Re: [PATCH] spi: Fix hung task timeout when initialization fails

[Mark Brown]

On Thu, May 01, 2014 at 04:33:58PM +0200, Geert Uytterhoeven wrote:
> On Thu, May 1, 2014 at 4:11 PM, Mark Brown <broonie@xxxxxxxxxx> wrote:

> master->kworker_task is set like this:

>     master->kworker_task = kthread_run(...)

> so it just contains the status of the creation of the kthread, not if it was
> killed, right? Hence we don't check if it was killed.

> So Ricardo's patch prevents the stopping and destruction of the thread
> if it failed to be _created_.

OK, so that means that the description is very confusing then since it's
talking about the issue being due to kthread_run being killed.

> What if it is killed? I suppose the kthread API handles that internally, as it
> could happen to any thread (e.g. OOM)?

I'm not 100% clear on this to be honest.  Since I'm at ELC I've not
investigated fully yet and I'm mostly going on the patch descriptions
here.

Attachment: signature.asc
Description: Digital signature