On 06/14/2019 01:34 AM, Andrew Morton wrote:
> On Thu, 13 Jun 2019 15:37:24 +0530 Anshuman Khandual <anshuman.khandual@xxxxxxx> wrote:
>
>> Architectures which support kprobes have very similar boilerplate around
>> calling kprobe_fault_handler(). Use a helper function in kprobes.h to unify
>> them, based on the x86 code.
>>
>> This changes the behaviour for other architectures when preemption is
>> enabled. Previously, they would have disabled preemption while calling the
>> kprobe handler. However, preemption would be disabled if this fault was
>> due to a kprobe, so we know the fault was not due to a kprobe handler and
>> can simply return failure.
>>
>> This behaviour was introduced in the commit a980c0ef9f6d ("x86/kprobes:
>> Refactor kprobes_fault() like kprobe_exceptions_notify()")
>>
>> ...
>>
>> --- a/arch/arm/mm/fault.c
>> +++ b/arch/arm/mm/fault.c
>> @@ -30,28 +30,6 @@
>>
>> #ifdef CONFIG_MMU
>>
>> -#ifdef CONFIG_KPROBES
>> -static inline int notify_page_fault(struct pt_regs *regs, unsigned int fsr)
>
> Some architectures make this `static inline'. Others make it
> `nokprobes_inline', others make it `static inline __kprobes'. The
> latter seems weird - why try to put an inline function into
> .kprobes.text?
>
> So.. what's the best thing to do here? You chose `static
> nokprobe_inline' - is that the best approach, if so why? Does
> kprobe_page_fault() actually need to be inlined?
Matthew had suggested that (nokprobe_-inline) based on current x86
implementation. But every architecture already had an inlined definition
which I did not want to deviate from.
>
> Also, some architectures had notify_page_fault returning int, others
> bool. You chose bool and that seems appropriate and all callers are OK
> with that.
I would believe so. No one has complained yet :)
