[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ZxJ2NxXpqowd73om@smile.fi.intel.com>
Date: Fri, 18 Oct 2024 17:52:39 +0300
From: Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
To: Baoquan He <bhe@...hat.com>
Cc: Gregory Price <gourry@...rry.net>, kexec@...ts.infradead.org,
linux-kernel@...r.kernel.org, akpm@...ux-foundation.org,
bhelgaas@...gle.com, ilpo.jarvinen@...ux.intel.com,
mika.westerberg@...ux.intel.com, ying.huang@...el.com,
tglx@...utronix.de, takahiro.akashi@...aro.org,
stable@...r.kernel.org
Subject: Re: [PATCH] resource,kexec: walk_system_ram_res_rev must retain
resource flags
On Fri, Oct 18, 2024 at 05:51:09PM +0300, Andy Shevchenko wrote:
> On Fri, Oct 18, 2024 at 09:52:47PM +0800, Baoquan He wrote:
> > On 10/18/24 at 03:22pm, Andy Shevchenko wrote:
> > > On Fri, Oct 18, 2024 at 10:18:42AM +0800, Baoquan He wrote:
> > > > On 10/17/24 at 03:03pm, Gregory Price wrote:
> > > > > walk_system_ram_res_rev() erroneously discards resource flags when
> > > > > passing the information to the callback.
> > > > >
> > > > > This causes systems with IORESOURCE_SYSRAM_DRIVER_MANAGED memory to
> > > > > have these resources selected during kexec to store kexec buffers
> > > > > if that memory happens to be at placed above normal system ram.
> > > >
> > > > Sorry about that. I haven't checked IORESOURCE_SYSRAM_DRIVER_MANAGED
> > > > memory carefully, wondering if res could be set as
> > > > 'IORESOURCE_SYSTEM_RAM | IORESOURCE_BUSY' plus
> > > > IORESOURCE_SYSRAM_DRIVER_MANAGED in iomem_resource tree.
> > > >
> > > > Anyway, the change in this patch is certainly better. Thanks.
> > >
> > > Can we get more test cases in the respective module, please?
> >
> > Do you mean testing CXL memory in kexec/kdump? No, we can't. Kexec/kdump
> > test cases basically is system testing, not unit test or module test. It
> > needs run system and then jump to 2nd kernel, vm can be used but it
> > can't cover many cases existing only on baremetal. Currenly, Redhat's
> > CKI is heavily relied on to test them, however I am not sure if system
> > with CXL support is available in our LAB.
> >
> > Not sure if I got you right.
>
> I meant since we touch resource.c, we should really touch resource_kunit.c
> *in addition to*.
And to be more clear, there is no best time to add test cases than
as early as possible. So, can we add the test cases to the (new) APIs,
so we want have an issue like the one this patch fixes?
--
With Best Regards,
Andy Shevchenko
Powered by blists - more mailing lists