[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [Qemu-block] [PATCH for-2.6] crypto: Avoid memory leak

From: Eric Blake
Subject: Re: [Qemu-devel] [Qemu-block] [PATCH for-2.6] crypto: Avoid memory leak on failure
Date: Thu, 7 Apr 2016 10:23:53 -0600
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.7.1

On 04/07/2016 09:14 AM, Markus Armbruster wrote:
> Max Reitz <address@hidden> writes:
>> On 01.04.2016 17:57, Eric Blake wrote:
>>> Commit 7836857 introduced a memory leak due to invalid use of
>>> Error vs. visit_type_end().  If visiting the intermediate
>>> members fails, we clear the error and unconditionally use
>>> visit_end_struct() on the same error object; but if that
>>> cleanup succeeds, we then skip the qapi_free call.
>> It's not really a memleak. Due to skipping those conditional branches
>> after the "out" label, a non-null value will be returned. In order to
>> determine whether the function call failed, the callers of these
>> functions do not use the errp value but the return value. Therefore,
>> they will think the call succeeded when actually it did not.
> Please amend the commit message accordingly.

Too late; already merged as 95c3df5a.  [And welcome back - hope you
don't mind the backlog...]

(Locally it looks like a memory leak; it is only the wider analysis that
shows that the caller is not leaking things, but where the bug then
shifts to being a potential for the caller to abort if it tries to set
an error into the already-set errp)

Eric Blake   eblake redhat com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org

Attachment: signature.asc
Description: OpenPGP digital signature

reply via email to

[Prev in Thread] Current Thread [Next in Thread]