[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20241028175008.lgdGYYtp@linutronix.de>
Date: Mon, 28 Oct 2024 18:50:08 +0100
From: Sebastian Andrzej Siewior <bigeasy@...utronix.de>
To: linux-kernel@...r.kernel.org
Cc: André Almeida <andrealmeid@...lia.com>,
Darren Hart <dvhart@...radead.org>,
Davidlohr Bueso <dave@...olabs.net>, Ingo Molnar <mingo@...hat.com>,
Juri Lelli <juri.lelli@...hat.com>,
Peter Zijlstra <peterz@...radead.org>,
Valentin Schneider <vschneid@...hat.com>,
Waiman Long <longman@...hat.com>
Subject: Re: [RFC v2 PATCH 0/4] futex: Add support task local hash maps.
On 2024-10-28 13:13:54 [+0100], To linux-kernel@...r.kernel.org wrote:
> from time to time and I haven't seen it using the hashmap. Need to do
> more testing.
booted gnome and did a few things:
- Total allocations: 632
- tasks which never used their allocated futex hash: 2
gpg-agent and systemd-journal
- Tasks which did not terminate within the measurement: 85
This includes gpg-agent and systemd-journal
- Top5 users of the private hash:
- firefox-esr-3786 used 215985
- gnome-software-2343 used 121296
- chromium-3369 used 65796
- chromium-3209 used 34639
- Isolated used 34211
This looks like we could attach the private futex hashmap directly on
fork instead of delaying it to the first usage.
Side note: If someone is waiting for a thread to exit via pthread_join()
then glibc uses here futex() with op 0x109. I would have expected a
private flag.
Sebastian
Powered by blists - more mailing lists