Re: [PATCH 7/7] xen/arm32: traps: Dump more information for hypervisor data abort

[Julien Grall <julien@xxxxxxx>]

On 16/08/2022 08:28, Bertrand Marquis wrote:
> Hi Julien,
Hi Bertrand,


> > On 15 Aug 2022, at 18:04, Julien Grall <julien@xxxxxxx> wrote:
> >
> >
> >
> > On 15/08/2022 17:39, Bertrand Marquis wrote:
> > > Hi Julien,
> > Hi Bertrand,
> >
> > > > On 12 Aug 2022, at 20:24, Julien Grall <julien@xxxxxxx> wrote:
> > > >
> > > > From: Julien Grall <jgrall@xxxxxxxxxx>
> > > >
> > > > Unlike arm64, on arm32 there are no extra information dumped (e.g.
> > > > page table walk) for hypervisor data abort.
> > > The code in arch/arm/traps.c has nothing arm32 specific like that so
> > > could you explain this statement ?
> > > Here the arm32 code will call the generic function which has only
> > > something specific for BRK handling but the rest is generic.
> > The statement is not related to the code but the console output. On arm64, a 
> > data abort will decode the HSR and provide a dump of the page-table walk.
> >
> > This doesn't happen on arm32 because Xen will call do_unexpected_trap(). So the 
> > only information we have is the HSR and FAR. This is not very helpful for 
> > debugging page-table walk.
> >
> > After this patch, the same information will be printed on arm32 and arm64.
> Ok then this is what I understood. Your commit message is maybe a bit unclear.
>
> I would add a sentence like that: Call do_trap_hyp_sync for hypervisor data 
> aborts on arm32 to have the same information than on arm64.
Below the new commit message:

    Unlike arm64, on arm32 there are no extra information dumped (e.g.
    page table walk) for hypervisor data abort.

    For data abort, the HSR will be set properly and so call
    do_trap_hyp_sync() instead of do_unexpected_trap() on arm32 to have
    the print the same information as arm64.


> This can be done on commit so feel free to add my:
>
> Reviewed-by: Bertrand Marquis <bertrand.marquis@xxxxxxx>
Thanks! I have committed the series.

Cheers,

--
Julien Grall