Re: [RFC PATCH v5 0/3] printk: new ringbuffer implementation

From: Prarit Bhargava
Date: Fri Dec 06 2019 - 09:17:05 EST


John Ogness <john.ogness@xxxxxxxxxxxxx> wrote:
> Hi Prarit,
>
> On 2019-12-05, Prarit Bhargava <prarit@xxxxxxxxxx> wrote:
> > Based on the comments there is going to be a v6 but in any case I am
> > starting testing of this patchset on several large core systems across
> > multiple architectures (x86_64, ARM64, s390, ppc64le). Some of those
> > systems are known to fail boot due to the large amount of printk output so
> > it will be good to see if these changes resolve those issues.
>
> Right now the patches only include the ringbuffer as a separate entity
> with a test module. So they do not yet have any effect on printk.
>
> If you apply the patches and then build the "modules" target, you will
> have a new test_prb.ko module. Loading that module will start some heavy
> testing of the ringbuffer. As long as the testing is successful, the
> module will keep testing. During this time the machine will be very
> slow, but should still respond.
>
> The test can be stopped by unloading the module. If the test stops on
> its own, then a problem was found. The output of the test is put into
> the ftrace buffer.
>
> It would be nice if you could run the test module on some fat machines,
> at least for a few minutes to see if anything explodes. ARM64 and
> ppc64le will probably be the most interesting, due to memory barrier
> testing.
>

I've run the module overnight on all 4 arches I mentioned above. I didn't
see any failures but IIUC the module test runs at max. I'm going to put a
load test on these systems that introduces a variable load to interfere
with the prbtest module to see if that kicks anything.

> Otherwise I will definitely be reaching out to you when we are ready to
> perform actual printk testing with the newly agreed up semantics
> (lockless, per-console printing threads, synchronous panic
> consoles). Thanks for your help with this.
>

np :) but I should be the one thanking you ;)

P.

> John Ogness