Re: [PATCH v4 tip/perf/core 0/4] uprobes,mm: speculative lockless VMA-to-uprobe lookup

[Andrii Nakryiko <andrii.nakryiko@gmail.com>]

On Wed, Nov 20, 2024 at 8:03 AM Ingo Molnar <mingo@kernel.org> wrote:
>
>
> * Peter Zijlstra <peterz@infradead.org> wrote:
>
> > On Wed, Nov 20, 2024 at 07:40:15AM -0800, Andrii Nakryiko wrote:
> > > Linus,
> > >
> > > I'm not sure what's going on here, this patch set seems to be in some
> > > sort of "ignore list" on Peter's side with no indication on its
> > > destiny.
> >
> > *sigh* it is not, but my inbox is like drinking from a firehose :/
>
> And I've been considering that particular series WIP for two reasons:
>
>  1) Oleg was still unconvinced about patch 5/5 in the v2 discussion.
>     Upon re-reading it I think he might have come around and has agreed
>     to the current approach - but sending a v3 & not seeing Oleg object
>     would ascertain that.

Is this about Liao's siglock patch set? We are at v4 (!) already (see
[0]) with Oleg's Acked-by added.

>
>  2) There was a build failure reported against -v2 at:
>
>        https://lore.kernel.org/all/202410050745.2Nuvusy4-lkp@intel.com/t.mbox.gz
>
>     We cannot and will not merge patches with build failures.

This one is about this patch set (speculative uprobe lookup), right?
It is already at v4 ([1]), while you are mentioning v2 as the reason
for this to not yet be applied. Those build failures were fixed *a
long time ago*, v4 itself has been sitting idle for almost a month
(since Oct 27). If there are any other problems, do bring them up,
don't wait for weeks.

>
> Andrii did get some other uprobes scalability work merged in v6.13:
>
>     - Switch to RCU Tasks Trace flavor for better performance (Andrii Nakryiko)
>
>     - Massively increase uretprobe SMP scalability by SRCU-protecting
>       the uretprobe lifetime (Andrii Nakryiko)
>
> So we've certainly not been ignoring his patches, to the contrary ...

Yes, and as I mentioned, this one is a) ready, reviewed, tested and b)
complements the other work you mention. It removes mmap_lock which
limits scalability of the rest of the work. Is there some rule that I
get to land only two patch sets in a single release?

>
> Thanks,
>
>         Ingo


[0] https://lore.kernel.org/linux-trace-kernel/20241022073141.3291245-1-liaochang1@huawei.com/
[1] https://lore.kernel.org/linux-trace-kernel/20241028010818.2487581-1-andrii@kernel.org/
[2] https://lore.kernel.org/linux-trace-kernel/CAEf4BzYPajbgyvcvm7z1EiPgkee1D1r=a8gaqxzd7k13gh9Uzw@mail.gmail.com/
[3] https://lore.kernel.org/linux-trace-kernel/CAEf4Bza=pwrZvd+3dz-a7eiAQMk9rwBDO1Kk_iwXSCM70CAARw@mail.gmail.com/