vac.c

这是一个同样来自贝尔实验室的和UNIX有着渊源的操作系统, 其简洁的设计和实现易于我们学习和理解

#include "stdinc.h"
#include "vac.h"
#include "dat.h"
#include "fns.h"

typedef struct Sink Sink;
typedef struct MetaSink MetaSink;
typedef struct DirSink DirSink;

struct Sink {
	VtSession *z;
	VtEntry dir;
	uchar *buf;
	uchar *pbuf[VtPointerDepth+1];
};

struct DirSink {
	Sink *sink;
	MetaSink *msink;
	ulong nentry;
	uchar *buf;
	uchar *p;	/* current pointer */
	uchar *ep;	/* end pointer */
};

struct MetaSink {
	Sink *sink;
	uchar *buf;
	int maxindex;
	int nindex;
	uchar *rp;	/* start of current record */
	uchar *p;	/* current pointer */
	uchar *ep;	/* end pointer */
};

static void usage(void);
static int strpCmp(void*, void*);
static void warn(char *fmt, ...);
static void cleanup(void);
static u64int unittoull(char *s);
static int vac(VtSession *z, char *argv[]);
static void vacFile(DirSink *dsink, char *lname, char *sname, VacFile*);
static void vacStdin(DirSink *dsink, char *name, VacFile *vf);
static void vacData(DirSink *dsink, int fd, char *lname, VacFile*, Dir*);
static void vacDir(DirSink *dsink, int fd, char *lname, char *sname, VacFile*);
static int vacMerge(DirSink *dsink, char *lname, char *sname);

Sink *sinkAlloc(VtSession *z, int psize, int dsize);
void sinkWrite(Sink *k, uchar *data, int n);
void sinkWriteScore(Sink *k, uchar *score, int n);
void sinkClose(Sink *k);
void sinkFree(Sink *k);

DirSink *dirSinkAlloc(VtSession *z, int psize, int dsize);
void dirSinkWrite(DirSink *k, VtEntry*);
void dirSinkWriteSink(DirSink *k, Sink*);
int dirSinkWriteFile(DirSink *k, VacFile *vf);
void dirSinkClose(DirSink *k);
void dirSinkFree(DirSink *k);

MetaSink *metaSinkAlloc(VtSession *z, int psize, int dsize);
void metaSinkPutc(MetaSink *k, int c);
void metaSinkPutString(MetaSink *k, char *s);
void metaSinkPutUint32(MetaSink *k, ulong x);
void metaSinkPutUint64(MetaSink *k, uvlong x);
void metaSinkWrite(MetaSink *k, uchar *data, int n);
void metaSinkWriteDir(MetaSink *ms, VacDir *vd);
void metaSinkEOR(MetaSink *k);
void metaSinkClose(MetaSink *k);
void metaSinkFree(MetaSink *k);
void plan9ToVacDir(VacDir*, Dir*, ulong entry, uvlong qid);

enum {
	Debug = 1,
	Version = 8,
	BlockSize = 8*1024,
	MaxExclude = 1000,
};

struct {
	ulong	file;
	ulong	sfile;
	ulong	data;
	ulong	sdata;
	ulong	skip;
	ulong	meta;
} stats;

int bsize = BlockSize;
int maxbsize;
char *oname, *dfile;
int verbose;
uvlong fileid = 1;
int qdiff;
char *exclude[MaxExclude];
int nexclude;
int nowrite;
int merge;
char *isi;

void
main(int argc, char *argv[])
{
	VtSession *z;
	char *host = nil;
	int statsFlag = 0;

	atexit(cleanup);

	ARGBEGIN{
	default:
		usage();
	case 'b':
		bsize = unittoull(EARGF(usage()));
		if(bsize == ~0)
			usage();
		break;
	case 'd':
		dfile = EARGF(usage());
		break;
	case 'e':
		if(nexclude >= MaxExclude)
			sysfatal("too many exclusions");
		exclude[nexclude++] = EARGF(usage());
		break;
	case 'f':
		oname = EARGF(usage());
		break;
	case 'h':
		host = EARGF(usage());
		break;
	case 'i':
		isi = EARGF(usage());
		break;
	case 'n':
		nowrite++;
		break;
	case 'm':
		merge++;
		break;
	case 'q':
		qdiff++;
		break;
	case 's':
		statsFlag++;
		break;
	case 'v':
		verbose++;
		break;
	}ARGEND;

