rights reserved.
+ * (c) UNIX System Laboratories, Inc.
+ * All or some portions of this file are derived from material licensed
+ * to the University of California by American Telephone and Telegraph
+ * Co. or Unix System Laboratories, Inc. and are reproduced herein with
+ * the permission of UNIX System Laboratories, Inc.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * 3. All advertising materials mentioning features or use of this software
+ * must display the following acknowledgement:
+ * This product includes software developed by the University of
+ * California, Berkeley and its contributors.
+ * 4. Neither the name of the University nor the names of its contributors
+ * may be used to endorse or promote products derived from this software
+ * without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * @(#)vfs_syscalls.c 8.28 (Berkeley) 12/10/94
+ */
+
+#include <sys/param.h>
+#include <sys/systm.h>
+#include <sys/namei.h>
+#include <sys/filedesc.h>
+#include <sys/kernel.h>
+#include <sys/file.h>
+#include <sys/stat.h>
+#include <sys/vnode.h>
+#include <sys/mount.h>
+#include <sys/proc.h>
+#include <sys/uio.h>
+#include <sys/malloc.h>
+#include <sys/dirent.h>
+
+#include <sys/syscallargs.h>
+
+#include <vm/vm.h>
+#include <sys/sysctl.h>
+
+static int change_dir __P((struct nameidata *ndp, struct proc *p));
+
+/*
+ * Virtual File System System Calls
+ */
+
+/*
+ * Mount a file system.
+ */
+/* ARGSUSED */
+sys_mount(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_mount_args /* {
+ syscallarg(char *) type;
+ syscallarg(char *) path;
+ syscallarg(int) flags;
+ syscallarg(caddr_t) data;
+ } */ *uap = v;
+ register struct vnode *vp;
+ register struct mount *mp;
+ int error, flag;
+ u_long fsindex;
+ char fstypename[MFSNAMELEN];
+ struct vattr va;
+ struct nameidata nd;
+
+ /*
+ * Get vnode to be covered
+ */
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ vp = nd.ni_vp;
+ if (SCARG(uap, flags) & MNT_UPDATE) {
+ if ((vp->v_flag & VROOT) == 0) {
+ vput(vp);
+ return (EINVAL);
+ }
+ mp = vp->v_mount;
+ flag = mp->mnt_flag;
+ /*
+ * We only allow the filesystem to be reloaded if it
+ * is currently mounted read-only.
+ */
+ if ((SCARG(uap, flags) & MNT_RELOAD) &&
+ ((mp->mnt_flag & MNT_RDONLY) == 0)) {
+ vput(vp);
+ return (EOPNOTSUPP); /* Needs translation */
+ }
+ mp->mnt_flag |=
+ SCARG(uap, flags) & (MNT_RELOAD | MNT_FORCE | MNT_UPDATE);
+ /*
+ * Only root, or the user that did the original mount is
+ * permitted to update it.
+ */
+ if (mp->mnt_stat.f_owner != p->p_ucred->cr_uid &&
+ (error = suser(p->p_ucred, &p->p_acflag))) {
+ vput(vp);
+ return (error);
+ }
+ /*
+ * Do not allow NFS export by non-root users. Silently
+ * enforce MNT_NOSUID and MNT_NODEV for non-root users.
+ */
+ if (p->p_ucred->cr_uid != 0) {
+ if (SCARG(uap, flags) & MNT_EXPORTED) {
+ vput(vp);
+ return (EPERM);
+ }
+ SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
+ }
+ VOP_UNLOCK(vp);
+ goto update;
+ }
+ /*
+ * If the user is not root, ensure that they own the directory
+ * onto which we are attempting to mount.
+ */
+ if ((error = VOP_GETATTR(vp, &va, p->p_ucred, p)) ||
+ (va.va_uid != p->p_ucred->cr_uid &&
+ (error = suser(p->p_ucred, &p->p_acflag)))) {
+ vput(vp);
+ return (error);
+ }
+ /*
+ * Do not allow NFS export by non-root users. Silently
+ * enforce MNT_NOSUID and MNT_NODEV for non-root users.
+ */
+ if (p->p_ucred->cr_uid != 0) {
+ if (SCARG(uap, flags) & MNT_EXPORTED) {
+ vput(vp);
+ return (EPERM);
+ }
+ SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
+ }
+ if (error = vinvalbuf(vp, V_SAVE, p->p_ucred, p, 0, 0))
+ return (error);
+ if (vp->v_type != VDIR) {
+ vput(vp);
+ return (ENOTDIR);
+ }
+ if (error = copyinstr(SCARG(uap, type), fstypename, MFSNAMELEN,
+ (size_t *)0)) {
+#if defined(COMPAT_09) || defined(COMPAT_43)
+ /*
+ * Historically filesystem types were identified by number.
+ * If we get an integer for the filesystem type instead of a
+ * string, we check to see if it matches one of the historic
+ * filesystem types.
+ */
+ fsindex = (u_long)SCARG(uap, type);
+ if (fsindex >= nvfssw || vfssw[fsindex] == NULL) {
+ vput(vp);
+ return (ENODEV);
+ }
+ strncpy(fstypename, vfssw[fsindex]->vfs_name, MFSNAMELEN);
+#else
+ vput(vp);
+ return (error);
+#endif
+ }
+ for (fsindex = 0; fsindex < nvfssw; fsindex++)
+ if (vfssw[fsindex] != NULL &&
+ !strncmp(vfssw[fsindex]->vfs_name, fstypename, MFSNAMELEN))
+ break;
+ if (fsindex >= nvfssw) {
+ vput(vp);
+ return (ENODEV);
+ }
+ if (vp->v_mountedhere != NULL) {
+ vput(vp);
+ return (EBUSY);
+ }
+
+ /*
+ * Allocate and initialize the file system.
+ */
+ mp = (struct mount *)malloc((u_long)sizeof(struct mount),
+ M_MOUNT, M_WAITOK);
+ bzero((char *)mp, (u_long)sizeof(struct mount));
+ mp->mnt_op = vfssw[fsindex];
+ if (error = vfs_lock(mp)) {
+ free((caddr_t)mp, M_MOUNT);
+ vput(vp);
+ return (error);
+ }
+ /* Do this early in case we block later. */
+ vfssw[fsindex]->vfs_refcount++;
+ vp->v_mountedhere = mp;
+ mp->mnt_vnodecovered = vp;
+ mp->mnt_stat.f_owner = p->p_ucred->cr_uid;
+update:
+ /*
+ * Set the mount level flags.
+ */
+ if (SCARG(uap, flags) & MNT_RDONLY)
+ mp->mnt_flag |= MNT_RDONLY;
+ else if (mp->mnt_flag & MNT_RDONLY)
+ mp->mnt_flag |= MNT_WANTRDWR;
+ mp->mnt_flag &=~ (MNT_NOSUID | MNT_NOEXEC | MNT_NODEV |
+ MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC);
+ mp->mnt_flag |= SCARG(uap, flags) & (MNT_NOSUID | MNT_NOEXEC |
+ MNT_NODEV | MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC);
+ /*
+ * Mount the filesystem.
+ */
+ error = VFS_MOUNT(mp, SCARG(uap, path), SCARG(uap, data), &nd, p);
+ if (mp->mnt_flag & MNT_UPDATE) {
+ vrele(vp);
+ if (mp->mnt_flag & MNT_WANTRDWR)
+ mp->mnt_flag &= ~MNT_RDONLY;
+ mp->mnt_flag &=~
+ (MNT_UPDATE | MNT_RELOAD | MNT_FORCE | MNT_WANTRDWR);
+ if (error)
+ mp->mnt_flag = flag;
+ return (error);
+ }
+ /*
+ * Put the new filesystem on the mount list after root.
+ */
+ cache_purge(vp);
+ if (!error) {
+ TAILQ_INSERT_TAIL(&mountlist, mp, mnt_list);
+ checkdirs(vp);
+ VOP_UNLOCK(vp);
+ vfs_unlock(mp);
+ (void) VFS_STATFS(mp, &mp->mnt_stat, p);
+ error = VFS_START(mp, 0, p);
+ } else {
+ mp->mnt_vnodecovered->v_mountedhere = (struct mount *)0;
+ vfssw[fsindex]->vfs_refcount--;
+ vfs_unlock(mp);
+ free((caddr_t)mp, M_MOUNT);
+ vput(vp);
+ }
+ return (error);
+}
+
+/*
+ * Scan all active processes to see if any of them have a current
+ * or root directory onto which the new filesystem has just been
+ * mounted. If so, replace them with the new mount point.
+ */
+checkdirs(olddp)
+ struct vnode *olddp;
+{
+ struct filedesc *fdp;
+ struct vnode *newdp;
+ struct proc *p;
+
+ if (olddp->v_usecount == 1)
+ return;
+ if (VFS_ROOT(olddp->v_mountedhere, &newdp))
+ panic("mount: lost mount");
+ for (p = allproc.lh_first; p != 0; p = p->p_list.le_next) {
+ fdp = p->p_fd;
+ if (fdp->fd_cdir == olddp) {
+ vrele(fdp->fd_cdir);
+ VREF(newdp);
+ fdp->fd_cdir = newdp;
+ }
+ if (fdp->fd_rdir == olddp) {
+ vrele(fdp->fd_rdir);
+ VREF(newdp);
+ fdp->fd_rdir = newdp;
+ }
+ }
+ if (rootvnode == olddp) {
+ vrele(rootvnode);
+ VREF(newdp);
+ rootvnode = newdp;
+ }
+ vput(newdp);
+}
+
+/*
+ * Unmount a file system.
+ *
+ * Note: unmount takes a path to the vnode mounted on as argument,
+ * not special file (as before).
+ */
+/* ARGSUSED */
+sys_unmount(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_unmount_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) flags;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct mount *mp;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ vp = nd.ni_vp;
+ mp = vp->v_mount;
+
+ /*
+ * Only root, or the user that did the original mount is
+ * permitted to unmount this filesystem.
+ */
+ if ((mp->mnt_stat.f_owner != p->p_ucred->cr_uid) &&
+ (error = suser(p->p_ucred, &p->p_acflag))) {
+ vput(vp);
+ return (error);
+ }
+
+ /*
+ * Don't allow unmounting the root file system.
+ */
+ if (mp->mnt_flag & MNT_ROOTFS) {
+ vput(vp);
+ return (EINVAL);
+ }
+
+ /*
+ * Must be the root of the filesystem
+ */
+ if ((vp->v_flag & VROOT) == 0) {
+ vput(vp);
+ return (EINVAL);
+ }
+ vput(vp);
+ return (dounmount(mp, SCARG(uap, flags), p));
+}
+
+/*
+ * Do the actual file system unmount.
+ */
+dounmount(mp, flags, p)
+ register struct mount *mp;
+ int flags;
+ struct proc *p;
+{
+ struct vnode *coveredvp;
+ int error;
+
+ coveredvp = mp->mnt_vnodecovered;
+ if (vfs_busy(mp))
+ return (EBUSY);
+ mp->mnt_flag |= MNT_UNMOUNT;
+ if (error = vfs_lock(mp))
+ return (error);
+
+ mp->mnt_flag &=~ MNT_ASYNC;
+ vnode_pager_umount(mp); /* release cached vnodes */
+ cache_purgevfs(mp); /* remove cache entries for this file sys */
+ if ((error = VFS_SYNC(mp, MNT_WAIT, p->p_ucred, p)) == 0 ||
+ (flags & MNT_FORCE))
+ error = VFS_UNMOUNT(mp, flags, p);
+ mp->mnt_flag &= ~MNT_UNMOUNT;
+ vfs_unbusy(mp);
+ if (error) {
+ vfs_unlock(mp);
+ } else {
+ TAILQ_REMOVE(&mountlist, mp, mnt_list);
+ if (coveredvp != NULLVP) {
+ vrele(coveredvp);
+ coveredvp->v_mountedhere = (struct mount *)0;
+ }
+ mp->mnt_op->vfs_refcount--;
+ vfs_unlock(mp);
+ if (mp->mnt_vnodelist.lh_first != NULL)
+ panic("unmount: dangling vnode");
+ free((caddr_t)mp, M_MOUNT);
+ }
+ return (error);
+}
+
+/*
+ * Sync each mounted filesystem.
+ */
+#ifdef DEBUG
+int syncprt = 0;
+struct ctldebug debug0 = { "syncprt", &syncprt };
+#endif
+
+/* ARGSUSED */
+sys_sync(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct mount *mp, *nmp;
+ int asyncflag;
+
+ for (mp = mountlist.cqh_first; mp != (void *)&mountlist; mp = nmp) {
+ /*
+ * Get the next pointer in case we hang on vfs_busy
+ * while we are being unmounted.
+ */
+ nmp = mp->mnt_list.cqe_next;
+ /*
+ * The lock check below is to avoid races with mount
+ * and unmount.
+ */
+ if ((mp->mnt_flag & (MNT_MLOCK|MNT_RDONLY|MNT_MPBUSY)) == 0 &&
+ !vfs_busy(mp)) {
+ asyncflag = mp->mnt_flag & MNT_ASYNC;
+ mp->mnt_flag &= ~MNT_ASYNC;
+ VFS_SYNC(mp, MNT_NOWAIT, p->p_ucred, p);
+ if (asyncflag)
+ mp->mnt_flag |= MNT_ASYNC;
+ /*
+ * Get the next pointer again, as the next filesystem
+ * might have been unmounted while we were sync'ing.
+ */
+ nmp = mp->mnt_list.cqe_next;
+ vfs_unbusy(mp);
+ }
+ }
+#ifdef DEBUG
+ if (syncprt)
+ vfs_bufstats();
+#endif /* DEBUG */
+ return (0);
+}
+
+/*
+ * Change filesystem quotas.
+ */
+/* ARGSUSED */
+sys_quotactl(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_quotactl_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) cmd;
+ syscallarg(int) uid;
+ syscallarg(caddr_t) arg;
+ } */ *uap = v;
+ register struct mount *mp;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ mp = nd.ni_vp->v_mount;
+ vrele(nd.ni_vp);
+ return (VFS_QUOTACTL(mp, SCARG(uap, cmd), SCARG(uap, uid),
+ SCARG(uap, arg), p));
+}
+
+/*
+ * Get filesystem statistics.
+ */
+/* ARGSUSED */
+sys_statfs(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_statfs_args /* {
+ syscallarg(char *) path;
+ syscallarg(struct statfs *) buf;
+ } */ *uap = v;
+ register struct mount *mp;
+ register struct statfs *sp;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ mp = nd.ni_vp->v_mount;
+ sp = &mp->mnt_stat;
+ vrele(nd.ni_vp);
+ if (error = VFS_STATFS(mp, sp, p))
+ return (error);
+ sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
+ return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
+}
+
+/*
+ * Get filesystem statistics.
+ */
+/* ARGSUSED */
+sys_fstatfs(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_fstatfs_args /* {
+ syscallarg(int) fd;
+ syscallarg(struct statfs *) buf;
+ } */ *uap = v;
+ struct file *fp;
+ struct mount *mp;
+ register struct statfs *sp;
+ int error;
+
+ if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
+ return (error);
+ mp = ((struct vnode *)fp->f_data)->v_mount;
+ sp = &mp->mnt_stat;
+ if (error = VFS_STATFS(mp, sp, p))
+ return (error);
+ sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
+ return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
+}
+
+/*
+ * Get statistics on all filesystems.
+ */
+sys_getfsstat(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_getfsstat_args /* {
+ syscallarg(struct statfs *) buf;
+ syscallarg(long) bufsize;
+ syscallarg(int) flags;
+ } */ *uap = v;
+ register struct mount *mp, *nmp;
+ register struct statfs *sp;
+ caddr_t sfsp;
+ long count, maxcount, error;
+
+ maxcount = SCARG(uap, bufsize) / sizeof(struct statfs);
+ sfsp = (caddr_t)SCARG(uap, buf);
+ for (count = 0,
+ mp = mountlist.cqh_first; mp != (void *)&mountlist; mp = nmp) {
+ nmp = mp->mnt_list.cqe_next;
+ if (sfsp && count < maxcount &&
+ ((mp->mnt_flag & MNT_MLOCK) == 0)) {
+ sp = &mp->mnt_stat;
+ /*
+ * If MNT_NOWAIT is specified, do not refresh the
+ * fsstat cache. MNT_WAIT overrides MNT_NOWAIT.
+ */
+ if (((SCARG(uap, flags) & MNT_NOWAIT) == 0 ||
+ (SCARG(uap, flags) & MNT_WAIT)) &&
+ (error = VFS_STATFS(mp, sp, p)))
+ continue;
+ sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
+ if (error = copyout((caddr_t)sp, sfsp, sizeof(*sp)))
+ return (error);
+ sfsp += sizeof(*sp);
+ }
+ count++;
+ }
+ if (sfsp && count > maxcount)
+ *retval = maxcount;
+ else
+ *retval = count;
+ return (0);
+}
+
+/*
+ * Change current working directory to a given file descriptor.
+ */
+/* ARGSUSED */
+sys_fchdir(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_fchdir_args /* {
+ syscallarg(int) fd;
+ } */ *uap = v;
+ register struct filedesc *fdp = p->p_fd;
+ struct vnode *vp, *tdp;
+ struct mount *mp;
+ struct file *fp;
+ int error;
+
+ if (error = getvnode(fdp, SCARG(uap, fd), &fp))
+ return (error);
+ vp = (struct vnode *)fp->f_data;
+ VREF(vp);
+ VOP_LOCK(vp);
+ if (vp->v_type != VDIR)
+ error = ENOTDIR;
+ else
+ error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
+ while (!error && (mp = vp->v_mountedhere) != NULL) {
+ if (mp->mnt_flag & MNT_MLOCK) {
+ mp->mnt_flag |= MNT_MWAIT;
+ sleep((caddr_t)mp, PVFS);
+ continue;
+ }
+ if (error = VFS_ROOT(mp, &tdp))
+ break;
+ vput(vp);
+ vp = tdp;
+ }
+ VOP_UNLOCK(vp);
+ if (error) {
+ vrele(vp);
+ return (error);
+ }
+ vrele(fdp->fd_cdir);
+ fdp->fd_cdir = vp;
+ return (0);
+}
+
+/*
+ * Change current working directory (``.'').
+ */
+/* ARGSUSED */
+sys_chdir(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_chdir_args /* {
+ syscallarg(char *) path;
+ } */ *uap = v;
+ register struct filedesc *fdp = p->p_fd;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if (error = change_dir(&nd, p))
+ return (error);
+ vrele(fdp->fd_cdir);
+ fdp->fd_cdir = nd.ni_vp;
+ return (0);
+}
+
+/*
+ * Change notion of root (``/'') directory.
+ */
+/* ARGSUSED */
+sys_chroot(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_chroot_args /* {
+ syscallarg(char *) path;
+ } */ *uap = v;
+ register struct filedesc *fdp = p->p_fd;
+ int error;
+ struct nameidata nd;
+
+ if (error = suser(p->p_ucred, &p->p_acflag))
+ return (error);
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if (error = change_dir(&nd, p))
+ return (error);
+ if (fdp->fd_rdir != NULL)
+ vrele(fdp->fd_rdir);
+ fdp->fd_rdir = nd.ni_vp;
+ return (0);
+}
+
+/*
+ * Common routine for chroot and chdir.
+ */
+static int
+change_dir(ndp, p)
+ register struct nameidata *ndp;
+ struct proc *p;
+{
+ struct vnode *vp;
+ int error;
+
+ if (error = namei(ndp))
+ return (error);
+ vp = ndp->ni_vp;
+ if (vp->v_type != VDIR)
+ error = ENOTDIR;
+ else
+ error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
+ VOP_UNLOCK(vp);
+ if (error)
+ vrele(vp);
+ return (error);
+}
+
+/*
+ * Check permissions, allocate an open file structure,
+ * and call the device open routine if any.
+ */
+sys_open(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_open_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) flags;
+ syscallarg(int) mode;
+ } */ *uap = v;
+ register struct filedesc *fdp = p->p_fd;
+ register struct file *fp;
+ register struct vnode *vp;
+ int flags, cmode;
+ struct file *nfp;
+ int type, indx, error;
+ struct flock lf;
+ struct nameidata nd;
+ extern struct fileops vnops;
+
+ if (error = falloc(p, &nfp, &indx))
+ return (error);
+ fp = nfp;
+ flags = FFLAGS(SCARG(uap, flags));
+ cmode = ((SCARG(uap, mode) &~ fdp->fd_cmask) & ALLPERMS) &~ S_ISTXT;
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ p->p_dupfd = -indx - 1; /* XXX check for fdopen */
+ if (error = vn_open(&nd, flags, cmode)) {
+ ffree(fp);
+ if ((error == ENODEV || error == ENXIO) &&
+ p->p_dupfd >= 0 && /* XXX from fdopen */
+ (error =
+ dupfdopen(fdp, indx, p->p_dupfd, flags, error)) == 0) {
+ *retval = indx;
+ return (0);
+ }
+ if (error == ERESTART)
+ error = EINTR;
+ fdp->fd_ofiles[indx] = NULL;
+ return (error);
+ }
+ p->p_dupfd = 0;
+ vp = nd.ni_vp;
+ fp->f_flag = flags & FMASK;
+ fp->f_type = DTYPE_VNODE;
+ fp->f_ops = &vnops;
+ fp->f_data = (caddr_t)vp;
+ if (flags & (O_EXLOCK | O_SHLOCK)) {
+ lf.l_whence = SEEK_SET;
+ lf.l_start = 0;
+ lf.l_len = 0;
+ if (flags & O_EXLOCK)
+ lf.l_type = F_WRLCK;
+ else
+ lf.l_type = F_RDLCK;
+ type = F_FLOCK;
+ if ((flags & FNONBLOCK) == 0)
+ type |= F_WAIT;
+ VOP_UNLOCK(vp);
+ if (error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type)) {
+ (void) vn_close(vp, fp->f_flag, fp->f_cred, p);
+ ffree(fp);
+ fdp->fd_ofiles[indx] = NULL;
+ return (error);
+ }
+ VOP_LOCK(vp);
+ fp->f_flag |= FHASLOCK;
+ }
+ VOP_UNLOCK(vp);
+ *retval = indx;
+ return (0);
+}
+
+/*
+ * Create a special file.
+ */
+/* ARGSUSED */
+sys_mknod(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_mknod_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) mode;
+ syscallarg(int) dev;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct vattr vattr;
+ int error;
+ int whiteout;
+ struct nameidata nd;
+
+ if (error = suser(p->p_ucred, &p->p_acflag))
+ return (error);
+ NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ vp = nd.ni_vp;
+ if (vp != NULL)
+ error = EEXIST;
+ else {
+ VATTR_NULL(&vattr);
+ vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
+ vattr.va_rdev = SCARG(uap, dev);
+ whiteout = 0;
+
+ switch (SCARG(uap, mode) & S_IFMT) {
+ case S_IFMT: /* used by badsect to flag bad sectors */
+ vattr.va_type = VBAD;
+ break;
+ case S_IFCHR:
+ vattr.va_type = VCHR;
+ break;
+ case S_IFBLK:
+ vattr.va_type = VBLK;
+ break;
+ case S_IFWHT:
+ whiteout = 1;
+ break;
+ default:
+ error = EINVAL;
+ break;
+ }
+ }
+ if (!error) {
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ if (whiteout) {
+ error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE);
+ if (error)
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ vput(nd.ni_dvp);
+ } else {
+ error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp,
+ &nd.ni_cnd, &vattr);
+ }
+ } else {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+ if (vp)
+ vrele(vp);
+ }
+ return (error);
+}
+
+/*
+ * Create a named pipe.
+ */
+/* ARGSUSED */
+sys_mkfifo(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_mkfifo_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) mode;
+ } */ *uap = v;
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+
+#ifndef FIFO
+ return (EOPNOTSUPP);
+#else
+ NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ if (nd.ni_vp != NULL) {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == nd.ni_vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+ vrele(nd.ni_vp);
+ return (EEXIST);
+ }
+ VATTR_NULL(&vattr);
+ vattr.va_type = VFIFO;
+ vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ return (VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr));
+#endif /* FIFO */
+}
+
+/*
+ * Make a hard file link.
+ */
+/* ARGSUSED */
+sys_link(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_link_args /* {
+ syscallarg(char *) path;
+ syscallarg(char *) link;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct nameidata nd;
+ int error;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ vp = nd.ni_vp;
+ if (vp->v_type != VDIR ||
+ (error = suser(p->p_ucred, &p->p_acflag)) == 0) {
+ nd.ni_cnd.cn_nameiop = CREATE;
+ nd.ni_cnd.cn_flags = LOCKPARENT;
+ nd.ni_dirp = SCARG(uap, link);
+ if ((error = namei(&nd)) == 0) {
+ if (nd.ni_vp != NULL)
+ error = EEXIST;
+ if (!error) {
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred,
+ LEASE_WRITE);
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
+ } else {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == nd.ni_vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+ if (nd.ni_vp)
+ vrele(nd.ni_vp);
+ }
+ }
+ }
+ vrele(vp);
+ return (error);
+}
+
+/*
+ * Make a symbolic link.
+ */
+/* ARGSUSED */
+sys_symlink(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_symlink_args /* {
+ syscallarg(char *) path;
+ syscallarg(char *) link;
+ } */ *uap = v;
+ struct vattr vattr;
+ char *path;
+ int error;
+ struct nameidata nd;
+
+ MALLOC(path, char *, MAXPATHLEN, M_NAMEI, M_WAITOK);
+ if (error = copyinstr(SCARG(uap, path), path, MAXPATHLEN, (size_t *)0))
+ goto out;
+ NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, link), p);
+ if (error = namei(&nd))
+ goto out;
+ if (nd.ni_vp) {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == nd.ni_vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+ vrele(nd.ni_vp);
+ error = EEXIST;
+ goto out;
+ }
+ VATTR_NULL(&vattr);
+ vattr.va_mode = ACCESSPERMS &~ p->p_fd->fd_cmask;
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr, path);
+out:
+ FREE(path, M_NAMEI);
+ return (error);
+}
+
+/*
+ * Delete a whiteout from the filesystem.
+ */
+/* ARGSUSED */
+sys_undelete(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_undelete_args /* {
+ syscallarg(char *) path;
+ } */ *uap = v;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, DELETE, LOCKPARENT|DOWHITEOUT, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ error = namei(&nd);
+ if (error)
+ return (error);
+
+ if (nd.ni_vp != NULLVP || !(nd.ni_cnd.cn_flags & ISWHITEOUT)) {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == nd.ni_vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+ if (nd.ni_vp)
+ vrele(nd.ni_vp);
+ return (EEXIST);
+ }
+
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ if (error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE))
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ vput(nd.ni_dvp);
+ return (error);
+}
+
+/*
+ * Delete a name from the filesystem.
+ */
+/* ARGSUSED */
+sys_unlink(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_unlink_args /* {
+ syscallarg(char *) path;
+ } */ *uap = v;
+ register struct vnode *vp;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, DELETE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ VOP_LOCK(vp);
+
+ if (vp->v_type != VDIR ||
+ (error = suser(p->p_ucred, &p->p_acflag)) == 0) {
+ /*
+ * The root of a mounted filesystem cannot be deleted.
+ */
+ if (vp->v_flag & VROOT)
+ error = EBUSY;
+ else
+ (void)vnode_pager_uncache(vp);
+ }
+
+ if (!error) {
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ error = VOP_REMOVE(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
+ } else {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+ if (vp != NULLVP)
+ vput(vp);
+ }
+ return (error);
+}
+
+/*
+ * Reposition read/write file offset.
+ */
+sys_lseek(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_lseek_args /* {
+ syscallarg(int) fd;
+ syscallarg(int) pad;
+ syscallarg(off_t) offset;
+ syscallarg(int) whence;
+ } */ *uap = v;
+ struct ucred *cred = p->p_ucred;
+ register struct filedesc *fdp = p->p_fd;
+ register struct file *fp;
+ struct vattr vattr;
+ int error;
+
+ if ((u_int)SCARG(uap, fd) >= fdp->fd_nfiles ||
+ (fp = fdp->fd_ofiles[SCARG(uap, fd)]) == NULL)
+ return (EBADF);
+ if (fp->f_type != DTYPE_VNODE)
+ return (ESPIPE);
+ switch (SCARG(uap, whence)) {
+ case L_INCR:
+ fp->f_offset += SCARG(uap, offset);
+ break;
+ case L_XTND:
+ if (error =
+ VOP_GETATTR((struct vnode *)fp->f_data, &vattr, cred, p))
+ return (error);
+ fp->f_offset = SCARG(uap, offset) + vattr.va_size;
+ break;
+ case L_SET:
+ fp->f_offset = SCARG(uap, offset);
+ break;
+ default:
+ return (EINVAL);
+ }
+ *(off_t *)retval = fp->f_offset;
+ return (0);
+}
+
+/*
+ * Check access permissions.
