Bug 1197128 (CVE-2022-0742)

As far as I can see, recent commit 2d3916f31891 ("ipv6: fix skb drops in
igmp6_event_query() and igmp6_event_report()"), present in mainline 5.17-rc7
and stable 5.16.13, should address the issue as described in the initial
comment.

        introduced              f185de28d9ae    v5.13-rc1
        fixed                   2d3916f31891    v5.17-rc7

The offending commit has not been backported into any pre-5.13 branch and
stable and master received the fix already so only SLE15-SP4 should need it.

From: "sirdarckcat ." <sirdarckcat@chromium.org>
Subject: [oss-security] CVE-2022-0742: Remote Denial of Service on Linux Kernel >=5.13 icmp6

Flooding icmp6 messages of type 130 or 131 is enough to exploit a
memory leak in the kernel and cause the host to go out-of-memory. The
volume of traffic doesn't need to be particularly high. Note that
since the vulnerability was introduced recently (5.13) only 5.15's
stable was affected.

This vulnerability was found/fixed by Eric Dumazet.

CVE will land on MITRE's website sometime this week.

This was fixed on https://kernel.dance/2d3916f3189172d5c69d33065c3c21119fe539fc
 "the commit landed on upstream on": [
  {
   "tags": "tags/v5.17-rc7~18^2"
  }
 ],
 "the commit was backported to": [
  {
   "tags": "tags/v5.16.13~140",
   "commit": "5ed9983ce67341b405cf6fda826e29aed26a7371"
  },
  {
   "tags": "tags/v5.15.27~216",
   "commit": "771aca9bc70709771f66c3e7c00ce87339aa1790"
  }
 ],
 "the commit fixes a bug introduced by": [
  {
   "fixes": "f185de28d9ae (\"mld: add new workqueues for process mld events\")"
  }
 ],
 "the buggy commit landed on upstream on": [
  {
   "tags": "tags/v5.13-rc1~94^2~371^2~1",
   "commit": "f185de28d9ae6c978135993769352e523ee8df06"
  }
 ],

Patch: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d3916f3189172d5c69d33065c3c21119fe539fc
ipv6: fix skb drops in igmp6_event_query() and igmp6_event_report()
While investigating on why a synchronize_net() has been added recently
in ipv6_mc_down(), I found that igmp6_event_query() and igmp6_event_report()
might drop skbs in some cases.

Discussion about removing synchronize_net() from ipv6_mc_down()
will happen in a different thread.

Fixes: f185de28d9ae ("mld: add new workqueues for process mld events")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Taehee Yoo <ap420073@gmail.com>
Cc: Cong Wang <xiyou.wangcong@gmail.com>
Cc: David Ahern <dsahern@kernel.org>
Link: https://lore.kernel.org/r/20220303173728.937869-1-eric.dumazet@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Timeline:

Following https://about.google/appsecurity/ policy:

Feb 23, 2022  - Discovery / Shared with network upstream maintainers,
reproduced, patch confirmed to work, CVE reserved
Feb 25, 2022 - security@kernel.org decides fix/disclosure timeline
Mar 3, 2022 - Patch lands on mainline (Linus tree)
2d3916f3189172d5c69d33065c3c21119fe539fc
Mar 8, 2022 - Patch lands on stable (5.15/5.16)
Mar 15, 2022 - This email is sent (public disclosure of vuln details)

The fix has been submitted to

    SLE15-SP4               86ba959dfd70 *

No other branch needs it.

Reassigning back to security team.

I wonder if this should rather go to SLE15-SP4-GA?

(In reply to Vlastimil Babka from comment #5)
> I wonder if this should rather go to SLE15-SP4-GA?

Michal resubmitted to GA, I merged.

yes, definitely should be fixed in GA