Re: [PATCH] KVM: PPC: Fix unknown SPR error message in emulation

[Scott Wood <scottwood@freescale.com>]

On 04/26/2012 08:25 AM, Alexander Graf wrote:
> 
> On 24.04.2012, at 11:23, Mihai Caraman wrote:
> 
>> mtspr/mfspr emulation prints an error message for unknown SPRs. The message
>> was badly formatted displaying the hex value without 0x prefix. Use decimal
>> representation in accordance with the manuals, though the Linux headers
>> annoyingly use hex.
>>
>> Signed-off-by: Mihai Caraman <mihai.caraman@freescale.com>
>> ---
>> arch/powerpc/kvm/emulate.c |    4 ++--
>> 1 files changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/arch/powerpc/kvm/emulate.c b/arch/powerpc/kvm/emulate.c
>> index afc9154..06d12c4 100644
>> --- a/arch/powerpc/kvm/emulate.c
>> +++ b/arch/powerpc/kvm/emulate.c
>> @@ -296,7 +296,7 @@ int kvmppc_emulate_instruction(struct kvm_run *run, struct kvm_vcpu *vcpu)
>> 			default:
>> 				emulated = kvmppc_core_emulate_mfspr(vcpu, sprn, rt);
>> 				if (emulated == EMULATE_FAIL) {
>> -					printk("mfspr: unknown spr %x\n", sprn);
>> +					printk("mfspr: unknown spr %u\n", sprn);
> 
> This means that if an older kernel threw an error on let's say
> SPR_VRSAVE, we got an error saying that it couldn't find "100", while

Eww.  Perhaps we should grep for bare %x and fix them all at once?  And
not let any more in...

> with new kernels we'd get 256. However, we don't have any indication
> if we're on an old or new kernel, making user failures pretty hard to
> debug.
> 
> So either we change the message to hex always, with 0x prefixed, or
> we rephrase it to give us some indication if the user is running a
> patched kernel.

How about:
printk("mfspr: unknown spr %u (dec)\n", sprn);

Or if we just care about having some sort of marker that those familiar
with the history will understand, and don't want something nicer-looking
that doesn't make it look like appending (dec) should be normal practice:

printk("mfspr: unknown spr #%u\n", sprn);

-Scott