X-Git-Url: https://git.rrq.au/?a=blobdiff_plain;f=fusefile.c;h=01f4c58400f0d4a2197797ddebb6a415e27b808a;hb=74baf758e5465bb0675fb7a8e134e29467cfa7b9;hp=9c4b9648d226d11fb6a42c43e56ab5ac78447079;hpb=7059a8375f259a9d739fe6a642d0162ee83c30c8;p=rrq%2Ffusefile.git

diff --git a/fusefile.c b/fusefile.c
index 9c4b964..01f4c58 100644
--- a/fusefile.c
+++ b/fusefile.c
@@ -34,12 +34,18 @@
 #include <time.h>
 #include <errno.h>
 
+struct Region {
+    off_t beg;
+    off_t end;
+};
+
 struct Source {
     char *filename;
     ssize_t from;
     ssize_t to;
     ssize_t start; // starting position in concatenated file
     int fd;
+    int dirty;
 };
 
 static struct {
@@ -53,7 +59,224 @@ static struct {
     time_t mtime;
     time_t ctime;
 } times;
-    
+
+/**
+ * Overlay
+ */
+static struct {
+    struct Source source;
+    struct Region *table;
+    size_t count;
+    size_t limit;
+} overlay;
+
+static void usage();
+
+/**
+ * Find the nearest overlay.table region below pos. Returns the index,
+ * or -1 if there is none, i.e. pos < overlay.table[0].
+ */
+static ssize_t overlay_prior_fragment(off_t pos) {
+    size_t lo = 0, hi = overlay.count;
+    while ( lo < hi ) {
+	size_t m = ( lo + hi ) / 2;
+	if ( m == lo ) {
+	    return overlay.table[m].beg <= pos? m : -1;
+	}
+	if ( overlay.table[m].beg <= pos ) {
+	    lo = m;
+	} else {
+	    hi = m;
+	}
+    }
+    return -1;
+}
+
+/**
+ * Save the entry count for overlay.table as 64-bit integer
+ * immediately following the overlay content at the index
+ * corresponding to the fused file size.
+ */
+static void overlay_save_count() {
+    lseek( overlay.source.fd, overlay.source.to, SEEK_SET );
+    size_t size = sizeof( overlay.count );
+    char *p = (char *) &overlay.count ;
+    while ( size > 0 ) {
+	size_t n = write( overlay.source.fd, p, size );
+	if ( n < 0 ) {
+	    perror( overlay.source.filename );
+	    exit( 1 );
+	}
+	size -= n;
+	p += n;
+    }
+    if ( overlay.source.dirty++ > 1000 ) {
+	fsync( overlay.source.fd );
+	overlay.source.dirty = 0;
+    }
+}
+
+/**
+ * Update the on-disk cache of overlay.table between the given
+ * indexes. The table is laid out immediately following the table
+ * count with each region saved as two 64-bit unsigned integers.
+ */
+static void overlay_save_table(size_t lo,size_t hi) {
+    char *p = (char *) &overlay.table[ lo ];
+    size_t pos =  overlay.source.to + sizeof( overlay.count ) +
+	lo * sizeof( struct Region );
+    size_t size = ( hi - lo ) * sizeof( struct Region );
+    if ( pos != lseek( overlay.source.fd, pos, SEEK_SET ) ) {
+	fprintf( stderr, "%s: seek error\n", overlay.source.filename );
+	exit( 1 );
+    }
+    while ( size > 0 ) {
+	size_t n = write( overlay.source.fd, p, size );
+	if ( n < 0 ) {
+	    perror( overlay.source.filename );
+	    exit( 1 );
+	}
+	size -= n;
+	p += n;
+    }
+    if ( overlay.source.dirty++ > 1000 ) {
+	fsync( overlay.source.fd );
+	overlay.source.dirty = 0;
