Re: [PATCH] perf: Fix sibling iteration

[Peter Zijlstra]

On Fri, Mar 16, 2018 at 12:59:34PM +0300, Alexey Budankov wrote:
> Hi,
> On 15.03.2018 20:01, Peter Zijlstra wrote:
> > Subject: perf: Fix sibling iteration
> > From: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> > Date: Thu Mar 15 17:36:56 CET 2018
> > 
> > Mark noticed that the change to sibling_list changed some iteration
> > semantics; because previously we used group_list as list entry,
> > sibling events would always have an empty sibling_list.
> > 
> > But because we now use sibling_list for both list head and list entry,
> > siblings will report as having siblings.
> > 
> > Fix this with a custom for_each_sibling_event() iterator.
> > 
> > Suggested-by: Mark Rutland <mark.rutland@xxxxxxx>
> > Reported-by: Mark Rutland <mark.rutland@xxxxxxx>
> > Fixes: 8343aae66167 ("perf/core: Remove perf_event::group_entry")
> > Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
> > ---
> 
> Applied to tip repo and run testing on Fedora 27/x86_64 (client skylake 8 cores):

There is at least one more known issue with that patch, but neither Mark
nor me could reproduce so far, so we don't know if we're right about the
cause.