# [PATCH] fetch-pack: check for valid commit from server 5 messages from 2011-08-18 to 2011-08-19. Participants: Nguyễn Thái Ngọc Duy, Shawn Pearce, Nguyen Thai Ngoc Duy, Junio C Hamano. Thread: https://gitlist.dev/t/28153 ## Nguyễn Thái Ngọc Duy, 2011-08-18 13:36 Subject: [PATCH] fetch-pack: check for valid commit from server Message-ID: <1313674563-12755-1-git-send-email-pclouds@gmail.com> URL: https://gitlist.dev/e/1313674563-12755-1-git-send-email-pclouds%40gmail.com ``` A malicious server can return ACK with non-existent SHA-1 or not a commit. lookup_commit() in this case may return NULL. Do not let fetch-pack crash by accessing NULL address in this case. Signed-off-by: Nguyễn Thái Ngọc Duy --- However it raises another question, what if the other end returns a valid commit, but not the one in "have" line fetch-pack sent? Are we OK with that? builtin/fetch-pack.c | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-) diff --git a/builtin/fetch-pack.c b/builtin/fetch-pack.c index 4367984..3c871c2 100644 --- a/builtin/fetch-pack.c +++ b/builtin/fetch-pack.c @@ -395,6 +395,8 @@ static int find_common(int fd[2], unsigned char *result_sha1, case ACK_continue: { struct commit *commit = lookup_commit(result_sha1); + if (!commit) + die("invalid commit %s", sha1_to_hex(result_sha1)); if (args.stateless_rpc && ack == ACK_common && !(commit->object.flags & COMMON)) { -- 1.7.4.74.g639db ``` ## Shawn Pearce, 2011-08-18 21:32 Subject: Re: [PATCH] fetch-pack: check for valid commit from server Message-ID: URL: https://gitlist.dev/e/CAJo%3DhJvN9W00ncG53933yiqT%2B%2BQCxGLUqp8KtG%3DBsvKD81wWyA%40mail.gmail.com In-Reply-To: <1313674563-12755-1-git-send-email-pclouds@gmail.com> ``` 2011/8/18 Nguyễn Thái Ngọc Duy : > A malicious server can return ACK with non-existent SHA-1 or not a > commit. lookup_commit() in this case may return NULL. Do not let > fetch-pack crash by accessing NULL address in this case. > > Signed-off-by: Nguyễn Thái Ngọc Duy > --- >  However it raises another question, what if the other end returns a >  valid commit, but not the one in "have" line fetch-pack sent? Are we >  OK with that? Not really. The server is not supposed to return a SHA-1 in the ACK line unless the client said it first in a have line. So aborting with an error is reasonable thing for a client to do. > +                                       if (!commit) > +                                               die("invalid commit %s", sha1_to_hex(result_sha1)); Maybe: die("server ACK contained unknown commit %s", sha1_to_hex(result_sha1)); is more specific to the problem. Just curious, did you see this on a particular server somewhere? -- Shawn. ``` ## Nguyen Thai Ngoc Duy, 2011-08-19 04:19 Subject: Re: [PATCH] fetch-pack: check for valid commit from server Message-ID: URL: https://gitlist.dev/e/CACsJy8AhQ2EqdZRyBD3-E1Sx52Jz_OhSVoGUbCmsC2LKMiYZhw%40mail.gmail.com In-Reply-To: ``` 2011/8/19 Shawn Pearce : > Just curious, did you see this on a particular server somewhere? No. I was looking for lookup_commit() call sites without proper NULL handling. -- Duy ``` ## Nguyễn Thái Ngọc Duy, 2011-08-19 15:11 Subject: [PATCH] fetch-pack: check for valid commit from server Message-ID: <1313766693-20798-1-git-send-email-pclouds@gmail.com> URL: https://gitlist.dev/e/1313766693-20798-1-git-send-email-pclouds%40gmail.com In-Reply-To: <1313674563-12755-1-git-send-email-pclouds@gmail.com> ``` A malicious server can return ACK with non-existent SHA-1 or not a commit. lookup_commit() in this case may return NULL. Do not let fetch-pack crash by accessing NULL address in this case. Signed-off-by: Nguyễn Thái Ngọc Duy --- 2011/8/19 Shawn Pearce : > 2011/8/18 Nguyễn Thái Ngọc Duy : >> However it raises another question, what if the other end returns a >> valid commit, but not the one in "have" line fetch-pack sent? Are we >> OK with that? > > Not really. The server is not supposed to return a SHA-1 in the ACK > line unless the client said it first in a have line. So aborting with > an error is reasonable thing for a client to do. I assumed I could check result_sha1 against sha1. If it did not match, fetch-pack would abort. But I was wrong because fetch-pack would send a few have lines before receiving the first ack (which carries sha1 of some 'have' line in the middle, not the last 'have'). I'd leave it here if anyone wants to tackle it. builtin/fetch-pack.c | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-) diff --git a/builtin/fetch-pack.c b/builtin/fetch-pack.c index 4367984..561f1a3 100644 --- a/builtin/fetch-pack.c +++ b/builtin/fetch-pack.c @@ -395,6 +395,9 @@ static int find_common(int fd[2], unsigned char *result_sha1, case ACK_continue: { struct commit *commit = lookup_commit(result_sha1); + if (!commit) + die("server ACK contained unknown commit %s", + sha1_to_hex(result_sha1)); if (args.stateless_rpc && ack == ACK_common && !(commit->object.flags & COMMON)) { -- 1.7.4.74.g639db ``` ## Junio C Hamano, 2011-08-19 20:18 Subject: Re: [PATCH] fetch-pack: check for valid commit from server Message-ID: <7v62lt2m0q.fsf@alter.siamese.dyndns.org> URL: https://gitlist.dev/e/7v62lt2m0q.fsf%40alter.siamese.dyndns.org In-Reply-To: <1313766693-20798-1-git-send-email-pclouds@gmail.com> ``` Thanks, but already queued the one from yesterday in 'maint' ;-). ```