+    }
+}
+
+/**
+ * Insert a new region at index p, with previous portion [p,count]
+ * moved up to make space.
+ */
+static void overlay_insert(size_t p,off_t beg,off_t end) {
+    size_t bytes;
+    // Grow the table if needed
+    if ( overlay.count >= overlay.limit ) {
+	overlay.limit = overlay.count + 10;
+	bytes = overlay.limit * sizeof( struct Region );
+	overlay.table = overlay.table?
+	    realloc( overlay.table, bytes ) : malloc( bytes );
+    }
+    bytes = ( overlay.count++ - p ) * sizeof( struct Region );
+    if ( bytes ) {
+	memmove( (char*) &overlay.table[ p+1 ],
+		 (char*) &overlay.table[ p ],
+		 bytes );
+    }
+    overlay.table[ p ].beg = beg;
+    overlay.table[ p ].end = end;
+    overlay_save_count();
+}
+
+/**
+ * Delete the region entry at p by moving the portion [p+1,count]
+ * down.
+ */
+static void overlay_delete(size_t p) {
+    size_t bytes = ( --overlay.count - p ) * sizeof( struct Region );
+    if ( bytes ) {
+	memmove( (char*) &overlay.table[ p ],
+		 (char*) &overlay.table[ p+1 ],
+		 bytes );
+    }
+}
+
+/**
+ * Mark the given region as updated, i.e. written to the overlay. The
+ * mark region may attach to prior marked regions ro be a new,
+ * separate region. If attaching, it causes the prior regions to
+ * expand and the table adjusted by deleting any regions that become
+ * fully contained in other regions.
+ */
+static void overlay_mark(off_t beg,off_t end) {
+#if DEBUG
+    fprintf( stderr, "overlay_mark( %ld, %ld )\n", beg, end );
+#endif
+    int deleted = 0;
+    ssize_t q;
+    ssize_t p = overlay_prior_fragment( beg );
+    // p is the nearest region below or at beg (or -1)
+    if ( p >= 0 && beg <= overlay.table[p].end ) {
+	// p overlaps mark region
+	if ( end <= overlay.table[p].end ) {
+	    // region p covers mark region already
+#if DEBUG
+	    fprintf( stderr, "overlay covering ( %ld %ld )\n",
+		     overlay.table[p].beg, overlay.table[p].end );
+#endif
+	    return;
+	}
+	// the new mark region extends region p
+	overlay.table[p].end = end;
+	q = p+1;
+	while ( q < overlay.count &&
+		overlay.table[q].beg <= overlay.table[p].end ) {
+	    // Extended region merges with subsequent region
+	    if ( overlay.table[p].end < overlay.table[q].end ) {
+		overlay.table[p].end = overlay.table[q].end;
+	    }
+	    overlay_delete( q );
+	    deleted++;
+	}
+	if ( deleted ) {
+	    overlay_save_count();
+	    q = overlay.count;
+	}
+	overlay_save_table( p, q );
+#if DEBUG
+	fprintf( stderr, "overlay expand ( %ld %ld ) deleted %d\n",
+		 overlay.table[p].beg, overlay.table[p].end, deleted );
+#endif
+	return;
+    }
+    // The prior region p does not expand into new mark region
+    p++; // subsequent region 
+    if ( p >= overlay.count || end < overlay.table[p].beg ) {
+	// New mark region is a separate region at p
+	overlay_insert( p, beg, end );
+#if DEBUG
+	fprintf( stderr, "overlay new ( %ld %ld )\n",
+		 overlay.table[p].beg, overlay.table[p].end );