+ */
+sys_access(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_access_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) flags;
+ } */ *uap = v;
+ register struct ucred *cred = p->p_ucred;
+ register struct vnode *vp;
+ int error, flags, t_gid, t_uid;
+ struct nameidata nd;
+
+ t_uid = cred->cr_uid;
+ t_gid = cred->cr_gid;
+ cred->cr_uid = p->p_cred->p_ruid;
+ cred->cr_gid = p->p_cred->p_rgid;
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if (error = namei(&nd))
+ goto out1;
+ vp = nd.ni_vp;
+
+ /* Flags == 0 means only check for existence. */
+ if (SCARG(uap, flags)) {
+ flags = 0;
+ if (SCARG(uap, flags) & R_OK)
+ flags |= VREAD;
+ if (SCARG(uap, flags) & W_OK)
+ flags |= VWRITE;
+ if (SCARG(uap, flags) & X_OK)
+ flags |= VEXEC;
+ if ((flags & VWRITE) == 0 || (error = vn_writechk(vp)) == 0)
+ error = VOP_ACCESS(vp, flags, cred, p);
+ }
+ vput(vp);
+out1:
+ cred->cr_uid = t_uid;
+ cred->cr_gid = t_gid;
+ return (error);
+}
+
+/*
+ * Get file status; this version follows links.
+ */
+/* ARGSUSED */
+sys_stat(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_stat_args /* {
+ syscallarg(char *) path;
+ syscallarg(struct stat *) ub;
+ } */ *uap = v;
+ struct stat sb;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ error = vn_stat(nd.ni_vp, &sb, p);
+ vput(nd.ni_vp);
+ if (error)
+ return (error);
+ error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
+ return (error);
+}
+
+/*
+ * Get file status; this version does not follow links.
+ */
+/* ARGSUSED */
+sys_lstat(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_lstat_args /* {
+ syscallarg(char *) path;
+ syscallarg(struct stat *) ub;
+ } */ *uap = v;
+ int error;
+ struct vnode *vp, *dvp;
+ struct stat sb, sb1;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | LOCKPARENT, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ /*
+ * For symbolic links, always return the attributes of its
+ * containing directory, except for mode, size, and links.
+ */
+ vp = nd.ni_vp;
+ dvp = nd.ni_dvp;
+ if (vp->v_type != VLNK) {
+ if (dvp == vp)
+ vrele(dvp);
+ else
+ vput(dvp);
+ error = vn_stat(vp, &sb, p);
+ vput(vp);
+ if (error)
+ return (error);
+ } else {
+ error = vn_stat(dvp, &sb, p);
+ vput(dvp);
+ if (error) {
+ vput(vp);
+ return (error);
+ }
+ error = vn_stat(vp, &sb1, p);
+ vput(vp);
+ if (error)
+ return (error);
+ sb.st_mode &= ~S_IFDIR;
+ sb.st_mode |= S_IFLNK;
+ sb.st_nlink = sb1.st_nlink;
+ sb.st_size = sb1.st_size;
+ sb.st_blocks = sb1.st_blocks;
+ }
+ error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
+ return (error);
+}
+
+/*
+ * Get configurable pathname variables.
+ */
+/* ARGSUSED */
+sys_pathconf(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_pathconf_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) name;
+ } */ *uap = v;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ error = VOP_PATHCONF(nd.ni_vp, SCARG(uap, name), retval);
+ vput(nd.ni_vp);
+ return (error);
+}
+
+/*
+ * Return target name of a symbolic link.
+ */
+/* ARGSUSED */
+sys_readlink(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_readlink_args /* {
+ syscallarg(char *) path;
+ syscallarg(char *) buf;
+ syscallarg(int) count;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct iovec aiov;
+ struct uio auio;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ vp = nd.ni_vp;
+ if (vp->v_type != VLNK)
+ error = EINVAL;
+ else {
+ aiov.iov_base = SCARG(uap, buf);
+ aiov.iov_len = SCARG(uap, count);
+ auio.uio_iov = &aiov;
+ auio.uio_iovcnt = 1;
+ auio.uio_offset = 0;
+ auio.uio_rw = UIO_READ;
+ auio.uio_segflg = UIO_USERSPACE;
+ auio.uio_procp = p;
+ auio.uio_resid = SCARG(uap, count);
+ error = VOP_READLINK(vp, &auio, p->p_ucred);
+ }
+ vput(vp);
+ *retval = SCARG(uap, count) - auio.uio_resid;
+ return (error);
+}
+
+/*
+ * Change flags of a file given a path name.
+ */
+/* ARGSUSED */
+sys_chflags(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_chflags_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) flags;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ VOP_LOCK(vp);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+ VATTR_NULL(&vattr);
+ vattr.va_flags = SCARG(uap, flags);
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+ vput(vp);
+ return (error);
+}
+
+/*
+ * Change flags of a file given a file descriptor.
+ */
+/* ARGSUSED */
+sys_fchflags(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_fchflags_args /* {
+ syscallarg(int) fd;
+ syscallarg(int) flags;
+ } */ *uap = v;
+ struct vattr vattr;
+ struct vnode *vp;
+ struct file *fp;
+ int error;
+
+ if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
+ return (error);
+ vp = (struct vnode *)fp->f_data;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ VOP_LOCK(vp);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+ VATTR_NULL(&vattr);
+ vattr.va_flags = SCARG(uap, flags);
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+ VOP_UNLOCK(vp);
+ return (error);
+}
+
+/*
+ * Change mode of a file given path name.
+ */
+/* ARGSUSED */
+sys_chmod(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_chmod_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) mode;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ VOP_LOCK(vp);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+ VATTR_NULL(&vattr);
+ vattr.va_mode = SCARG(uap, mode) & ALLPERMS;
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+ vput(vp);
+ return (error);
+}
+
+/*
+ * Change mode of a file given a file descriptor.
+ */
+/* ARGSUSED */
+sys_fchmod(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_fchmod_args /* {
+ syscallarg(int) fd;
+ syscallarg(int) mode;
+ } */ *uap = v;
+ struct vattr vattr;
+ struct vnode *vp;
+ struct file *fp;
+ int error;
+
+ if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
+ return (error);
+ vp = (struct vnode *)fp->f_data;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ VOP_LOCK(vp);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+ VATTR_NULL(&vattr);
+ vattr.va_mode = SCARG(uap, mode) & ALLPERMS;
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+ VOP_UNLOCK(vp);
+ return (error);
+}
+
+/*
+ * Set ownership given a path name.
+ */
+/* ARGSUSED */
+sys_chown(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_chown_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) uid;
+ syscallarg(int) gid;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ VOP_LOCK(vp);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+ VATTR_NULL(&vattr);
+ vattr.va_uid = SCARG(uap, uid);
+ vattr.va_gid = SCARG(uap, gid);
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+ vput(vp);
+ return (error);
+}
+
+/*
+ * Set ownership given a file descriptor.
+ */
+/* ARGSUSED */
+sys_fchown(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_fchown_args /* {
+ syscallarg(int) fd;
+ syscallarg(int) uid;
+ syscallarg(int) gid;
+ } */ *uap = v;
+ struct vattr vattr;
+ struct vnode *vp;
+ struct file *fp;
+ int error;
+
+ if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
+ return (error);
+ vp = (struct vnode *)fp->f_data;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ VOP_LOCK(vp);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+ VATTR_NULL(&vattr);
+ vattr.va_uid = SCARG(uap, uid);
+ vattr.va_gid = SCARG(uap, gid);
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+ VOP_UNLOCK(vp);
+ return (error);
+}
+
+/*
+ * Set the access and modification times of a file.
+ */
+/* ARGSUSED */
+sys_utimes(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_utimes_args /* {
+ syscallarg(char *) path;
+ syscallarg(struct timeval *) tptr;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct timeval tv[2];
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+
+ VATTR_NULL(&vattr);
+ if (SCARG(uap, tptr) == NULL) {
+ microtime(&tv[0]);
+ tv[1] = tv[0];
+ vattr.va_vaflags |= VA_UTIMES_NULL;
+ } else if (error = copyin((caddr_t)SCARG(uap, tptr), (caddr_t)tv,
+ sizeof (tv)))
+ return (error);
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ VOP_LOCK(vp);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+ vattr.va_atime.ts_sec = tv[0].tv_sec;
+ vattr.va_atime.ts_nsec = tv[0].tv_usec * 1000;
+ vattr.va_mtime.ts_sec = tv[1].tv_sec;
+ vattr.va_mtime.ts_nsec = tv[1].tv_usec * 1000;
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+ vput(vp);
+ return (error);
+}
+
+/*
+ * Truncate a file given its path name.
+ */
+/* ARGSUSED */
+sys_truncate(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_truncate_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) pad;
+ syscallarg(off_t) length;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ VOP_LOCK(vp);
+ if (vp->v_type == VDIR)
+ error = EISDIR;
+ else if ((error = vn_writechk(vp)) == 0 &&
+ (error = VOP_ACCESS(vp, VWRITE, p->p_ucred, p)) == 0) {
+ VATTR_NULL(&vattr);
+ vattr.va_size = SCARG(uap, length);
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+ vput(vp);
+ return (error);
+}
+
+/*
+ * Truncate a file given a file descriptor.
+ */
+/* ARGSUSED */
+sys_ftruncate(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_ftruncate_args /* {
+ syscallarg(int) fd;
+ syscallarg(int) pad;
+ syscallarg(off_t) length;
+ } */ *uap = v;
+ struct vattr vattr;
+ struct vnode *vp;
+ struct file *fp;
+ int error;
+
+ if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
+ return (error);
+ if ((fp->f_flag & FWRITE) == 0)
+ return (EINVAL);
+ vp = (struct vnode *)fp->f_data;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ VOP_LOCK(vp);
+ if (vp->v_type == VDIR)
+ error = EISDIR;
+ else if ((error = vn_writechk(vp)) == 0) {
+ VATTR_NULL(&vattr);
+ vattr.va_size = SCARG(uap, length);
+ error = VOP_SETATTR(vp, &vattr, fp->f_cred, p);
+ }
+ VOP_UNLOCK(vp);
+ return (error);
+}
+
+/*
+ * Sync an open file.
+ */
+/* ARGSUSED */
+sys_fsync(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_fsync_args /* {
+ syscallarg(int) fd;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct file *fp;
+ int error;
+
+ if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
+ return (error);
+ vp = (struct vnode *)fp->f_data;
+ VOP_LOCK(vp);
+ error = VOP_FSYNC(vp, fp->f_cred, MNT_WAIT, p);
+ VOP_UNLOCK(vp);
+ return (error);
+}
+
+/*
+ * Rename files. Source and destination must either both be directories,
+ * or both not be directories. If target is a directory, it must be empty.
+ */
+/* ARGSUSED */
+sys_rename(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_rename_args /* {
+ syscallarg(char *) from;
+ syscallarg(char *) to;
+ } */ *uap = v;
+ register struct vnode *tvp, *fvp, *tdvp;
+ struct nameidata fromnd, tond;
+ int error;
+
+ NDINIT(&fromnd, DELETE, WANTPARENT | SAVESTART, UIO_USERSPACE,
+ SCARG(uap, from), p);
+ if (error = namei(&fromnd))
+ return (error);
+ fvp = fromnd.ni_vp;
+ NDINIT(&tond, RENAME, LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART,
+ UIO_USERSPACE, SCARG(uap, to), p);
+ if (error = namei(&tond)) {
+ VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
+ vrele(fromnd.ni_dvp);
+ vrele(fvp);
+ goto out1;
+ }
+ tdvp = tond.ni_dvp;
+ tvp = tond.ni_vp;
+ if (tvp != NULL) {
+ if (fvp->v_type == VDIR && tvp->v_type != VDIR) {
+ error = ENOTDIR;
+ goto out;
+ } else if (fvp->v_type != VDIR && tvp->v_type == VDIR) {
+ error = EISDIR;
+ goto out;
+ }
+ }
+ if (fvp == tdvp)
+ error = EINVAL;
+ /*
+ * If source is the same as the destination (that is the
+ * same inode number with the same name in the same directory),
+ * then there is nothing to do.
+ */
+ if (fvp == tvp && fromnd.ni_dvp == tdvp &&
+ fromnd.ni_cnd.cn_namelen == tond.ni_cnd.cn_namelen &&
+ !bcmp(fromnd.ni_cnd.cn_nameptr, tond.ni_cnd.cn_nameptr,
+ fromnd.ni_cnd.cn_namelen))
+ error = -1;
+out:
+ if (!error) {
+ VOP_LEASE(tdvp, p, p->p_ucred, LEASE_WRITE);
+ if (fromnd.ni_dvp != tdvp)
+ VOP_LEASE(fromnd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ if (tvp)
+ VOP_LEASE(tvp, p, p->p_ucred, LEASE_WRITE);
+ error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd,
+ tond.ni_dvp, tond.ni_vp, &tond.ni_cnd);
+ } else {
+ VOP_ABORTOP(tond.ni_dvp, &tond.ni_cnd);
+ if (tdvp == tvp)
+ vrele(tdvp);
+ else
+ vput(tdvp);
+ if (tvp)
+ vput(tvp);
+ VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
+ vrele(fromnd.ni_dvp);
+ vrele(fvp);
+ }
+ vrele(tond.ni_startdir);
+ FREE(tond.ni_cnd.cn_pnbuf, M_NAMEI);
+out1:
+ if (fromnd.ni_startdir)
+ vrele(fromnd.ni_startdir);
+ FREE(fromnd.ni_cnd.cn_pnbuf, M_NAMEI);
+ if (error == -1)
+ return (0);
+ return (error);
+}
+
+/*
+ * Make a directory file.
+ */
+/* ARGSUSED */
+sys_mkdir(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_mkdir_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) mode;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ vp = nd.ni_vp;
+ if (vp != NULL) {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+ vrele(vp);
+ return (EEXIST);
+ }
+ VATTR_NULL(&vattr);
+ vattr.va_type = VDIR;
+ vattr.va_mode = (SCARG(uap, mode) & ACCESSPERMS) &~ p->p_fd->fd_cmask;
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
+ if (!error)
+ vput(nd.ni_vp);
+ return (error);
+}
+
+/*
+ * Remove a directory file.
+ */
+/* ARGSUSED */
+sys_rmdir(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_rmdir_args /* {
+ syscallarg(char *) path;
+ } */ *uap = v;
+ register struct vnode *vp;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ vp = nd.ni_vp;
+ if (vp->v_type != VDIR) {
+ error = ENOTDIR;
+ goto out;
+ }
+ /*
+ * No rmdir "." please.
+ */
+ if (nd.ni_dvp == vp) {
+ error = EINVAL;
+ goto out;
+ }
+ /*
+ * The root of a mounted filesystem cannot be deleted.
+ */
+ if (vp->v_flag & VROOT)
+ error = EBUSY;
+out:
+ if (!error) {
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
+ } else {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+ vput(vp);
+ }
+ return (error);
+}
+
+/*
+ * Read a block of directory entries in a file system independent format.
+ */
+sys_getdirentries(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_getdirentries_args /* {
+ syscallarg(int) fd;
+ syscallarg(char *) buf;
+ syscallarg(u_int) count;
+ syscallarg(long *) basep;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct file *fp;
+ struct uio auio;
+ struct iovec aiov;
+ long loff;
+ int error, eofflag;
+
+ if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
+ return (error);
+ if ((fp->f_flag & FREAD) == 0)
+ return (EBADF);
+ vp = (struct vnode *)fp->f_data;
+unionread:
+ if (vp->v_type != VDIR)
+ return (EINVAL);
+ aiov.iov_base = SCARG(uap, buf);
+ aiov.iov_len = SCARG(uap, count);
+ auio.uio_iov = &aiov;
+ auio.uio_iovcnt = 1;
+ auio.uio_rw = UIO_READ;
+ auio.uio_segflg = UIO_USERSPACE;
+ auio.uio_procp = p;
+ auio.uio_resid = SCARG(uap, count);
+ VOP_LOCK(vp);
+ loff = auio.uio_offset = fp->f_offset;
+ error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, (u_long *)0, 0);
+ fp->f_offset = auio.uio_offset;
+ VOP_UNLOCK(vp);
+ if (error)
+ return (error);
+
+#ifdef UNION
+{
+ extern int (**union_vnodeop_p)();
+ extern struct vnode *union_dircache __P((struct vnode *));
+
+ if ((SCARG(uap, count) == auio.uio_resid) &&
+ (vp->v_op == union_vnodeop_p)) {
+ struct vnode *lvp;
+
+ lvp = union_dircache(vp);
+ if (lvp != NULLVP) {
+ struct vattr va;
+
+ /*
+ * If the directory is opaque,
+ * then don't show lower entries
+ */
+ error = VOP_GETATTR(vp, &va, fp->f_cred, p);
+ if (va.va_flags & OPAQUE) {
+ vput(lvp);
+ lvp = NULL;
+ }
+ }
+
+ if (lvp != NULLVP) {
+ error = VOP_OPEN(lvp, FREAD, fp->f_cred, p);
+ VOP_UNLOCK(lvp);
+
+ if (error) {
+ vrele(lvp);
+ return (error);
+ }
+ fp->f_data = (caddr_t) lvp;
+ fp->f_offset = 0;
+ error = vn_close(vp, FREAD, fp->f_cred, p);
+ if (error)
+ return (error);
+ vp = lvp;
+ goto unionread;
+ }
+ }
+}
+#endif /* UNION */
+
+ if ((SCARG(uap, count) == auio.uio_resid) &&
+ (vp->v_flag & VROOT) &&
+ (vp->v_mount->mnt_flag & MNT_UNION)) {
+ struct vnode *tvp = vp;
+ vp = vp->v_mount->mnt_vnodecovered;
+ VREF(vp);
+ fp->f_data = (caddr_t) vp;
+ fp->f_offset = 0;
+ vrele(tvp);
+ goto unionread;
+ }
+ error = copyout((caddr_t)&loff, (caddr_t)SCARG(uap, basep),
+ sizeof(long));
+ *retval = SCARG(uap, count) - auio.uio_resid;
+ return (error);
+}
+
+/*
+ * Set the mode mask for creation of filesystem nodes.
+ */
+int
+sys_umask(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_umask_args /* {
+ syscallarg(int) newmask;
+ } */ *uap = v;
+ register struct filedesc *fdp;
+
+ fdp = p->p_fd;
+ *retval = fdp->fd_cmask;
+ fdp->fd_cmask = SCARG(uap, newmask) & ALLPERMS;
+ return (0);
+}
+
+/*
+ * Void all references to file by ripping underlying filesystem
+ * away from vnode.
+ */
+/* ARGSUSED */
+sys_revoke(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_revoke_args /* {
+ syscallarg(char *) path;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if (error = namei(&nd))
+ return (error);
+ vp = nd.ni_vp;
+ if (vp->v_type != VCHR && vp->v_type != VBLK) {
+ error = EINVAL;
+ goto out;
+ }
+ if (error = VOP_GETATTR(vp, &vattr, p->p_ucred, p))
+ goto out;
+ if (p->p_ucred->cr_uid != vattr.va_uid &&
+ (error = suser(p->p_ucred, &p->p_acflag)))
+ goto out;
+ if (vp->v_usecount > 1 || (vp->v_flag & VALIASED))
+ vgoneall(vp);
+out:
+ vrele(vp);
+ return (error);
+}
+
+/*
+ * Convert a user file descriptor to a kernel file entry.
+ */
+getvnode(fdp, fd, fpp)
+ struct filedesc *fdp;
+ struct file **fpp;
+ int fd;
+{
+ struct file *fp;
+
+ if ((u_int)fd >= fdp->fd_nfiles ||
+ (fp = fdp->fd_ofiles[fd]) == NULL)
+ return (EBADF);
+ if (fp->f_type != DTYPE_VNODE)
+ return (EINVAL);
+ *fpp = fp;
+ return (0);
+}
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t9.2
@@ -1,0 +1,3216 @@
+/* $OpenBSD: t9.2,v 1.2 2013/12/01 16:40:56 krw Exp $ */
+/* $NetBSD: vfs_syscalls.c,v 1.71 1996/04/23 10:29:02 mycroft Exp $ */
+
+/*
+ * Copyright (c) 1989, 1993
+ * The Regents of the University of California. All rights reserved.
+ * (c) UNIX System Laboratories, Inc.
+ * All or some portions of this file are derived from material licensed
+ * to the University of California by American Telephone and Telegraph
+ * Co. or Unix System Laboratories, Inc. and are reproduced herein with
+ * the permission of UNIX System Laboratories, Inc.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * 3. Neither the name of the University nor the names of its contributors
+ * may be used to endorse or promote products derived from this software
+ * without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * @(#)vfs_syscalls.c 8.28 (Berkeley) 12/10/94
+ */
+
+#include <sys/param.h>
+#include <sys/systm.h>
+#include <sys/namei.h>
+#include <sys/filedesc.h>
+#include <sys/kernel.h>
+#include <sys/file.h>
+#include <sys/stat.h>
+#include <sys/vnode.h>
+#include <sys/mount.h>
+#include <sys/proc.h>
+#include <sys/uio.h>
+#include <sys/malloc.h>
+#include <sys/dirent.h>
+#include <sys/extattr.h>
+
+#include <sys/syscallargs.h>
+
+#include <uvm/uvm_extern.h>
+#include <sys/sysctl.h>
+
+extern int suid_clear;
+int usermount = 0; /* sysctl: by default, users may not mount */
+
+static int change_dir(struct nameidata *, struct proc *);
+
+void checkdirs(struct vnode *);
+
+/*
+ * Redirection info so we don't have to include the union fs routines in
+ * the kernel directly. This way, we can build unionfs as an LKM. The
+ * pointer gets filled in later, when we modload the LKM, or when the
+ * compiled-in unionfs code gets initialized. For now, we just set
+ * it to a stub routine.
+ */
+
+int (*union_check_p)(struct proc *, struct vnode **,
+ struct file *, struct uio, int *) = NULL;
+
+/*
+ * Virtual File System System Calls
+ */
+
+/*
+ * Mount a file system.
+ */
+/* ARGSUSED */
+int
+sys_mount(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_mount_args /* {
+ syscallarg(char *) type;
+ syscallarg(char *) path;
+ syscallarg(int) flags;
+ syscallarg(void *) data;
+ } */ *uap = v;
+ register struct vnode *vp;
+ register struct mount *mp;
+ int error, flag = 0;
+#ifdef COMPAT_43
+ u_long fstypenum = 0;
+#endif
+ char fstypename[MFSNAMELEN];
+ char fspath[MNAMELEN];
+ struct vattr va;
+ struct nameidata nd;
+ struct vfsconf *vfsp;
+ struct timeval tv;
+
+ if (usermount == 0 && (error = suser(p->p_ucred, &p->p_acflag)))
+ return (error);
+
+ /*
+ * Mount points must fit in MNAMELEN, not MAXPATHLEN.
+ */
+ error = copyinstr(SCARG(uap, path), fspath, MNAMELEN, NULL);
+ if (error)
+ return(error);
+
+ /*
+ * Get vnode to be covered
+ */
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE, fspath, p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ if (SCARG(uap, flags) & MNT_UPDATE) {
+ if ((vp->v_flag & VROOT) == 0) {
+ vput(vp);
+ return (EINVAL);
+ }
+ mp = vp->v_mount;
+ flag = mp->mnt_flag;
+ /*
+ * We only allow the filesystem to be reloaded if it
+ * is currently mounted read-only.
+ */
+ if ((SCARG(uap, flags) & MNT_RELOAD) &&
+ ((mp->mnt_flag & MNT_RDONLY) == 0)) {
+ vput(vp);
+ return (EOPNOTSUPP); /* Needs translation */
+ }
+ mp->mnt_flag |=
+ SCARG(uap, flags) & (MNT_RELOAD | MNT_UPDATE);
+ /*
+ * Only root, or the user that did the original mount is
+ * permitted to update it.
+ */
+ if (mp->mnt_stat.f_owner != p->p_ucred->cr_uid &&
+ (error = suser(p->p_ucred, &p->p_acflag))) {
+ vput(vp);
+ return (error);
+ }
+ /*
+ * Do not allow NFS export by non-root users. Silently
+ * enforce MNT_NOSUID and MNT_NODEV for non-root users.
+ */
+ if (p->p_ucred->cr_uid != 0) {
+ if (SCARG(uap, flags) & MNT_EXPORTED) {
+ vput(vp);
+ return (EPERM);
+ }
+ SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
+ }
+ if ((error = vfs_busy(mp, LK_NOWAIT, 0, p)) != 0) {
+ vput(vp);
+ return (error);
+ }
+ VOP_UNLOCK(vp, 0, p);
+ goto update;
+ }
+ /*
+ * If the user is not root, ensure that they own the directory
+ * onto which we are attempting to mount.
+ */
+ if ((error = VOP_GETATTR(vp, &va, p->p_ucred, p)) ||
+ (va.va_uid != p->p_ucred->cr_uid &&
+ (error = suser(p->p_ucred, &p->p_acflag)))) {
+ vput(vp);
+ return (error);
+ }
+ /*
+ * Do not allow NFS export by non-root users. Silently
+ * enforce MNT_NOSUID and MNT_NODEV for non-root users.
+ */
+ if (p->p_ucred->cr_uid != 0) {
+ if (SCARG(uap, flags) & MNT_EXPORTED) {
+ vput(vp);
+ return (EPERM);
+ }
+ SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
+ }
+ if ((error = vinvalbuf(vp, V_SAVE, p->p_ucred, p, 0, 0)) != 0)
+ return (error);
+ if (vp->v_type != VDIR) {
+ vput(vp);
+ return (ENOTDIR);
+ }
+ error = copyinstr(SCARG(uap, type), fstypename, MFSNAMELEN, NULL);
+ if (error) {
+#ifdef COMPAT_43
+ /*
+ * Historically filesystem types were identified by number.
+ * If we get an integer for the filesystem type instead of a
+ * string, we check to see if it matches one of the historic
+ * filesystem types.
+ */
+ fstypenum = (u_long)SCARG(uap, type);
+
+ for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next)
+ if (vfsp->vfc_typenum == fstypenum)
+ break;
+ if (vfsp == NULL) {
+ vput(vp);
+ return (ENODEV);
+ }
+ strncpy(fstypename, vfsp->vfc_name, MFSNAMELEN);
+
+#else
+ vput(vp);
+ return (error);
+#endif
+ }
+ for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next) {
+ if (!strcmp(vfsp->vfc_name, fstypename))
+ break;
+ }
+
+ if (vfsp == NULL) {
+ vput(vp);
+ return (EOPNOTSUPP);
+ }
+
+ if (vp->v_mountedhere != NULL) {
+ vput(vp);
+ return (EBUSY);
+ }
+
+ /*
+ * Allocate and initialize the file system.
+ */
+ mp = (struct mount *)malloc((u_long)sizeof(struct mount),
+ M_MOUNT, M_WAITOK);
+ bzero((char *)mp, (u_long)sizeof(struct mount));
+ lockinit(&mp->mnt_lock, PVFS, "vfslock", 0, 0);
+ /* This error never happens, but it makes auditing easier */
+ if ((error = vfs_busy(mp, LK_NOWAIT, 0, p)))
+ return (error);
+ mp->mnt_op = vfsp->vfc_vfsops;
+ mp->mnt_vfc = vfsp;
+ mp->mnt_flag |= (vfsp->vfc_flags & MNT_VISFLAGMASK);
+ strncpy(mp->mnt_stat.f_fstypename, vfsp->vfc_name, MFSNAMELEN);
+ mp->mnt_vnodecovered = vp;
+ mp->mnt_stat.f_owner = p->p_ucred->cr_uid;
+update:
+ /*
+ * Set the mount level flags.
+ */
+ if (SCARG(uap, flags) & MNT_RDONLY)
+ mp->mnt_flag |= MNT_RDONLY;
+ else if (mp->mnt_flag & MNT_RDONLY)
+ mp->mnt_flag |= MNT_WANTRDWR;
+ mp->mnt_flag &=~ (MNT_NOSUID | MNT_NOEXEC | MNT_NODEV |
+ MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | MNT_SOFTDEP |
+ MNT_NOATIME | MNT_FORCE);
+ mp->mnt_flag |= SCARG(uap, flags) & (MNT_NOSUID | MNT_NOEXEC |
+ MNT_NODEV | MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC |
+ MNT_SOFTDEP | MNT_NOATIME | MNT_FORCE);
+ /*
+ * Mount the filesystem.
