From: Jeff King <peff@peff.net>
To: "brian m. carlson" <sandals@crustytoothpaste.net>
Cc: git@vger.kernel.org, "Nguyễn Thái Ngọc Duy" <pclouds@gmail.com>,
"Junio C Hamano" <gitster@pobox.com>
Subject: Re: [PATCH v2 06/21] builtin/receive-pack: fix incorrect pointer arithmetic
Date: Tue, 28 Mar 2017 02:51:23 -0400 [thread overview]
Message-ID: <20170328065123.f4uswxxznv3wyqaz@sigill.intra.peff.net> (raw)
In-Reply-To: <20170326160143.769630-7-sandals@crustytoothpaste.net>
On Sun, Mar 26, 2017 at 04:01:28PM +0000, brian m. carlson wrote:
> If we had already processed the last newline in a push certificate, we
> would end up subtracting NULL from the end-of-certificate pointer when
> computing the length of the line. This would have resulted in an
> absurdly large length, and possibly a buffer overflow. Instead,
> subtract the beginning-of-certificate pointer from the
> end-of-certificate pointer, which is what's expected.
>
> Note that this situation should never occur, since not only do we
> require the certificate to be newline terminated, but the signature will
> only be read from the beginning of a line. Nevertheless, it seems
> prudent to correct it.
I think you can trigger it with carefully-crafted input. Try this on the
client side:
diff --git a/send-pack.c b/send-pack.c
index 66e652f7e..dd18c9a33 100644
--- a/send-pack.c
+++ b/send-pack.c
@@ -311,8 +311,7 @@ static int generate_push_cert(struct strbuf *req_buf,
if (!update_seen)
goto free_return;
- if (sign_buffer(&cert, &cert, signing_key))
- die(_("failed to sign the push certificate"));
+ strbuf_rtrim(&cert);
packet_buf_write(req_buf, "push-cert%c%s", 0, cap_string);
for (cp = cert.buf; cp < cert.buf + cert.len; cp = np) {
We omit the signature entirely, which causes the parser to treat the
end of the string as the end-of-cert (we still find the end because the
push-cert-end is in its own pkt-line; you could also just issue a flush,
which has the same effect).
And then the rtrim means that the cert doesn't actually end in a
newline. Running t5534 with this patch causes receive-pack to segfault.
It's not an overflow on writing the buffer, though; the nonsense size is
passed into a FLEX_ALLOC_MEM(). In my test case, I ended up allocating a
1.4GB buffer (which just happened to be the mod-2^32 residue of my "eoc"
pointer), and the segfault comes from trying to read an unallocated
page.
So I don't think it's exploitable in any interesting way.
> diff --git a/builtin/receive-pack.c b/builtin/receive-pack.c
> index feafb076a4..116f3177a1 100644
> --- a/builtin/receive-pack.c
> +++ b/builtin/receive-pack.c
> @@ -1524,7 +1524,7 @@ static void queue_commands_from_cert(struct command **tail,
>
> while (boc < eoc) {
> const char *eol = memchr(boc, '\n', eoc - boc);
> - tail = queue_command(tail, boc, eol ? eol - boc : eoc - eol);
> + tail = queue_command(tail, boc, eol ? eol - boc : eoc - boc);
> boc = eol ? eol + 1 : eoc;
> }
> }
The patch itself is obviously an improvement. It may be worth graduating
separately from the rest of the series.