+#endif
+	overlay_save_table( p, overlay.count );
+	return;
+    }
+    // New marks start before and overlap with region p => change p
+    // and handle any subsequent regions being covered
+    overlay.table[p].beg = beg;
+    q = p+1;
+    if ( overlay.table[p].end < end ) {
+	overlay.table[p].end = end;
+	while ( q < overlay.count &&
+		overlay.table[q].beg <= overlay.table[p].end ) {
+	    if ( overlay.table[p].end < overlay.table[q].end ) {
+		overlay.table[p].end = overlay.table[q].end;
+	    }
+	    overlay_delete( q );
+	    deleted++;
+	}
+	if ( deleted ) {
+	    overlay_save_count();
+	    q = overlay.count;
+	}
+    }
+    overlay_save_table( p, q );
+#if DEBUG
+    fprintf( stderr, "overlay before ( %ld %ld ) deleted %d\n",
+	     overlay.table[p].beg, overlay.table[p].end, deleted );
+#endif
+}
+
+static void setup_overlay(char *filename) {
+    overlay.source.filename = filename;
+    overlay.source.fd = open( filename, O_RDWR | O_CREAT, S_IRUSR | S_IWUSR );
+    if ( overlay.source.fd < 0 ) {
+	perror( filename );
+	usage();
+    }
+}
+
 #if DEBUG
 static void print_source(struct Source *p) {
     fprintf( stderr, "%p { %s, %ld, %ld, %ld, %d }\n",
@@ -117,10 +340,10 @@ static int setup_sources(char **argv,int i,int n) {
 	    } else if ( RANGE( sscanf( range, "%d:%n", &a, &c ), 1 )) {
 		p->from = ( a < 0 )? ( p->to + a ) : a;
 	    } else if ( RANGE( sscanf( range, "%d%n", &a, &c ), 1 )) {
-		if ( a < 0 ) {
-		    p->from = p->to + a;
+		if ( a >= 0 ) {
+		    p->from = a;
 		} else {
-		    p->to = p->from + a;
+		    p->from = p->to + a;
 		}
 	    } else if ( RANGE( sscanf( range, ":%n", &c), 0 ) ) {
 		// to end from start
@@ -129,8 +352,16 @@ static int setup_sources(char **argv,int i,int n) {
 		return 1;
 	    }
 	}
-	if ( p->from >= p->to ||
-	     p->from >= filestat.st_size || p->to > filestat.st_size ) {
+	if ( ( filestat.st_mode &  S_IFMT ) == S_IFCHR ) {
+	    filestat.st_size = p->to; // Pretend size of character device
+	}
+	if ( p->from < 0 ) {
+	    p->from = 0;
+	}
+	if ( p->to > filestat.st_size ) {
+	    p->to = filestat.st_size;
+	}
+	if ( p->from >= p->to || p->from >= filestat.st_size ) {
 	    fprintf( stderr, "** BAD RANGE: %s [%ld:%ld]\n",
 		     argv[i], p->from, p->to );
 	    return 1;
@@ -192,17 +423,62 @@ static int find_source(off_t offset) {
     if ( offset >= sources.size ) {
 	return -1;
     }
+#if DEBUG
+    fprintf( stderr, "find_source( %ld )\n", offset );
+#endif
     while ( lo + 1 < hi ) {
 	int m = ( lo + hi ) / 2;
 	if ( offset < sources.array[ m ].start ) {
+#if DEBUG
+	    fprintf( stderr, "  offset < [%d].start: %ld\n",
+		     m, sources.array[ m ].start );
+#endif
 	    hi = m;
 	} else {
+#if DEBUG
+	    fprintf( stderr, "  offset >= [%d].start: %ld\n",
+		     m, sources.array[ m ].start );
+#endif
 	    lo = m;
 	}
     }
+#if DEBUG
+    fprintf( stderr, "found %d\n", lo );
+#endif
     return lo;
 }
 