+ */
+ error = VFS_MOUNT(mp, SCARG(uap, path), SCARG(uap, data), &nd, p);
+ if (!error) {
+ microtime(&tv);
+ mp->mnt_stat.f_ctime = tv.tv_sec;
+ }
+ if (mp->mnt_flag & MNT_UPDATE) {
+ vrele(vp);
+ if (mp->mnt_flag & MNT_WANTRDWR)
+ mp->mnt_flag &= ~MNT_RDONLY;
+ mp->mnt_flag &=~
+ (MNT_UPDATE | MNT_RELOAD | MNT_FORCE | MNT_WANTRDWR);
+ if (error)
+ mp->mnt_flag = flag;
+
+ if ((mp->mnt_flag & MNT_RDONLY) == 0) {
+ if (mp->mnt_syncer == NULL)
+ error = vfs_allocate_syncvnode(mp);
+ } else {
+ if (mp->mnt_syncer != NULL)
+ vgone(mp->mnt_syncer);
+ mp->mnt_syncer = NULL;
+ }
+
+ vfs_unbusy(mp, p);
+ return (error);
+ }
+
+ vp->v_mountedhere = mp;
+
+ /*
+ * Put the new filesystem on the mount list after root.
+ */
+ cache_purge(vp);
+ if (!error) {
+ vfsp->vfc_refcount++;
+ simple_lock(&mountlist_slock);
+ TAILQ_INSERT_TAIL(&mountlist, mp, mnt_list);
+ simple_unlock(&mountlist_slock);
+ checkdirs(vp);
+ VOP_UNLOCK(vp, 0, p);
+ if ((mp->mnt_flag & MNT_RDONLY) == 0)
+ error = vfs_allocate_syncvnode(mp);
+ vfs_unbusy(mp, p);
+ (void) VFS_STATFS(mp, &mp->mnt_stat, p);
+ if ((error = VFS_START(mp, 0, p)) != 0)
+ vrele(vp);
+ } else {
+ mp->mnt_vnodecovered->v_mountedhere = (struct mount *)0;
+ vfs_unbusy(mp, p);
+ free((caddr_t)mp, M_MOUNT);
+ vput(vp);
+ }
+ return (error);
+}
+
+/*
+ * Scan all active processes to see if any of them have a current
+ * or root directory onto which the new filesystem has just been
+ * mounted. If so, replace them with the new mount point.
+ */
+void
+checkdirs(olddp)
+ struct vnode *olddp;
+{
+ struct filedesc *fdp;
+ struct vnode *newdp;
+ struct proc *p;
+
+ if (olddp->v_usecount == 1)
+ return;
+ if (VFS_ROOT(olddp->v_mountedhere, &newdp))
+ panic("mount: lost mount");
+ for (p = LIST_FIRST(&allproc); p != 0; p = LIST_NEXT(p, p_list)) {
+ fdp = p->p_fd;
+ if (fdp->fd_cdir == olddp) {
+ vrele(fdp->fd_cdir);
+ VREF(newdp);
+ fdp->fd_cdir = newdp;
+ }
+ if (fdp->fd_rdir == olddp) {
+ vrele(fdp->fd_rdir);
+ VREF(newdp);
+ fdp->fd_rdir = newdp;
+ }
+ }
+ if (rootvnode == olddp) {
+ vrele(rootvnode);
+ VREF(newdp);
+ rootvnode = newdp;
+ }
+ vput(newdp);
+}
+
+/*
+ * Unmount a file system.
+ *
+ * Note: unmount takes a path to the vnode mounted on as argument,
+ * not special file (as before).
+ */
+/* ARGSUSED */
+int
+sys_unmount(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_unmount_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) flags;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct mount *mp;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ mp = vp->v_mount;
+
+ /*
+ * Only root, or the user that did the original mount is
+ * permitted to unmount this filesystem.
+ */
+ if ((mp->mnt_stat.f_owner != p->p_ucred->cr_uid) &&
+ (error = suser(p->p_ucred, &p->p_acflag))) {
+ vput(vp);
+ return (error);
+ }
+
+ /*
+ * Don't allow unmounting the root file system.
+ */
+ if (mp->mnt_flag & MNT_ROOTFS) {
+ vput(vp);
+ return (EINVAL);
+ }
+
+ /*
+ * Must be the root of the filesystem
+ */
+ if ((vp->v_flag & VROOT) == 0) {
+ vput(vp);
+ return (EINVAL);
+ }
+ vput(vp);
+
+ if (vfs_busy(mp, LK_EXCLUSIVE, NULL, p))
+ return (EBUSY);
+
+ return (dounmount(mp, SCARG(uap, flags), p, vp));
+}
+
+/*
+ * Do the actual file system unmount.
+ */
+int
+dounmount(struct mount *mp, int flags, struct proc *p, struct vnode *olddp)
+{
+ struct vnode *coveredvp;
+ struct proc *p2;
+ int error;
+ int hadsyncer = 0;
+
+ mp->mnt_flag &=~ MNT_ASYNC;
+ cache_purgevfs(mp); /* remove cache entries for this file sys */
+ if (mp->mnt_syncer != NULL) {
+ hadsyncer = 1;
+ vgone(mp->mnt_syncer);
+ mp->mnt_syncer = NULL;
+ }
+ if (((mp->mnt_flag & MNT_RDONLY) ||
+ (error = VFS_SYNC(mp, MNT_WAIT, p->p_ucred, p)) == 0) ||
+ (flags & MNT_FORCE))
+ error = VFS_UNMOUNT(mp, flags, p);
+ simple_lock(&mountlist_slock);
+ if (error) {
+ if ((mp->mnt_flag & MNT_RDONLY) == 0 && hadsyncer)
+ (void) vfs_allocate_syncvnode(mp);
+ lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK,
+ &mountlist_slock, p);
+ return (error);
+ }
+ TAILQ_REMOVE(&mountlist, mp, mnt_list);
+ if ((coveredvp = mp->mnt_vnodecovered) != NULLVP) {
+ if (olddp) {
+ /*
+ * Try to put processes back in a real directory
+ * after a forced unmount.
+ * XXX We're not holding a ref on olddp, which may
+ * change, so compare id numbers.
+ */
+ LIST_FOREACH(p2, &allproc, p_list) {
+ struct filedesc *fdp = p2->p_fd;
+ if (fdp->fd_cdir &&
+ fdp->fd_cdir->v_id == olddp->v_id) {
+ vrele(fdp->fd_cdir);
+ vref(coveredvp);
+ fdp->fd_cdir = coveredvp;
+ }
+ if (fdp->fd_rdir &&
+ fdp->fd_rdir->v_id == olddp->v_id) {
+ vrele(fdp->fd_rdir);
+ vref(coveredvp);
+ fdp->fd_rdir = coveredvp;
+ }
+ }
+ }
+ coveredvp->v_mountedhere = NULL;
+ vrele(coveredvp);
+ }
+ mp->mnt_vfc->vfc_refcount--;
+ if (mp->mnt_vnodelist.lh_first != NULL)
+ panic("unmount: dangling vnode");
+ lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK, &mountlist_slock, p);
+ free((caddr_t)mp, M_MOUNT);
+ return (0);
+}
+
+/*
+ * Sync each mounted filesystem.
+ */
+#ifdef DEBUG
+int syncprt = 0;
+struct ctldebug debug0 = { "syncprt", &syncprt };
+#endif
+
+/* ARGSUSED */
+int
+sys_sync(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct mount *mp, *nmp;
+ int asyncflag;
+
+ simple_lock(&mountlist_slock);
+ TAILQ_FOREACH_REVERSE_SAFE(mp, &mountlist, mnt_list, nmp) {
+ if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock, p))
+ continue;
+ if ((mp->mnt_flag & MNT_RDONLY) == 0) {
+ asyncflag = mp->mnt_flag & MNT_ASYNC;
+ mp->mnt_flag &= ~MNT_ASYNC;
+ uvm_vnp_sync(mp);
+ VFS_SYNC(mp, MNT_NOWAIT, p->p_ucred, p);
+ if (asyncflag)
+ mp->mnt_flag |= MNT_ASYNC;
+ }
+ simple_lock(&mountlist_slock);
+ vfs_unbusy(mp, p);
+ }
+ simple_unlock(&mountlist_slock);
+
+#ifdef DEBUG
+ if (syncprt)
+ vfs_bufstats();
+#endif /* DEBUG */
+ return (0);
+}
+
+/*
+ * Change filesystem quotas.
+ */
+/* ARGSUSED */
+int
+sys_quotactl(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_quotactl_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) cmd;
+ syscallarg(int) uid;
+ syscallarg(caddr_t) arg;
+ } */ *uap = v;
+ register struct mount *mp;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ mp = nd.ni_vp->v_mount;
+ vrele(nd.ni_vp);
+ return (VFS_QUOTACTL(mp, SCARG(uap, cmd), SCARG(uap, uid),
+ SCARG(uap, arg), p));
+}
+
+/*
+ * Get filesystem statistics.
+ */
+/* ARGSUSED */
+int
+sys_statfs(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_statfs_args /* {
+ syscallarg(char *) path;
+ syscallarg(struct statfs *) buf;
+ } */ *uap = v;
+ register struct mount *mp;
+ register struct statfs *sp;
+ int error;
+ struct nameidata nd;
+ struct statfs sb;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ mp = nd.ni_vp->v_mount;
+ sp = &mp->mnt_stat;
+ vrele(nd.ni_vp);
+ if ((error = VFS_STATFS(mp, sp, p)) != 0)
+ return (error);
+ sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
+#if notyet
+ if (mp->mnt_flag & MNT_SOFTDEP)
+ sp->f_eflags = STATFS_SOFTUPD;
+#endif
+ /* Don't let non-root see filesystem id (for NFS security) */
+ if (suser(p->p_ucred, &p->p_acflag)) {
+ bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
+ sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
+ sp = &sb;
+ }
+ return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
+}
+
+/*
+ * Get filesystem statistics.
+ */
+/* ARGSUSED */
+int
+sys_fstatfs(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_fstatfs_args /* {
+ syscallarg(int) fd;
+ syscallarg(struct statfs *) buf;
+ } */ *uap = v;
+ struct file *fp;
+ struct mount *mp;
+ struct statfs *sp;
+ int error;
+ struct statfs sb;
+
+ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+ mp = ((struct vnode *)fp->f_data)->v_mount;
+ sp = &mp->mnt_stat;
+ error = VFS_STATFS(mp, sp, p);
+ FRELE(fp);
+ if (error)
+ return (error);
+ sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
+#if notyet
+ if (mp->mnt_flag & MNT_SOFTDEP)
+ sp->f_eflags = STATFS_SOFTUPD;
+#endif
+ /* Don't let non-root see filesystem id (for NFS security) */
+ if (suser(p->p_ucred, &p->p_acflag)) {
+ bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
+ sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
+ sp = &sb;
+ }
+ return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
+}
+
+/*
+ * Get statistics on all filesystems.
+ */
+int
+sys_getfsstat(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_getfsstat_args /* {
+ syscallarg(struct statfs *) buf;
+ syscallarg(size_t) bufsize;
+ syscallarg(int) flags;
+ } */ *uap = v;
+ register struct mount *mp *nmp;
+ register struct statfs *sp;
+ struct statfs sb;
+ caddr_t sfsp;
+ size_t count, maxcount;
+ int error, flags = SCARG(uap, flags);
+
+ maxcount = SCARG(uap, bufsize) / sizeof(struct statfs);
+ sfsp = (caddr_t)SCARG(uap, buf);
+ count = 0;
+ simple_lock(&mountlist_slock);
+ TAILQ_FOREACH_REVERSE_SAFE(mp, &mountlist, mnt_list, nmp) {
+ if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock, p))
+ continue;
+ if (sfsp && count < maxcount) {
+ sp = &mp->mnt_stat;
+
+ /* Refresh stats unless MNT_NOWAIT is specified */
+ if (flags != MNT_NOWAIT &&
+ flags != MNT_LAZY &&
+ (flags == MNT_WAIT ||
+ flags == 0) &&
+ (error = VFS_STATFS(mp, sp, p))) {
+ simple_lock(&mountlist_slock);
+ vfs_unbusy(mp, p);
+ continue;
+ }
+
+ sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
+#if notyet
+ if (mp->mnt_flag & MNT_SOFTDEP)
+ sp->f_eflags = STATFS_SOFTUPD;
+#endif
+ if (suser(p->p_ucred, &p->p_acflag)) {
+ bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
+ sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
+ sp = &sb;
+ }
+ error = copyout((caddr_t)sp, sfsp, sizeof(*sp));
+ if (error) {
+ vfs_unbusy(mp, p);
+ return (error);
+ }
+ sfsp += sizeof(*sp);
+ }
+ count++;
+ simple_lock(&mountlist_slock);
+ vfs_unbusy(mp, p);
+ }
+ simple_unlock(&mountlist_slock);
+ if (sfsp && count > maxcount)
+ *retval = maxcount;
+ else
+ *retval = count;
+ return (0);
+}
+
+/*
+ * Change current working directory to a given file descriptor.
+ */
+/* ARGSUSED */
+int
+sys_fchdir(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_fchdir_args /* {
+ syscallarg(int) fd;
+ } */ *uap = v;
+ struct filedesc *fdp = p->p_fd;
+ struct vnode *vp, *tdp;
+ struct mount *mp;
+ struct file *fp;
+ int error;
+
+ if ((error = getvnode(fdp, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+ vp = (struct vnode *)fp->f_data;
+ VREF(vp);
+ FRELE(fp);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_type != VDIR)
+ error = ENOTDIR;
+ else
+ error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
+
+ while (!error && (mp = vp->v_mountedhere) != NULL) {
+ if (vfs_busy(mp, 0, 0, p))
+ continue;
+ error = VFS_ROOT(mp, &tdp);
+ vfs_unbusy(mp, p);
+ if (error)
+ break;
+ vput(vp);
+ vp = tdp;
+ }
+ if (error) {
+ vput(vp);
+ return (error);
+ }
+ VOP_UNLOCK(vp, 0, p);
+ vrele(fdp->fd_cdir);
+ fdp->fd_cdir = vp;
+ return (0);
+}
+
+/*
+ * Change current working directory (``.'').
+ */
+/* ARGSUSED */
+int
+sys_chdir(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_chdir_args /* {
+ syscallarg(char *) path;
+ } */ *uap = v;
+ register struct filedesc *fdp = p->p_fd;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if ((error = change_dir(&nd, p)) != 0)
+ return (error);
+ vrele(fdp->fd_cdir);
+ fdp->fd_cdir = nd.ni_vp;
+ return (0);
+}
+
+/*
+ * Change notion of root (``/'') directory.
+ */
+/* ARGSUSED */
+int
+sys_chroot(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_chroot_args /* {
+ syscallarg(char *) path;
+ } */ *uap = v;
+ register struct filedesc *fdp = p->p_fd;
+ int error;
+ struct nameidata nd;
+
+ if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
+ return (error);
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if ((error = change_dir(&nd, p)) != 0)
+ return (error);
+ if (fdp->fd_rdir != NULL) {
+ /*
+ * A chroot() done inside a changed root environment does
+ * an automatic chdir to avoid the out-of-tree experience.
+ */
+ vrele(fdp->fd_rdir);
+ vrele(fdp->fd_cdir);
+ VREF(nd.ni_vp);
+ fdp->fd_cdir = nd.ni_vp;
+ }
+ fdp->fd_rdir = nd.ni_vp;
+ return (0);
+}
+
+/*
+ * Common routine for chroot and chdir.
+ */
+static int
+change_dir(ndp, p)
+ register struct nameidata *ndp;
+ struct proc *p;
+{
+ struct vnode *vp;
+ int error;
+
+ if ((error = namei(ndp)) != 0)
+ return (error);
+ vp = ndp->ni_vp;
+ if (vp->v_type != VDIR)
+ error = ENOTDIR;
+ else
+ error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
+ if (error)
+ vput(vp);
+ else
+ VOP_UNLOCK(vp, 0, p);
+ return (error);
+}
+
+/*
+ * Check permissions, allocate an open file structure,
+ * and call the device open routine if any.
+ */
+int
+sys_open(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_open_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) flags;
+ syscallarg(int) mode;
+ } */ *uap = v;
+ struct filedesc *fdp = p->p_fd;
+ struct file *fp;
+ struct vnode *vp;
+ struct vattr vattr;
+ int flags, cmode;
+ int type, indx, error, localtrunc = 0;
+ struct flock lf;
+ struct nameidata nd;
+
+ if ((error = falloc(p, &fp, &indx)) != 0)
+ return (error);
+
+ flags = FFLAGS(SCARG(uap, flags));
+ cmode = ((SCARG(uap, mode) &~ fdp->fd_cmask) & ALLPERMS) &~ S_ISTXT;
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ p->p_dupfd = -indx - 1; /* XXX check for fdopen */
+ if ((flags & O_TRUNC) && (flags & (O_EXLOCK | O_SHLOCK))) {
+ localtrunc = 1;
+ flags &= ~O_TRUNC; /* Must do truncate ourselves */
+ }
+ if ((error = vn_open(&nd, flags, cmode)) != 0) {
+ if ((error == ENODEV || error == ENXIO) &&
+ p->p_dupfd >= 0 && /* XXX from fdopen */
+ (error =
+ dupfdopen(fdp, indx, p->p_dupfd, flags, error)) == 0) {
+ closef(fp, p);
+ *retval = indx;
+ return (0);
+ }
+ if (error == ERESTART)
+ error = EINTR;
+ fdremove(fdp, indx);
+ closef(fp, p);
+ return (error);
+ }
+ p->p_dupfd = 0;
+ vp = nd.ni_vp;
+ fp->f_flag = flags & FMASK;
+ fp->f_type = DTYPE_VNODE;
+ fp->f_ops = &vnops;
+ fp->f_data = (caddr_t)vp;
+ if (flags & (O_EXLOCK | O_SHLOCK)) {
+ lf.l_whence = SEEK_SET;
+ lf.l_start = 0;
+ lf.l_len = 0;
+ if (flags & O_EXLOCK)
+ lf.l_type = F_WRLCK;
+ else
+ lf.l_type = F_RDLCK;
+ type = F_FLOCK;
+ if ((flags & FNONBLOCK) == 0)
+ type |= F_WAIT;
+ VOP_UNLOCK(vp, 0, p);
+ error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type);
+ if (error) {
+ /* closef will vn_close the file for us. */
+ fdremove(fdp, indx);
+ closef(fp, p);
+ return (error);
+ }
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ fp->f_flag |= FHASLOCK;
+ }
+ if (localtrunc) {
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ if ((fp->f_flag & FWRITE) == 0)
+ error = EACCES;
+ else if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else if (vp->v_type == VDIR)
+ error = EISDIR;
+ else if ((error = vn_writechk(vp)) == 0) {
+ VATTR_NULL(&vattr);
+ vattr.va_size = 0;
+ error = VOP_SETATTR(vp, &vattr, fp->f_cred, p);
+ }
+ if (error) {
+ VOP_UNLOCK(vp, 0, p);
+ /* closef will close the file for us. */
+ fdremove(fdp, indx);
+ closef(fp, p);
+ return (error);
+ }
+ }
+ VOP_UNLOCK(vp, 0, p);
+ *retval = indx;
+ FILE_SET_MATURE(fp);
+ return (0);
+}
+
+/*
+ * Get file handle system call
+ */
+int
+sys_getfh(p, v, retval)
+ struct proc *p;
+ register void *v;
+ register_t *retval;
+{
+ register struct sys_getfh_args /* {
+ syscallarg(char *) fname;
+ syscallarg(fhandle_t *) fhp;
+ } */ *uap = v;
+ register struct vnode *vp;
+ fhandle_t fh;
+ int error;
+ struct nameidata nd;
+
+ /*
+ * Must be super user
+ */
+ error = suser(p->p_ucred, &p->p_acflag);
+ if (error)
+ return (error);
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, fname), p);
+ error = namei(&nd);
+ if (error)
+ return (error);
+ vp = nd.ni_vp;
+ bzero((caddr_t)&fh, sizeof(fh));
+ fh.fh_fsid = vp->v_mount->mnt_stat.f_fsid;
+ error = VFS_VPTOFH(vp, &fh.fh_fid);
+ vput(vp);
+ if (error)
+ return (error);
+ error = copyout((caddr_t)&fh, (caddr_t)SCARG(uap, fhp), sizeof (fh));
+ return (error);
+}
+
+/*
+ * Open a file given a file handle.
+ *
+ * Check permissions, allocate an open file structure,
+ * and call the device open routine if any.
+ */
+int
+sys_fhopen(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_fhopen_args /* {
+ syscallarg(const fhandle_t *) fhp;
+ syscallarg(int) flags;
+ } */ *uap = v;
+ struct filedesc *fdp = p->p_fd;
+ struct file *fp;
+ struct vnode *vp = NULL;
+ struct mount *mp;
+ struct ucred *cred = p->p_ucred;
+ int flags;
+ int type, indx, error=0;
+ struct flock lf;
+ struct vattr va;
+ fhandle_t fh;
+
+ /*
+ * Must be super user
+ */
+ if ((error = suser(p->p_ucred, &p->p_acflag)))
+ return (error);
+
+ flags = FFLAGS(SCARG(uap, flags));
+ if ((flags & (FREAD | FWRITE)) == 0)
+ return (EINVAL);
+ if ((flags & O_CREAT))
+ return (EINVAL);
+
+ if ((error = falloc(p, &fp, &indx)) != 0)
+ return (error);
+
+ if ((error = copyin(SCARG(uap, fhp), &fh, sizeof(fhandle_t))) != 0)
+ goto bad;
+
+ if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL) {
+ error = ESTALE;
+ goto bad;
+ }
+
+ if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)) != 0) {
+ vp = NULL; /* most likely unnecessary sanity for bad: */
+ goto bad;
+ }
+
+ /* Now do an effective vn_open */
+
+ if (vp->v_type == VSOCK) {
+ error = EOPNOTSUPP;
+ goto bad;
+ }
+ if (flags & FREAD) {
+ if ((error = VOP_ACCESS(vp, VREAD, cred, p)) != 0)
+ goto bad;
+ }
+ if (flags & (FWRITE | O_TRUNC)) {
+ if (vp->v_type == VDIR) {
+ error = EISDIR;
+ goto bad;
+ }
+ if ((error = vn_writechk(vp)) != 0 ||
+ (error = VOP_ACCESS(vp, VWRITE, cred, p)) != 0)
+ goto bad;
+ }
+ if (flags & O_TRUNC) {
+ VOP_UNLOCK(vp, 0, p); /* XXX */
+ VOP_LEASE(vp, p, cred, LEASE_WRITE);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p); /* XXX */
+ VATTR_NULL(&va);
+ va.va_size = 0;
+ if ((error = VOP_SETATTR(vp, &va, cred, p)) != 0)
+ goto bad;
+ }
+ if ((error = VOP_OPEN(vp, flags, cred, p)) != 0)
+ goto bad;
+ if (flags & FWRITE)
+ vp->v_writecount++;
+
+ /* done with modified vn_open, now finish what sys_open does. */
+
+ fp->f_flag = flags & FMASK;
+ fp->f_type = DTYPE_VNODE;
+ fp->f_ops = &vnops;
+ fp->f_data = (caddr_t)vp;
+ if (flags & (O_EXLOCK | O_SHLOCK)) {
+ lf.l_whence = SEEK_SET;
+ lf.l_start = 0;
+ lf.l_len = 0;
+ if (flags & O_EXLOCK)
+ lf.l_type = F_WRLCK;
+ else
+ lf.l_type = F_RDLCK;
+ type = F_FLOCK;
+ if ((flags & FNONBLOCK) == 0)
+ type |= F_WAIT;
+ VOP_UNLOCK(vp, 0, p);
+ error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type);
+ if (error) {
+ /* closef will vn_close the file for us. */
+ fdremove(fdp, indx);
+ closef(fp, p);
+ return (error);
+ }
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ fp->f_flag |= FHASLOCK;
+ }
+ VOP_UNLOCK(vp, 0, p);
+ *retval = indx;
+ FILE_SET_MATURE(fp);
+ return (0);
+
+bad:
+ fdremove(fdp, indx);
+ closef(fp, p);
+ if (vp != NULL)
+ vput(vp);
+ return (error);
+}
+
+/* ARGSUSED */
+int
+sys_fhstat(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_fhstat_args /* {
+ syscallarg(const fhandle_t *) fhp;
+ syscallarg(struct stat *) sb;
+ } */ *uap = v;
+ struct stat sb;
+ int error;
+ fhandle_t fh;
+ struct mount *mp;
+ struct vnode *vp;
+
+ /*
+ * Must be super user
+ */
+ if ((error = suser(p->p_ucred, &p->p_acflag)))
+ return (error);
+
+ if ((error = copyin(SCARG(uap, fhp), &fh, sizeof(fhandle_t))) != 0)
+ return (error);
+
+ if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
+ return (ESTALE);
+ if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
+ return (error);
+ error = vn_stat(vp, &sb, p);
+ vput(vp);
+ if (error)
+ return (error);
+ error = copyout(&sb, SCARG(uap, sb), sizeof(sb));
+ return (error);
+}
+
+/* ARGSUSED */
+int
+sys_fhstatfs(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_fhstatfs_args /*
+ syscallarg(const fhandle_t *) fhp;
+ syscallarg(struct statfs *) buf;
+ } */ *uap = v;
+ struct statfs sp;
+ fhandle_t fh;
+ struct mount *mp;
+ struct vnode *vp;
+ int error;
+
+ /*
+ * Must be super user
+ */
+ if ((error = suser(p->p_ucred, &p->p_acflag)))
+ return (error);
+
+ if ((error = copyin(SCARG(uap, fhp), &fh, sizeof(fhandle_t))) != 0)
+ return (error);
+
+ if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
+ return (ESTALE);
+ if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
+ return (error);
+ mp = vp->v_mount;
+ vput(vp);
+ if ((error = VFS_STATFS(mp, &sp, p)) != 0)
+ return (error);
+ sp.f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
+ return (copyout(&sp, SCARG(uap, buf), sizeof(sp)));
+}
+
+/*
+ * Create a special file.
+ */
+/* ARGSUSED */
+int
+sys_mknod(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_mknod_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) mode;
+ syscallarg(int) dev;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct vattr vattr;
+ int error;
+ int whiteout = 0;
+ struct nameidata nd;
+
+ if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
+ return (error);
+ if (p->p_fd->fd_rdir)
+ return (EINVAL);
+ NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ if (vp != NULL)
+ error = EEXIST;
+ else {
+ VATTR_NULL(&vattr);
+ vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
+ vattr.va_rdev = SCARG(uap, dev);
+ whiteout = 0;
+
+ switch (SCARG(uap, mode) & S_IFMT) {
+ case S_IFMT: /* used by badsect to flag bad sectors */
+ vattr.va_type = VBAD;
+ break;
+ case S_IFCHR:
+ vattr.va_type = VCHR;
+ break;
+ case S_IFBLK:
+ vattr.va_type = VBLK;
+ break;
+ case S_IFWHT:
+ whiteout = 1;
+ break;
+ default:
+ error = EINVAL;
+ break;
+ }
+ }
+ if (!error) {
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ if (whiteout) {
+ error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE);
+ if (error)
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ vput(nd.ni_dvp);
+ } else {
+ error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp,
+ &nd.ni_cnd, &vattr);
+ }
+ } else {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+ if (vp)
+ vrele(vp);
+ }
+ return (error);
+}
+
+/*
+ * Create a named pipe.
+ */
+/* ARGSUSED */
+int
+sys_mkfifo(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+#ifndef FIFO
+ return (EOPNOTSUPP);
+#else
+ register struct sys_mkfifo_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) mode;
+ } */ *uap = v;
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ if (nd.ni_vp != NULL) {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == nd.ni_vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+ vrele(nd.ni_vp);
+ return (EEXIST);
+ }
+ VATTR_NULL(&vattr);
+ vattr.va_type = VFIFO;
+ vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ return (VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr));
+#endif /* FIFO */
+}
+
+/*
+ * Make a hard file link.
+ */
+/* ARGSUSED */
+int
+sys_link(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_link_args /* {
+ syscallarg(char *) path;
+ syscallarg(char *) link;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct nameidata nd;
+ int error;
+ int flags;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+
+ flags = LOCKPARENT;
+ if (vp->v_type == VDIR) {
+ flags |= STRIPSLASHES;
+ }
+
+ NDINIT(&nd, CREATE, flags, UIO_USERSPACE, SCARG(uap, link), p);
+ if ((error = namei(&nd)) != 0)
+ goto out;
+ if (nd.ni_vp) {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == nd.ni_vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+ vrele(nd.ni_vp);
+ error = EEXIST;
+ goto out;
+ }
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
+out:
+ vrele(vp);
+ return (error);
+}
+
+/*
+ * Make a symbolic link.
