Re: [PATCH v2 2/7] xen/x86: merge 2 hvm_event_... functions into 1

[Razvan Cojocaru <rcojocaru@bitdefender.com>]

On 02/10/2016 07:04 PM, Corneliu ZUZU wrote:
>>> @@ -27,9 +33,8 @@ bool_t hvm_event_cr(unsigned int index, unsigned
>>> long value,
>>>   #define hvm_event_crX(what, new, old) \
>>>       hvm_event_cr(VM_EVENT_X86_##what, new, old)
>>>   void hvm_event_msr(unsigned int msr, uint64_t value);
>>> -/* Called for current VCPU: returns -1 if no listener */
>>> -int hvm_event_int3(unsigned long rip);
>>> -int hvm_event_single_step(unsigned long rip);
>>> +int hvm_event_breakpoint(unsigned long rip,
>>> +                         enum hvm_event_breakpoint_type type);
>> I guess the comment was here for a reason, and this reason doesn't
>> go away with this code folding. But I'll leave it to the VM event code
>> maintainers to judge.
>>
>> Jan
> 
> That comment seemed & still seems wrong to me, I don't see any code
> paths out of which that function would return -1.

That seems to be true. Those functions return whatever hvm_event_traps()
returns, which is 0 on success, 1 (maybe the minus is a typo?) if
there's no ring, or whatever value vm_event_claim_slot() returns.
Vm_event_claim_slot()'s documentation says that it can only return 0 (on
success), -ENOSYS or -EBUSY, none of which translate to -1 (and the code
seems to agree with that claim).

Maybe I'm missing some macro wizardry here, but I don't think so - it
looks like the comment is stale. Tamas, maybe you remember more, should
those functions return -1 if no listener is present?


Thanks,
Razvan