From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Message-ID: <53CFC571.1090709@kernel.dk> Date: Wed, 23 Jul 2014 16:23:45 +0200 From: Jens Axboe MIME-Version: 1.0 Subject: Re: fio crash when using verify References: <20140722205947.GD14599@beardog.cce.hp.com> <53CF8A01.2090001@kernel.dk> <20140723142204.GK14599@beardog.cce.hp.com> In-Reply-To: <20140723142204.GK14599@beardog.cce.hp.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit To: scameron@beardog.cce.hp.com Cc: fio@vger.kernel.org List-ID: On 2014-07-23 16:22, scameron@beardog.cce.hp.com wrote: > On Wed, Jul 23, 2014 at 12:10:09PM +0200, Jens Axboe wrote: >> On 2014-07-22 22:59, scameron@beardog.cce.hp.com wrote: >>> >>> Hi Jens, >>> >>> I'm running into a crash with fio today, seems to be easily reproducible. >>> >>> [root@msablackburn saperf]# fio --version >>> fio-2.1.11-2-g90811 >>> >>> (top of current git tree.) >>> >>> >>> [root@msablackburn saperf]# fio ./2drive_sdr_verify.fio >>> 4_KiB_RR_drive_r: (g=0): rw=randwrite, bs=4K-4K/4K-4K/4K-4K, >>> ioengine=libaio, iodepth=170 >>> 4_KiB_RR_drive_s: (g=0): rw=randwrite, bs=4K-4K/4K-4K/4K-4K, >>> ioengine=libaio, iodepth=170 >>> fio-2.1.11-2-g90811 >>> Starting 2 threads >>> fio: io_u.c:1315: __get_io_u: Assertion `io_u->flags & IO_U_F_FREE' >>> failed.a 58m:59s] >>> Aborted (core dumped) >> >> Looks like there's a long standing race in the async verification. Does >> this patch fix it for you? > > So far so good. Been running for about 30 minutes now when previously it > would crash within a minute or two. Good, that's a lot better... The fix has been committed, btw, that's how much confidence I have in my fixes ;-) Let me know how it goes when it has run longer. -- Jens Axboe