1394331 - Missing return value check of mozilla::Vector::reserve() could lead to memory corruption

Status: RESOLVED FIXED

(Core :: XPCOM, defect)

Description

[Julian Hector [:tedd] [:jhector]]

The ParamTraits<> specialization of mozilla::HangStack is missing a return value check after the call to reserve() [1], combined with for looping following immediately after, this could lead to memory corruption due to insufficient memory alloction.

If |length| is specified as for example -1, mozilla::Vector::reserve() is likely to fail here [2] meaning the reserved memory for the vector stays the same. However, the for loop operates on unsigned integer leading to an iteration of 0 to size_t max.

Calls to infallibleAppend() [3] rely on a successful call to reserve().

I came across this while checking the diff between my local branch and upstream for changes to ParamTraits<>, I don't know where this is used, but if the deserialization happens in the parent, a compromised child process could potentially leverage as a privilege escalation vector.

[1] http://searchfox.org/mozilla-central/rev/18c16ebf818abb86805ce08a6e537e4cd826f044/toolkit/components/backgroundhangmonitor/HangStack.cpp#253-254
[2] http://searchfox.org/mozilla-central/rev/18c16ebf818abb86805ce08a6e537e4cd826f044/mfbt/Vector.h#1085
[3] http://searchfox.org/mozilla-central/rev/18c16ebf818abb86805ce08a6e537e4cd826f044/mfbt/Vector.h#707-715

Comment 1

[Nika Layzell [:nika] (ni? for response)]

Attachment: Check allocation of hang stack vector

Comment 2

[Nathan Froyd [:froydnj]]

Comment on attachment 8901959 [details] [diff] [review]
Check allocation of hang stack vector

Review of attachment 8901959 [details] [diff] [review]:
-----------------------------------------------------------------

r=me with the below fixed.

::: toolkit/components/backgroundhangmonitor/HangStack.cpp
@@ +256,5 @@
>    if (!ReadParam(aMsg, aIter, &length)) {
>      return false;
>    }
>  
> +  if (!aResult->reserve(length)) {

reserve() and any other functions in HangStack should be made MOZ_MUST_USE.

Comment 3

[Nika Layzell [:nika] (ni? for response)]

Attachment: Check allocation of hang stack vector

Added MOZ_MUST_USE annotations

Comment 4

[Nika Layzell [:nika] (ni? for response)]

Comment on attachment 8901964 [details] [diff] [review]
Check allocation of hang stack vector

oops - just carrying over the flag.

[Security approval request comment]
How easily could an exploit be constructed based on the patch? Other vulnerability (takeover of content process) would be required.

Do comments in the patch, the check-in comment, or tests included in the patch paint a bulls-eye on the security problem? Yes, the security problem is pretty obvious by looking at the patch.

Which older supported branches are affected by this flaw? None, only nightly

If not all supported branches, which bug introduced the flaw? bug 1367406

Do you have backports for the affected branches? If not, how different, hard to create, and risky will they be? N/A

How likely is this patch to cause regressions; how much testing does it need? Very unlikely, no testing.

Comment 5

[Andrew McCreight [:mccr8]]

Security issues that only affect Nightly don't need sec-approval.

Comment 6

[Nika Layzell [:nika] (ni? for response)]

Comment on attachment 8901964 [details] [diff] [review]
Check allocation of hang stack vector

(In reply to Andrew McCreight [:mccr8] from comment #5)
> Security issues that only affect Nightly don't need sec-approval.

Oops - didn't realize that - thanks.

Comment 7

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/2c46386c1b19