threads / patch / 9638

patchDon't segfault if we failed to inflate a packed delta

Subject: [PATCH] Don't segfault if we failed to inflate a packed delta

## tl;dr

5 messages between Aug 25, 2007 and Aug 25, 2007. Diffs are folded; open one to read it.

replies: 4people: 3as markdown or json

Shawn O. Pearce· Aug 25, 2007, 07:26 UTC · lore

Under some types of packfile corruption the zlib stream holding the data for a delta within a packfile may fail to inflate, due to say a CRC failure within the compressed data itself. When this occurs the unpack_compressed_entry function will return NULL as a signal to the caller that the data is not available. Unfortunately we then tried to use that NULL as though it referenced a memory location where a delta was stored and tried to apply it to the delta base. Loading a byte from the NULL address typically causes a SIGSEGV.

cate on #git noticed this failure in `git fsck --full` where the call to verify_pack() first noticed that the packfile was corrupt by finding that the packfile's SHA-1 did not match the raw data of the file. After finding this fsck went ahead and tried to verify every object within the packfile, even though the packfile was already known to be bad. If we are going to shovel bad data at the delta unpacking code, we better handle it correctly.

Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
---
 sha1_file.c |    4 ++++
 1 files changed, 4 insertions(+), 0 deletions(-)

Show changes to sha1_file.c +4 −0

diff --git a/sha1_file.c b/sha1_file.c
index b219d4d..9978a58 100644
--- a/sha1_file.c
+++ b/sha1_file.c
@@ -1570,6 +1570,10 @@ static void *unpack_delta_entry(struct packed_git *p,
 		    (uintmax_t)base_offset, p->pack_name);
 
 	delta_data = unpack_compressed_entry(p, w_curs, curpos, delta_size);
+	if (!delta_data)
+		die("failed to unpack compressed delta"
+		    " at %"PRIuMAX" from %s",
+		    (uintmax_t)curpos, p->pack_name);
 	result = patch_delta(base, base_size,
 			     delta_data, delta_size,
 			     sizep);

-- 
1.5.3.rc6.17.g1911

Junio C Hamano· Aug 25, 2007, 08:30 UTC · re: Shawn O. Pearce · lore

[PATCH 1/2] blame: check return value from read_sha1_file()

(Trivial #1).

Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin-blame.c |   11 +++++++++++
 1 files changed, 11 insertions(+), 0 deletions(-)

Show changes to builtin-blame.c +11 −0

diff --git a/builtin-blame.c b/builtin-blame.c
index 0519339..dc88a95 100644
--- a/builtin-blame.c
+++ b/builtin-blame.c
@@ -98,6 +98,10 @@ static char *fill_origin_blob(struct origin *o, mmfile_t *file)
 		num_read_blob++;
 		file->ptr = read_sha1_file(o->blob_sha1, &type,
 					   (unsigned long *)(&(file->size)));
+		if (!file->ptr)
+			die("Cannot read blob %s for path %s",
+			    sha1_to_hex(o->blob_sha1),
+			    o->path);
 		o->file = *file;
 	}
 	else
@@ -1384,6 +1388,9 @@ static void get_commit_info(struct commit *commit,
 		unsigned long size;
 		commit->buffer =
 			read_sha1_file(commit->object.sha1, &type, &size);
+		if (!commit->buffer)
+			die("Cannot read commit %s",
+			    sha1_to_hex(commit->object.sha1));
 	}
 	ret->author = author_buf;
 	get_ac_line(commit->buffer, "\nauthor ",
@@ -2382,6 +2389,10 @@ int cmd_blame(int argc, const char **argv, const char *prefix)
 
 		sb.final_buf = read_sha1_file(o->blob_sha1, &type,
 					      &sb.final_buf_size);
+		if (!sb.final_buf)
+			die("Cannot read blob %s for path %s",
+			    sha1_to_hex(o->blob_sha1),
+			    path);
 	}
 	num_read_blob++;
 	lno = prepare_lines(&sb);

-- 
1.5.3.rc6.23.g0058

Junio C Hamano· Aug 25, 2007, 08:30 UTC · re: Shawn O. Pearce · lore

[PATCH 2/2] pack-objects: check return value from read_sha1_file()

(Trivial #2).

Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin-pack-objects.c |    6 ++++++
 1 files changed, 6 insertions(+), 0 deletions(-)

Show changes to builtin-pack-objects.c +6 −0

diff --git a/builtin-pack-objects.c b/builtin-pack-objects.c
index 77481df..9b3ef94 100644
--- a/builtin-pack-objects.c
+++ b/builtin-pack-objects.c
@@ -1356,6 +1356,9 @@ static int try_delta(struct unpacked *trg, struct unpacked *src,
 	/* Load data if not already done */
 	if (!trg->data) {
 		trg->data = read_sha1_file(trg_entry->idx.sha1, &type, &sz);
+		if (!trg->data)
+			die("object %s cannot be read",
+			    sha1_to_hex(trg_entry->idx.sha1));
 		if (sz != trg_size)
 			die("object %s inconsistent object length (%lu vs %lu)",
 			    sha1_to_hex(trg_entry->idx.sha1), sz, trg_size);
@@ -1363,6 +1366,9 @@ static int try_delta(struct unpacked *trg, struct unpacked *src,
 	}
 	if (!src->data) {
 		src->data = read_sha1_file(src_entry->idx.sha1, &type, &sz);
+		if (!src->data)
+			die("object %s cannot be read",
+			    sha1_to_hex(src_entry->idx.sha1));
 		if (sz != src_size)
 			die("object %s inconsistent object length (%lu vs %lu)",
 			    sha1_to_hex(src_entry->idx.sha1), sz, src_size);

-- 
1.5.3.rc6.23.g0058

Linus Torvalds· Aug 25, 2007, 15:19 UTC · re: Shawn O. Pearce · lore

Re: [PATCH] Don't segfault if we failed to inflate a packed delta

On Sat, 25 Aug 2007, Shawn O. Pearce wrote:

Show 8 quoted lines

> 
> cate on #git noticed this failure in `git fsck --full` where the
> call to verify_pack() first noticed that the packfile was corrupt
> by finding that the packfile's SHA-1 did not match the raw data of
> the file.  After finding this fsck went ahead and tried to verify
> every object within the packfile, even though the packfile was
> already known to be bad.  If we are going to shovel bad data at
> the delta unpacking code, we better handle it correctly.

Hmm. We should actually make "unpack_entry()" return print an error and return NULL for these cases, rather than die, I think.

Most of the callers seem to already check for NULL (not "load_tree()" in fast-import.c), but for something like fsck, while "die()" is obviously better than a SIGSEGV, we should probably continue and try to see what else we find.

(Although, to be honest, it might not matter. If your pack-file is corrupt enough for this to trigger, there's seldom anything interesting fsck will tell, so in practical terms this probably isn't a big deal).

		Linus

Junio C Hamano· Aug 25, 2007, 17:56 UTC · re: Linus Torvalds · lore

Re: [PATCH] Don't segfault if we failed to inflate a packed delta

Linus Torvalds <torvalds@linux-foundation.org> writes:

Show 11 quoted lines

> Hmm. We should actually make "unpack_entry()" return print an error and 
> return NULL for these cases, rather than die, I think.
>
> Most of the callers seem to already check for NULL (not "load_tree()" in 
> fast-import.c), but for something like fsck, while "die()" is obviously 
> better than a SIGSEGV, we should probably continue and try to see what 
> else we find.
>
> (Although, to be honest, it might not matter. If your pack-file is corrupt 
> enough for this to trigger, there's seldom anything interesting fsck will 
> tell, so in practical terms this probably isn't a big deal).

But if the delta does not apply we already die(). Last night I and Shawn actually were wondering if it makes sense to have unpack_entry() die -- quite the opposite of being nicer to fsck.