From: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
To: Jason Baron <jbaron@redhat.com>
Cc: linux-kernel@vger.kernel.org, fweisbec@gmail.com, mingo@elte.hu,
laijs@cn.fujitsu.com, rostedt@goodmis.org, peterz@infradead.org,
jiayingz@google.com, mbligh@google.com, roland@redhat.com,
fche@redhat.com
Subject: Re: [PATCH 0/3] tracepoints: delay argument evaluation
Date: Tue, 19 May 2009 17:17:04 -0400 [thread overview]
Message-ID: <20090519211704.GA3325@Krystal> (raw)
In-Reply-To: <cover.1242763826.git.jbaron@redhat.com>
* Jason Baron (jbaron@redhat.com) wrote:
> hi,
>
> After disassembling some of the tracepoints, I've noticed that arguments that
> are passed as macros or that perform dereferences, evaluate prior to the
> tracepoint on/off check. This means that we are needlessly impacting the
> off case.
>
> I am proposing to fix this by adding a macro that first checks for on/off and
> then calls 'trace_##name', preserving type checking. Thus, callsites have to
> move from:
>
> trace_block_bio_complete(md->queue, bio);
>
> to:
>
> tracepoint_call(block_bio_complete, md->queue, bio);
>
I knew this limitation in the first place, but decided it was not worth
uglifying the tracepoint call site for it.
The expected use is to pass a pointer or a value as tracepoint argument
and dereference it in the callback attached to it.
Is there any _real_ added value for going through this API change pain ?
Mathieu
> I've tried '__always_inline', but that did not fix this issue. Obviously this
> change will require changes to all the callsites. But, that shouldn't be
> very hard, I've already included the scheduler and block changes with this
> patch. I think its important to minimize code execution in the off case, and
> thus going through all the callsites is well worth it. If we agree on this
> change, I can change the rest in very short order.
>
> Below I'm also showing the assembly in the 'dec_pending()' function before and
> after this change to show the difference it makes. The arguments to the
> tracepoint are as above, 'md->queue' and 'bio'. Notice the 2 extra instructions,
> before the initial 'je', that could be moved after the 'je'.
>
> before:
>
> ffffffff8137b2a3: 83 3d de 90 4b 00 00 cmpl $0x0,0x4b90de(%rip) # ffffffff81834388 <__tracepoint_block_bio_complete+0x8>
> ffffffff8137b2aa: 49 8b 45 50 mov 0x50(%r13),%rax
> ffffffff8137b2ae: 48 89 45 d0 mov %rax,-0x30(%rbp)
> ffffffff8137b2b2: 74 1f je ffffffff8137b2d3 <dec_pending+0x101>
> ffffffff8137b2b4: 48 8b 1d d5 90 4b 00 mov 0x4b90d5(%rip),%rbx # ffffffff81834390 <__tracepoint_block_bio_complete+0x10>
> ffffffff8137b2bb: 48 85 db test %rbx,%rbx
> ffffffff8137b2be: 74 13 je ffffffff8137b2d3 <dec_pending+0x101>
> ffffffff8137b2c0: 4c 89 f6 mov %r14,%rsi
> ffffffff8137b2c3: 48 8b 7d d0 mov -0x30(%rbp),%rdi
> ffffffff8137b2c7: ff 13 callq *(%rbx)
> ffffffff8137b2c9: 48 83 c3 08 add $0x8,%rbx
> ffffffff8137b2cd: 48 83 3b 00 cmpq $0x0,(%rbx)
> ffffffff8137b2d1: eb eb jmp ffffffff8137b2be <dec_pending+0xec>
> ffffffff8137b2d3: 44 89 fe mov %r15d,%esi
>
> after:
>
> ffffffff8137b2a3: 83 3d de 90 4b 00 00 cmpl $0x0,0x4b90de(%rip) # ffffffff81834388 <__tracepoint_block_bio_complete+0x8>
