[PATCH v2 0/2] arm64: Fix the call trace when running kgdb test suite

Will Deacon will at kernel.org
Thu Jul 9 09:03:48 EDT 2020


On Thu, Jul 09, 2020 at 08:55:04PM +0800, Kevin Hao wrote:
> On Wed, Jul 08, 2020 at 11:02:34PM +0100, Will Deacon wrote:
> > On Fri, 17 Apr 2020 18:32:10 +0800, Kevin Hao wrote:
> > > This fixes a call trace when running the kgdb test suite:
> > >     # echo kgdbts=V1 > /sys/module/kgdbts/parameters/kgdbts
> > > 
> > > v2:
> > >   - Add a patch to fix a typo in el1_dbg()
> > >   - Use the method as suggested by Mark.
> > > 
> > > [...]
> > 
> > Applied to arm64 (for-next/fixes), thanks!
> > 
> > [1/1] arm64: entry: Fix the typo in the comment of el1_dbg()
> >       https://git.kernel.org/arm64/c/b8c1c9fe6a04
> 
> How about patch 2 ("arm64: debug: Always update the IRQ tracing in debug_exception_enter())"?
> It has been dangled in the mail list for almost 3 months without any explicitly objections.

No, sorry, I really think it's the wrong direction of travel. Its adding
bodges to try to handle the fact that we disabled interrupts during the
step, and we really shouldn't be doing that (See my reply to Doug).

As I said, I would prefer some feedback on my proposal for dealing with
this properly [1] rather than more futile attempts to build on top of the
mess that we have at the moment.

Will

[1] https://lore.kernel.org/r/20200626095551.GA9312@willie-the-truck



More information about the linux-arm-kernel mailing list