	if(bsize < 512)
		bsize = 512;
	if(bsize > VtMaxLumpSize)
		bsize = VtMaxLumpSize;
	maxbsize = bsize;

	vtAttach();

	fmtinstall('V', vtScoreFmt);
	fmtinstall('R', vtErrFmt);

	z = vtDial(host, 0);
	if(z == nil)
		vtFatal("could not connect to server: %R");

	if(!vtConnect(z, 0))
		vtFatal("vtConnect: %R");

	qsort(exclude, nexclude, sizeof(char*), strpCmp);

	vac(z, argv);
	if(!vtSync(z))
		fprint(2,
	"%s: warning: could not ask server to flush pending writes: %R\n",
			argv0);

	if(statsFlag)
		fprint(2, "%s: files %ld:%ld data %ld:%ld:%ld meta %ld\n",
			argv0, stats.file, stats.sfile,
			stats.data, stats.skip, stats.sdata, stats.meta);
//packetStats();
	vtClose(z);
	vtDetach();

	exits(0);
}

static void
usage(void)
{
	fprint(2, "usage: %s [-amqsv] [-h host] [-d vacfile] [-b blocksize] [-i name] [-e exclude] [-f vacfile] file ... \n", argv0);
	exits("usage");
}

static int
strpCmp(void *p0, void *p1)
{
	return strcmp(*(char**)p0, *(char**)p1);
}


int
readBlock(int fd, uchar *buf, int n)
{
	int m, t = 0;

	while(t < n){
		m = read(fd, buf+t, n-t);
		if(m < 0)
			return -1;
		if(m == 0)
			break;
		t += m;
	}
	return t;
}

int
vacWrite(VtSession *z, uchar score[VtScoreSize], int type, uchar *buf, int n)
{
assert(n > 0);
	if(nowrite) {
		vtSha1(score, buf, n);
		return 1;
	}
	if(!vtWrite(z, score, type, buf, n))
		return 0;
	if(!vtSha1Check(score, buf, n)) {
		uchar score2[VtScoreSize];

		vtSha1(score2, buf, n);
		fprint(2, "%s: vtSha1Check: n = %d %V %V\n",
			argv0, n, score, score2);
		vtSetError("vtSha1Check failed");
		return 0;
	}
	return 1;
}


static int
vac(VtSession *z, char *argv[])
{
	DirSink *dsink, *ds;
	MetaSink *ms;
	VtRoot root;
	uchar score[VtScoreSize], buf[VtRootSize];
	char cwd[2048];
	int cd, i;
	char *cp2, *cp;
	VacFS *fs;
	VacFile *vff;
	int fd;
	Dir *dir;
	VacDir vd;

	if(getwd(cwd, sizeof(cwd)) == 0)
		sysfatal("can't find current directory: %r");

	dsink = dirSinkAlloc(z, bsize, bsize);

	fs = nil;
	if(dfile != nil) {
		fs = vfsOpen(z, dfile, 1, 10000);
		if(fs == nil)
			fprint(2, "%s: could not open diff: %s: %s\n",
				argv0, dfile, vtGetError());
	}


	if(oname != nil) {
		fd = create(oname, OWRITE, 0666);
		if(fd < 0)
			sysfatal("could not create file: %s: %r", oname);
	} else
		fd = 1;

	dir = dirfstat(fd);
	if(dir == nil)
		sysfatal("dirfstat failed: %r");

	for(; *argv; argv++) {
		cp2 = *argv;
		cd = 0;
		for (cp = *argv; *cp; cp++)
			if (*cp == '/')
				cp2 = cp;
		if (cp2 != *argv) {
			*cp2 = '\0';
			if (chdir(*argv) < 0)
				sysfatal("can't cd to %s: %r", *argv);
			*cp2 = '/';
			cp2++;
			cd = 1;
		}
		vff = nil;
		if(fs)
			vff = vfOpen(fs, cp2);
		vacFile(dsink, argv[0], cp2, vff);
		if(vff)
			vfDecRef(vff);
		if(cd && chdir(cwd) < 0)
			sysfatal("can't cd back to %s: %r", cwd);
	}

	if(isi) {
		vff = nil;
		if(fs)
			vff = vfOpen(fs, isi);
		vacStdin(dsink, isi, vff);
		if(vff)
			vfDecRef(vff);
	}

	dirSinkClose(dsink);

