[PATCH] Don't coredump on bad refs in update-server-info.

Shawn O. Pearce· Jan 31, 2007, 07:24 UTC · lore

Apparently if we are unable to parse an object update-server-info coredumps, as it doesn't bother to check the return value of its call to parse_object.

Instead of coredumping, skip the ref.

Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
---

 I don't know how often folks might run into this.  I intentionally
 broke pack reading in one of my topics and discovered that
 update-server-info crashed with a segfault when the pack access code
 was unable to decompress the object behind the ref.  That's less
 than optimal.

 server-info.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

Show changes to server-info.c +2 −0

diff --git a/server-info.c b/server-info.c
index 6cd38be..f9be5a7 100644
--- a/server-info.c
+++ b/server-info.c
@@ -10,6 +10,8 @@ static FILE *info_ref_fp;
 static int add_info_ref(const char *path, const unsigned char *sha1, int flag, void *cb_data)
 {
 	struct object *o = parse_object(sha1);
+	if (!o)
+		return -1;
 
 	fprintf(info_ref_fp, "%s	%s\n", sha1_to_hex(sha1), path);
 	if (o->type == OBJ_TAG) {

-- 
1.5.0.rc2.81.g73a2

Jakub Narebski· Jan 31, 2007, 08:20 UTC · re: Shawn O. Pearce · lore

Re: [PATCH] Don't coredump on bad refs in update-server-info.

Shawn O. Pearce wrote:

> Apparently if we are unable to parse an object update-server-info
> coredumps, as it doesn't bother to check the return value of its

"Apparently if we are unable to parse an object _then_ update-server-info
 coredumps" is a bit more easier to understand.

-- 
Jakub Narebski
Warsaw, Poland
ShadeHawk on #git

patchDon't coredump on bad refs in update-server-info.