linux-kernel - Re: [PATCH] dmaengine: dmatest: fix timeout caused by kthread

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date: Thu, 25 Apr 2024 14:35:44 +0530
From: Vinod Koul <vkoul@...nel.org>
To: Jie Hai <haijie1@...wei.com>
Cc: dmaengine@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] dmaengine: dmatest: fix timeout caused by kthread_stop

On 23-04-24, 11:39, Jie Hai wrote:
> Hi, Vinod Koul,
> 
> Stop an ongoing test by
> "echo 0 > /sys/module/dmatest/parameters/run".
> If the current code is executed inside the while loop
> "while (!(kthread_should_stop() ||
>  (params->iterations &&
>  total_tests >= params->iterations)))"
> and before the call of "wait_event_freezable_timeout",
> the "wait_event_freezable_timeout" will be interrupted
> and result in  "time out" for the test even if the test
> is not completed.
> 
> 
> 
> Operations to the problem is as follows,
> and the failures are probabilistic:
> 
> modprobe hisi_dma
> modprobe dmatest
> 
> echo 0 > /sys/module/dmatest/parameters/iterations
> echo "dma0chan0" > /sys/module/dmatest/parameters/channel
> echo "dma0chan1" > /sys/module/dmatest/parameters/channel
> echo "dma0chan2" > /sys/module/dmatest/parameters/channel
> echo 1 > /sys/module/dmatest/parameters/run
> echo 0 > /sys/module/dmatest/parameters/run
> 
> dmesg:
> 
> [52575.636992] dmatest: Added 1 threads using dma0chan0
> [52575.637555] dmatest: Added 1 threads using dma0chan1
> [52575.638044] dmatest: Added 1 threads using dma0chan2
> [52581.020355] dmatest: Started 1 threads using dma0chan0
> [52581.020585] dmatest: Started 1 threads using dma0chan1
> [52581.020814] dmatest: Started 1 threads using dma0chan2
> [52587.705782] dmatest: dma0chan0-copy0: result #57691: 'test timed out'
> with src_off=0xfe6 dst_off=0x89 len=0x1d9a (0)
> [52587.706527] dmatest: dma0chan0-copy0: summary 57691 tests, 1 failures
> 51179.98 iops 411323 KB/s (0)
> [52587.707028] dmatest: dma0chan1-copy0: result #63178: 'test timed out'
> with src_off=0xdf dst_off=0x6ab len=0x389e (0)
> [52587.707767] dmatest: dma0chan1-copy0: summary 63178 tests, 1 failures
> 62851.60 iops 503835 KB/s (0)
> [52587.708376] dmatest: dma0chan2-copy0: result #60527: 'test timed out'
> with src_off=0x10e dst_off=0x58 len=0x3ea4 (0)
> [52587.708951] dmatest: dma0chan2-copy0: summary 60527 tests, 1 failures
> 52403.78 iops 420014 KB/s (0)


This is usefel in the commitlog, please add this and update the patchset

> 
> 
> On 2024/4/18 1:18, Vinod Koul wrote:
> > On 20-07-23, 19:41, Jie Hai wrote:
> > > The change introduced by commit a7c01fa93aeb ("signal: break
> > > out of wait loops on kthread_stop()") causes dmatest aborts
> > > any ongoing tests and possible failure on the tests. This patch
> > 
> > Have you see this failure? Any log of that..
> > 
> > > use wait_event_timeout instead of wait_event_freezable_timeout
> > > to avoid interrupting ongoing tests by signal brought by
> > > kthread_stop().
> > > 
> > > Signed-off-by: Jie Hai <haijie1@...wei.com>
> > > ---
> > >   drivers/dma/dmatest.c | 2 +-
> > >   1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > > diff --git a/drivers/dma/dmatest.c b/drivers/dma/dmatest.c
> > > index ffe621695e47..c06b8b16645a 100644
> > > --- a/drivers/dma/dmatest.c
> > > +++ b/drivers/dma/dmatest.c
> > > @@ -827,7 +827,7 @@ static int dmatest_func(void *data)
> > >   		} else {
> > >   			dma_async_issue_pending(chan);
> > > -			wait_event_freezable_timeout(thread->done_wait,
> > > +			ret = wait_event_timeout(thread->done_wait,
> > >   					done->done,
> > >   					msecs_to_jiffies(params->timeout));
> > > -- 
> > > 2.33.0
> > 

-- 
~Vinod