	/* build meta information for the root */
	ms = metaSinkAlloc(z, bsize, bsize);
	/* fake into a directory */
	dir->mode |= (dir->mode&0444)>>2;
	dir->qid.type |= QTDIR;
	dir->mode |= DMDIR;
	plan9ToVacDir(&vd, dir, 0, fileid++);
	if(strcmp(vd.elem, "/") == 0){
		vtMemFree(vd.elem);
		vd.elem = vtStrDup("root");
	}
	metaSinkWriteDir(ms, &vd);
	vdCleanup(&vd);
	metaSinkClose(ms);

	ds = dirSinkAlloc(z, bsize, bsize);
	dirSinkWriteSink(ds, dsink->sink);
	dirSinkWriteSink(ds, dsink->msink->sink);
	dirSinkWriteSink(ds, ms->sink);
	dirSinkClose(ds);

	memset(&root, 0, sizeof(root));
	root.version = VtRootVersion;
	strncpy(root.name, dir->name, sizeof(root.name));
	root.name[sizeof(root.name)-1] = 0;
	free(dir);
	sprint(root.type, "vac");
	memmove(root.score, ds->sink->dir.score, VtScoreSize);
	root.blockSize = maxbsize;
	if(fs != nil)
		vfsGetScore(fs, root.prev);

	metaSinkFree(ms);
	dirSinkFree(ds);
	dirSinkFree(dsink);
	if(fs != nil)
		vfsClose(fs);

	vtRootPack(&root, buf);
	if(!vacWrite(z, score, VtRootType, buf, VtRootSize))
		vtFatal("vacWrite failed: %s", vtGetError());

	fprint(fd, "vac:");
	for(i=0; i<VtScoreSize; i++)
		fprint(fd, "%.2x", score[i]);
	fprint(fd, "\n");

	/* avoid remove at cleanup */
	oname = nil;
	return 1;
}

static int
isExcluded(char *name)
{
	int bot, top, i, x;

	bot = 0;
	top = nexclude;
	while(bot < top) {
		i = (bot+top)>>1;
		x = strcmp(exclude[i], name);
		if(x == 0)
			return 1;
		if(x < 0)
			bot = i + 1;
		else /* x > 0 */
			top = i;
	}
	return 0;
}

static void
vacFile(DirSink *dsink, char *lname, char *sname, VacFile *vf)
{
	int fd;
	Dir *dir;
	Dir fake;
	VacDir vd;
	ulong entry;

	if(isExcluded(lname)) {
		warn("excluding: %s", lname);
		return;
	}

	if(merge && vacMerge(dsink, lname, sname))
		return;

	fd = open(sname, OREAD);
	if(fd < 0) {
		warn("could not open file: %s: %s", lname, vtOSError());

		/*
		 * fake up dsink & vf contents so we don't explode later.
		 * I'm not certain that this is needed, but it seems like
		 * a wise precaution.
		 */
		entry = dsink->nentry;
		/* pretend it's a plain file */
		dir = &fake;
		nulldir(dir);
		dir->type = 'M';
		dir->dev = 10;
		dir->qid = (Qid){ 10, 2, QTFILE};
		dir->mode = 0664;
		dir->atime = dir->mtime = time(nil);
		dir->length = 0;
		dir->name = sname;
		dir->uid = dir->gid = dir->muid = "missing";

		vacData(dsink, fd, lname, vf, dir);
		plan9ToVacDir(&vd, dir, entry, fileid++);
		metaSinkWriteDir(dsink->msink, &vd);
		vdCleanup(&vd);
		return;
	}

	if(verbose)
		fprint(2, "%s\n", lname);

	dir = dirfstat(fd);
	if(dir == nil) {
		warn("can't stat %s: %r", lname);
		close(fd);
		return;
	}

	entry = dsink->nentry;

	if(dir->mode & DMDIR)
		vacDir(dsink, fd, lname, sname, vf);
	else
		vacData(dsink, fd, lname, vf, dir);

	plan9ToVacDir(&vd, dir, entry, fileid++);
	metaSinkWriteDir(dsink->msink, &vd);
	vdCleanup(&vd);

	free(dir);
	close(fd);
}

static void
vacStdin(DirSink *dsink, char *name, VacFile *vf)
{
	Dir *dir;
	VacDir vd;
	ulong entry;

	if(verbose)
		fprint(2, "%s\n", "<stdio>");

	dir = dirfstat(0);
	if(dir == nil) {
		warn("can't stat <stdio>: %r");
		return;
	}

	entry = dsink->nentry;

	vacData(dsink, 0, "<stdin>", vf, dir);

	plan9ToVacDir(&vd, dir, entry, fileid++);
	vd.elem = vtStrDup(name);
	metaSinkWriteDir(dsink->msink, &vd);
	vdCleanup(&vd);

	free(dir);
}

static ulong
vacDataSkip(Sink *sink, VacFile *vf, int fd, ulong blocks, uchar *buf, char *lname)
{
	int n;
	ulong i;
	uchar score[VtScoreSize];

