Re: [PATCH v6 bpf-next 1/3] perf: enable branch record for software events

[Song Liu <songliubraving@fb.com>]

> On Sep 10, 2021, at 11:40 AM, Peter Zijlstra <peterz@infradead.org> wrote:
> 
> On Fri, Sep 10, 2021 at 06:27:36PM +0000, Song Liu wrote:
> 
>> This works great and saves 3 entries! We have the following now:
> 
> Yay!
> 
>> ID: 0 from bpf_get_branch_snapshot+18 to intel_pmu_snapshot_branch_stack+0
> 
> is unavoidable, we need to end up in intel_pmu_snapshot_branch_stack()
> eventually.
> 
>> ID: 1 from __brk_limit+477143934 to bpf_get_branch_snapshot+0
> 
> could be elided by having the JIT emit the call to
> intel_pmu_snapshot_branch_stack directly, instead of laundering it
> through that helper I suppose.

Yep, some JIT magic could save one entry here. 

> 
>> ID: 2 from __brk_limit+477192263 to __brk_limit+477143880  # trampoline 
>> ID: 3 from __bpf_prog_enter+34 to __brk_limit+477192251
> 
> -ENOCLUE
> 
>> ID: 4 from migrate_disable+60 to __bpf_prog_enter+9
>> ID: 5 from __bpf_prog_enter+4 to migrate_disable+0
> 
> I suppose we can reduce that to a single branch if we inline
> migrate_disable() here, that thing unfortunately needs one branch
> itself.

To inline migrate_disable, we may need expose this_rq() in include/, or 
use some other alternatives. I am planning to optimize that after this
set gets in.

Thanks,
Song

> 
>> ID: 6 from bpf_testmod_loop_test+20 to __bpf_prog_enter+0
> 
> And this is the first branch out of the test program, giving 7 entries
> now, of which we can remove at least 2 more with a bit of elbow greace,
> right?
> 
>> ID: 7 from bpf_testmod_loop_test+20 to bpf_testmod_loop_test+13
>> ID: 8 from bpf_testmod_loop_test+20 to bpf_testmod_loop_test+13
>> 
>> I will fold this in and send v7. 
> 
> Excellent.