-Peff
next prev parent reply other threads:[~2017-03-28 6:51 UTC|newest]
Thread overview: 40+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-03-26 16:01 [PATCH v2 00/21] object_id part 7 brian m. carlson
2017-03-26 16:01 ` [PATCH v2 01/21] Define new hash-size constants for allocating memory brian m. carlson
2017-03-26 16:01 ` [PATCH v2 02/21] Convert GIT_SHA1_HEXSZ used for allocation to GIT_MAX_HEXSZ brian m. carlson
2017-03-26 16:01 ` [PATCH v2 03/21] Convert GIT_SHA1_RAWSZ used for allocation to GIT_MAX_RAWSZ brian m. carlson
2017-03-26 16:01 ` [PATCH v2 04/21] builtin/diff: convert to struct object_id brian m. carlson
2017-03-26 16:01 ` [PATCH v2 05/21] builtin/pull: convert portions " brian m. carlson
2017-03-26 16:01 ` [PATCH v2 06/21] builtin/receive-pack: fix incorrect pointer arithmetic brian m. carlson
2017-03-28 6:51 ` Jeff King [this message]
2017-03-28 16:58 ` Junio C Hamano
2017-03-26 16:01 ` [PATCH v2 07/21] builtin/receive-pack: convert portions to struct object_id brian m. carlson
2017-03-28 7:07 ` Jeff King
2017-03-29 23:21 ` brian m. carlson
2017-03-30 1:37 ` Jeff King
2017-03-26 16:01 ` [PATCH v2 08/21] fsck: convert init_skiplist " brian m. carlson
2017-03-26 16:01 ` [PATCH v2 09/21] parse-options-cb: convert sha1_array_append caller " brian m. carlson
2017-03-26 16:01 ` [PATCH v2 10/21] test-sha1-array: convert most code " brian m. carlson
2017-03-26 16:01 ` [PATCH v2 11/21] sha1_name: convert struct disambiguate_state to object_id brian m. carlson
2017-03-26 16:01 ` [PATCH v2 12/21] sha1_name: convert disambiguate_hint_fn to take object_id brian m. carlson
2017-03-26 16:01 ` [PATCH v2 13/21] submodule: convert check_for_new_submodule_commits to object_id brian m. carlson
2017-03-26 16:01 ` [PATCH v2 14/21] builtin/pull: convert to struct object_id brian m. carlson
2017-03-26 16:01 ` [PATCH v2 15/21] sha1-array: convert internal storage for struct sha1_array to object_id brian m. carlson
2017-03-28 7:24 ` Jeff King
2017-03-26 16:01 ` [PATCH v2 16/21] Make sha1_array_append take a struct object_id * brian m. carlson
2017-03-28 7:26 ` Jeff King
2017-03-28 17:27 ` Junio C Hamano
2017-03-29 0:06 ` brian m. carlson
2017-03-29 15:14 ` Junio C Hamano
2017-03-29 22:28 ` brian m. carlson
2017-03-26 16:01 ` [PATCH v2 17/21] Convert remaining callers of sha1_array_lookup to object_id brian m. carlson
2017-03-26 16:01 ` [PATCH v2 18/21] Convert sha1_array_lookup to take struct object_id brian m. carlson
2017-03-26 16:01 ` [PATCH v2 19/21] Convert sha1_array_for_each_unique and for_each_abbrev to object_id brian m. carlson
2017-03-26 16:01 ` [PATCH v2 20/21] Rename sha1_array to oid_array brian m. carlson
2017-03-26 16:01 ` [PATCH v2 21/21] Documentation: update and rename api-sha1-array.txt brian m. carlson
2017-03-28 7:31 ` [PATCH v2 00/21] object_id part 7 Jeff King
2017-03-28 11:13 ` brian m. carlson
2017-03-28 17:35 ` Jeff King
2017-03-28 17:42 ` Jeff King
2017-03-28 19:40 ` Junio C Hamano
2017-03-28 20:00 ` Jeff King
2017-03-28 17:32 ` Junio C Hamano
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
List information: http://vger.kernel.org/majordomo-info.html
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20170328065123.f4uswxxznv3wyqaz@sigill.intra.peff.net \
--to=peff@peff.net \
--cc=git@vger.kernel.org \
--cc=gitster@pobox.com \
--cc=pclouds@gmail.com \
--cc=sandals@crustytoothpaste.net \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
Code repositories for project(s) associated with this public inbox
https://80x24.org/mirrors/git.git
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).