> ffffffff8137b2aa: 74 27 je ffffffff8137b2d3 <dec_pending+0x101>
> ffffffff8137b2ac: 49 8b 45 50 mov 0x50(%r13),%rax
> ffffffff8137b2b0: 48 8b 1d d9 90 4b 00 mov 0x4b90d9(%rip),%rbx # ffffffff81834390 <__tracepoint_block_bio_complete+0x10>
> ffffffff8137b2b7: 48 89 45 d0 mov %rax,-0x30(%rbp)
> ffffffff8137b2bb: 48 85 db test %rbx,%rbx
> ffffffff8137b2be: 74 13 je ffffffff8137b2d3 <dec_pending+0x101>
> ffffffff8137b2c0: 4c 89 f6 mov %r14,%rsi
> ffffffff8137b2c3: 48 8b 7d d0 mov -0x30(%rbp),%rdi
> ffffffff8137b2c7: ff 13 callq *(%rbx)
> ffffffff8137b2c9: 48 83 c3 08 add $0x8,%rbx
> ffffffff8137b2cd: 48 83 3b 00 cmpq $0x0,(%rbx)
> ffffffff8137b2d1: eb eb jmp ffffffff8137b2be <dec_pending+0xec>
> ffffffff8137b2d3: 44 89 fe mov %r15d,%esi
>
>
> thanks,
>
> -Jason
>
>
>
> Jason Baron (3):
> -add wrapper so we don't have argument resolution overhead
> -add scheduler wrapper calls
> -add block layer trace wrappers
>
> block/blk-core.c | 27 ++++++++++++++-------------
> block/elevator.c | 6 +++---
> drivers/md/dm.c | 7 ++++---
> fs/bio.c | 2 +-
> include/linux/tracepoint.h | 20 +++++++++++++++++++-
> kernel/exit.c | 6 +++---
> kernel/fork.c | 2 +-
> kernel/kthread.c | 4 ++--
> kernel/sched.c | 10 +++++-----
> kernel/signal.c | 2 +-
> mm/bounce.c | 2 +-
> 11 files changed, 54 insertions(+), 34 deletions(-)
>
--
Mathieu Desnoyers
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F BA06 3F25 A8FE 3BAE 9A68
next prev parent reply other threads:[~2009-05-19 21:17 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-05-19 21:03 [PATCH 0/3] tracepoints: delay argument evaluation Jason Baron
2009-05-19 21:03 ` [PATCH 1/3] tracepoints: add tracepoint_call() to optimize tracepoints disabled Jason Baron
2009-05-19 21:03 ` [PATCH 2/3] tracepoints: convert scheduler tracepoints to 'tracepoint_call' api Jason Baron
2009-05-19 21:03 ` [PATCH 3/3] tracepoints: convert block " Jason Baron
2009-05-19 21:17 ` Mathieu Desnoyers [this message]
2009-05-19 22:16 ` [PATCH 0/3] tracepoints: delay argument evaluation Jason Baron
2009-05-19 22:25 ` Roland McGrath
2009-05-19 22:31 ` Steven Rostedt
2009-05-19 22:36 ` Steven Rostedt
2009-05-19 23:52 ` Frederic Weisbecker
2009-05-20 0:33 ` Mathieu Desnoyers
2009-05-20 0:42 ` Steven Rostedt
2009-05-20 7:01 ` Roland McGrath
2009-05-20 7:12 ` Peter Zijlstra
[not found] ` <20090520072750.DA9A0FC38D@magilla.sf.frob.com>
2009-05-20 7:38 ` Peter Zijlstra
2009-05-20 9:18 ` Roland McGrath
2009-05-20 7:33 ` Ingo Molnar
2009-05-20 15:42 ` Jason Baron
2009-05-21 1:49 ` Jiaying Zhang
2009-05-21 1:59 ` Li Zefan
2009-05-21 2:15 ` Jiaying Zhang
2009-05-21 2:41 ` Li Zefan
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=20090519211704.GA3325@Krystal \
--to=mathieu.desnoyers@polymtl.ca \
--cc=fche@redhat.com \
--cc=fweisbec@gmail.com \
--cc=jbaron@redhat.com \
--cc=jiayingz@google.com \
--cc=laijs@cn.fujitsu.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mbligh@google.com \
--cc=mingo@elte.hu \
--cc=peterz@infradead.org \
--cc=roland@redhat.com \
--cc=rostedt@goodmis.org \
/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).