Re: [PATCH 1/2] trace, RAS: Add basic RAS trace event

[Mauro Carvalho Chehab <m.chehab@samsung.com>]

Em Thu, 06 Mar 2014 16:26:33 +0100
Borislav Petkov <bp@alien8.de> escreveu:

> On Thu, Mar 06, 2014 at 10:06:53AM -0300, Mauro Carvalho Chehab wrote:
> > For example PCIe and memory errors are not x86-specific. Also, as ACPI 
> > may also be used on ARM, we may also start to have APEI errors there:
> > 	https://lwn.net/Articles/574439/
> > 	https://wiki.linaro.org/LEG/Engineering/Kernel/ACPI
> > 
> > So, better to think on that on a long term.
> 
> kernel/ras/ could also be used in that case but I guess drivers/ras/ is
> fine too.

Both work for me, although drivers/ras seems more adequate, IMHO,
as I expect that we'll have there both subsystem code and drivers.

> 
> > In order to put all RAS drivers under the same place. We may
> > eventually have a subdir there for EDAC, and one per RAS report
> > mechanism, in order to keep it cleaner.
> 
> That doesn't bring any advantages - edac drivers are just fine in
> drivers/edac/. And without benefits for a move, it would be a senseless
> code churn only.

No, it won't bring any technical advantage. Err... if an EDAC driver
or core would depend on something at /drivers/ras, then we may need
to add some extra early init glue, in order to be sure that the code
at /drivers/ras will be initialized before /drivers/edac, or otherwise
it would fail with both are compiled builtin.

-- 

Cheers,
Mauro