+static int overlay_merge(char *buf,off_t beg,off_t end) {
+#if DEBUG
+    fprintf( stderr, "merge %ld %ld\n", beg, end );
+#endif
+    // Find nearest overlay data before or at beg
+    ssize_t p = overlay_prior_fragment( beg );
+    if ( p < 0 ) {
+	p = 0;
+    }
+    for ( ; p < overlay.count && overlay.table[p].beg < end; p++ ) {
+	if ( overlay.table[p].end < beg ) {
+	    continue;
+	}
+	if ( overlay.table[p].beg > beg ) {
+	    size_t delta = overlay.table[p].beg - beg;
+	    buf += delta;
+	    beg += delta;
+	}
+	size_t size = ( overlay.table[p].end <= end )?
+	    ( overlay.table[p].end - beg ) : ( end - beg ); 
+	lseek( overlay.source.fd, beg, SEEK_SET );
+	while ( size > 0 ) {
+	    size_t n = read( overlay.source.fd, buf, size );
+	    size -= n;
+	    buf += n;
+	    beg += n; //
+	}
+    }
+    return 0;
+}
+
 // Read <size> bytes from <offset> in file
 static int fusefile_read(const char *path, char *buf, size_t size,
 			 off_t off, struct fuse_file_info *fi)
@@ -216,10 +492,10 @@ static int fusefile_read(const char *path, char *buf, size_t size,
 #if DEBUG
     fprintf( stderr, "read %ld %ld\n", off, size );
 #endif
-    size_t rr = 0;
+    size_t rr = 0; // total reading
     while ( size > 0 ) {
 #if DEBUG
-	fprintf( stderr, "find_source %ld %ld\n", off, size );
+	fprintf( stderr, "  find_source %ld %ld\n", off, size );
 #endif
 	int i = find_source( off );
 	if ( i < 0 ) {
@@ -237,16 +513,24 @@ static int fusefile_read(const char *path, char *buf, size_t size,
 	if ( n > size ) {
 	    n = size;
 	}
+	if ( sources.array[i].dirty ) {
+	    fsync( sources.array[i].fd );
+	    sources.array[i].dirty = 0;
+	}
+#if DEBUG
+	fprintf( stderr, "  seek fd=%d to %ld\n", sources.array[i].fd, b );
+#endif
 	if ( lseek( sources.array[i].fd, b, SEEK_SET ) < 0 ) {
 	    perror( sources.array[i].filename );
 	    return -ENOENT;
 	}
 #if DEBUG
-	fprintf( stderr, "get %ld bytes at %ld\n", n, rr );
+	fprintf( stderr, "  now read %ld from fd=%d\n",
+		 n, sources.array[i].fd );
 #endif
 	ssize_t r = read( sources.array[i].fd, buf + rr, n );
 #if DEBUG
-	fprintf( stderr, "got %ld bytes\n", r );
+	fprintf( stderr, "  got %ld bytes\n", r );
 #endif
 	if ( r < 0 ) {
 	    perror( sources.array[i].filename );
@@ -255,10 +539,23 @@ static int fusefile_read(const char *path, char *buf, size_t size,
 	if ( r == 0 ) {
 	    break;
 	}
+	if ( overlay.source.filename ) {
+	    if ( overlay.source.dirty ) {
+		fsync( overlay.source.fd );
+		overlay.source.dirty = 0;
+	    }
+	    int x = overlay_merge( buf + rr, off + rr, off + rr + r );
+	    if ( x ) {
+		return x;
+	    }
+	}
 	rr += r;
 	off += r;
 	size -= r;
     }
+#if DEBUG
+    fprintf( stderr, "  total reading %ld bytes\n", rr );
+#endif
     return rr;
 }
 