+ */
+/* ARGSUSED */
+int
+sys_symlink(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_symlink_args /* {
+ syscallarg(char *) path;
+ syscallarg(char *) link;
+ } */ *uap = v;
+ struct vattr vattr;
+ char *path;
+ int error;
+ struct nameidata nd;
+
+ MALLOC(path, char *, MAXPATHLEN, M_NAMEI, M_WAITOK);
+ error = copyinstr(SCARG(uap, path), path, MAXPATHLEN, NULL);
+ if (error)
+ goto out;
+ NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, link), p);
+ if ((error = namei(&nd)) != 0)
+ goto out;
+ if (nd.ni_vp) {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == nd.ni_vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+ vrele(nd.ni_vp);
+ error = EEXIST;
+ goto out;
+ }
+ VATTR_NULL(&vattr);
+ vattr.va_mode = ACCESSPERMS &~ p->p_fd->fd_cmask;
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr, path);
+out:
+ FREE(path, M_NAMEI);
+ return (error);
+}
+
+/*
+ * Delete a whiteout from the filesystem.
+ */
+/* ARGSUSED */
+int
+sys_undelete(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_undelete_args /* {
+ syscallarg(char *) path;
+ } */ *uap = v;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, DELETE, LOCKPARENT|DOWHITEOUT, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ error = namei(&nd);
+ if (error)
+ return (error);
+
+ if (nd.ni_vp != NULLVP || !(nd.ni_cnd.cn_flags & ISWHITEOUT)) {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == nd.ni_vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+ if (nd.ni_vp)
+ vrele(nd.ni_vp);
+ return (EEXIST);
+ }
+
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ if ((error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE)) != 0)
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ vput(nd.ni_dvp);
+ return (error);
+}
+
+/*
+ * Delete a name from the filesystem.
+ */
+/* ARGSUSED */
+int
+sys_unlink(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_unlink_args /* {
+ syscallarg(char *) path;
+ } */ *uap = v;
+ register struct vnode *vp;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+
+ /*
+ * The root of a mounted filesystem cannot be deleted.
+ */
+ if (vp->v_flag & VROOT) {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+ vput(vp);
+ error = EBUSY;
+ goto out;
+ }
+
+ (void)uvm_vnp_uncache(vp);
+
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ error = VOP_REMOVE(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
+out:
+ return (error);
+}
+
+/*
+ * Reposition read/write file offset.
+ */
+int
+sys_lseek(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_lseek_args /* {
+ syscallarg(int) fd;
+ syscallarg(int) pad;
+ syscallarg(off_t) offset;
+ syscallarg(int) whence;
+ } */ *uap = v;
+ struct ucred *cred = p->p_ucred;
+ register struct filedesc *fdp = p->p_fd;
+ register struct file *fp;
+ struct vattr vattr;
+ struct vnode *vp;
+ int error, special;
+
+ if ((fp = fd_getfile(fdp, SCARG(uap, fd))) == NULL)
+ return (EBADF);
+ if (fp->f_type != DTYPE_VNODE)
+ return (ESPIPE);
+ vp = (struct vnode *)fp->f_data;
+ if (vp->v_type == VFIFO)
+ return (ESPIPE);
+ if (vp->v_type == VCHR)
+ special = 1;
+ else
+ special = 0;
+ switch (SCARG(uap, whence)) {
+ case SEEK_CUR:
+ if (!special && fp->f_offset + SCARG(uap, offset) < 0)
+ return (EINVAL);
+ fp->f_offset += SCARG(uap, offset);
+ break;
+ case SEEK_END:
+ error = VOP_GETATTR((struct vnode *)fp->f_data, &vattr,
+ cred, p);
+ if (error)
+ return (error);
+ if (!special && (off_t)vattr.va_size + SCARG(uap, offset) < 0)
+ return (EINVAL);
+ fp->f_offset = SCARG(uap, offset) + vattr.va_size;
+ break;
+ case SEEK_SET:
+ if (!special && SCARG(uap, offset) < 0)
+ return (EINVAL);
+ fp->f_offset = SCARG(uap, offset);
+ break;
+ default:
+ return (EINVAL);
+ }
+ *(off_t *)retval = fp->f_offset;
+ return (0);
+}
+
+/*
+ * Check access permissions.
+ */
+int
+sys_access(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_access_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) flags;
+ } */ *uap = v;
+ register struct ucred *cred = p->p_ucred;
+ register struct vnode *vp;
+ int error, flags, t_gid, t_uid;
+ struct nameidata nd;
+
+ if (SCARG(uap, flags) & ~(R_OK | W_OK | X_OK))
+ return (EINVAL);
+ t_uid = cred->cr_uid;
+ t_gid = cred->cr_gid;
+ cred->cr_uid = p->p_cred->p_ruid;
+ cred->cr_gid = p->p_cred->p_rgid;
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ goto out1;
+ vp = nd.ni_vp;
+
+ /* Flags == 0 means only check for existence. */
+ if (SCARG(uap, flags)) {
+ flags = 0;
+ if (SCARG(uap, flags) & R_OK)
+ flags |= VREAD;
+ if (SCARG(uap, flags) & W_OK)
+ flags |= VWRITE;
+ if (SCARG(uap, flags) & X_OK)
+ flags |= VEXEC;
+ if ((flags & VWRITE) == 0 || (error = vn_writechk(vp)) == 0)
+ error = VOP_ACCESS(vp, flags, cred, p);
+ }
+ vput(vp);
+out1:
+ cred->cr_uid = t_uid;
+ cred->cr_gid = t_gid;
+ return (error);
+}
+
+/*
+ * Get file status; this version follows links.
+ */
+/* ARGSUSED */
+int
+sys_stat(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_stat_args /* {
+ syscallarg(char *) path;
+ syscallarg(struct stat *) ub;
+ } */ *uap = v;
+ struct stat sb;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ error = vn_stat(nd.ni_vp, &sb, p);
+ vput(nd.ni_vp);
+ if (error)
+ return (error);
+ /* Don't let non-root see generation numbers (for NFS security) */
+ if (suser(p->p_ucred, &p->p_acflag))
+ sb.st_gen = 0;
+ error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
+ return (error);
+}
+
+/*
+ * Get file status; this version does not follow links.
+ */
+/* ARGSUSED */
+int
+sys_lstat(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_lstat_args /* {
+ syscallarg(char *) path;
+ syscallarg(struct stat *) ub;
+ } */ *uap = v;
+ struct stat sb;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ error = vn_stat(nd.ni_vp, &sb, p);
+ vput(nd.ni_vp);
+ if (error)
+ return (error);
+ /* Don't let non-root see generation numbers (for NFS security) */
+ if (suser(p->p_ucred, &p->p_acflag))
+ sb.st_gen = 0;
+ error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
+ return (error);
+}
+
+/*
+ * Get configurable pathname variables.
+ */
+/* ARGSUSED */
+int
+sys_pathconf(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_pathconf_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) name;
+ } */ *uap = v;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ error = VOP_PATHCONF(nd.ni_vp, SCARG(uap, name), retval);
+ vput(nd.ni_vp);
+ return (error);
+}
+
+/*
+ * Return target name of a symbolic link.
+ */
+/* ARGSUSED */
+int
+sys_readlink(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_readlink_args /* {
+ syscallarg(char *) path;
+ syscallarg(char *) buf;
+ syscallarg(size_t) count;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct iovec aiov;
+ struct uio auio;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ if (vp->v_type != VLNK)
+ error = EINVAL;
+ else {
+ aiov.iov_base = SCARG(uap, buf);
+ aiov.iov_len = SCARG(uap, count);
+ auio.uio_iov = &aiov;
+ auio.uio_iovcnt = 1;
+ auio.uio_offset = 0;
+ auio.uio_rw = UIO_READ;
+ auio.uio_segflg = UIO_USERSPACE;
+ auio.uio_procp = p;
+ auio.uio_resid = SCARG(uap, count);
+ error = VOP_READLINK(vp, &auio, p->p_ucred);
+ }
+ vput(vp);
+ *retval = SCARG(uap, count) - auio.uio_resid;
+ return (error);
+}
+
+/*
+ * Change flags of a file given a path name.
+ */
+/* ARGSUSED */
+int
+sys_chflags(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_chflags_args /* {
+ syscallarg(char *) path;
+ syscallarg(unsigned int) flags;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else if (SCARG(uap, flags) == VNOVAL)
+ error = EINVAL;
+ else {
+ if (suser(p->p_ucred, &p->p_acflag)) {
+ if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
+ goto out;
+ if (vattr.va_type == VCHR || vattr.va_type == VBLK) {
+ error = EINVAL;
+ goto out;
+ }
+ }
+ VATTR_NULL(&vattr);
+ vattr.va_flags = SCARG(uap, flags);
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+out:
+ vput(vp);
+ return (error);
+}
+
+/*
+ * Change flags of a file given a file descriptor.
+ */
+/* ARGSUSED */
+int
+sys_fchflags(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_fchflags_args /* {
+ syscallarg(int) fd;
+ syscallarg(unsigned int) flags;
+ } */ *uap = v;
+ struct vattr vattr;
+ struct vnode *vp;
+ struct file *fp;
+ int error;
+
+ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+ vp = (struct vnode *)fp->f_data;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else if (SCARG(uap, flags) == VNOVAL)
+ error = EINVAL;
+ else {
+ if (suser(p->p_ucred, &p->p_acflag)) {
+ if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p))
+ != 0)
+ goto out;
+ if (vattr.va_type == VCHR || vattr.va_type == VBLK) {
+ error = EINVAL;
+ goto out;
+ }
+ }
+ VATTR_NULL(&vattr);
+ vattr.va_flags = SCARG(uap, flags);
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+out:
+ VOP_UNLOCK(vp, 0, p);
+ FRELE(fp);
+ return (error);
+}
+
+/*
+ * Change mode of a file given path name.
+ */
+/* ARGSUSED */
+int
+sys_chmod(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_chmod_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) mode;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+
+ if (SCARG(uap, mode) & ~(S_IFMT | ALLPERMS))
+ return (EINVAL);
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+ VATTR_NULL(&vattr);
+ vattr.va_mode = SCARG(uap, mode) & ALLPERMS;
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+ vput(vp);
+ return (error);
+}
+
+/*
+ * Change mode of a file given a file descriptor.
+ */
+/* ARGSUSED */
+int
+sys_fchmod(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_fchmod_args /* {
+ syscallarg(int) fd;
+ syscallarg(int) mode;
+ } */ *uap = v;
+ struct vattr vattr;
+ struct vnode *vp;
+ struct file *fp;
+ int error;
+
+ if (SCARG(uap, mode) & ~(S_IFMT | ALLPERMS))
+ return (EINVAL);
+
+ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+ vp = (struct vnode *)fp->f_data;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+ VATTR_NULL(&vattr);
+ vattr.va_mode = SCARG(uap, mode) & ALLPERMS;
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+ VOP_UNLOCK(vp, 0, p);
+ FRELE(fp);
+ return (error);
+}
+
+/*
+ * Set ownership given a path name.
+ */
+/* ARGSUSED */
+int
+sys_chown(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_chown_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) uid;
+ syscallarg(int) gid;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+ u_short mode;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+ if ((SCARG(uap, uid) != -1 || SCARG(uap, gid) != -1) &&
+ (suser(p->p_ucred, &p->p_acflag) || suid_clear)) {
+ error = VOP_GETATTR(vp, &vattr, p->p_ucred, p);
+ if (error)
+ goto out;
+ mode = vattr.va_mode & ~(VSUID | VSGID);
+ if (mode == vattr.va_mode)
+ mode = VNOVAL;
+ }
+ else
+ mode = VNOVAL;
+ VATTR_NULL(&vattr);
+ vattr.va_uid = SCARG(uap, uid);
+ vattr.va_gid = SCARG(uap, gid);
+ vattr.va_mode = mode;
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+out:
+ vput(vp);
+ return (error);
+}
+
+/*
+ * Set ownership given a path name, without following links.
+ */
+/* ARGSUSED */
+int
+sys_lchown(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_lchown_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) uid;
+ syscallarg(int) gid;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+ u_short mode;
+
+ NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+ if ((SCARG(uap, uid) != -1 || SCARG(uap, gid) != -1) &&
+ (suser(p->p_ucred, &p->p_acflag) || suid_clear)) {
+ error = VOP_GETATTR(vp, &vattr, p->p_ucred, p);
+ if (error)
+ goto out;
+ mode = vattr.va_mode & ~(VSUID | VSGID);
+ if (mode == vattr.va_mode)
+ mode = VNOVAL;
+ }
+ else
+ mode = VNOVAL;
+ VATTR_NULL(&vattr);
+ vattr.va_uid = SCARG(uap, uid);
+ vattr.va_gid = SCARG(uap, gid);
+ vattr.va_mode = mode;
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+out:
+ vput(vp);
+ return (error);
+}
+
+/*
+ * Set ownership given a file descriptor.
+ */
+/* ARGSUSED */
+int
+sys_fchown(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_fchown_args /* {
+ syscallarg(int) fd;
+ syscallarg(int) uid;
+ syscallarg(int) gid;
+ } */ *uap = v;
+ struct vnode *vp;
+ struct vattr vattr;
+ int error;
+ struct file *fp;
+ u_short mode;
+
+ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+ vp = (struct vnode *)fp->f_data;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+ if ((SCARG(uap, uid) != -1 || SCARG(uap, gid) != -1) &&
+ (suser(p->p_ucred, &p->p_acflag) || suid_clear)) {
+ error = VOP_GETATTR(vp, &vattr, p->p_ucred, p);
+ if (error)
+ goto out;
+ mode = vattr.va_mode & ~(VSUID | VSGID);
+ if (mode == vattr.va_mode)
+ mode = VNOVAL;
+ } else
+ mode = VNOVAL;
+ VATTR_NULL(&vattr);
+ vattr.va_uid = SCARG(uap, uid);
+ vattr.va_gid = SCARG(uap, gid);
+ vattr.va_mode = mode;
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+out:
+ VOP_UNLOCK(vp, 0, p);
+ FRELE(fp);
+ return (error);
+}
+
+/*
+ * Set the access and modification times given a path name.
+ */
+/* ARGSUSED */
+int
+sys_utimes(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_utimes_args /* {
+ syscallarg(char *) path;
+ syscallarg(struct timeval *) tptr;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct timeval tv[2];
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+
+ VATTR_NULL(&vattr);
+ if (SCARG(uap, tptr) == NULL) {
+ microtime(&tv[0]);
+ tv[1] = tv[0];
+ vattr.va_vaflags |= VA_UTIMES_NULL;
+ } else {
+ error = copyin((caddr_t)SCARG(uap, tptr), (caddr_t)tv,
+ sizeof (tv));
+ if (error)
+ return (error);
+ /* XXX workaround timeval matching the VFS constant VNOVAL */
+ if (tv[0].tv_sec == VNOVAL)
+ tv[0].tv_sec = VNOVAL - 1;
+ if (tv[1].tv_sec == VNOVAL)
+ tv[1].tv_sec = VNOVAL - 1;
+ }
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+ vattr.va_atime.tv_sec = tv[0].tv_sec;
+ vattr.va_atime.tv_nsec = tv[0].tv_usec * 1000;
+ vattr.va_mtime.tv_sec = tv[1].tv_sec;
+ vattr.va_mtime.tv_nsec = tv[1].tv_usec * 1000;
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+ vput(vp);
+ return (error);
+}
+
+
+/*
+ * Set the access and modification times given a file descriptor.
+ */
+/* ARGSUSED */
+int
+sys_futimes(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_futimes_args /* {
+ syscallarg(int) fd;
+ syscallarg(struct timeval *) tptr;
+ } */ *uap = v;
+ struct vnode *vp;
+ struct timeval tv[2];
+ struct vattr vattr;
+ int error;
+ struct file *fp;
+
+ VATTR_NULL(&vattr);
+ if (SCARG(uap, tptr) == NULL) {
+ microtime(&tv[0]);
+ tv[1] = tv[0];
+ vattr.va_vaflags |= VA_UTIMES_NULL;
+ } else {
+ error = copyin((caddr_t)SCARG(uap, tptr), (caddr_t)tv,
+ sizeof (tv));
+ if (error)
+ return (error);
+ /* XXX workaround timeval matching the VFS constant VNOVAL */
+ if (tv[0].tv_sec == VNOVAL)
+ tv[0].tv_sec = VNOVAL - 1;
+ if (tv[1].tv_sec == VNOVAL)
+ tv[1].tv_sec = VNOVAL - 1;
+ }
+ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+ vp = (struct vnode *)fp->f_data;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+ vattr.va_atime.tv_sec = tv[0].tv_sec;
+ vattr.va_atime.tv_nsec = tv[0].tv_usec * 1000;
+ vattr.va_mtime.tv_sec = tv[1].tv_sec;
+ vattr.va_mtime.tv_nsec = tv[1].tv_usec * 1000;
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+ VOP_UNLOCK(vp, 0, p);
+ FRELE(fp);
+ return (error);
+}
+
+/*
+ * Truncate a file given its path name.
+ */
+/* ARGSUSED */
+int
+sys_truncate(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_truncate_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) pad;
+ syscallarg(off_t) length;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_type == VDIR)
+ error = EISDIR;
+ else if ((error = vn_writechk(vp)) == 0 &&
+ (error = VOP_ACCESS(vp, VWRITE, p->p_ucred, p)) == 0) {
+ VATTR_NULL(&vattr);
+ vattr.va_size = SCARG(uap, length);
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+ vput(vp);
+ return (error);
+}
+
+/*
+ * Truncate a file given a file descriptor.
+ */
+/* ARGSUSED */
+int
+sys_ftruncate(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_ftruncate_args /* {
+ syscallarg(int) fd;
+ syscallarg(int) pad;
+ syscallarg(off_t) length;
+ } */ *uap = v;
+ struct vattr vattr;
+ struct vnode *vp;
+ struct file *fp;
+ int error;
+
+ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+ if ((fp->f_flag & FWRITE) == 0) {
+ error = EINVAL;
+ goto bad;
+ }
+ vp = (struct vnode *)fp->f_data;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_type == VDIR)
+ error = EISDIR;
+ else if ((error = vn_writechk(vp)) == 0) {
+ VATTR_NULL(&vattr);
+ vattr.va_size = SCARG(uap, length);
+ error = VOP_SETATTR(vp, &vattr, fp->f_cred, p);
+ }
+ VOP_UNLOCK(vp, 0, p);
+bad:
+ FRELE(fp);
+ return (error);
+}
+
+/*
+ * Sync an open file.
+ */
+/* ARGSUSED */
+int
+sys_fsync(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_fsync_args /* {
+ syscallarg(int) fd;
+ } */ *uap = v;
+ struct vnode *vp;
+ struct file *fp;
+ int error;
+
+ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+ vp = (struct vnode *)fp->f_data;
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ error = VOP_FSYNC(vp, fp->f_cred, MNT_WAIT, p);
+#ifdef FFS_SOFTUPDATES
+ if (error == 0 && vp->v_mount && (vp->v_mount->mnt_flag & MNT_SOFTDEP))
+ error = softdep_fsync(vp);
+#endif
+
+ VOP_UNLOCK(vp, 0, p);
+ FRELE(fp);
+ return (error);
+}
+
+/*
+ * Rename files. Source and destination must either both be directories,
+ * or both not be directories. If target is a directory, it must be empty.
+ */
+/* ARGSUSED */
+int
+sys_rename(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_rename_args /* {
+ syscallarg(char *) from;
+ syscallarg(char *) to;
+ } */ *uap = v;
+ register struct vnode *tvp, *fvp, *tdvp;
+ struct nameidata fromnd, tond;
+ int error;
+ int flags;
+
+ NDINIT(&fromnd, DELETE, WANTPARENT | SAVESTART, UIO_USERSPACE,
+ SCARG(uap, from), p);
+ if ((error = namei(&fromnd)) != 0)
+ return (error);
+ fvp = fromnd.ni_vp;
+
+ flags = LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART;
+ /*
+ * rename("foo/", "bar/"); is OK
+ */
+ if (fvp->v_type == VDIR)
+ flags |= STRIPSLASHES;
+
+ NDINIT(&tond, RENAME, flags,
+ UIO_USERSPACE, SCARG(uap, to), p);
+ if ((error = namei(&tond)) != 0) {
+ VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
+ vrele(fromnd.ni_dvp);
+ vrele(fvp);
+ goto out1;
+ }
+ tdvp = tond.ni_dvp;
+ tvp = tond.ni_vp;
+ if (tvp != NULL) {
+ if (fvp->v_type == VDIR && tvp->v_type != VDIR) {
+ error = ENOTDIR;
+ goto out;
+ } else if (fvp->v_type != VDIR && tvp->v_type == VDIR) {
+ error = EISDIR;
+ goto out;
+ }
+ }
+ if (fvp == tdvp)
+ error = EINVAL;
+ /*
+ * If source is the same as the destination (that is the
+ * same inode number)
+ */
+ if (fvp == tvp)
+ error = -1;
+out:
+ if (!error) {
+ VOP_LEASE(tdvp, p, p->p_ucred, LEASE_WRITE);
+ if (fromnd.ni_dvp != tdvp)
+ VOP_LEASE(fromnd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ if (tvp) {
+ (void)uvm_vnp_uncache(tvp);
+ VOP_LEASE(tvp, p, p->p_ucred, LEASE_WRITE);
+ }
+ error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd,
+ tond.ni_dvp, tond.ni_vp, &tond.ni_cnd);
+ } else {
+ VOP_ABORTOP(tond.ni_dvp, &tond.ni_cnd);
+ if (tdvp == tvp)
+ vrele(tdvp);
+ else
+ vput(tdvp);
+ if (tvp)
+ vput(tvp);
+ VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
+ vrele(fromnd.ni_dvp);
+ vrele(fvp);
+ }
+ vrele(tond.ni_startdir);
+ FREE(tond.ni_cnd.cn_pnbuf, M_NAMEI);
+out1:
+ if (fromnd.ni_startdir)
+ vrele(fromnd.ni_startdir);
+ FREE(fromnd.ni_cnd.cn_pnbuf, M_NAMEI);
+ if (error == -1)
+ return (0);
+ return (error);
+}
+
+/*
+ * Make a directory file.
+ */
+/* ARGSUSED */
+int
+sys_mkdir(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_mkdir_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) mode;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, CREATE, LOCKPARENT | STRIPSLASHES,
+ UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ if (vp != NULL) {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+ vrele(vp);
+ return (EEXIST);
+ }
+ VATTR_NULL(&vattr);
+ vattr.va_type = VDIR;
+ vattr.va_mode = (SCARG(uap, mode) & ACCESSPERMS) &~ p->p_fd->fd_cmask;
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr);
+ if (!error)
+ vput(nd.ni_vp);
+ return (error);
+}
+
+/*
+ * Remove a directory file.
+ */
+/* ARGSUSED */
+int
+sys_rmdir(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_rmdir_args /* {
+ syscallarg(char *) path;
+ } */ *uap = v;
+ register struct vnode *vp;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ if (vp->v_type != VDIR) {
+ error = ENOTDIR;
+ goto out;
+ }
+ /*
+ * No rmdir "." please.
+ */
+ if (nd.ni_dvp == vp) {
+ error = EBUSY;
+ goto out;
+ }
+ /*
+ * The root of a mounted filesystem cannot be deleted.
+ */
+ if (vp->v_flag & VROOT)
+ error = EBUSY;
+out:
+ if (!error) {
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
+ } else {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+ vput(vp);
+ }
+ return (error);
+}
+
+/*
+ * Read a block of directory entries in a file system independent format.
+ */
+int
+sys_getdirentries(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_getdirentries_args /* {
+ syscallarg(int) fd;
+ syscallarg(char *) buf;
+ syscallarg(int) count;
+ syscallarg(long *) basep;
+ } */ *uap = v;
+ struct vnode *vp;
+ struct file *fp;
+ struct uio auio;
+ struct iovec aiov;
+ long loff;
+ int error, eofflag;
+
+ if (SCARG(uap, count) < 0)
+ return EINVAL;
+ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+ if ((fp->f_flag & FREAD) == 0) {
+ error = EBADF;
+ goto bad;
+ }
+ vp = (struct vnode *)fp->f_data;
+unionread:
+ if (vp->v_type != VDIR) {
+ error = EINVAL;
+ goto bad;
+ }
+ aiov.iov_base = SCARG(uap, buf);
+ aiov.iov_len = SCARG(uap, count);
+ auio.uio_iov = &aiov;
+ auio.uio_iovcnt = 1;
+ auio.uio_rw = UIO_READ;
+ auio.uio_segflg = UIO_USERSPACE;
+ auio.uio_procp = p;
+ auio.uio_resid = SCARG(uap, count);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ loff = auio.uio_offset = fp->f_offset;
+ error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, 0, 0);
+ fp->f_offset = auio.uio_offset;
+ VOP_UNLOCK(vp, 0, p);
+ if (error)
+ goto bad;
+ if ((SCARG(uap, count) == auio.uio_resid) &&
+ union_check_p &&
+ (union_check_p(p, &vp, fp, auio, &error) != 0))
+ goto unionread;
+ if (error)
+ goto bad;
+
+ if ((SCARG(uap, count) == auio.uio_resid) &&
+ (vp->v_flag & VROOT) &&
+ (vp->v_mount->mnt_flag & MNT_UNION)) {
+ struct vnode *tvp = vp;
+ vp = vp->v_mount->mnt_vnodecovered;
+ VREF(vp);
+ fp->f_data = (caddr_t) vp;
+ fp->f_offset = 0;
+ vrele(tvp);
+ goto unionread;
+ }
+ error = copyout((caddr_t)&loff, (caddr_t)SCARG(uap, basep),
+ sizeof(long));
+ *retval = SCARG(uap, count) - auio.uio_resid;
+bad:
+ FRELE(fp);
+ return (error);
+}
+
+/*
+ * Set the mode mask for creation of filesystem nodes.
+ */
+int
+sys_umask(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_umask_args /* {
+ syscallarg(int) newmask;
+ } */ *uap = v;
+ register struct filedesc *fdp;
+
+ fdp = p->p_fd;
+ *retval = fdp->fd_cmask;
+ fdp->fd_cmask = SCARG(uap, newmask) & ACCESSPERMS;
+ return (0);
+}
+
+/*
+ * Void all references to file by ripping underlying filesystem
+ * away from vnode.
+ */
+/* ARGSUSED */
+int
+sys_revoke(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ register struct sys_revoke_args /* {
+ syscallarg(char *) path;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
+ goto out;
+ if (p->p_ucred->cr_uid != vattr.va_uid &&
+ (error = suser(p->p_ucred, &p->p_acflag)))
+ goto out;
+ if (vp->v_usecount > 1 || (vp->v_flag & (VALIASED | VLAYER)))
+ VOP_REVOKE(vp, REVOKEALL);
+out:
+ vrele(vp);
+ return (error);
+}
+
+/*
+ * Convert a user file descriptor to a kernel file entry.
+ *
+ * On return *fpp is FREF:ed.
+ */
+int
+getvnode(fdp, fd, fpp)
+ struct filedesc *fdp;
+ struct file **fpp;
+ int fd;
+{
+ struct file *fp;
+
+ if ((fp = fd_getfile(fdp, fd)) == NULL)
+ return (EBADF);
+ if (fp->f_type != DTYPE_VNODE)
+ return (EINVAL);
+ FREF(fp);
+ *fpp = fp;
+
+ return (0);
+}
+
+/*
+ * Positional read system call.
+ */
+int
+sys_pread(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_pread_args /* {
+ syscallarg(int) fd;
+ syscallarg(void *) buf;
+ syscallarg(size_t) nbyte;
+ syscallarg(int) pad;
+ syscallarg(off_t) offset;
+ } */ *uap = v;
+ struct filedesc *fdp = p->p_fd;
+ struct file *fp;
+ struct vnode *vp;
+ off_t offset;
+ int fd = SCARG(uap, fd);
+
+ if ((fp = fd_getfile(fdp, fd)) == NULL)
+ return (EBADF);
+ if ((fp->f_flag & FREAD) == 0)
+ return (EBADF);
+
+ vp = (struct vnode *)fp->f_data;
+ if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
+ return (ESPIPE);
+ }
+
+ offset = SCARG(uap, offset);
+
+ FREF(fp);
+
+ /* dofileread() will FRELE the descriptor for us */
+ return (dofileread(p, fd, fp, SCARG(uap, buf), SCARG(uap, nbyte),
+ &offset, retval));
+}
+
+/*
+ * Positional scatter read system call.
+ */
+int
+sys_preadv(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_preadv_args /* {
+ syscallarg(int) fd;
+ syscallarg(const struct iovec *) iovp;
+ syscallarg(int) iovcnt;
+ syscallarg(int) pad;
+ syscallarg(off_t) offset;
+ } */ *uap = v;
+ struct filedesc *fdp = p->p_fd;
+ struct file *fp;
+ struct vnode *vp;
+ off_t offset;
+ int fd = SCARG(uap, fd);
+
+ if ((fp = fd_getfile(fdp, fd)) == NULL)
+ return (EBADF);
+ if ((fp->f_flag & FREAD) == 0)
+ return (EBADF);
+
+ vp = (struct vnode *)fp->f_data;
+ if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
+ return (ESPIPE);
+ }
+
+ FREF(fp);
+
+ offset = SCARG(uap, offset);
+
+ /* dofilereadv() will FRELE the descriptor for us */
+ return (dofilereadv(p, fd, fp, SCARG(uap, iovp), SCARG(uap, iovcnt),
+ &offset, retval));
+}
+
+/*
+ * Positional write system call.
