[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sat, 12 Jan 2013 00:11:18 +0100
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: SIGALRM and new 'done' method
On 11 Jan, 2013, at 23:59 , Claudio André <claudioandre.br@...il.com> wrote:
> Em 11-01-2013 20:39, magnum escreveu:
>> On another thought it looks from your patch like you have that commit. The problem then, is that bench_running is not set for some reason. It should be!
>>
>> magnum
>
> I tried a fake pull request and looked at bench.c and common-opencl.c. Everything seems to be ok.
> ---
>
> Previous to the patch, I put a printf("V: %d", bench_running) and i can see a lot of zeros and (at the end) some ones. I am not sure, but seems to me the error happened when bench_running change to 1.
>
> Anyway, we can deal with it latter when 'done' got created on bleeding. The ugly hack works for now.
Yes, I just realized this is a race condition and your printf confirms that. It's a chain of events and conditions leading up to this. Your hack is not a good solution. I will look at it.
This only affects OS_TIMER=0 platforms (including all AMD GPU builds).
magnum
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
