linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Linus Torvalds <torvalds@linux-foundation.org>
To: Frederic Weisbecker <frederic@kernel.org>
Cc: LKML <linux-kernel@vger.kernel.org>,
	Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
	Peter Zijlstra <peterz@infradead.org>,
	"David S . Miller" <davem@davemloft.net>,
	Mauro Carvalho Chehab <mchehab+samsung@kernel.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	"Paul E . McKenney" <paulmck@linux.vnet.ibm.com>,
	Frederic Weisbecker <fweisbec@gmail.com>,
	Pavan Kondeti <pkondeti@codeaurora.org>,
	Ingo Molnar <mingo@kernel.org>,
	Joel Fernandes <joel@joelfernandes.org>
Subject: Re: [PATCH 00/37] softirq: Per vector masking v3
Date: Fri, 1 Mar 2019 08:51:38 -0800	[thread overview]
Message-ID: <CAHk-=wgOZuGZaVOOiC=drG6ykVkOGk8RRXZ_CrPBMXHKjTg0dg@mail.gmail.com> (raw)
In-Reply-To: <20190301034536.GA19200@lenoir>

On Thu, Feb 28, 2019 at 7:45 PM Frederic Weisbecker <frederic@kernel.org> wrote:
>
> Numbers are indeed missing. In fact this patchset mostly just brings an
> infrastructure. We have yet to pinpoint the most latency-inducing
> softirq disabled sites and make them disable only the vectors that
> are involved in a given lock.

Note that I think we pretty much know that already: the people who
have had issues have never actually really had issues with the actual
"disable softirq" paths, they've all been about actually *running* the
softirq's (and that in turn being a latency issue for running _other_
softirqs, and in particular for delaying them into softirqd).

Now, it may well be that yes, we'll have "block softirqs" code that
has issues too, but it's absolutely been swamped by the latencies for
actually running them so far.

Note that this is all really fairly independent of the whole masking
logic. Yes, the masking logic comes into play too (allowing you to run
a subset of softirq's at a time), but on the whole the complaints I've
seen have not been "the networking softirq takes so long that it
delays USB tasklet handling", but they have been along the lines of
"the networking softirq gets invoked so often that it then floods the
system and triggers softirqd, and _that_ then makes tasklet handling
latency go up insanely".

See the difference? Not the latency of softirq's disabled, but the
latency of one group of softirqs causing problems for another when
they all get batched together (and soft-scheduled to another context
together).

                  Linus

  reply	other threads:[~2019-03-01 16:52 UTC|newest]

Thread overview: 44+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-02-28 17:12 [PATCH 00/37] softirq: Per vector masking v3 Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 01/37] locking/lockdep: Move valid_state() inside CONFIG_TRACE_IRQFLAGS && CONFIG_PROVE_LOCKING Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 02/37] locking/lockdep: Use expanded masks on find_usage_*() functions Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 03/37] locking/lockdep: Introduce struct lock_usage Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 04/37] locking/lockdep: Convert usage_mask to u64 Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 05/37] locking/lockdep: Introduce lock usage mask iterator Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 06/37] locking/lockdep: Test all incompatible scenario at once in check_irq_usage() Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 07/37] locking/lockdep: Prepare valid_state() to handle plain masks Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 08/37] locking/lockdep: Prepare check_usage_*() " Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 09/37] locking/lockdep: Prepare state_verbose() to handle all softirqs Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 10/37] locking/lockdep: Make mark_lock() fastpath to work with multiple usage at once Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 11/37] locking/lockdep: Save stack trace for each softirq vector involved Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 12/37] locking/lockdep: Report all usages on mark_lock() verbosity mode Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 13/37] softirq: Macrofy softirq vectors Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 14/37] locking/lockdep: Define per vector softirq lock usage states Frederic Weisbecker
2019-04-09 12:03   ` Peter Zijlstra
2019-02-28 17:12 ` [PATCH 15/37] softirq: Pass softirq vector number to lockdep on vector execution Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 16/37] x86: Revert "x86/irq: Demote irq_cpustat_t::__softirq_pending to u16" Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 17/37] arch/softirq: Rename softirq_pending fields to softirq_data Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 18/37] softirq: Normalize softirq_pending naming scheme Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 19/37] softirq: Convert softirq_pending_*() to set/clear mask scheme Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 20/37] softirq: Introduce disabled softirq vectors bits Frederic Weisbecker
2019-03-01 11:29   ` Sebastian Andrzej Siewior
2019-02-28 17:12 ` [PATCH 21/37] softirq: Rename _local_bh_enable() to local_bh_enable_no_softirq() Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 22/37] softirq: Move vectors bits to bottom_half.h Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 23/37] x86: Init softirq enabled field Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 24/37] parisc: " Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 25/37] powerpc: " Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 26/37] softirq: Init softirq enabled field for default irq_stat definition Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 27/37] softirq: Check enabled vectors before processing Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 28/37] softirq: Remove stale comment Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 29/37] softirq: Uninline !CONFIG_TRACE_IRQFLAGS __local_bh_disable_ip() Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 30/37] softirq: Prepare for mixing all/per-vector masking Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 31/37] softirq: Support per vector masking Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 32/37] locking/lockdep: Remove redundant softirqs on check Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 33/37] locking/lockdep: Update check_flags() according to new layout Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 34/37] locking/lockdep: Branch the new vec-finegrained softirq masking to lockdep Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 35/37] softirq: Allow to soft interrupt vector-specific masked contexts Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 36/37] locking: Introduce spin_[un]lock_bh_mask() Frederic Weisbecker
2019-02-28 17:12 ` [PATCH 37/37] net: Make softirq vector masking finegrained on release_sock() Frederic Weisbecker
2019-02-28 17:33 ` [PATCH 00/37] softirq: Per vector masking v3 Linus Torvalds
2019-03-01  3:45   ` Frederic Weisbecker
2019-03-01 16:51     ` Linus Torvalds [this message]
2019-03-08 15:30       ` Frederic Weisbecker

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CAHk-=wgOZuGZaVOOiC=drG6ykVkOGk8RRXZ_CrPBMXHKjTg0dg@mail.gmail.com' \
    --to=torvalds@linux-foundation.org \
    --cc=bigeasy@linutronix.de \
    --cc=davem@davemloft.net \
    --cc=frederic@kernel.org \
    --cc=fweisbec@gmail.com \
    --cc=joel@joelfernandes.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mchehab+samsung@kernel.org \
    --cc=mingo@kernel.org \
    --cc=paulmck@linux.vnet.ibm.com \
    --cc=peterz@infradead.org \
    --cc=pkondeti@codeaurora.org \
    --cc=tglx@linutronix.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body. 
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).