+ */
+int
+sys_pwrite(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_pwrite_args /* {
+ syscallarg(int) fd;
+ syscallarg(const void *) buf;
+ syscallarg(size_t) nbyte;
+ syscallarg(int) pad;
+ syscallarg(off_t) offset;
+ } */ *uap = v;
+ struct filedesc *fdp = p->p_fd;
+ struct file *fp;
+ struct vnode *vp;
+ off_t offset;
+ int fd = SCARG(uap, fd);
+
+ if ((fp = fd_getfile(fdp, fd)) == NULL)
+ return (EBADF);
+ if ((fp->f_flag & FWRITE) == 0)
+ return (EBADF);
+
+ vp = (struct vnode *)fp->f_data;
+ if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
+ return (ESPIPE);
+ }
+
+ FREF(fp);
+
+ offset = SCARG(uap, offset);
+
+ /* dofilewrite() will FRELE the descriptor for us */
+ return (dofilewrite(p, fd, fp, SCARG(uap, buf), SCARG(uap, nbyte),
+ &offset, retval));
+}
+
+
+/*
+ * Positional gather write system call.
+ */
+int
+sys_pwritev(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_pwritev_args /* {
+ syscallarg(int) fd;
+ syscallarg(const struct iovec *) iovp;
+ syscallarg(int) iovcnt;
+ syscallarg(int) pad;
+ syscallarg(off_t) offset;
+ } */ *uap = v;
+ struct filedesc *fdp = p->p_fd;
+ struct file *fp;
+ struct vnode *vp;
+ off_t offset;
+ int fd = SCARG(uap, fd);
+
+ if ((fp = fd_getfile(fdp, fd)) == NULL)
+ return (EBADF);
+ if ((fp->f_flag & FWRITE) == 0)
+ return (EBADF);
+
+ vp = (struct vnode *)fp->f_data;
+ if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
+ return (ESPIPE);
+ }
+
+ FREF(fp);
+
+ offset = SCARG(uap, offset);
+
+ /* dofilewritev() will FRELE the descriptor for us */
+ return (dofilewritev(p, fd, fp, SCARG(uap, iovp), SCARG(uap, iovcnt),
+ &offset, retval));
+}
+
+#ifdef UFS_EXTATTR
+/*
+ * Syscall to push extended attribute configuration information into the
+ * VFS. Accepts a path, which it converts to a mountpoint, as well as
+ * a command (int cmd), and attribute name and misc data. For now, the
+ * attribute name is left in userspace for consumption by the VFS_op.
+ * It will probably be changed to be copied into sysspace by the
+ * syscall in the future, once issues with various consumers of the
+ * attribute code have raised their hands.
+ *
+ * Currently this is used only by UFS Extended Attributes.
+ */
+int
+sys_extattrctl(struct proc *p, void *v, register_t *reval)
+{
+ struct sys_extattrctl_args /* {
+ syscallarg(const char *) path;
+ syscallarg(int) cmd;
+ syscallarg(const char *) filename;
+ syscallarg(int) attrnamespace;
+ syscallarg(const char *) attrname;
+ } */ *uap = v;
+ struct vnode *filename_vp;
+ struct nameidata nd;
+ struct mount *mp;
+ char attrname[EXTATTR_MAXNAMELEN];
+ int error;
+
+ /*
+ * SCARG(uap, attrname) not always defined. We check again later
+ * when we invoke the VFS call so as to pass in NULL there if needed.
+ */
+ if (SCARG(uap, attrname) != NULL) {
+ error = copyinstr(SCARG(uap, attrname), attrname,
+ EXTATTR_MAXNAMELEN, NULL);
+ if (error)
+ return (error);
+ }
+
+ /*
+ * SCARG(uap, filename) not always defined. If it is, grab
+ * a vnode lock, which VFS_EXTATTRCTL() will later release.
+ */
+ filename_vp = NULL;
+ if (SCARG(uap, filename) != NULL) {
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, filename), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+ filename_vp = nd.ni_vp;
+ }
+
+ /* SCARG(uap, path) always defined. */
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0) {
+ if (filename_vp != NULL)
+ vput(filename_vp);
+ return (error);
+ }
+
+ mp = nd.ni_vp->v_mount;
+ if (error) {
+ if (filename_vp != NULL)
+ vput(filename_vp);
+ return (error);
+ }
+
+ if (SCARG(uap, attrname) != NULL) {
+ error = VFS_EXTATTRCTL(mp, SCARG(uap, cmd), filename_vp,
+ SCARG(uap, attrnamespace), attrname, p);
+ } else {
+ error = VFS_EXTATTRCTL(mp, SCARG(uap, cmd), filename_vp,
+ SCARG(uap, attrnamespace), NULL, p);
+ }
+
+ /*
+ * VFS_EXTATTRCTL will have unlocked, but not de-ref'd,
+ * filename_vp, so vrele it if it is defined.
+ */
+ if (filename_vp != NULL)
+ vrele(filename_vp);
+
+ return (error);
+}
+
+/*-
+ * Set a named extended attribute on a file or directory
+ *
+ * Arguments: unlocked vnode "vp", attribute namespace "attrnamespace",
+ * kernelspace string pointer "attrname", userspace buffer
+ * pointer "data", buffer length "nbytes", thread "td".
+ * Returns: 0 on success, an error number otherwise
+ * Locks: none
+ * References: vp must be a valid reference for the duration of the call
+ */
+static int
+extattr_set_vp(struct vnode *vp, int attrnamespace, const char *attrname,
+ void *data, size_t nbytes, struct proc *p, register_t *retval)
+{
+ struct uio auio;
+ struct iovec aiov;
+ ssize_t cnt;
+ int error;
+
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+
+ aiov.iov_base = data;
+ aiov.iov_len = nbytes;
+ auio.uio_iov = &aiov;
+ auio.uio_iovcnt = 1;
+ auio.uio_offset = 0;
+ if (nbytes > INT_MAX) {
+ error = EINVAL;
+ goto done;
+ }
+ auio.uio_resid = nbytes;
+ auio.uio_rw = UIO_WRITE;
+ auio.uio_segflg = UIO_USERSPACE;
+ auio.uio_procp = p;
+ cnt = nbytes;
+
+ error = VOP_SETEXTATTR(vp, attrnamespace, attrname, &auio,
+ p->p_ucred, p);
+ cnt -= auio.uio_resid;
+ retval[0] = cnt;
+
+done:
+ VOP_UNLOCK(vp, 0, p);
+ return (error);
+}
+
+int
+sys_extattr_set_file(struct proc *p, void *v, register_t *retval)
+{
+ struct sys_extattr_set_file_args /* {
+ syscallarg(const char *) path;
+ syscallarg(int) attrnamespace;
+ syscallarg(const char *) attrname;
+ syscallarg(void *) data;
+ syscallarg(size_t) nbytes;
+ } */ *uap = v;
+ struct nameidata nd;
+ char attrname[EXTATTR_MAXNAMELEN];
+ int error;
+
+ error = copyinstr(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN,
+ NULL);
+ if (error)
+ return (error);
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+
+ error = extattr_set_vp(nd.ni_vp, SCARG(uap, attrnamespace), attrname,
+ SCARG(uap, data), SCARG(uap, nbytes), p, retval);
+
+ vrele(nd.ni_vp);
+ return (error);
+}
+
+int
+sys_extattr_set_fd(struct proc *p, void *v, register_t *retval)
+{
+ struct sys_extattr_set_fd_args /* {
+ syscallarg(int) fd;
+ syscallarg(int) attrnamespace;
+ syscallarg(const char *) attrname;
+ syscallarg(struct iovec *) iovp;
+ syscallarg(int) iovcnt;
+ } */ *uap = v;
+ struct file *fp;
+ char attrname[EXTATTR_MAXNAMELEN];
+ int error;
+
+ error = copyinstr(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN,
+ NULL);
+ if (error)
+ return (error);
+
+ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+
+ error = extattr_set_vp((struct vnode *)fp->f_data,
+ SCARG(uap, attrnamespace), attrname, SCARG(uap, data),
+ SCARG(uap, nbytes), p, retval);
+ FRELE(fp);
+
+ return (error);
+}
+
+/*-
+ * Get a named extended attribute on a file or directory
+ *
+ * Arguments: unlocked vnode "vp", attribute namespace "attrnamespace",
+ * kernelspace string pointer "attrname", userspace buffer
+ * pointer "data", buffer length "nbytes", thread "td".
+ * Returns: 0 on success, an error number otherwise
+ * Locks: none
+ * References: vp must be a valid reference for the duration of the call
+ */
+static int
+extattr_get_vp(struct vnode *vp, int attrnamespace, const char *attrname,
+ void *data, size_t nbytes, struct proc *p, register_t *retval)
+{
+ struct uio auio;
+ struct iovec aiov;
+ ssize_t cnt;
+ size_t size;
+ int error;
+
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_READ);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+
+ /*
+ * Slightly unusual semantics: if the user provides a NULL data
+ * pointer, they don't want to receive the data, just the
+ * maximum read length.
+ */
+ if (data != NULL) {
+ aiov.iov_base = data;
+ aiov.iov_len = nbytes;
+ auio.uio_iov = &aiov;
+ auio.uio_offset = 0;
+ if (nbytes > INT_MAX) {
+ error = EINVAL;
+ goto done;
+ }
+ auio.uio_resid = nbytes;
+ auio.uio_rw = UIO_READ;
+ auio.uio_segflg = UIO_USERSPACE;
+ auio.uio_procp = p;
+ cnt = nbytes;
+ error = VOP_GETEXTATTR(vp, attrnamespace, attrname, &auio,
+ NULL, p->p_ucred, p);
+ cnt -= auio.uio_resid;
+ retval[0] = cnt;
+ } else {
+ error = VOP_GETEXTATTR(vp, attrnamespace, attrname, NULL,
+ &size, p->p_ucred, p);
+ retval[0] = size;
+ }
+done:
+ VOP_UNLOCK(vp, 0, p);
+ return (error);
+}
+
+int
+sys_extattr_get_file(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_extattr_get_file_args /* {
+ syscallarg(const char *) path;
+ syscallarg(int) attrnamespace;
+ syscallarg(const char *) attrname;
+ syscallarg(void *) data;
+ syscallarg(size_t) nbytes;
+ } */ *uap = v;
+ struct nameidata nd;
+ char attrname[EXTATTR_MAXNAMELEN];
+ int error;
+
+ error = copyinstr(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN,
+ NULL);
+ if (error)
+ return (error);
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return (error);
+
+ error = extattr_get_vp(nd.ni_vp, SCARG(uap, attrnamespace), attrname,
+ SCARG(uap, data), SCARG(uap, nbytes), p, retval);
+
+ vrele(nd.ni_vp);
+ return (error);
+}
+
+int
+sys_extattr_get_fd(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_extattr_get_fd_args /* {
+ syscallarg(int) fd;
+ syscallarg(int) attrnamespace;
+ syscallarg(const char *) attrname;
+ syscallarg(void *) data;
+ syscallarg(size_t) nbytes;
+ } */ *uap = v;
+ struct file *fp;
+ char attrname[EXTATTR_MAXNAMELEN];
+ int error;
+
+ error = copyinstr(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN,
+ NULL);
+ if (error)
+ return (error);
+
+ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+
+ error = extattr_get_vp((struct vnode *)fp->f_data,
+ SCARG(uap, attrnamespace), attrname, SCARG(uap, data),
+ SCARG(uap, nbytes), p, retval);
+ FRELE(fp);
+
+ return (error);
+}
+
+/*
+ * extattr_delete_vp(): Delete a named extended attribute on a file or
+ * directory
+ *
+ * Arguments: unlocked vnode "vp", attribute namespace "attrnamespace",
+ * kernelspace string pointer "attrname", proc "p"
+ * Returns: 0 on success, an error number otherwise
+ * Locks: none
+ * References: vp must be a valid reference for the duration of the call
+ */
+static int
+extattr_delete_vp(struct vnode *vp, int attrnamespace, const char *attrname,
+ struct proc *p)
+{
+ int error;
+
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+
+ error = VOP_SETEXTATTR(vp, attrnamespace, attrname, NULL,
+ p->p_ucred, p);
+
+ VOP_UNLOCK(vp, 0, p);
+ return (error);
+}
+
+int
+sys_extattr_delete_file(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_extattr_delete_file_args /* {
+ syscallarg(int) fd;
+ syscallarg(int) attrnamespace;
+ syscallarg(const char *) attrname;
+ } */ *uap = v;
+ struct nameidata nd;
+ char attrname[EXTATTR_MAXNAMELEN];
+ int error;
+
+ error = copyinstr(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN,
+ NULL);
+ if (error)
+ return(error);
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ if ((error = namei(&nd)) != 0)
+ return(error);
+
+ error = extattr_delete_vp(nd.ni_vp, SCARG(uap, attrnamespace),
+ attrname, p);
+
+ vrele(nd.ni_vp);
+ return(error);
+}
+
+int
+sys_extattr_delete_fd(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+{
+ struct sys_extattr_delete_fd_args /* {
+ syscallarg(int) fd;
+ syscallarg(int) attrnamespace;
+ syscallarg(const char *) attrname;
+ } */ *uap = v;
+ struct file *fp;
+ char attrname[EXTATTR_MAXNAMELEN];
+ int error;
+
+ error = copyinstr(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN,
+ NULL);
+ if (error)
+ return (error);
+
+ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+
+ error = extattr_delete_vp((struct vnode *)fp->f_data,
+ SCARG(uap, attrnamespace), attrname, p);
+ FRELE(fp);
+
+ return (error);
+}
+#endif
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff-t9.expected
@@ -1,0 +1,3182 @@
+--- diff-t9.1
++++ diff-t9.2
+@@ -1,4 +1,5 @@
+-/* $NetBSD: vfs_syscalls.c,v 1.57 1995/10/07 06:28:51 mycroft Exp $ */
++/* $OpenBSD: t9.2,v 1.2 2013/12/01 16:40:56 krw Exp $ */
++/* $NetBSD: vfs_syscalls.c,v 1.71 1996/04/23 10:29:02 mycroft Exp $ */
+
+ /*
+ * Copyright (c) 1989, 1993
+@@ -17,11 +18,7 @@
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+- * 3. All advertising materials mentioning features or use of this software
+- * must display the following acknowledgement:
+- * This product includes software developed by the University of
+- * California, Berkeley and its contributors.
+- * 4. Neither the name of the University nor the names of its contributors
++ * 3. Neither the name of the University nor the names of its contributors
+ * may be used to endorse or promote products derived from this software
+ * without specific prior written permission.
+ *
+@@ -53,13 +50,30 @@
+ #include <sys/uio.h>
+ #include <sys/malloc.h>
+ #include <sys/dirent.h>
++#include <sys/extattr.h>
+
+ #include <sys/syscallargs.h>
+
+-#include <vm/vm.h>
++#include <uvm/uvm_extern.h>
+ #include <sys/sysctl.h>
+
+-static int change_dir __P((struct nameidata *ndp, struct proc *p));
++extern int suid_clear;
++int usermount = 0; /* sysctl: by default, users may not mount */
++
++static int change_dir(struct nameidata *, struct proc *);
++
++void checkdirs(struct vnode *);
++
++/*
++ * Redirection info so we don't have to include the union fs routines in
++ * the kernel directly. This way, we can build unionfs as an LKM. The
++ * pointer gets filled in later, when we modload the LKM, or when the
++ * compiled-in unionfs code gets initialized. For now, we just set
++ * it to a stub routine.
++ */
++
++int (*union_check_p)(struct proc *, struct vnode **,
++ struct file *, struct uio, int *) = NULL;
+
+ /*
+ * Virtual File System System Calls
+@@ -69,6 +83,7 @@
+ * Mount a file system.
+ */
+ /* ARGSUSED */
++int
+ sys_mount(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -78,22 +93,36 @@
+ syscallarg(char *) type;
+ syscallarg(char *) path;
+ syscallarg(int) flags;
+- syscallarg(caddr_t) data;
++ syscallarg(void *) data;
+ } */ *uap = v;
+ register struct vnode *vp;
+ register struct mount *mp;
+- int error, flag;
+- u_long fsindex;
++ int error, flag = 0;
++#ifdef COMPAT_43
++ u_long fstypenum = 0;
++#endif
+ char fstypename[MFSNAMELEN];
++ char fspath[MNAMELEN];
+ struct vattr va;
+ struct nameidata nd;
++ struct vfsconf *vfsp;
++ struct timeval tv;
++
++ if (usermount == 0 && (error = suser(p->p_ucred, &p->p_acflag)))
++ return (error);
++
++ /*
++ * Mount points must fit in MNAMELEN, not MAXPATHLEN.
++ */
++ error = copyinstr(SCARG(uap, path), fspath, MNAMELEN, NULL);
++ if (error)
++ return(error);
+
+ /*
+ * Get vnode to be covered
+ */
+- NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+- SCARG(uap, path), p);
+- if (error = namei(&nd))
++ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE, fspath, p);
++ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ if (SCARG(uap, flags) & MNT_UPDATE) {
+@@ -113,7 +142,7 @@
+ return (EOPNOTSUPP); /* Needs translation */
+ }
+ mp->mnt_flag |=
+- SCARG(uap, flags) & (MNT_RELOAD | MNT_FORCE | MNT_UPDATE);
++ SCARG(uap, flags) & (MNT_RELOAD | MNT_UPDATE);
+ /*
+ * Only root, or the user that did the original mount is
+ * permitted to update it.
+@@ -134,7 +163,11 @@
+ }
+ SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
+ }
+- VOP_UNLOCK(vp);
++ if ((error = vfs_busy(mp, LK_NOWAIT, 0, p)) != 0) {
++ vput(vp);
++ return (error);
++ }
++ VOP_UNLOCK(vp, 0, p);
+ goto update;
+ }
+ /*
+@@ -143,7 +176,7 @@
+ */
+ if ((error = VOP_GETATTR(vp, &va, p->p_ucred, p)) ||
+ (va.va_uid != p->p_ucred->cr_uid &&
+- (error = suser(p->p_ucred, &p->p_acflag)))) {
++ (error = suser(p->p_ucred, &p->p_acflag)))) {
+ vput(vp);
+ return (error);
+ }
+@@ -158,40 +191,47 @@
+ }
+ SCARG(uap, flags) |= MNT_NOSUID | MNT_NODEV;
+ }
+- if (error = vinvalbuf(vp, V_SAVE, p->p_ucred, p, 0, 0))
++ if ((error = vinvalbuf(vp, V_SAVE, p->p_ucred, p, 0, 0)) != 0)
+ return (error);
+ if (vp->v_type != VDIR) {
+ vput(vp);
+ return (ENOTDIR);
+ }
+- if (error = copyinstr(SCARG(uap, type), fstypename, MFSNAMELEN,
+- (size_t *)0)) {
+-#if defined(COMPAT_09) || defined(COMPAT_43)
++ error = copyinstr(SCARG(uap, type), fstypename, MFSNAMELEN, NULL);
++ if (error) {
++#ifdef COMPAT_43
+ /*
+ * Historically filesystem types were identified by number.
+ * If we get an integer for the filesystem type instead of a
+ * string, we check to see if it matches one of the historic
+ * filesystem types.
+- */
+- fsindex = (u_long)SCARG(uap, type);
+- if (fsindex >= nvfssw || vfssw[fsindex] == NULL) {
++ */
++ fstypenum = (u_long)SCARG(uap, type);
++
++ for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next)
++ if (vfsp->vfc_typenum == fstypenum)
++ break;
++ if (vfsp == NULL) {
+ vput(vp);
+ return (ENODEV);
+ }
+- strncpy(fstypename, vfssw[fsindex]->vfs_name, MFSNAMELEN);
++ strncpy(fstypename, vfsp->vfc_name, MFSNAMELEN);
++
+ #else
+ vput(vp);
+ return (error);
+ #endif
+ }
+- for (fsindex = 0; fsindex < nvfssw; fsindex++)
+- if (vfssw[fsindex] != NULL &&
+- !strncmp(vfssw[fsindex]->vfs_name, fstypename, MFSNAMELEN))
++ for (vfsp = vfsconf; vfsp; vfsp = vfsp->vfc_next) {
++ if (!strcmp(vfsp->vfc_name, fstypename))
+ break;
+- if (fsindex >= nvfssw) {
++ }
++
++ if (vfsp == NULL) {
+ vput(vp);
+- return (ENODEV);
++ return (EOPNOTSUPP);
+ }
++
+ if (vp->v_mountedhere != NULL) {
+ vput(vp);
+ return (EBUSY);
+@@ -203,15 +243,14 @@
+ mp = (struct mount *)malloc((u_long)sizeof(struct mount),
+ M_MOUNT, M_WAITOK);
+ bzero((char *)mp, (u_long)sizeof(struct mount));
+- mp->mnt_op = vfssw[fsindex];
+- if (error = vfs_lock(mp)) {
+- free((caddr_t)mp, M_MOUNT);
+- vput(vp);
+- return (error);
+- }
+- /* Do this early in case we block later. */
+- vfssw[fsindex]->vfs_refcount++;
+- vp->v_mountedhere = mp;
++ lockinit(&mp->mnt_lock, PVFS, "vfslock", 0, 0);
++ /* This error never happens, but it makes auditing easier */
++ if ((error = vfs_busy(mp, LK_NOWAIT, 0, p)))
++ return (error);
++ mp->mnt_op = vfsp->vfc_vfsops;
++ mp->mnt_vfc = vfsp;
++ mp->mnt_flag |= (vfsp->vfc_flags & MNT_VISFLAGMASK);
++ strncpy(mp->mnt_stat.f_fstypename, vfsp->vfc_name, MFSNAMELEN);
+ mp->mnt_vnodecovered = vp;
+ mp->mnt_stat.f_owner = p->p_ucred->cr_uid;
+ update:
+@@ -223,13 +262,19 @@
+ else if (mp->mnt_flag & MNT_RDONLY)
+ mp->mnt_flag |= MNT_WANTRDWR;
+ mp->mnt_flag &=~ (MNT_NOSUID | MNT_NOEXEC | MNT_NODEV |
+- MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC);
++ MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | MNT_SOFTDEP |
++ MNT_NOATIME | MNT_FORCE);
+ mp->mnt_flag |= SCARG(uap, flags) & (MNT_NOSUID | MNT_NOEXEC |
+- MNT_NODEV | MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC);
++ MNT_NODEV | MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC |
++ MNT_SOFTDEP | MNT_NOATIME | MNT_FORCE);
+ /*
+ * Mount the filesystem.
+ */
+ error = VFS_MOUNT(mp, SCARG(uap, path), SCARG(uap, data), &nd, p);
++ if (!error) {
++ microtime(&tv);
++ mp->mnt_stat.f_ctime = tv.tv_sec;
++ }
+ if (mp->mnt_flag & MNT_UPDATE) {
+ vrele(vp);
+ if (mp->mnt_flag & MNT_WANTRDWR)
+@@ -238,23 +283,42 @@
+ (MNT_UPDATE | MNT_RELOAD | MNT_FORCE | MNT_WANTRDWR);
+ if (error)
+ mp->mnt_flag = flag;
++
++ if ((mp->mnt_flag & MNT_RDONLY) == 0) {
++ if (mp->mnt_syncer == NULL)
++ error = vfs_allocate_syncvnode(mp);
++ } else {
++ if (mp->mnt_syncer != NULL)
++ vgone(mp->mnt_syncer);
++ mp->mnt_syncer = NULL;
++ }
++
++ vfs_unbusy(mp, p);
+ return (error);
+ }
++
++ vp->v_mountedhere = mp;
++
+ /*
+ * Put the new filesystem on the mount list after root.
+ */
+ cache_purge(vp);
+ if (!error) {
++ vfsp->vfc_refcount++;
++ simple_lock(&mountlist_slock);
+ TAILQ_INSERT_TAIL(&mountlist, mp, mnt_list);
++ simple_unlock(&mountlist_slock);
+ checkdirs(vp);
+- VOP_UNLOCK(vp);
+- vfs_unlock(mp);
++ VOP_UNLOCK(vp, 0, p);
++ if ((mp->mnt_flag & MNT_RDONLY) == 0)
++ error = vfs_allocate_syncvnode(mp);
++ vfs_unbusy(mp, p);
+ (void) VFS_STATFS(mp, &mp->mnt_stat, p);
+- error = VFS_START(mp, 0, p);
++ if ((error = VFS_START(mp, 0, p)) != 0)
++ vrele(vp);
+ } else {
+ mp->mnt_vnodecovered->v_mountedhere = (struct mount *)0;
+- vfssw[fsindex]->vfs_refcount--;
+- vfs_unlock(mp);
++ vfs_unbusy(mp, p);
+ free((caddr_t)mp, M_MOUNT);
+ vput(vp);
+ }
+@@ -266,6 +330,7 @@
+ * or root directory onto which the new filesystem has just been
+ * mounted. If so, replace them with the new mount point.
+ */
++void
+ checkdirs(olddp)
+ struct vnode *olddp;
+ {
+@@ -277,7 +342,7 @@
+ return;
+ if (VFS_ROOT(olddp->v_mountedhere, &newdp))
+ panic("mount: lost mount");
+- for (p = allproc.lh_first; p != 0; p = p->p_list.le_next) {
++ for (p = LIST_FIRST(&allproc); p != 0; p = LIST_NEXT(p, p_list)) {
+ fdp = p->p_fd;
+ if (fdp->fd_cdir == olddp) {
+ vrele(fdp->fd_cdir);
+@@ -305,6 +370,7 @@
+ * not special file (as before).
+ */
+ /* ARGSUSED */
++int
+ sys_unmount(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -321,7 +387,7 @@
+
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+- if (error = namei(&nd))
++ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ mp = vp->v_mount;
+@@ -352,50 +418,77 @@
+ return (EINVAL);
+ }
+ vput(vp);
+- return (dounmount(mp, SCARG(uap, flags), p));
++
++ if (vfs_busy(mp, LK_EXCLUSIVE, NULL, p))
++ return (EBUSY);
++
++ return (dounmount(mp, SCARG(uap, flags), p, vp));
+ }
+
+ /*
+ * Do the actual file system unmount.
+ */
+-dounmount(mp, flags, p)
+- register struct mount *mp;
+- int flags;
+- struct proc *p;
++int
++dounmount(struct mount *mp, int flags, struct proc *p, struct vnode *olddp)
+ {
+ struct vnode *coveredvp;
++ struct proc *p2;
+ int error;
++ int hadsyncer = 0;
+
+- coveredvp = mp->mnt_vnodecovered;
+- if (vfs_busy(mp))
+- return (EBUSY);
+- mp->mnt_flag |= MNT_UNMOUNT;
+- if (error = vfs_lock(mp))
+- return (error);
+-
+- mp->mnt_flag &=~ MNT_ASYNC;
+- vnode_pager_umount(mp); /* release cached vnodes */
+- cache_purgevfs(mp); /* remove cache entries for this file sys */
+- if ((error = VFS_SYNC(mp, MNT_WAIT, p->p_ucred, p)) == 0 ||
+- (flags & MNT_FORCE))
+- error = VFS_UNMOUNT(mp, flags, p);
+- mp->mnt_flag &= ~MNT_UNMOUNT;
+- vfs_unbusy(mp);
+- if (error) {
+- vfs_unlock(mp);
+- } else {
+- TAILQ_REMOVE(&mountlist, mp, mnt_list);
+- if (coveredvp != NULLVP) {
+- vrele(coveredvp);
+- coveredvp->v_mountedhere = (struct mount *)0;
+- }
+- mp->mnt_op->vfs_refcount--;
+- vfs_unlock(mp);
+- if (mp->mnt_vnodelist.lh_first != NULL)
+- panic("unmount: dangling vnode");
+- free((caddr_t)mp, M_MOUNT);
+- }
+- return (error);
++ mp->mnt_flag &=~ MNT_ASYNC;
++ cache_purgevfs(mp); /* remove cache entries for this file sys */
++ if (mp->mnt_syncer != NULL) {
++ hadsyncer = 1;
++ vgone(mp->mnt_syncer);
++ mp->mnt_syncer = NULL;
++ }
++ if (((mp->mnt_flag & MNT_RDONLY) ||
++ (error = VFS_SYNC(mp, MNT_WAIT, p->p_ucred, p)) == 0) ||
++ (flags & MNT_FORCE))
++ error = VFS_UNMOUNT(mp, flags, p);
++ simple_lock(&mountlist_slock);
++ if (error) {
++ if ((mp->mnt_flag & MNT_RDONLY) == 0 && hadsyncer)
++ (void) vfs_allocate_syncvnode(mp);
++ lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK,
++ &mountlist_slock, p);
++ return (error);
++ }
++ TAILQ_REMOVE(&mountlist, mp, mnt_list);
++ if ((coveredvp = mp->mnt_vnodecovered) != NULLVP) {
++ if (olddp) {
++ /*
++ * Try to put processes back in a real directory
++ * after a forced unmount.