	/* skip blocks for append only files */
	if(seek(fd, (blocks-1)*bsize, 0) != (blocks-1)*bsize) {
		warn("error seeking: %s", lname);
		goto Err;
	}
	n = readBlock(fd, buf, bsize);
	if(n < bsize) {
		warn("error checking append only file: %s", lname);
		goto Err;
	}
	if(!vfGetBlockScore(vf, blocks-1, score) ||
	    !vtSha1Check(score, buf, n)) {
		warn("last block of append file did not match: %s", lname);
		goto Err;
	}

	for(i=0; i<blocks; i++) {
		if(!vfGetBlockScore(vf, i, score)) {
			warn("could not get score: %s: %lud", lname, i);
			seek(fd, i*bsize, 0);
			return i;
		}
		stats.skip++;
		sinkWriteScore(sink, score, bsize);
	}

	return i;
Err:
	seek(fd, 0, 0);
	return 0;
}

static void
vacData(DirSink *dsink, int fd, char *lname, VacFile *vf, Dir *dir)
{
	uchar *buf;
	Sink *sink;
	int n;
	uchar score[VtScoreSize];
	ulong block, same;
	VacDir vd;
	ulong vfblocks;

	vfblocks = 0;
	if(vf != nil && qdiff) {
		vfGetDir(vf, &vd);
		if(vd.mtime == dir->mtime && vd.size == dir->length &&
		    (!vd.plan9 ||
		     /* vd.p9path == dir->qid.path && */
		     vd.p9version == dir->qid.vers))
		if(dirSinkWriteFile(dsink, vf)) {
			stats.sfile++;
			vdCleanup(&vd);
			return;
		}

		/* look for an append only file */
		if((dir->mode&DMAPPEND) && vd.size < dir->length &&
		    vd.plan9 && vd.p9path == dir->qid.path)
			vfblocks = vd.size/bsize;

		vdCleanup(&vd);
	}
	stats.file++;

	buf = vtMemAlloc(bsize);
	sink = sinkAlloc(dsink->sink->z, bsize, bsize);
	block = 0;
	same = stats.sdata+stats.skip;

	if(vfblocks > 1)
		block += vacDataSkip(sink, vf, fd, vfblocks, buf, lname);

if(0) fprint(2, "vacData: %s: %ld\n", lname, block);
	for(;;) {
		n = readBlock(fd, buf, bsize);
		if(0 && n < 0)
			warn("file truncated due to read error: %s: %s",
				lname, vtOSError());
		if(n <= 0)
			break;
		if(vf != nil && vfGetBlockScore(vf, block, score) &&
		    vtSha1Check(score, buf, n)) {
			stats.sdata++;
			sinkWriteScore(sink, score, n);
		} else
			sinkWrite(sink, buf, n);
		block++;
	}
	same = stats.sdata+stats.skip - same;
	if(same && (dir->mode&DMAPPEND))
		if(0)fprint(2, "%s: total %lud same %lud:%lud diff %lud\n",
			lname, block, same, vfblocks, block-same);

	sinkClose(sink);
	dirSinkWriteSink(dsink, sink);
	sinkFree(sink);
	free(buf);
}

static void
vacDir(DirSink *dsink, int fd, char *lname, char *sname, VacFile *vf)
{
	Dir *dirs;
	char *ln, *sn;
	char *name;
	int i, nd;
	DirSink *ds;
	VacFile *vvf;

	/*
	 * if we could see the score underlying dir, we could quickly
	 * short-circuit further directory descent if vf (see vacfs(vf)->score)
	 * and dir had identical scores.
	 */

	ds = dirSinkAlloc(dsink->sink->z, bsize, bsize);
	while((nd = dirread(fd, &dirs)) > 0){
		for(i = 0; i < nd; i++){
			name = dirs[i].name;
			/* check for bad file names */
			if(name[0] == 0 || name[0] == '/' ||
			    strcmp(name, ".") == 0 || strcmp(name, "..") == 0)
				continue;