Re: [PATCH v4 5/9] perf/amd/ibs: Don't allow freq mode event creation through ->config interface

[Ravi Bangoria]

On 17-Jan-25 2:39 AM, Peter Zijlstra wrote:
> On Wed, Jan 15, 2025 at 05:44:34AM +0000, Ravi Bangoria wrote:
>> Most perf_event_attr->config bits directly maps to IBS_{FETCH|OP}_CTL
>> MSR. Since the sample period is programmed in these control registers,
>> IBS PMU driver allows opening an IBS event by setting sample period
>> value directly in perf_event_attr->config instead of using explicit
>> perf_event_attr->sample_period interface.
>>
>> However, this logic is not applicable for freq mode events since the
>> semantics of control register fields are applicable only to fixed
>> sample period whereas the freq mode event adjusts sample period after
>> each and every sample. Currently, IBS driver (unintentionally) allows
>> creating freq mode event via ->config interface, which is semantically
>> wrong as well as detrimental because it can be misused to bypass
>> perf_event_max_sample_rate checks.
>>
>> Don't allow freq mode event creation through perf_event_attr->config
>> interface.
> 
> So this is the case where:
> 
> perf_event_attr = {
> 	.freq = 1.
> 	.sample_freq = 0,
> }
> 
> Right, where we would then take the period from the .config fields?
> 
> So that seems sensible enough; but does this not break existing users?
> Or are we lucky enough to not have any users of this borken behaviour,
> unlike the earlier broken behaviour that we now have to live with?

I'm not aware of any existing tools using the ->config interface.

Also, I can't think of any advantage of the ->config interface over
stranded one. So, if anyone is using it and start seeing an error, they
can switch to { attr->freq=1, attr->sample_freq=xxx } and there is no
functional breakage.

Thanks for the review,
Ravi