++ * XXX We're not holding a ref on olddp, which may
++ * change, so compare id numbers.
++ */
++ LIST_FOREACH(p2, &allproc, p_list) {
++ struct filedesc *fdp = p2->p_fd;
++ if (fdp->fd_cdir &&
++ fdp->fd_cdir->v_id == olddp->v_id) {
++ vrele(fdp->fd_cdir);
++ vref(coveredvp);
++ fdp->fd_cdir = coveredvp;
++ }
++ if (fdp->fd_rdir &&
++ fdp->fd_rdir->v_id == olddp->v_id) {
++ vrele(fdp->fd_rdir);
++ vref(coveredvp);
++ fdp->fd_rdir = coveredvp;
++ }
++ }
++ }
++ coveredvp->v_mountedhere = NULL;
++ vrele(coveredvp);
++ }
++ mp->mnt_vfc->vfc_refcount--;
++ if (mp->mnt_vnodelist.lh_first != NULL)
++ panic("unmount: dangling vnode");
++ lockmgr(&mp->mnt_lock, LK_RELEASE | LK_INTERLOCK, &mountlist_slock, p);
++ free((caddr_t)mp, M_MOUNT);
++ return (0);
+ }
+
+ /*
+@@ -407,6 +500,7 @@
+ #endif
+
+ /* ARGSUSED */
++int
+ sys_sync(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -415,31 +509,23 @@
+ register struct mount *mp, *nmp;
+ int asyncflag;
+
+- for (mp = mountlist.cqh_first; mp != (void *)&mountlist; mp = nmp) {
+- /*
+- * Get the next pointer in case we hang on vfs_busy
+- * while we are being unmounted.
+- */
+- nmp = mp->mnt_list.cqe_next;
+- /*
+- * The lock check below is to avoid races with mount
+- * and unmount.
+- */
+- if ((mp->mnt_flag & (MNT_MLOCK|MNT_RDONLY|MNT_MPBUSY)) == 0 &&
+- !vfs_busy(mp)) {
++ simple_lock(&mountlist_slock);
++ TAILQ_FOREACH_REVERSE_SAFE(mp, &mountlist, mnt_list, nmp) {
++ if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock, p))
++ continue;
++ if ((mp->mnt_flag & MNT_RDONLY) == 0) {
+ asyncflag = mp->mnt_flag & MNT_ASYNC;
+ mp->mnt_flag &= ~MNT_ASYNC;
++ uvm_vnp_sync(mp);
+ VFS_SYNC(mp, MNT_NOWAIT, p->p_ucred, p);
+ if (asyncflag)
+ mp->mnt_flag |= MNT_ASYNC;
+- /*
+- * Get the next pointer again, as the next filesystem
+- * might have been unmounted while we were sync'ing.
+- */
+- nmp = mp->mnt_list.cqe_next;
+- vfs_unbusy(mp);
+ }
++ simple_lock(&mountlist_slock);
++ vfs_unbusy(mp, p);
+ }
++ simple_unlock(&mountlist_slock);
++
+ #ifdef DEBUG
+ if (syncprt)
+ vfs_bufstats();
+@@ -451,6 +537,7 @@
+ * Change filesystem quotas.
+ */
+ /* ARGSUSED */
++int
+ sys_quotactl(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -467,7 +554,7 @@
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+- if (error = namei(&nd))
++ if ((error = namei(&nd)) != 0)
+ return (error);
+ mp = nd.ni_vp->v_mount;
+ vrele(nd.ni_vp);
+@@ -479,6 +566,7 @@
+ * Get filesystem statistics.
+ */
+ /* ARGSUSED */
++int
+ sys_statfs(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -492,16 +580,27 @@
+ register struct statfs *sp;
+ int error;
+ struct nameidata nd;
++ struct statfs sb;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+- if (error = namei(&nd))
++ if ((error = namei(&nd)) != 0)
+ return (error);
+ mp = nd.ni_vp->v_mount;
+ sp = &mp->mnt_stat;
+ vrele(nd.ni_vp);
+- if (error = VFS_STATFS(mp, sp, p))
++ if ((error = VFS_STATFS(mp, sp, p)) != 0)
+ return (error);
+ sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
++#if notyet
++ if (mp->mnt_flag & MNT_SOFTDEP)
++ sp->f_eflags = STATFS_SOFTUPD;
++#endif
++ /* Don't let non-root see filesystem id (for NFS security) */
++ if (suser(p->p_ucred, &p->p_acflag)) {
++ bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
++ sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
++ sp = &sb;
++ }
+ return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
+ }
+
+@@ -509,33 +608,48 @@
+ * Get filesystem statistics.
+ */
+ /* ARGSUSED */
++int
+ sys_fstatfs(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+ {
+- register struct sys_fstatfs_args /* {
++ struct sys_fstatfs_args /* {
+ syscallarg(int) fd;
+ syscallarg(struct statfs *) buf;
+ } */ *uap = v;
+ struct file *fp;
+ struct mount *mp;
+- register struct statfs *sp;
++ struct statfs *sp;
+ int error;
++ struct statfs sb;
+
+- if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
++ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+ mp = ((struct vnode *)fp->f_data)->v_mount;
+ sp = &mp->mnt_stat;
+- if (error = VFS_STATFS(mp, sp, p))
++ error = VFS_STATFS(mp, sp, p);
++ FRELE(fp);
++ if (error)
+ return (error);
+ sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
++#if notyet
++ if (mp->mnt_flag & MNT_SOFTDEP)
++ sp->f_eflags = STATFS_SOFTUPD;
++#endif
++ /* Don't let non-root see filesystem id (for NFS security) */
++ if (suser(p->p_ucred, &p->p_acflag)) {
++ bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
++ sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
++ sp = &sb;
++ }
+ return (copyout((caddr_t)sp, (caddr_t)SCARG(uap, buf), sizeof(*sp)));
+ }
+
+ /*
+ * Get statistics on all filesystems.
+ */
++int
+ sys_getfsstat(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -543,37 +657,59 @@
+ {
+ register struct sys_getfsstat_args /* {
+ syscallarg(struct statfs *) buf;
+- syscallarg(long) bufsize;
++ syscallarg(size_t) bufsize;
+ syscallarg(int) flags;
+ } */ *uap = v;
+- register struct mount *mp, *nmp;
++ register struct mount *mp *nmp;
+ register struct statfs *sp;
++ struct statfs sb;
+ caddr_t sfsp;
+- long count, maxcount, error;
++ size_t count, maxcount;
++ int error, flags = SCARG(uap, flags);
+
+ maxcount = SCARG(uap, bufsize) / sizeof(struct statfs);
+ sfsp = (caddr_t)SCARG(uap, buf);
+- for (count = 0,
+- mp = mountlist.cqh_first; mp != (void *)&mountlist; mp = nmp) {
+- nmp = mp->mnt_list.cqe_next;
+- if (sfsp && count < maxcount &&
+- ((mp->mnt_flag & MNT_MLOCK) == 0)) {
++ count = 0;
++ simple_lock(&mountlist_slock);
++ TAILQ_FOREACH_REVERSE_SAFE(mp, &mountlist, mnt_list, nmp) {
++ if (vfs_busy(mp, LK_NOWAIT, &mountlist_slock, p))
++ continue;
++ if (sfsp && count < maxcount) {
+ sp = &mp->mnt_stat;
+- /*
+- * If MNT_NOWAIT is specified, do not refresh the
+- * fsstat cache. MNT_WAIT overrides MNT_NOWAIT.
+- */
+- if (((SCARG(uap, flags) & MNT_NOWAIT) == 0 ||
+- (SCARG(uap, flags) & MNT_WAIT)) &&
+- (error = VFS_STATFS(mp, sp, p)))
+- continue;
++
++ /* Refresh stats unless MNT_NOWAIT is specified */
++ if (flags != MNT_NOWAIT &&
++ flags != MNT_LAZY &&
++ (flags == MNT_WAIT ||
++ flags == 0) &&
++ (error = VFS_STATFS(mp, sp, p))) {
++ simple_lock(&mountlist_slock);
++ vfs_unbusy(mp, p);
++ continue;
++ }
++
+ sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
+- if (error = copyout((caddr_t)sp, sfsp, sizeof(*sp)))
++#if notyet
++ if (mp->mnt_flag & MNT_SOFTDEP)
++ sp->f_eflags = STATFS_SOFTUPD;
++#endif
++ if (suser(p->p_ucred, &p->p_acflag)) {
++ bcopy((caddr_t)sp, (caddr_t)&sb, sizeof(sb));
++ sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
++ sp = &sb;
++ }
++ error = copyout((caddr_t)sp, sfsp, sizeof(*sp));
++ if (error) {
++ vfs_unbusy(mp, p);
+ return (error);
++ }
+ sfsp += sizeof(*sp);
+ }
+ count++;
++ simple_lock(&mountlist_slock);
++ vfs_unbusy(mp, p);
+ }
++ simple_unlock(&mountlist_slock);
+ if (sfsp && count > maxcount)
+ *retval = maxcount;
+ else
+@@ -585,6 +721,7 @@
+ * Change current working directory to a given file descriptor.
+ */
+ /* ARGSUSED */
++int
+ sys_fchdir(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -593,37 +730,38 @@
+ struct sys_fchdir_args /* {
+ syscallarg(int) fd;
+ } */ *uap = v;
+- register struct filedesc *fdp = p->p_fd;
++ struct filedesc *fdp = p->p_fd;
+ struct vnode *vp, *tdp;
+ struct mount *mp;
+ struct file *fp;
+ int error;
+
+- if (error = getvnode(fdp, SCARG(uap, fd), &fp))
++ if ((error = getvnode(fdp, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+ vp = (struct vnode *)fp->f_data;
+ VREF(vp);
+- VOP_LOCK(vp);
++ FRELE(fp);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_type != VDIR)
+ error = ENOTDIR;
+ else
+ error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
++
+ while (!error && (mp = vp->v_mountedhere) != NULL) {
+- if (mp->mnt_flag & MNT_MLOCK) {
+- mp->mnt_flag |= MNT_MWAIT;
+- sleep((caddr_t)mp, PVFS);
++ if (vfs_busy(mp, 0, 0, p))
+ continue;
+- }
+- if (error = VFS_ROOT(mp, &tdp))
++ error = VFS_ROOT(mp, &tdp);
++ vfs_unbusy(mp, p);
++ if (error)
+ break;
+ vput(vp);
+ vp = tdp;
+ }
+- VOP_UNLOCK(vp);
+ if (error) {
+- vrele(vp);
++ vput(vp);
+ return (error);
+ }
++ VOP_UNLOCK(vp, 0, p);
+ vrele(fdp->fd_cdir);
+ fdp->fd_cdir = vp;
+ return (0);
+@@ -633,6 +771,7 @@
+ * Change current working directory (``.'').
+ */
+ /* ARGSUSED */
++int
+ sys_chdir(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -647,7 +786,7 @@
+
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+- if (error = change_dir(&nd, p))
++ if ((error = change_dir(&nd, p)) != 0)
+ return (error);
+ vrele(fdp->fd_cdir);
+ fdp->fd_cdir = nd.ni_vp;
+@@ -658,6 +797,7 @@
+ * Change notion of root (``/'') directory.
+ */
+ /* ARGSUSED */
++int
+ sys_chroot(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -670,14 +810,22 @@
+ int error;
+ struct nameidata nd;
+
+- if (error = suser(p->p_ucred, &p->p_acflag))
++ if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
+ return (error);
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+- if (error = change_dir(&nd, p))
++ if ((error = change_dir(&nd, p)) != 0)
+ return (error);
+- if (fdp->fd_rdir != NULL)
++ if (fdp->fd_rdir != NULL) {
++ /*
++ * A chroot() done inside a changed root environment does
++ * an automatic chdir to avoid the out-of-tree experience.
++ */
+ vrele(fdp->fd_rdir);
++ vrele(fdp->fd_cdir);
++ VREF(nd.ni_vp);
++ fdp->fd_cdir = nd.ni_vp;
++ }
+ fdp->fd_rdir = nd.ni_vp;
+ return (0);
+ }
+@@ -693,16 +841,17 @@
+ struct vnode *vp;
+ int error;
+
+- if (error = namei(ndp))
++ if ((error = namei(ndp)) != 0)
+ return (error);
+ vp = ndp->ni_vp;
+ if (vp->v_type != VDIR)
+ error = ENOTDIR;
+ else
+ error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p);
+- VOP_UNLOCK(vp);
+ if (error)
+- vrele(vp);
++ vput(vp);
++ else
++ VOP_UNLOCK(vp, 0, p);
+ return (error);
+ }
+
+@@ -710,45 +859,50 @@
+ * Check permissions, allocate an open file structure,
+ * and call the device open routine if any.
+ */
++int
+ sys_open(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+ {
+- register struct sys_open_args /* {
++ struct sys_open_args /* {
+ syscallarg(char *) path;
+ syscallarg(int) flags;
+ syscallarg(int) mode;
+ } */ *uap = v;
+- register struct filedesc *fdp = p->p_fd;
+- register struct file *fp;
+- register struct vnode *vp;
++ struct filedesc *fdp = p->p_fd;
++ struct file *fp;
++ struct vnode *vp;
++ struct vattr vattr;
+ int flags, cmode;
+- struct file *nfp;
+- int type, indx, error;
++ int type, indx, error, localtrunc = 0;
+ struct flock lf;
+ struct nameidata nd;
+- extern struct fileops vnops;
+
+- if (error = falloc(p, &nfp, &indx))
++ if ((error = falloc(p, &fp, &indx)) != 0)
+ return (error);
+- fp = nfp;
++
+ flags = FFLAGS(SCARG(uap, flags));
+ cmode = ((SCARG(uap, mode) &~ fdp->fd_cmask) & ALLPERMS) &~ S_ISTXT;
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+ p->p_dupfd = -indx - 1; /* XXX check for fdopen */
+- if (error = vn_open(&nd, flags, cmode)) {
+- ffree(fp);
++ if ((flags & O_TRUNC) && (flags & (O_EXLOCK | O_SHLOCK))) {
++ localtrunc = 1;
++ flags &= ~O_TRUNC; /* Must do truncate ourselves */
++ }
++ if ((error = vn_open(&nd, flags, cmode)) != 0) {
+ if ((error == ENODEV || error == ENXIO) &&
+ p->p_dupfd >= 0 && /* XXX from fdopen */
+ (error =
+ dupfdopen(fdp, indx, p->p_dupfd, flags, error)) == 0) {
++ closef(fp, p);
+ *retval = indx;
+ return (0);
+ }
+ if (error == ERESTART)
+ error = EINTR;
+- fdp->fd_ofiles[indx] = NULL;
++ fdremove(fdp, indx);
++ closef(fp, p);
+ return (error);
+ }
+ p->p_dupfd = 0;
+@@ -768,189 +922,463 @@
+ type = F_FLOCK;
+ if ((flags & FNONBLOCK) == 0)
+ type |= F_WAIT;
+- VOP_UNLOCK(vp);
+- if (error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type)) {
+- (void) vn_close(vp, fp->f_flag, fp->f_cred, p);
+- ffree(fp);
+- fdp->fd_ofiles[indx] = NULL;
++ VOP_UNLOCK(vp, 0, p);
++ error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type);
++ if (error) {
++ /* closef will vn_close the file for us. */
++ fdremove(fdp, indx);
++ closef(fp, p);
+ return (error);
+ }
+- VOP_LOCK(vp);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ fp->f_flag |= FHASLOCK;
+ }
+- VOP_UNLOCK(vp);
++ if (localtrunc) {
++ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
++ if ((fp->f_flag & FWRITE) == 0)
++ error = EACCES;
++ else if (vp->v_mount->mnt_flag & MNT_RDONLY)
++ error = EROFS;
++ else if (vp->v_type == VDIR)
++ error = EISDIR;
++ else if ((error = vn_writechk(vp)) == 0) {
++ VATTR_NULL(&vattr);
++ vattr.va_size = 0;
++ error = VOP_SETATTR(vp, &vattr, fp->f_cred, p);
++ }
++ if (error) {
++ VOP_UNLOCK(vp, 0, p);
++ /* closef will close the file for us. */
++ fdremove(fdp, indx);
++ closef(fp, p);
++ return (error);
++ }
++ }
++ VOP_UNLOCK(vp, 0, p);
+ *retval = indx;
++ FILE_SET_MATURE(fp);
+ return (0);
+ }
+
+ /*
+- * Create a special file.
++ * Get file handle system call
+ */
+-/* ARGSUSED */
+-sys_mknod(p, v, retval)
++int
++sys_getfh(p, v, retval)
+ struct proc *p;
+- void *v;
++ register void *v;
+ register_t *retval;
+ {
+- register struct sys_mknod_args /* {
+- syscallarg(char *) path;
+- syscallarg(int) mode;
+- syscallarg(int) dev;
++ register struct sys_getfh_args /* {
++ syscallarg(char *) fname;
++ syscallarg(fhandle_t *) fhp;
+ } */ *uap = v;
+ register struct vnode *vp;
+- struct vattr vattr;
++ fhandle_t fh;
+ int error;
+- int whiteout;
+ struct nameidata nd;
+
+- if (error = suser(p->p_ucred, &p->p_acflag))
++ /*
++ * Must be super user
++ */
++ error = suser(p->p_ucred, &p->p_acflag);
++ if (error)
+ return (error);
+- NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
+- if (error = namei(&nd))
++ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
++ SCARG(uap, fname), p);
++ error = namei(&nd);
++ if (error)
+ return (error);
+ vp = nd.ni_vp;
+- if (vp != NULL)
+- error = EEXIST;
+- else {
+- VATTR_NULL(&vattr);
+- vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
+- vattr.va_rdev = SCARG(uap, dev);
+- whiteout = 0;
+-
+- switch (SCARG(uap, mode) & S_IFMT) {
+- case S_IFMT: /* used by badsect to flag bad sectors */
+- vattr.va_type = VBAD;
+- break;
+- case S_IFCHR:
+- vattr.va_type = VCHR;
+- break;
+- case S_IFBLK:
+- vattr.va_type = VBLK;
+- break;
+- case S_IFWHT:
+- whiteout = 1;
+- break;
+- default:
+- error = EINVAL;
+- break;
+- }
+- }
+- if (!error) {
+- VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+- if (whiteout) {
+- error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE);
+- if (error)
+- VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+- vput(nd.ni_dvp);
+- } else {
+- error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp,
+- &nd.ni_cnd, &vattr);
+- }
+- } else {
+- VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+- if (nd.ni_dvp == vp)
+- vrele(nd.ni_dvp);
+- else
+- vput(nd.ni_dvp);
+- if (vp)
+- vrele(vp);
+- }
++ bzero((caddr_t)&fh, sizeof(fh));
++ fh.fh_fsid = vp->v_mount->mnt_stat.f_fsid;
++ error = VFS_VPTOFH(vp, &fh.fh_fid);
++ vput(vp);
++ if (error)
++ return (error);
++ error = copyout((caddr_t)&fh, (caddr_t)SCARG(uap, fhp), sizeof (fh));
+ return (error);
+ }
+
+ /*
+- * Create a named pipe.
++ * Open a file given a file handle.
++ *
++ * Check permissions, allocate an open file structure,
++ * and call the device open routine if any.
+ */
+-/* ARGSUSED */
+-sys_mkfifo(p, v, retval)
++int
++sys_fhopen(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+ {
+- register struct sys_mkfifo_args /* {
+- syscallarg(char *) path;
+- syscallarg(int) mode;
++ register struct sys_fhopen_args /* {
++ syscallarg(const fhandle_t *) fhp;
++ syscallarg(int) flags;
+ } */ *uap = v;
+- struct vattr vattr;
+- int error;
+- struct nameidata nd;
++ struct filedesc *fdp = p->p_fd;
++ struct file *fp;
++ struct vnode *vp = NULL;
++ struct mount *mp;
++ struct ucred *cred = p->p_ucred;
++ int flags;
++ int type, indx, error=0;
++ struct flock lf;
++ struct vattr va;
++ fhandle_t fh;
+
+-#ifndef FIFO
+- return (EOPNOTSUPP);
+-#else
+- NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
+- if (error = namei(&nd))
++ /*
++ * Must be super user
++ */
++ if ((error = suser(p->p_ucred, &p->p_acflag)))
+ return (error);
+- if (nd.ni_vp != NULL) {
+- VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+- if (nd.ni_dvp == nd.ni_vp)
+- vrele(nd.ni_dvp);
+- else
+- vput(nd.ni_dvp);
+- vrele(nd.ni_vp);
+- return (EEXIST);
+- }
+- VATTR_NULL(&vattr);
+- vattr.va_type = VFIFO;
+- vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
+- VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+- return (VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr));
+-#endif /* FIFO */
+-}
+
+-/*
+- * Make a hard file link.
+- */
+-/* ARGSUSED */
+-sys_link(p, v, retval)
+- struct proc *p;
+- void *v;
+- register_t *retval;
+-{
+- register struct sys_link_args /* {
+- syscallarg(char *) path;
+- syscallarg(char *) link;
+- } */ *uap = v;
+- register struct vnode *vp;
+- struct nameidata nd;
+- int error;
++ flags = FFLAGS(SCARG(uap, flags));
++ if ((flags & (FREAD | FWRITE)) == 0)
++ return (EINVAL);
++ if ((flags & O_CREAT))
++ return (EINVAL);
+
+- NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+- if (error = namei(&nd))
++ if ((error = falloc(p, &fp, &indx)) != 0)
+ return (error);
+- vp = nd.ni_vp;
+- if (vp->v_type != VDIR ||
+- (error = suser(p->p_ucred, &p->p_acflag)) == 0) {
+- nd.ni_cnd.cn_nameiop = CREATE;
+- nd.ni_cnd.cn_flags = LOCKPARENT;
+- nd.ni_dirp = SCARG(uap, link);
+- if ((error = namei(&nd)) == 0) {
+- if (nd.ni_vp != NULL)
+- error = EEXIST;
+- if (!error) {
+- VOP_LEASE(nd.ni_dvp, p, p->p_ucred,
+- LEASE_WRITE);
+- VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+- error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
+- } else {
+- VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+- if (nd.ni_dvp == nd.ni_vp)
+- vrele(nd.ni_dvp);
+- else
+- vput(nd.ni_dvp);
+- if (nd.ni_vp)
+- vrele(nd.ni_vp);
+- }
+- }
++
++ if ((error = copyin(SCARG(uap, fhp), &fh, sizeof(fhandle_t))) != 0)
++ goto bad;
++
++ if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL) {
++ error = ESTALE;
++ goto bad;
+ }
+- vrele(vp);
+- return (error);
+-}
++
++ if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)) != 0) {
++ vp = NULL; /* most likely unnecessary sanity for bad: */
++ goto bad;
++ }
++
++ /* Now do an effective vn_open */
++
++ if (vp->v_type == VSOCK) {
++ error = EOPNOTSUPP;
++ goto bad;
++ }
++ if (flags & FREAD) {
++ if ((error = VOP_ACCESS(vp, VREAD, cred, p)) != 0)
++ goto bad;
++ }
++ if (flags & (FWRITE | O_TRUNC)) {
++ if (vp->v_type == VDIR) {
++ error = EISDIR;
++ goto bad;
++ }
++ if ((error = vn_writechk(vp)) != 0 ||
++ (error = VOP_ACCESS(vp, VWRITE, cred, p)) != 0)
++ goto bad;
++ }
++ if (flags & O_TRUNC) {
++ VOP_UNLOCK(vp, 0, p); /* XXX */
++ VOP_LEASE(vp, p, cred, LEASE_WRITE);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p); /* XXX */
++ VATTR_NULL(&va);
++ va.va_size = 0;
++ if ((error = VOP_SETATTR(vp, &va, cred, p)) != 0)
++ goto bad;
++ }
++ if ((error = VOP_OPEN(vp, flags, cred, p)) != 0)
++ goto bad;
++ if (flags & FWRITE)
++ vp->v_writecount++;
++
++ /* done with modified vn_open, now finish what sys_open does. */
++
++ fp->f_flag = flags & FMASK;
++ fp->f_type = DTYPE_VNODE;
++ fp->f_ops = &vnops;
++ fp->f_data = (caddr_t)vp;
++ if (flags & (O_EXLOCK | O_SHLOCK)) {
++ lf.l_whence = SEEK_SET;
++ lf.l_start = 0;
++ lf.l_len = 0;
++ if (flags & O_EXLOCK)
++ lf.l_type = F_WRLCK;
++ else
++ lf.l_type = F_RDLCK;
++ type = F_FLOCK;
++ if ((flags & FNONBLOCK) == 0)
++ type |= F_WAIT;
++ VOP_UNLOCK(vp, 0, p);
++ error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type);
++ if (error) {
++ /* closef will vn_close the file for us. */
++ fdremove(fdp, indx);
++ closef(fp, p);
++ return (error);
++ }
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
++ fp->f_flag |= FHASLOCK;
++ }
++ VOP_UNLOCK(vp, 0, p);
++ *retval = indx;
++ FILE_SET_MATURE(fp);
++ return (0);
++
++bad:
++ fdremove(fdp, indx);
++ closef(fp, p);
++ if (vp != NULL)
++ vput(vp);
++ return (error);
++}
++
++/* ARGSUSED */
++int
++sys_fhstat(p, v, retval)
++ struct proc *p;
++ void *v;
++ register_t *retval;
++{
++ register struct sys_fhstat_args /* {
++ syscallarg(const fhandle_t *) fhp;
++ syscallarg(struct stat *) sb;
++ } */ *uap = v;
++ struct stat sb;
++ int error;
++ fhandle_t fh;
++ struct mount *mp;
++ struct vnode *vp;
++
++ /*
++ * Must be super user
++ */
++ if ((error = suser(p->p_ucred, &p->p_acflag)))
++ return (error);
++
++ if ((error = copyin(SCARG(uap, fhp), &fh, sizeof(fhandle_t))) != 0)
++ return (error);
++
++ if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
++ return (ESTALE);
++ if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
++ return (error);
++ error = vn_stat(vp, &sb, p);
++ vput(vp);
++ if (error)
++ return (error);
++ error = copyout(&sb, SCARG(uap, sb), sizeof(sb));
++ return (error);
++}
++
++/* ARGSUSED */
++int
++sys_fhstatfs(p, v, retval)
++ struct proc *p;
++ void *v;
++ register_t *retval;
++{
++ register struct sys_fhstatfs_args /*
++ syscallarg(const fhandle_t *) fhp;
++ syscallarg(struct statfs *) buf;
++ } */ *uap = v;
++ struct statfs sp;
++ fhandle_t fh;
++ struct mount *mp;
++ struct vnode *vp;
++ int error;
++
++ /*
++ * Must be super user
++ */
++ if ((error = suser(p->p_ucred, &p->p_acflag)))
++ return (error);
++
++ if ((error = copyin(SCARG(uap, fhp), &fh, sizeof(fhandle_t))) != 0)
++ return (error);
++
++ if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
++ return (ESTALE);
++ if ((error = VFS_FHTOVP(mp, &fh.fh_fid, &vp)))
++ return (error);
++ mp = vp->v_mount;
++ vput(vp);
++ if ((error = VFS_STATFS(mp, &sp, p)) != 0)
++ return (error);
++ sp.f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
++ return (copyout(&sp, SCARG(uap, buf), sizeof(sp)));
++}
++
++/*
++ * Create a special file.
++ */
++/* ARGSUSED */
++int
++sys_mknod(p, v, retval)
++ struct proc *p;
++ void *v;
++ register_t *retval;
++{
++ register struct sys_mknod_args /* {
++ syscallarg(char *) path;
++ syscallarg(int) mode;
++ syscallarg(int) dev;
++ } */ *uap = v;
++ register struct vnode *vp;
++ struct vattr vattr;
++ int error;
++ int whiteout = 0;
++ struct nameidata nd;
++
++ if ((error = suser(p->p_ucred, &p->p_acflag)) != 0)
++ return (error);
++ if (p->p_fd->fd_rdir)
++ return (EINVAL);
++ NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
++ if ((error = namei(&nd)) != 0)
++ return (error);
++ vp = nd.ni_vp;
++ if (vp != NULL)
++ error = EEXIST;
++ else {
++ VATTR_NULL(&vattr);
++ vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
++ vattr.va_rdev = SCARG(uap, dev);
++ whiteout = 0;
++
++ switch (SCARG(uap, mode) & S_IFMT) {
++ case S_IFMT: /* used by badsect to flag bad sectors */
++ vattr.va_type = VBAD;
++ break;
++ case S_IFCHR:
++ vattr.va_type = VCHR;
++ break;
++ case S_IFBLK:
++ vattr.va_type = VBLK;
++ break;
++ case S_IFWHT:
++ whiteout = 1;
++ break;
++ default:
++ error = EINVAL;
++ break;
++ }
++ }
++ if (!error) {
++ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
++ if (whiteout) {
++ error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE);
++ if (error)
++ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
++ vput(nd.ni_dvp);
++ } else {
++ error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp,
++ &nd.ni_cnd, &vattr);
++ }
++ } else {
++ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
++ if (nd.ni_dvp == vp)
++ vrele(nd.ni_dvp);
++ else
++ vput(nd.ni_dvp);
++ if (vp)
++ vrele(vp);
++ }
++ return (error);
++}
++
++/*
++ * Create a named pipe.