@@ -280,6 +577,30 @@ int fusefile_poll(const char *path, struct fuse_file_info *fi,
     return 0;
 }
 
+static void overlay_load() {
+    lseek( overlay.source.fd, overlay.source.to, SEEK_SET );
+    size_t x = 0;
+    size_t size = sizeof( overlay.count );
+    if ( read( overlay.source.fd, &x, size ) != size ) {
+	return;
+    }
+#if DEBUG
+    fprintf( stderr, "overlay: %s with %ld regions\n",
+	     overlay.source.filename, x );
+#endif
+    struct Region f = { 0, 0 };
+    size = sizeof( struct Region );
+    while ( x-- > 0 ) {
+	if ( read( overlay.source.fd, &f, size ) != size ) {
+	    fprintf( stderr, "%s: bad meta data\n", overlay.source.filename );
+	    exit( 1 );
+	}
+#if DEBUG
+	fprintf( stderr, "overlay region: %ld %ld\n", f.beg, f.end );
+#endif
+	overlay_mark( f.beg, f.end );
+    }
+}
 
 /**
  * Write a full block of data over the sources at the offset
@@ -288,12 +609,16 @@ static int write_block(off_t off,const char *buf,size_t size) {
 #if DEBUG
     fprintf( stderr, "write_block( %ld, ?, %ld )\n", off, size );
 #endif
+    if ( overlay.source.filename ) {
+	overlay_mark( off, off + size ); // Mark region as written
+    }
     while ( size > 0 ) {
 	int index = find_source( off ); // index of source file
 	if ( index < 0 ) {
 	    return -EIO; // past EOF
 	}
-	struct Source *source = &sources.array[ index ];
+	struct Source *source = overlay.source.filename?
+	    &overlay.source :  &sources.array[ index ];
 	off_t from = off - source->start + source->from;
 	off_t max = source->to - from;
 	if ( lseek( source->fd, from, SEEK_SET ) < 0 ) {
@@ -311,6 +636,10 @@ static int write_block(off_t off,const char *buf,size_t size) {
 	    size -= n;
 	    off += n;
 	}
+	if ( source->dirty++ >= 1000 ) {
+	    fsync( source->fd );
+	    source->dirty = 0;
+	}
     }
     return 0;
 }
@@ -374,6 +703,20 @@ static void fusefile_destroy(void *data) {
     }
 }
 
+static void fsync_all_dirty() {
+    int i = 0;
+    for ( ; i < sources.count; i++ ) {
+	if ( sources.array[i].dirty ) {
+	    fsync( sources.array[i].fd );
+	    sources.array[i].dirty = 0;
+	}
+    }
+    if ( overlay.source.filename && overlay.source.dirty ) {
+	fsync( overlay.source.fd );
+	overlay.source.dirty = 0;
+    }
+}
+
 static int fusefile_flush(const char *path, struct fuse_file_info *info) {
 #if DEBUG
     fprintf( stderr, "fusefile_flush( %s )\n", path );
@@ -381,6 +724,7 @@ static int fusefile_flush(const char *path, struct fuse_file_info *info) {
     if ( strcmp( path, "/" ) != 0 ) {
 	return -ENOENT;
     }
+    fsync_all_dirty();
     return 0;
 }
 
@@ -401,6 +745,7 @@ static int fusefile_fsync(const char *path, int x, struct fuse_file_info *fi) {
     if ( strcmp( path, "/" ) != 0 ) {
 	return -ENOENT;
     }
+    fsync_all_dirty();
     return 0;
 }
 
@@ -451,7 +796,7 @@ static struct fuse_operations fusefile_oper = {
 static void usage() {
     char *usage =
 "Usage: fusefile [ <fuse options> ] <mount> <file/from-to> ... \n"
-"Mounts a virtual, read-only file that is a concatenation of file fragments\n"
+"Mounts a virtual, file that is a concatenation of file fragments\n"
 	;
     fprintf( stderr, "%s", usage );
     exit( 1 );
@@ -510,9 +855,22 @@ int main(int argc, char *argv[])
     }
     fuseargc = i;
     mnt = argv[ i++ ]; // First non-option argument is the mount pount
+    char *overlaytag = "-overlay:";
+    int overlaytagsize = strlen( overlaytag );
+    if ( strncmp( argv[i], overlaytag, overlaytagsize ) == 0 ) {
+	// consume "-overlay:filename"
+	setup_overlay( argv[i++] + overlaytagsize ); // Need a writable file
+	if ( i >= argc ) {
+	    usage();
+	}
+    }
     if ( setup_sources( argv, i, argc-i ) ) {
 	return 1;
     }
+    if ( overlay.source.filename ) {
+	overlay.source.to = sources.size; // Register total size.
+	overlay_load();
+    }
     if ( stat( mnt, &stbuf ) == -1 ) {
 	int fd = open( mnt, O_CREAT | O_RDWR, S_IRUSR | S_IWUSR );
 	if ( fd < 0 ) {
@@ -534,6 +892,16 @@ int main(int argc, char *argv[])
 	times.ctime = stbuf.st_ctime;
     }
 
+    {
+	int fd = open( mnt, O_RDWR, S_IRUSR | S_IWUSR );
+	if ( fd < 0 ) {
+	    perror( mnt );
+	    return 1;
+	}
+	if ( lseek( fd, sources.size, SEEK_SET ) < 0 ) {
+	    return -EIO;
+	}
+    }
     fuseargc = setup_argv( fuseargc, &argv );
     struct fuse_args args = FUSE_ARGS_INIT( fuseargc, argv );
     if ( fuse_parse_cmdline( &args, &mnt, &mt, &fg ) ) {