[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [patches] Track resolv.conf, avoid need for explicit res_init() or nscd in long-running applications
- To: Paul Pluzhnikov <ppluzhnikov@xxxxxxxxxx>
- Subject: Re: [patches] Track resolv.conf, avoid need for explicit res_init() or nscd in long-running applications
- From: Maxim Kuvyrkov <maxim@xxxxxxxxxxxxxxxx>
- Date: Thu, 23 Jun 2011 15:12:27 +0400
On Jun 21, 2011, at 12:53 PM, Maxim Kuvyrkov wrote:
> On Jun 20, 2011, at 9:38 PM, Paul Pluzhnikov wrote:
>
>> On Wed, May 25, 2011 at 6:47 PM, Paul Pluzhnikov <ppluzhnikov@xxxxxxxxxx> wrote:
>>> On Wed, May 25, 2011 at 6:20 PM, Paul Pluzhnikov <ppluzhnikov@xxxxxxxxxx> wrote:
>>>
>>>> We've just tracked a bug to this patch:
>>>> http://www.eglibc.org/cgi-bin/viewcvs.cgi/trunk/libc/resolv/res_libc.c?rev=9102&r1=8608&r2=9102
>>>
>>> Oops. I see this has already been noted:
>>> http://www.eglibc.org/archives/patches/msg00901.html
>>>
>>> Sorry about the noise.
>>
>> It has been pointed out to me that the fix above is still buggy :-(
>>
>> Effectively it's doing double-checked locking. Consider:
>
...
> The analysis and the patch look correct to me. I'll apply the fix in couple of days to give Joseph and everyone else time to comment.
I've checked in Paul's patch to trunk and backported it to EGLIBC 2.14 and EGLIBC 2.13 branches. EGLIBC 2.12 and earlier do not need this fix as they don't have original Aurelien's fix to resolv/res_libc.c.
Thanks,
--
Maxim Kuvyrkov
CodeSourcery / Mentor Graphics