++ */
++/* ARGSUSED */
++int
++sys_mkfifo(p, v, retval)
++ struct proc *p;
++ void *v;
++ register_t *retval;
++{
++#ifndef FIFO
++ return (EOPNOTSUPP);
++#else
++ register struct sys_mkfifo_args /* {
++ syscallarg(char *) path;
++ syscallarg(int) mode;
++ } */ *uap = v;
++ struct vattr vattr;
++ int error;
++ struct nameidata nd;
++
++ NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
++ if ((error = namei(&nd)) != 0)
++ return (error);
++ if (nd.ni_vp != NULL) {
++ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
++ if (nd.ni_dvp == nd.ni_vp)
++ vrele(nd.ni_dvp);
++ else
++ vput(nd.ni_dvp);
++ vrele(nd.ni_vp);
++ return (EEXIST);
++ }
++ VATTR_NULL(&vattr);
++ vattr.va_type = VFIFO;
++ vattr.va_mode = (SCARG(uap, mode) & ALLPERMS) &~ p->p_fd->fd_cmask;
++ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
++ return (VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr));
++#endif /* FIFO */
++}
++
++/*
++ * Make a hard file link.
++ */
++/* ARGSUSED */
++int
++sys_link(p, v, retval)
++ struct proc *p;
++ void *v;
++ register_t *retval;
++{
++ register struct sys_link_args /* {
++ syscallarg(char *) path;
++ syscallarg(char *) link;
++ } */ *uap = v;
++ register struct vnode *vp;
++ struct nameidata nd;
++ int error;
++ int flags;
++
++ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
++ if ((error = namei(&nd)) != 0)
++ return (error);
++ vp = nd.ni_vp;
++
++ flags = LOCKPARENT;
++ if (vp->v_type == VDIR) {
++ flags |= STRIPSLASHES;
++ }
++
++ NDINIT(&nd, CREATE, flags, UIO_USERSPACE, SCARG(uap, link), p);
++ if ((error = namei(&nd)) != 0)
++ goto out;
++ if (nd.ni_vp) {
++ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
++ if (nd.ni_dvp == nd.ni_vp)
++ vrele(nd.ni_dvp);
++ else
++ vput(nd.ni_dvp);
++ vrele(nd.ni_vp);
++ error = EEXIST;
++ goto out;
++ }
++ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
++ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
++ error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
++out:
++ vrele(vp);
++ return (error);
++}
+
+ /*
+ * Make a symbolic link.
+ */
+ /* ARGSUSED */
++int
+ sys_symlink(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -966,10 +1394,11 @@
+ struct nameidata nd;
+
+ MALLOC(path, char *, MAXPATHLEN, M_NAMEI, M_WAITOK);
+- if (error = copyinstr(SCARG(uap, path), path, MAXPATHLEN, (size_t *)0))
++ error = copyinstr(SCARG(uap, path), path, MAXPATHLEN, NULL);
++ if (error)
+ goto out;
+ NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, link), p);
+- if (error = namei(&nd))
++ if ((error = namei(&nd)) != 0)
+ goto out;
+ if (nd.ni_vp) {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+@@ -994,6 +1423,7 @@
+ * Delete a whiteout from the filesystem.
+ */
+ /* ARGSUSED */
++int
+ sys_undelete(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1023,7 +1453,7 @@
+ }
+
+ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+- if (error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE))
++ if ((error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE)) != 0)
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ vput(nd.ni_dvp);
+ return (error);
+@@ -1033,6 +1463,7 @@
+ * Delete a name from the filesystem.
+ */
+ /* ARGSUSED */
++int
+ sys_unlink(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1045,42 +1476,39 @@
+ int error;
+ struct nameidata nd;
+
+- NDINIT(&nd, DELETE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
+- if (error = namei(&nd))
++ NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
++ SCARG(uap, path), p);
++ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+- VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+- VOP_LOCK(vp);
+-
+- if (vp->v_type != VDIR ||
+- (error = suser(p->p_ucred, &p->p_acflag)) == 0) {
+- /*
+- * The root of a mounted filesystem cannot be deleted.
+- */
+- if (vp->v_flag & VROOT)
+- error = EBUSY;
+- else
+- (void)vnode_pager_uncache(vp);
+- }
+
+- if (!error) {
+- VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+- error = VOP_REMOVE(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
+- } else {
++ /*
++ * The root of a mounted filesystem cannot be deleted.
++ */
++ if (vp->v_flag & VROOT) {
+ VOP_ABORTOP(nd.ni_dvp, &nd.ni_cnd);
+ if (nd.ni_dvp == vp)
+ vrele(nd.ni_dvp);
+ else
+ vput(nd.ni_dvp);
+- if (vp != NULLVP)
+- vput(vp);
++ vput(vp);
++ error = EBUSY;
++ goto out;
+ }
++
++ (void)uvm_vnp_uncache(vp);
++
++ VOP_LEASE(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
++ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
++ error = VOP_REMOVE(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
++out:
+ return (error);
+ }
+
+ /*
+ * Reposition read/write file offset.
+ */
++int
+ sys_lseek(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1096,24 +1524,38 @@
+ register struct filedesc *fdp = p->p_fd;
+ register struct file *fp;
+ struct vattr vattr;
+- int error;
++ struct vnode *vp;
++ int error, special;
+
+- if ((u_int)SCARG(uap, fd) >= fdp->fd_nfiles ||
+- (fp = fdp->fd_ofiles[SCARG(uap, fd)]) == NULL)
++ if ((fp = fd_getfile(fdp, SCARG(uap, fd))) == NULL)
+ return (EBADF);
+ if (fp->f_type != DTYPE_VNODE)
+ return (ESPIPE);
++ vp = (struct vnode *)fp->f_data;
++ if (vp->v_type == VFIFO)
++ return (ESPIPE);
++ if (vp->v_type == VCHR)
++ special = 1;
++ else
++ special = 0;
+ switch (SCARG(uap, whence)) {
+- case L_INCR:
++ case SEEK_CUR:
++ if (!special && fp->f_offset + SCARG(uap, offset) < 0)
++ return (EINVAL);
+ fp->f_offset += SCARG(uap, offset);
+ break;
+- case L_XTND:
+- if (error =
+- VOP_GETATTR((struct vnode *)fp->f_data, &vattr, cred, p))
++ case SEEK_END:
++ error = VOP_GETATTR((struct vnode *)fp->f_data, &vattr,
++ cred, p);
++ if (error)
+ return (error);
++ if (!special && (off_t)vattr.va_size + SCARG(uap, offset) < 0)
++ return (EINVAL);
+ fp->f_offset = SCARG(uap, offset) + vattr.va_size;
+ break;
+- case L_SET:
++ case SEEK_SET:
++ if (!special && SCARG(uap, offset) < 0)
++ return (EINVAL);
+ fp->f_offset = SCARG(uap, offset);
+ break;
+ default:
+@@ -1126,6 +1568,7 @@
+ /*
+ * Check access permissions.
+ */
++int
+ sys_access(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1140,13 +1583,15 @@
+ int error, flags, t_gid, t_uid;
+ struct nameidata nd;
+
++ if (SCARG(uap, flags) & ~(R_OK | W_OK | X_OK))
++ return (EINVAL);
+ t_uid = cred->cr_uid;
+ t_gid = cred->cr_gid;
+ cred->cr_uid = p->p_cred->p_ruid;
+ cred->cr_gid = p->p_cred->p_rgid;
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+- if (error = namei(&nd))
++ if ((error = namei(&nd)) != 0)
+ goto out1;
+ vp = nd.ni_vp;
+
+@@ -1173,6 +1618,7 @@
+ * Get file status; this version follows links.
+ */
+ /* ARGSUSED */
++int
+ sys_stat(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1188,12 +1634,15 @@
+
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+- if (error = namei(&nd))
++ if ((error = namei(&nd)) != 0)
+ return (error);
+ error = vn_stat(nd.ni_vp, &sb, p);
+ vput(nd.ni_vp);
+ if (error)
+ return (error);
++ /* Don't let non-root see generation numbers (for NFS security) */
++ if (suser(p->p_ucred, &p->p_acflag))
++ sb.st_gen = 0;
+ error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
+ return (error);
+ }
+@@ -1202,6 +1651,7 @@
+ * Get file status; this version does not follow links.
+ */
+ /* ARGSUSED */
++int
+ sys_lstat(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1211,47 +1661,21 @@
+ syscallarg(char *) path;
+ syscallarg(struct stat *) ub;
+ } */ *uap = v;
++ struct stat sb;
+ int error;
+- struct vnode *vp, *dvp;
+- struct stat sb, sb1;
+ struct nameidata nd;
+
+- NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | LOCKPARENT, UIO_USERSPACE,
++ NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+- if (error = namei(&nd))
++ if ((error = namei(&nd)) != 0)
+ return (error);
+- /*
+- * For symbolic links, always return the attributes of its
+- * containing directory, except for mode, size, and links.
+- */
+- vp = nd.ni_vp;
+- dvp = nd.ni_dvp;
+- if (vp->v_type != VLNK) {
+- if (dvp == vp)
+- vrele(dvp);
+- else
+- vput(dvp);
+- error = vn_stat(vp, &sb, p);
+- vput(vp);
+- if (error)
+- return (error);
+- } else {
+- error = vn_stat(dvp, &sb, p);
+- vput(dvp);
+- if (error) {
+- vput(vp);
+- return (error);
+- }
+- error = vn_stat(vp, &sb1, p);
+- vput(vp);
+- if (error)
+- return (error);
+- sb.st_mode &= ~S_IFDIR;
+- sb.st_mode |= S_IFLNK;
+- sb.st_nlink = sb1.st_nlink;
+- sb.st_size = sb1.st_size;
+- sb.st_blocks = sb1.st_blocks;
+- }
++ error = vn_stat(nd.ni_vp, &sb, p);
++ vput(nd.ni_vp);
++ if (error)
++ return (error);
++ /* Don't let non-root see generation numbers (for NFS security) */
++ if (suser(p->p_ucred, &p->p_acflag))
++ sb.st_gen = 0;
+ error = copyout((caddr_t)&sb, (caddr_t)SCARG(uap, ub), sizeof (sb));
+ return (error);
+ }
+@@ -1260,6 +1684,7 @@
+ * Get configurable pathname variables.
+ */
+ /* ARGSUSED */
++int
+ sys_pathconf(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1274,7 +1699,7 @@
+
+ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+- if (error = namei(&nd))
++ if ((error = namei(&nd)) != 0)
+ return (error);
+ error = VOP_PATHCONF(nd.ni_vp, SCARG(uap, name), retval);
+ vput(nd.ni_vp);
+@@ -1285,6 +1710,7 @@
+ * Return target name of a symbolic link.
+ */
+ /* ARGSUSED */
++int
+ sys_readlink(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1293,7 +1719,7 @@
+ register struct sys_readlink_args /* {
+ syscallarg(char *) path;
+ syscallarg(char *) buf;
+- syscallarg(int) count;
++ syscallarg(size_t) count;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct iovec aiov;
+@@ -1303,7 +1729,7 @@
+
+ NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+- if (error = namei(&nd))
++ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ if (vp->v_type != VLNK)
+@@ -1329,6 +1755,7 @@
+ * Change flags of a file given a path name.
+ */
+ /* ARGSUSED */
++int
+ sys_chflags(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1336,7 +1763,7 @@
+ {
+ register struct sys_chflags_args /* {
+ syscallarg(char *) path;
+- syscallarg(int) flags;
++ syscallarg(unsigned int) flags;
+ } */ *uap = v;
+ register struct vnode *vp;
+ struct vattr vattr;
+@@ -1344,18 +1771,29 @@
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+- if (error = namei(&nd))
++ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+- VOP_LOCK(vp);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
++ else if (SCARG(uap, flags) == VNOVAL)
++ error = EINVAL;
+ else {
++ if (suser(p->p_ucred, &p->p_acflag)) {
++ if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
++ goto out;
++ if (vattr.va_type == VCHR || vattr.va_type == VBLK) {
++ error = EINVAL;
++ goto out;
++ }
++ }
+ VATTR_NULL(&vattr);
+ vattr.va_flags = SCARG(uap, flags);
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
++out:
+ vput(vp);
+ return (error);
+ }
+@@ -1364,33 +1802,47 @@
+ * Change flags of a file given a file descriptor.
+ */
+ /* ARGSUSED */
++int
+ sys_fchflags(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+ {
+- register struct sys_fchflags_args /* {
++ struct sys_fchflags_args /* {
+ syscallarg(int) fd;
+- syscallarg(int) flags;
++ syscallarg(unsigned int) flags;
+ } */ *uap = v;
+ struct vattr vattr;
+ struct vnode *vp;
+ struct file *fp;
+ int error;
+
+- if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
++ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+ vp = (struct vnode *)fp->f_data;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+- VOP_LOCK(vp);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
++ else if (SCARG(uap, flags) == VNOVAL)
++ error = EINVAL;
+ else {
++ if (suser(p->p_ucred, &p->p_acflag)) {
++ if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p))
++ != 0)
++ goto out;
++ if (vattr.va_type == VCHR || vattr.va_type == VBLK) {
++ error = EINVAL;
++ goto out;
++ }
++ }
+ VATTR_NULL(&vattr);
+ vattr.va_flags = SCARG(uap, flags);
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+- VOP_UNLOCK(vp);
++out:
++ VOP_UNLOCK(vp, 0, p);
++ FRELE(fp);
+ return (error);
+ }
+
+@@ -1398,6 +1850,7 @@
+ * Change mode of a file given path name.
+ */
+ /* ARGSUSED */
++int
+ sys_chmod(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1412,12 +1865,15 @@
+ int error;
+ struct nameidata nd;
+
++ if (SCARG(uap, mode) & ~(S_IFMT | ALLPERMS))
++ return (EINVAL);
++
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+- if (error = namei(&nd))
++ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+- VOP_LOCK(vp);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+@@ -1433,12 +1889,13 @@
+ * Change mode of a file given a file descriptor.
+ */
+ /* ARGSUSED */
++int
+ sys_fchmod(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+ {
+- register struct sys_fchmod_args /* {
++ struct sys_fchmod_args /* {
+ syscallarg(int) fd;
+ syscallarg(int) mode;
+ } */ *uap = v;
+@@ -1447,11 +1904,14 @@
+ struct file *fp;
+ int error;
+
+- if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
++ if (SCARG(uap, mode) & ~(S_IFMT | ALLPERMS))
++ return (EINVAL);
++
++ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+ vp = (struct vnode *)fp->f_data;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+- VOP_LOCK(vp);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+@@ -1459,7 +1919,8 @@
+ vattr.va_mode = SCARG(uap, mode) & ALLPERMS;
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+- VOP_UNLOCK(vp);
++ VOP_UNLOCK(vp, 0, p);
++ FRELE(fp);
+ return (error);
+ }
+
+@@ -1467,6 +1928,7 @@
+ * Set ownership given a path name.
+ */
+ /* ARGSUSED */
++int
+ sys_chown(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1481,21 +1943,87 @@
+ struct vattr vattr;
+ int error;
+ struct nameidata nd;
++ u_short mode;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+- if (error = namei(&nd))
++ if ((error = namei(&nd)) != 0)
++ return (error);
++ vp = nd.ni_vp;
++ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
++ if (vp->v_mount->mnt_flag & MNT_RDONLY)
++ error = EROFS;
++ else {
++ if ((SCARG(uap, uid) != -1 || SCARG(uap, gid) != -1) &&
++ (suser(p->p_ucred, &p->p_acflag) || suid_clear)) {
++ error = VOP_GETATTR(vp, &vattr, p->p_ucred, p);
++ if (error)
++ goto out;
++ mode = vattr.va_mode & ~(VSUID | VSGID);
++ if (mode == vattr.va_mode)
++ mode = VNOVAL;
++ }
++ else
++ mode = VNOVAL;
++ VATTR_NULL(&vattr);
++ vattr.va_uid = SCARG(uap, uid);
++ vattr.va_gid = SCARG(uap, gid);
++ vattr.va_mode = mode;
++ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
++ }
++out:
++ vput(vp);
++ return (error);
++}
++
++/*
++ * Set ownership given a path name, without following links.
++ */
++/* ARGSUSED */
++int
++sys_lchown(p, v, retval)
++ struct proc *p;
++ void *v;
++ register_t *retval;
++{
++ register struct sys_lchown_args /* {
++ syscallarg(char *) path;
++ syscallarg(int) uid;
++ syscallarg(int) gid;
++ } */ *uap = v;
++ register struct vnode *vp;
++ struct vattr vattr;
++ int error;
++ struct nameidata nd;
++ u_short mode;
++
++ NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
++ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+- VOP_LOCK(vp);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
++ if ((SCARG(uap, uid) != -1 || SCARG(uap, gid) != -1) &&
++ (suser(p->p_ucred, &p->p_acflag) || suid_clear)) {
++ error = VOP_GETATTR(vp, &vattr, p->p_ucred, p);
++ if (error)
++ goto out;
++ mode = vattr.va_mode & ~(VSUID | VSGID);
++ if (mode == vattr.va_mode)
++ mode = VNOVAL;
++ }
++ else
++ mode = VNOVAL;
+ VATTR_NULL(&vattr);
+ vattr.va_uid = SCARG(uap, uid);
+ vattr.va_gid = SCARG(uap, gid);
++ vattr.va_mode = mode;
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
++out:
+ vput(vp);
+ return (error);
+ }
+@@ -1504,42 +2032,58 @@
+ * Set ownership given a file descriptor.
+ */
+ /* ARGSUSED */
++int
+ sys_fchown(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+ {
+- register struct sys_fchown_args /* {
++ struct sys_fchown_args /* {
+ syscallarg(int) fd;
+ syscallarg(int) uid;
+ syscallarg(int) gid;
+ } */ *uap = v;
+- struct vattr vattr;
+ struct vnode *vp;
+- struct file *fp;
++ struct vattr vattr;
+ int error;
++ struct file *fp;
++ u_short mode;
+
+- if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
++ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+ vp = (struct vnode *)fp->f_data;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+- VOP_LOCK(vp);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
++ if ((SCARG(uap, uid) != -1 || SCARG(uap, gid) != -1) &&
++ (suser(p->p_ucred, &p->p_acflag) || suid_clear)) {
++ error = VOP_GETATTR(vp, &vattr, p->p_ucred, p);
++ if (error)
++ goto out;
++ mode = vattr.va_mode & ~(VSUID | VSGID);
++ if (mode == vattr.va_mode)
++ mode = VNOVAL;
++ } else
++ mode = VNOVAL;
+ VATTR_NULL(&vattr);
+ vattr.va_uid = SCARG(uap, uid);
+ vattr.va_gid = SCARG(uap, gid);
++ vattr.va_mode = mode;
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+- VOP_UNLOCK(vp);
++out:
++ VOP_UNLOCK(vp, 0, p);
++ FRELE(fp);
+ return (error);
+ }
+
+ /*
+- * Set the access and modification times of a file.
++ * Set the access and modification times given a path name.
+ */
+ /* ARGSUSED */
++int
+ sys_utimes(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1560,32 +2104,97 @@
+ microtime(&tv[0]);
+ tv[1] = tv[0];
+ vattr.va_vaflags |= VA_UTIMES_NULL;
+- } else if (error = copyin((caddr_t)SCARG(uap, tptr), (caddr_t)tv,
+- sizeof (tv)))
+- return (error);
++ } else {
++ error = copyin((caddr_t)SCARG(uap, tptr), (caddr_t)tv,
++ sizeof (tv));
++ if (error)
++ return (error);
++ /* XXX workaround timeval matching the VFS constant VNOVAL */
++ if (tv[0].tv_sec == VNOVAL)
++ tv[0].tv_sec = VNOVAL - 1;
++ if (tv[1].tv_sec == VNOVAL)
++ tv[1].tv_sec = VNOVAL - 1;
++ }
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+- if (error = namei(&nd))
++ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+- VOP_LOCK(vp);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_mount->mnt_flag & MNT_RDONLY)
+ error = EROFS;
+ else {
+- vattr.va_atime.ts_sec = tv[0].tv_sec;
+- vattr.va_atime.ts_nsec = tv[0].tv_usec * 1000;
+- vattr.va_mtime.ts_sec = tv[1].tv_sec;
+- vattr.va_mtime.ts_nsec = tv[1].tv_usec * 1000;
++ vattr.va_atime.tv_sec = tv[0].tv_sec;
++ vattr.va_atime.tv_nsec = tv[0].tv_usec * 1000;
++ vattr.va_mtime.tv_sec = tv[1].tv_sec;
++ vattr.va_mtime.tv_nsec = tv[1].tv_usec * 1000;
+ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
+ }
+ vput(vp);
+ return (error);
+ }
+
++
++/*
++ * Set the access and modification times given a file descriptor.
++ */
++/* ARGSUSED */
++int
++sys_futimes(p, v, retval)
++ struct proc *p;
++ void *v;
++ register_t *retval;
++{
++ register struct sys_futimes_args /* {
++ syscallarg(int) fd;
++ syscallarg(struct timeval *) tptr;
++ } */ *uap = v;
++ struct vnode *vp;
++ struct timeval tv[2];
++ struct vattr vattr;
++ int error;
++ struct file *fp;
++
++ VATTR_NULL(&vattr);
++ if (SCARG(uap, tptr) == NULL) {
++ microtime(&tv[0]);
++ tv[1] = tv[0];
++ vattr.va_vaflags |= VA_UTIMES_NULL;
++ } else {
++ error = copyin((caddr_t)SCARG(uap, tptr), (caddr_t)tv,
++ sizeof (tv));
++ if (error)
++ return (error);
++ /* XXX workaround timeval matching the VFS constant VNOVAL */
++ if (tv[0].tv_sec == VNOVAL)
++ tv[0].tv_sec = VNOVAL - 1;
++ if (tv[1].tv_sec == VNOVAL)
++ tv[1].tv_sec = VNOVAL - 1;
++ }
++ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
++ return (error);
++ vp = (struct vnode *)fp->f_data;
++ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
++ if (vp->v_mount->mnt_flag & MNT_RDONLY)
++ error = EROFS;
++ else {
++ vattr.va_atime.tv_sec = tv[0].tv_sec;
++ vattr.va_atime.tv_nsec = tv[0].tv_usec * 1000;
++ vattr.va_mtime.tv_sec = tv[1].tv_sec;
++ vattr.va_mtime.tv_nsec = tv[1].tv_usec * 1000;
++ error = VOP_SETATTR(vp, &vattr, p->p_ucred, p);
++ }
++ VOP_UNLOCK(vp, 0, p);
++ FRELE(fp);
++ return (error);
++}
++
+ /*
+ * Truncate a file given its path name.
+ */
+ /* ARGSUSED */
++int
+ sys_truncate(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1602,11 +2211,11 @@
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+- if (error = namei(&nd))
++ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+- VOP_LOCK(vp);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_type == VDIR)
+ error = EISDIR;
+ else if ((error = vn_writechk(vp)) == 0 &&
+@@ -1623,12 +2232,13 @@
+ * Truncate a file given a file descriptor.
+ */
+ /* ARGSUSED */
++int
+ sys_ftruncate(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+ {
+- register struct sys_ftruncate_args /* {
++ struct sys_ftruncate_args /* {
+ syscallarg(int) fd;
+ syscallarg(int) pad;
+ syscallarg(off_t) length;
+@@ -1638,13 +2248,15 @@
+ struct file *fp;
+ int error;
+
+- if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
++ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+- if ((fp->f_flag & FWRITE) == 0)
+- return (EINVAL);
++ if ((fp->f_flag & FWRITE) == 0) {
++ error = EINVAL;
++ goto bad;
++ }
+ vp = (struct vnode *)fp->f_data;
+ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
+- VOP_LOCK(vp);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ if (vp->v_type == VDIR)
+ error = EISDIR;
+ else if ((error = vn_writechk(vp)) == 0) {
+@@ -1652,7 +2264,9 @@
+ vattr.va_size = SCARG(uap, length);
+ error = VOP_SETATTR(vp, &vattr, fp->f_cred, p);
+ }
+- VOP_UNLOCK(vp);
++ VOP_UNLOCK(vp, 0, p);
++bad:
++ FRELE(fp);
+ return (error);
+ }
+
+@@ -1660,6 +2274,7 @@
+ * Sync an open file.
+ */
+ /* ARGSUSED */
++int
+ sys_fsync(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1668,16 +2283,22 @@
+ struct sys_fsync_args /* {
+ syscallarg(int) fd;
+ } */ *uap = v;
+- register struct vnode *vp;
++ struct vnode *vp;
+ struct file *fp;
+ int error;
+
+- if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
++ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
+ return (error);
+ vp = (struct vnode *)fp->f_data;
+- VOP_LOCK(vp);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ error = VOP_FSYNC(vp, fp->f_cred, MNT_WAIT, p);
+- VOP_UNLOCK(vp);
++#ifdef FFS_SOFTUPDATES
++ if (error == 0 && vp->v_mount && (vp->v_mount->mnt_flag & MNT_SOFTDEP))
++ error = softdep_fsync(vp);
++#endif
++
++ VOP_UNLOCK(vp, 0, p);
++ FRELE(fp);
+ return (error);
+ }
+
+@@ -1686,6 +2307,7 @@
+ * or both not be directories. If target is a directory, it must be empty.
+ */
+ /* ARGSUSED */
++int
+ sys_rename(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1698,15 +2320,24 @@
+ register struct vnode *tvp, *fvp, *tdvp;
+ struct nameidata fromnd, tond;
+ int error;
++ int flags;
+
+ NDINIT(&fromnd, DELETE, WANTPARENT | SAVESTART, UIO_USERSPACE,
+ SCARG(uap, from), p);
+- if (error = namei(&fromnd))
++ if ((error = namei(&fromnd)) != 0)
+ return (error);
+ fvp = fromnd.ni_vp;
+- NDINIT(&tond, RENAME, LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART,
++
++ flags = LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART;
++ /*
++ * rename("foo/", "bar/"); is OK
++ */
++ if (fvp->v_type == VDIR)
++ flags |= STRIPSLASHES;
++
++ NDINIT(&tond, RENAME, flags,
+ UIO_USERSPACE, SCARG(uap, to), p);
+- if (error = namei(&tond)) {
++ if ((error = namei(&tond)) != 0) {
+ VOP_ABORTOP(fromnd.ni_dvp, &fromnd.ni_cnd);
+ vrele(fromnd.ni_dvp);
+ vrele(fvp);
+@@ -1727,21 +2358,19 @@
+ error = EINVAL;
+ /*
+ * If source is the same as the destination (that is the
+- * same inode number with the same name in the same directory),
+- * then there is nothing to do.
++ * same inode number)
+ */
+- if (fvp == tvp && fromnd.ni_dvp == tdvp &&
+- fromnd.ni_cnd.cn_namelen == tond.ni_cnd.cn_namelen &&
+- !bcmp(fromnd.ni_cnd.cn_nameptr, tond.ni_cnd.cn_nameptr,
+- fromnd.ni_cnd.cn_namelen))
++ if (fvp == tvp)
+ error = -1;
+ out:
+ if (!error) {
+ VOP_LEASE(tdvp, p, p->p_ucred, LEASE_WRITE);
+ if (fromnd.ni_dvp != tdvp)
+ VOP_LEASE(fromnd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
+- if (tvp)
++ if (tvp) {
++ (void)uvm_vnp_uncache(tvp);
+ VOP_LEASE(tvp, p, p->p_ucred, LEASE_WRITE);
++ }
+ error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd,
+ tond.ni_dvp, tond.ni_vp, &tond.ni_cnd);
+ } else {
+@@ -1771,6 +2400,7 @@
+ * Make a directory file.
+ */
+ /* ARGSUSED */
++int
+ sys_mkdir(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1785,8 +2415,9 @@
+ int error;
+ struct nameidata nd;
+
+- NDINIT(&nd, CREATE, LOCKPARENT, UIO_USERSPACE, SCARG(uap, path), p);
+- if (error = namei(&nd))
++ NDINIT(&nd, CREATE, LOCKPARENT | STRIPSLASHES,
++ UIO_USERSPACE, SCARG(uap, path), p);
++ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ if (vp != NULL) {
+@@ -1812,6 +2443,7 @@
+ * Remove a directory file.
