[PATCH v3 08/20] arm64: entry.S: convert elX_irq

[James Morse]

Hi Julien,

On 11/10/17 18:13, Julien Thierry wrote:
> On 05/10/17 20:18, James Morse wrote:
>> Following our 'dai' order, irqs should be processed with debug and
>> serror exceptions unmasked.
>>  > Add a helper to unmask these two, (and fiq for good measure).
>>
>> Signed-off-by: James Morse <james.morse at arm.com>
>> ---
>>   arch/arm64/include/asm/assembler.h | 4 ++++
>>   arch/arm64/kernel/entry.S          | 4 ++--
>>   2 files changed, 6 insertions(+), 2 deletions(-)
>>
>> diff --git a/arch/arm64/include/asm/assembler.h
>> b/arch/arm64/include/asm/assembler.h
>> index c2a37e2f733c..7ffb2a629dc9 100644
>> --- a/arch/arm64/include/asm/assembler.h
>> +++ b/arch/arm64/include/asm/assembler.h
>> @@ -54,6 +54,10 @@
>>       msr    daif, \tmp
>>       .endm
>>   +    .macro enable_da_f
>> +    msr    daifclr, #(8 | 4 | 1)
>> +    .endm
>> +
> 
> We use this in irq entries because we are inheriting daif + we want to disable
> irqs while we process irqs right?

In this case not inheriting, we only do that for synchronous exceptions because
we can't mask them, keeping the flags 'the same' lets us ignore them.

Here we are unconditionally unmasking things for handling interrupts. If we
stick with this dai order we know its safe to unmask SError and enable Debug.


> I don't know if it's worth adding a comment but I find it easier to think about
> it like this.

If its not-obvious, there should be a comment:
/* IRQ is the lowest priority flag, unconditionally unmask the rest. */


(I was expecting flames for the hangman style naming!)

> Otherwise, for patches 3 to 8 (I don't have any comment on 3 to 7):
> Reviewed-by: Julien Thierry <julien.thierry at arm.com>

Thanks!

I'll post a v4 ~tomorrow with this and the renaming changes.



James