Re: Setting bs could lead to some double free or segfault

From: Jens Axboe <jens.axboe_at_oracle.com>
Date: Fri, 20 Jul 2007 12:42:06 +0200

On Fri, Jul 20 2007, Erwan Velu wrote:
> Erwan Velu wrote:
>> Note, that setting bs=4k __and__ direct=1 make things work perfectly (but
>> in that test I would like not beeing O_DIRECT ;))
>> So its sounds that a 4k bs with direct=0 generates the troubles.
> I found other value that could generates troubles (even if some are
> stupid I admit it but it might could help to find some limits :b ) :
> I was just checking this values to see what happens when a typo occurs
> whereas the 4k is a real case ;)
>
> bs size result
>
> 0k -> fio: pid=19381, got signal=8 + fio[19381] trap divide error
> rip:40dd89 rsp:7fffffc5d4e0 error:0
> -1k -> fio: pid=19396, err=12/file:memory.c:160, func=iomem allocation,
> error=Cannot allocate memory
>
> 4194000k -> job1: (g=0): rw=read,
> bs=4194000K-4194000K/4194000K-4194000K, ioengine=sync, iodepth=1
> 4195000k -> job1: (g=0): rw=read, bs=696K-696K/696K-696K, ioengine=sync,
> iodepth=1
>
> As you see after 4194000k, fio reports strange sizes.

It works here, can you generate the core dump like described yesterday
and show in gdb where it traps?

-- 
Jens Axboe
Received on Fri Jul 20 2007 - 12:42:06 CEST

This archive was generated by hypermail 2.2.0 : Fri Jul 20 2007 - 13:00:01 CEST