+ */
+ /* ARGSUSED */
++int
+ sys_rmdir(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -1826,7 +2458,7 @@
+
+ NDINIT(&nd, DELETE, LOCKPARENT | LOCKLEAF, UIO_USERSPACE,
+ SCARG(uap, path), p);
+- if (error = namei(&nd))
++ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+ if (vp->v_type != VDIR) {
+@@ -1837,7 +2469,7 @@
+ * No rmdir "." please.
+ */
+ if (nd.ni_dvp == vp) {
+- error = EINVAL;
++ error = EBUSY;
+ goto out;
+ }
+ /*
+@@ -1864,32 +2496,39 @@
+ /*
+ * Read a block of directory entries in a file system independent format.
+ */
++int
+ sys_getdirentries(p, v, retval)
+ struct proc *p;
+ void *v;
+ register_t *retval;
+ {
+- register struct sys_getdirentries_args /* {
++ struct sys_getdirentries_args /* {
+ syscallarg(int) fd;
+ syscallarg(char *) buf;
+- syscallarg(u_int) count;
++ syscallarg(int) count;
+ syscallarg(long *) basep;
+ } */ *uap = v;
+- register struct vnode *vp;
++ struct vnode *vp;
+ struct file *fp;
+ struct uio auio;
+ struct iovec aiov;
+ long loff;
+ int error, eofflag;
+
+- if (error = getvnode(p->p_fd, SCARG(uap, fd), &fp))
+- return (error);
+- if ((fp->f_flag & FREAD) == 0)
+- return (EBADF);
++ if (SCARG(uap, count) < 0)
++ return EINVAL;
++ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
++ return (error);
++ if ((fp->f_flag & FREAD) == 0) {
++ error = EBADF;
++ goto bad;
++ }
+ vp = (struct vnode *)fp->f_data;
+ unionread:
+- if (vp->v_type != VDIR)
+- return (EINVAL);
++ if (vp->v_type != VDIR) {
++ error = EINVAL;
++ goto bad;
++ }
+ aiov.iov_base = SCARG(uap, buf);
+ aiov.iov_len = SCARG(uap, count);
+ auio.uio_iov = &aiov;
+@@ -1898,57 +2537,19 @@
+ auio.uio_segflg = UIO_USERSPACE;
+ auio.uio_procp = p;
+ auio.uio_resid = SCARG(uap, count);
+- VOP_LOCK(vp);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+ loff = auio.uio_offset = fp->f_offset;
+- error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, (u_long *)0, 0);
++ error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, 0, 0);
+ fp->f_offset = auio.uio_offset;
+- VOP_UNLOCK(vp);
++ VOP_UNLOCK(vp, 0, p);
+ if (error)
+- return (error);
+-
+-#ifdef UNION
+-{
+- extern int (**union_vnodeop_p)();
+- extern struct vnode *union_dircache __P((struct vnode *));
+-
++ goto bad;
+ if ((SCARG(uap, count) == auio.uio_resid) &&
+- (vp->v_op == union_vnodeop_p)) {
+- struct vnode *lvp;
+-
+- lvp = union_dircache(vp);
+- if (lvp != NULLVP) {
+- struct vattr va;
+-
+- /*
+- * If the directory is opaque,
+- * then don't show lower entries
+- */
+- error = VOP_GETATTR(vp, &va, fp->f_cred, p);
+- if (va.va_flags & OPAQUE) {
+- vput(lvp);
+- lvp = NULL;
+- }
+- }
+-
+- if (lvp != NULLVP) {
+- error = VOP_OPEN(lvp, FREAD, fp->f_cred, p);
+- VOP_UNLOCK(lvp);
+-
+- if (error) {
+- vrele(lvp);
+- return (error);
+- }
+- fp->f_data = (caddr_t) lvp;
+- fp->f_offset = 0;
+- error = vn_close(vp, FREAD, fp->f_cred, p);
+- if (error)
+- return (error);
+- vp = lvp;
+- goto unionread;
+- }
+- }
+-}
+-#endif /* UNION */
++ union_check_p &&
++ (union_check_p(p, &vp, fp, auio, &error) != 0))
++ goto unionread;
++ if (error)
++ goto bad;
+
+ if ((SCARG(uap, count) == auio.uio_resid) &&
+ (vp->v_flag & VROOT) &&
+@@ -1964,6 +2565,8 @@
+ error = copyout((caddr_t)&loff, (caddr_t)SCARG(uap, basep),
+ sizeof(long));
+ *retval = SCARG(uap, count) - auio.uio_resid;
++bad:
++ FRELE(fp);
+ return (error);
+ }
+
+@@ -1983,7 +2586,7 @@
+
+ fdp = p->p_fd;
+ *retval = fdp->fd_cmask;
+- fdp->fd_cmask = SCARG(uap, newmask) & ALLPERMS;
++ fdp->fd_cmask = SCARG(uap, newmask) & ACCESSPERMS;
+ return (0);
+ }
+
+@@ -1992,6 +2595,7 @@
+ * away from vnode.
+ */
+ /* ARGSUSED */
++int
+ sys_revoke(p, v, retval)
+ struct proc *p;
+ void *v;
+@@ -2006,20 +2610,16 @@
+ struct nameidata nd;
+
+ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
+- if (error = namei(&nd))
++ if ((error = namei(&nd)) != 0)
+ return (error);
+ vp = nd.ni_vp;
+- if (vp->v_type != VCHR && vp->v_type != VBLK) {
+- error = EINVAL;
+- goto out;
+- }
+- if (error = VOP_GETATTR(vp, &vattr, p->p_ucred, p))
++ if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)) != 0)
+ goto out;
+ if (p->p_ucred->cr_uid != vattr.va_uid &&
+ (error = suser(p->p_ucred, &p->p_acflag)))
+ goto out;
+- if (vp->v_usecount > 1 || (vp->v_flag & VALIASED))
+- vgoneall(vp);
++ if (vp->v_usecount > 1 || (vp->v_flag & (VALIASED | VLAYER)))
++ VOP_REVOKE(vp, REVOKEALL);
+ out:
+ vrele(vp);
+ return (error);
+@@ -2027,7 +2627,10 @@
+
+ /*
+ * Convert a user file descriptor to a kernel file entry.
++ *
++ * On return *fpp is FREF:ed.
+ */
++int
+ getvnode(fdp, fd, fpp)
+ struct filedesc *fdp;
+ struct file **fpp;
+@@ -2035,11 +2638,579 @@
+ {
+ struct file *fp;
+
+- if ((u_int)fd >= fdp->fd_nfiles ||
+- (fp = fdp->fd_ofiles[fd]) == NULL)
++ if ((fp = fd_getfile(fdp, fd)) == NULL)
+ return (EBADF);
+ if (fp->f_type != DTYPE_VNODE)
+ return (EINVAL);
++ FREF(fp);
+ *fpp = fp;
++
+ return (0);
+ }
++
++/*
++ * Positional read system call.
++ */
++int
++sys_pread(p, v, retval)
++ struct proc *p;
++ void *v;
++ register_t *retval;
++{
++ struct sys_pread_args /* {
++ syscallarg(int) fd;
++ syscallarg(void *) buf;
++ syscallarg(size_t) nbyte;
++ syscallarg(int) pad;
++ syscallarg(off_t) offset;
++ } */ *uap = v;
++ struct filedesc *fdp = p->p_fd;
++ struct file *fp;
++ struct vnode *vp;
++ off_t offset;
++ int fd = SCARG(uap, fd);
++
++ if ((fp = fd_getfile(fdp, fd)) == NULL)
++ return (EBADF);
++ if ((fp->f_flag & FREAD) == 0)
++ return (EBADF);
++
++ vp = (struct vnode *)fp->f_data;
++ if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
++ return (ESPIPE);
++ }
++
++ offset = SCARG(uap, offset);
++
++ FREF(fp);
++
++ /* dofileread() will FRELE the descriptor for us */
++ return (dofileread(p, fd, fp, SCARG(uap, buf), SCARG(uap, nbyte),
++ &offset, retval));
++}
++
++/*
++ * Positional scatter read system call.
++ */
++int
++sys_preadv(p, v, retval)
++ struct proc *p;
++ void *v;
++ register_t *retval;
++{
++ struct sys_preadv_args /* {
++ syscallarg(int) fd;
++ syscallarg(const struct iovec *) iovp;
++ syscallarg(int) iovcnt;
++ syscallarg(int) pad;
++ syscallarg(off_t) offset;
++ } */ *uap = v;
++ struct filedesc *fdp = p->p_fd;
++ struct file *fp;
++ struct vnode *vp;
++ off_t offset;
++ int fd = SCARG(uap, fd);
++
++ if ((fp = fd_getfile(fdp, fd)) == NULL)
++ return (EBADF);
++ if ((fp->f_flag & FREAD) == 0)
++ return (EBADF);
++
++ vp = (struct vnode *)fp->f_data;
++ if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
++ return (ESPIPE);
++ }
++
++ FREF(fp);
++
++ offset = SCARG(uap, offset);
++
++ /* dofilereadv() will FRELE the descriptor for us */
++ return (dofilereadv(p, fd, fp, SCARG(uap, iovp), SCARG(uap, iovcnt),
++ &offset, retval));
++}
++
++/*
++ * Positional write system call.
++ */
++int
++sys_pwrite(p, v, retval)
++ struct proc *p;
++ void *v;
++ register_t *retval;
++{
++ struct sys_pwrite_args /* {
++ syscallarg(int) fd;
++ syscallarg(const void *) buf;
++ syscallarg(size_t) nbyte;
++ syscallarg(int) pad;
++ syscallarg(off_t) offset;
++ } */ *uap = v;
++ struct filedesc *fdp = p->p_fd;
++ struct file *fp;
++ struct vnode *vp;
++ off_t offset;
++ int fd = SCARG(uap, fd);
++
++ if ((fp = fd_getfile(fdp, fd)) == NULL)
++ return (EBADF);
++ if ((fp->f_flag & FWRITE) == 0)
++ return (EBADF);
++
++ vp = (struct vnode *)fp->f_data;
++ if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
++ return (ESPIPE);
++ }
++
++ FREF(fp);
++
++ offset = SCARG(uap, offset);
++
++ /* dofilewrite() will FRELE the descriptor for us */
++ return (dofilewrite(p, fd, fp, SCARG(uap, buf), SCARG(uap, nbyte),
++ &offset, retval));
++}
++
++
++/*
++ * Positional gather write system call.
++ */
++int
++sys_pwritev(p, v, retval)
++ struct proc *p;
++ void *v;
++ register_t *retval;
++{
++ struct sys_pwritev_args /* {
++ syscallarg(int) fd;
++ syscallarg(const struct iovec *) iovp;
++ syscallarg(int) iovcnt;
++ syscallarg(int) pad;
++ syscallarg(off_t) offset;
++ } */ *uap = v;
++ struct filedesc *fdp = p->p_fd;
++ struct file *fp;
++ struct vnode *vp;
++ off_t offset;
++ int fd = SCARG(uap, fd);
++
++ if ((fp = fd_getfile(fdp, fd)) == NULL)
++ return (EBADF);
++ if ((fp->f_flag & FWRITE) == 0)
++ return (EBADF);
++
++ vp = (struct vnode *)fp->f_data;
++ if (fp->f_type != DTYPE_VNODE || vp->v_type == VFIFO) {
++ return (ESPIPE);
++ }
++
++ FREF(fp);
++
++ offset = SCARG(uap, offset);
++
++ /* dofilewritev() will FRELE the descriptor for us */
++ return (dofilewritev(p, fd, fp, SCARG(uap, iovp), SCARG(uap, iovcnt),
++ &offset, retval));
++}
++
++#ifdef UFS_EXTATTR
++/*
++ * Syscall to push extended attribute configuration information into the
++ * VFS. Accepts a path, which it converts to a mountpoint, as well as
++ * a command (int cmd), and attribute name and misc data. For now, the
++ * attribute name is left in userspace for consumption by the VFS_op.
++ * It will probably be changed to be copied into sysspace by the
++ * syscall in the future, once issues with various consumers of the
++ * attribute code have raised their hands.
++ *
++ * Currently this is used only by UFS Extended Attributes.
++ */
++int
++sys_extattrctl(struct proc *p, void *v, register_t *reval)
++{
++ struct sys_extattrctl_args /* {
++ syscallarg(const char *) path;
++ syscallarg(int) cmd;
++ syscallarg(const char *) filename;
++ syscallarg(int) attrnamespace;
++ syscallarg(const char *) attrname;
++ } */ *uap = v;
++ struct vnode *filename_vp;
++ struct nameidata nd;
++ struct mount *mp;
++ char attrname[EXTATTR_MAXNAMELEN];
++ int error;
++
++ /*
++ * SCARG(uap, attrname) not always defined. We check again later
++ * when we invoke the VFS call so as to pass in NULL there if needed.
++ */
++ if (SCARG(uap, attrname) != NULL) {
++ error = copyinstr(SCARG(uap, attrname), attrname,
++ EXTATTR_MAXNAMELEN, NULL);
++ if (error)
++ return (error);
++ }
++
++ /*
++ * SCARG(uap, filename) not always defined. If it is, grab
++ * a vnode lock, which VFS_EXTATTRCTL() will later release.
++ */
++ filename_vp = NULL;
++ if (SCARG(uap, filename) != NULL) {
++ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_USERSPACE,
++ SCARG(uap, filename), p);
++ if ((error = namei(&nd)) != 0)
++ return (error);
++ filename_vp = nd.ni_vp;
++ }
++
++ /* SCARG(uap, path) always defined. */
++ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
++ if ((error = namei(&nd)) != 0) {
++ if (filename_vp != NULL)
++ vput(filename_vp);
++ return (error);
++ }
++
++ mp = nd.ni_vp->v_mount;
++ if (error) {
++ if (filename_vp != NULL)
++ vput(filename_vp);
++ return (error);
++ }
++
++ if (SCARG(uap, attrname) != NULL) {
++ error = VFS_EXTATTRCTL(mp, SCARG(uap, cmd), filename_vp,
++ SCARG(uap, attrnamespace), attrname, p);
++ } else {
++ error = VFS_EXTATTRCTL(mp, SCARG(uap, cmd), filename_vp,
++ SCARG(uap, attrnamespace), NULL, p);
++ }
++
++ /*
++ * VFS_EXTATTRCTL will have unlocked, but not de-ref'd,
++ * filename_vp, so vrele it if it is defined.
++ */
++ if (filename_vp != NULL)
++ vrele(filename_vp);
++
++ return (error);
++}
++
++/*-
++ * Set a named extended attribute on a file or directory
++ *
++ * Arguments: unlocked vnode "vp", attribute namespace "attrnamespace",
++ * kernelspace string pointer "attrname", userspace buffer
++ * pointer "data", buffer length "nbytes", thread "td".
++ * Returns: 0 on success, an error number otherwise
++ * Locks: none
++ * References: vp must be a valid reference for the duration of the call
++ */
++static int
++extattr_set_vp(struct vnode *vp, int attrnamespace, const char *attrname,
++ void *data, size_t nbytes, struct proc *p, register_t *retval)
++{
++ struct uio auio;
++ struct iovec aiov;
++ ssize_t cnt;
++ int error;
++
++ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
++
++ aiov.iov_base = data;
++ aiov.iov_len = nbytes;
++ auio.uio_iov = &aiov;
++ auio.uio_iovcnt = 1;
++ auio.uio_offset = 0;
++ if (nbytes > INT_MAX) {
++ error = EINVAL;
++ goto done;
++ }
++ auio.uio_resid = nbytes;
++ auio.uio_rw = UIO_WRITE;
++ auio.uio_segflg = UIO_USERSPACE;
++ auio.uio_procp = p;
++ cnt = nbytes;
++
++ error = VOP_SETEXTATTR(vp, attrnamespace, attrname, &auio,
++ p->p_ucred, p);
++ cnt -= auio.uio_resid;
++ retval[0] = cnt;
++
++done:
++ VOP_UNLOCK(vp, 0, p);
++ return (error);
++}
++
++int
++sys_extattr_set_file(struct proc *p, void *v, register_t *retval)
++{
++ struct sys_extattr_set_file_args /* {
++ syscallarg(const char *) path;
++ syscallarg(int) attrnamespace;
++ syscallarg(const char *) attrname;
++ syscallarg(void *) data;
++ syscallarg(size_t) nbytes;
++ } */ *uap = v;
++ struct nameidata nd;
++ char attrname[EXTATTR_MAXNAMELEN];
++ int error;
++
++ error = copyinstr(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN,
++ NULL);
++ if (error)
++ return (error);
++
++ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
++ if ((error = namei(&nd)) != 0)
++ return (error);
++
++ error = extattr_set_vp(nd.ni_vp, SCARG(uap, attrnamespace), attrname,
++ SCARG(uap, data), SCARG(uap, nbytes), p, retval);
++
++ vrele(nd.ni_vp);
++ return (error);
++}
++
++int
++sys_extattr_set_fd(struct proc *p, void *v, register_t *retval)
++{
++ struct sys_extattr_set_fd_args /* {
++ syscallarg(int) fd;
++ syscallarg(int) attrnamespace;
++ syscallarg(const char *) attrname;
++ syscallarg(struct iovec *) iovp;
++ syscallarg(int) iovcnt;
++ } */ *uap = v;
++ struct file *fp;
++ char attrname[EXTATTR_MAXNAMELEN];
++ int error;
++
++ error = copyinstr(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN,
++ NULL);
++ if (error)
++ return (error);
++
++ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
++ return (error);
++
++ error = extattr_set_vp((struct vnode *)fp->f_data,
++ SCARG(uap, attrnamespace), attrname, SCARG(uap, data),
++ SCARG(uap, nbytes), p, retval);
++ FRELE(fp);
++
++ return (error);
++}
++
++/*-
++ * Get a named extended attribute on a file or directory
++ *
++ * Arguments: unlocked vnode "vp", attribute namespace "attrnamespace",
++ * kernelspace string pointer "attrname", userspace buffer
++ * pointer "data", buffer length "nbytes", thread "td".
++ * Returns: 0 on success, an error number otherwise
++ * Locks: none
++ * References: vp must be a valid reference for the duration of the call
++ */
++static int
++extattr_get_vp(struct vnode *vp, int attrnamespace, const char *attrname,
++ void *data, size_t nbytes, struct proc *p, register_t *retval)
++{
++ struct uio auio;
++ struct iovec aiov;
++ ssize_t cnt;
++ size_t size;
++ int error;
++
++ VOP_LEASE(vp, p, p->p_ucred, LEASE_READ);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
++
++ /*
++ * Slightly unusual semantics: if the user provides a NULL data
++ * pointer, they don't want to receive the data, just the
++ * maximum read length.
++ */
++ if (data != NULL) {
++ aiov.iov_base = data;
++ aiov.iov_len = nbytes;
++ auio.uio_iov = &aiov;
++ auio.uio_offset = 0;
++ if (nbytes > INT_MAX) {
++ error = EINVAL;
++ goto done;
++ }
++ auio.uio_resid = nbytes;
++ auio.uio_rw = UIO_READ;
++ auio.uio_segflg = UIO_USERSPACE;
++ auio.uio_procp = p;
++ cnt = nbytes;
++ error = VOP_GETEXTATTR(vp, attrnamespace, attrname, &auio,
++ NULL, p->p_ucred, p);
++ cnt -= auio.uio_resid;
++ retval[0] = cnt;
++ } else {
++ error = VOP_GETEXTATTR(vp, attrnamespace, attrname, NULL,
++ &size, p->p_ucred, p);
++ retval[0] = size;
++ }
++done:
++ VOP_UNLOCK(vp, 0, p);
++ return (error);
++}
++
++int
++sys_extattr_get_file(p, v, retval)
++ struct proc *p;
++ void *v;
++ register_t *retval;
++{
++ struct sys_extattr_get_file_args /* {
++ syscallarg(const char *) path;
++ syscallarg(int) attrnamespace;
++ syscallarg(const char *) attrname;
++ syscallarg(void *) data;
++ syscallarg(size_t) nbytes;
++ } */ *uap = v;
++ struct nameidata nd;
++ char attrname[EXTATTR_MAXNAMELEN];
++ int error;
++
++ error = copyinstr(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN,
++ NULL);
++ if (error)
++ return (error);
++
++ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
++ if ((error = namei(&nd)) != 0)
++ return (error);
++
++ error = extattr_get_vp(nd.ni_vp, SCARG(uap, attrnamespace), attrname,
++ SCARG(uap, data), SCARG(uap, nbytes), p, retval);
++
++ vrele(nd.ni_vp);
++ return (error);
++}
++
++int
++sys_extattr_get_fd(p, v, retval)
++ struct proc *p;
++ void *v;
++ register_t *retval;
++{
++ struct sys_extattr_get_fd_args /* {
++ syscallarg(int) fd;
++ syscallarg(int) attrnamespace;
++ syscallarg(const char *) attrname;
++ syscallarg(void *) data;
++ syscallarg(size_t) nbytes;
++ } */ *uap = v;
++ struct file *fp;
++ char attrname[EXTATTR_MAXNAMELEN];
++ int error;
++
++ error = copyinstr(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN,
++ NULL);
++ if (error)
++ return (error);
++
++ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
++ return (error);
++
++ error = extattr_get_vp((struct vnode *)fp->f_data,
++ SCARG(uap, attrnamespace), attrname, SCARG(uap, data),
++ SCARG(uap, nbytes), p, retval);
++ FRELE(fp);
++
++ return (error);
++}
++
++/*
++ * extattr_delete_vp(): Delete a named extended attribute on a file or
++ * directory
++ *
++ * Arguments: unlocked vnode "vp", attribute namespace "attrnamespace",
++ * kernelspace string pointer "attrname", proc "p"
++ * Returns: 0 on success, an error number otherwise
++ * Locks: none
++ * References: vp must be a valid reference for the duration of the call
++ */
++static int
++extattr_delete_vp(struct vnode *vp, int attrnamespace, const char *attrname,
++ struct proc *p)
++{
++ int error;
++
++ VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
++ vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
++
++ error = VOP_SETEXTATTR(vp, attrnamespace, attrname, NULL,
++ p->p_ucred, p);
++
++ VOP_UNLOCK(vp, 0, p);
++ return (error);
++}
++
++int
++sys_extattr_delete_file(p, v, retval)
++ struct proc *p;
++ void *v;
++ register_t *retval;
++{
++ struct sys_extattr_delete_file_args /* {
++ syscallarg(int) fd;
++ syscallarg(int) attrnamespace;
++ syscallarg(const char *) attrname;
++ } */ *uap = v;
++ struct nameidata nd;
++ char attrname[EXTATTR_MAXNAMELEN];
++ int error;
++
++ error = copyinstr(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN,
++ NULL);
++ if (error)
++ return(error);
++
++ NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, path), p);
++ if ((error = namei(&nd)) != 0)
++ return(error);
++
++ error = extattr_delete_vp(nd.ni_vp, SCARG(uap, attrnamespace),
++ attrname, p);
++
++ vrele(nd.ni_vp);
++ return(error);
++}
++
++int
++sys_extattr_delete_fd(p, v, retval)
++ struct proc *p;
++ void *v;
++ register_t *retval;
++{
++ struct sys_extattr_delete_fd_args /* {
++ syscallarg(int) fd;
++ syscallarg(int) attrnamespace;
++ syscallarg(const char *) attrname;
++ } */ *uap = v;
++ struct file *fp;
++ char attrname[EXTATTR_MAXNAMELEN];
++ int error;
++
++ error = copyinstr(SCARG(uap, attrname), attrname, EXTATTR_MAXNAMELEN,
++ NULL);
++ if (error)
++ return (error);
++
++ if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
++ return (error);
++
++ error = extattr_delete_vp((struct vnode *)fp->f_data,
++ SCARG(uap, attrnamespace), attrname, p);
++ FRELE(fp);
++
++ return (error);
++}
++#endif
--- /dev/null
+++ b/sys/src/cmd/diff/test/diff.rc
@@ -1,0 +1,11 @@
+#!/bin/rc
+
+# tests with no line endings
+# are broken; fix them later
+# tests=`{seq 15}
+tests=(1 2 3 4 5 11 12 13 15)
+for(t in $tests){
+ echo ../$O.diff -u diff-t$t.1 diff-t$t.2
+ ../$O.diff -u diff-t$t.1 diff-t$t.2 > diff-t$t.out
+ cmp diff-t$t.out diff-t$t.expected #|| {echo 'failed '$t; exit failed}
+}
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t1.c
@@ -1,0 +1,10 @@
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t1.expected
@@ -1,0 +1,10 @@
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t1.l
@@ -1,0 +1,10 @@
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t1.r
@@ -1,0 +1,10 @@
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t2.c
@@ -1,0 +1,8 @@
+1
+2
+4
+5
+6
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t2.expected
@@ -1,0 +1,10 @@
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t2.l
@@ -1,0 +1,9 @@
+1
+2
+3
+4
+5
+6
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t2.r
@@ -1,0 +1,9 @@
+1
+2
+4
+5
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t3.c
@@ -1,0 +1,10 @@
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t3.expected
@@ -1,0 +1,16 @@
+1
+2
+3
+4
+<<<<<<<<<< merge-t3.l
+y
+========== original
+5
+========== merge-t3.r
+x
+>>>>>>>>>>
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t3.l
@@ -1,0 +1,10 @@
+1
+2
+3
+4
+y
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t3.r
@@ -1,0 +1,10 @@
+1
+2
+3
+4
+x
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t4.c
@@ -1,0 +1,10 @@
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t4.expected
@@ -1,0 +1,22 @@
+1
+2
+3
+<<<<<<<<<< merge-t4.l
+4
+x
+6
+7
+========== original
+4
+5
+6
+7
+========== merge-t4.r
+w
+x
+y
+z
+>>>>>>>>>>
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t4.l
@@ -1,0 +1,10 @@
+1
+2
+3
+4
+x
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t4.r
@@ -1,0 +1,10 @@
+1
+2
+3
+w
+x
+y
+z
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t5.c
@@ -1,0 +1,10 @@
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t5.expected
@@ -1,0 +1,25 @@
+1
+a
+b
+c
+2
+3
+<<<<<<<<<< merge-t5.l
+4
+x
+6
+7
+========== original
+4
+5
+6
+7
+========== merge-t5.r
+w
+x
+y
+z
+>>>>>>>>>>
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t5.l
@@ -1,0 +1,13 @@
+1
+a
+b
+c
+2
+3
+4
+x
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t5.r
@@ -1,0 +1,10 @@
+1
+2
+3
+w
+x
+y
+z
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t6.c
@@ -1,0 +1,10 @@
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t6.expected
@@ -1,0 +1,11 @@
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
+11
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t6.l
@@ -1,0 +1,11 @@
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
+11
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t6.r
@@ -1,0 +1,10 @@
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t7.c
@@ -1,0 +1,10 @@
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t7.expected
@@ -1,0 +1,12 @@
+0
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
+11
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t7.l
@@ -1,0 +1,12 @@
+0
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
+11
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t7.r
@@ -1,0 +1,10 @@
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t8.c
@@ -1,0 +1,1 @@
+foo
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t8.r
@@ -1,0 +1,1 @@
+foo
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t9.c
@@ -1,0 +1,2 @@
+1
+2
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t9.expected
@@ -1,0 +1,2 @@
+a
+b
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t9.l
@@ -1,0 +1,2 @@
+a
+2
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge-t9.r
@@ -1,0 +1,2 @@
+1
+b
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge.rc
@@ -1,0 +1,8 @@
+#!/bin/rc
+
+tests=`{seq 9}
+for(t in $tests){
+ echo ../$O.merge3 merge-t$t.l merge-t$t.c merge-t$t.r
+ ../$O.merge3 merge-t$t.l merge-t$t.c merge-t$t.r > merge-t$t.out
+ cmp merge-t$t.out merge-t$t.expected #|| {echo 'failed '$t; exit failed}
+}
--- /dev/null
+++ b/sys/src/cmd/diff/test/merge5.expected
@@ -1,0 +1,29 @@
+lc: (2,1)::(2,4)
+rc: (4,7)::(4,7)
+1
+a
+b
+c
+lc: (5,5)::(8,8)
+rc: (4,7)::(4,7)
+2
+3
+<<<<<<<<<< test/merge-t5.l
+4
+x
+6
+7
+========== original
+4
+5
+6
+7
+========== test/merge-t5.r
+w
+x
+y
+z
+>>>>>>>>>>
+8
+9
+10
--- /dev/null
+++ b/sys/src/cmd/diff/test/mkfile
@@ -1,0 +1,7 @@
+</$objtype/mkfile
+
+TEST=\
+ diff\
+ merge
+
+</sys/src/cmd/mktest