mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Linus Torvalds <torvalds@linux-foundation.org>
To: Ingo Molnar <mingo@elte.hu>
Cc: Eric Dumazet <eric.dumazet@gmail.com>,
	David Howells <dhowells@redhat.com>,
	akpm@linux-foundation.org, paulus@samba.org, arnd@arndb.de,
	linux-kernel@vger.kernel.org
Subject: Re: [patch] x86: atomic64_t: Improve atomic64_add_return()
Date: Fri, 3 Jul 2009 10:38:06 -0700 (PDT)	[thread overview]
Message-ID: <alpine.LFD.2.01.0907031035340.3210@localhost.localdomain> (raw)
In-Reply-To: <20090703120159.GB7161@elte.hu>



On Fri, 3 Jul 2009, Ingo Molnar wrote:
>  u64 atomic64_add_return(u64 delta, atomic64_t *ptr)
>  {
> +	/*
> +	 * Try first with a (probably incorrect) assumption about
> +	 * what we have there. We'll do two loops most likely,
> +	 * but we'll get an ownership MESI transaction straight away
> +	 * instead of a read transaction followed by a
> +	 * flush-for-ownership transaction:
> +	 */
> +	u64 old_val, new_val, real_val = 1ULL << 32;
>  
>  	do {
> +		old_val = real_val;
>  		new_val = old_val + delta;
>  
> +		real_val = atomic64_cmpxchg(ptr, old_val, new_val);
> +
> +	} while (real_val != old_val);

For the case where we actually have to keep looping until we succeed, 
we're probably better off starting with a good guess rather than a bad 
one.

So I'd suggest using 'real_val = 0', which gives smaller code, rather than 
the 1ull << 32 that is actively trying to be a bad guess.

It won't matter all that much (since 0 really isn't a much better guess 
than 1ull<<32), but I'd rather have a simple constant that doesn't matter, 
over an odd constant that makes no sense.

			Linus

  parent reply	other threads:[~2009-07-03 17:38 UTC|newest]

Thread overview: 79+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-06-30 21:24 [PATCH] FRV: Wire up new syscalls David Howells
2009-06-30 21:34 ` Ingo Molnar
2009-06-30 21:41   ` Arnd Bergmann
2009-06-30 21:54     ` Ingo Molnar
2009-07-01 11:28     ` David Howells
2009-07-01 11:54       ` Ingo Molnar
2009-07-01 12:19       ` David Howells
2009-07-01 12:36         ` Paul Mackerras
2009-07-01 12:41         ` David Howells
2009-07-01 13:13           ` Ingo Molnar
2009-07-01 14:10           ` David Howells
2009-07-01 14:49             ` Ingo Molnar
2009-07-01 16:47               ` [PATCH 1/2] FRV: Implement atomic64_t David Howells
2009-07-01 17:20                 ` Linus Torvalds
2009-07-01 21:11                   ` Ingo Molnar
2009-07-01 22:57                   ` [PATCH] x86: Code atomic(64)_read and atomic(64)_set in C not CPP [was Re: FRV: Implement atomic64_t] Paul Mackerras
2009-07-02  7:21                     ` [tip:x86/urgent] x86: Code atomic(64)_read and atomic(64)_set in C not CPP tip-bot for Paul Mackerras
2009-07-02  7:21                     ` [PATCH] x86: Code atomic(64)_read and atomic(64)_set in C not CPP [was Re: FRV: Implement atomic64_t] Ingo Molnar
2009-07-01 23:46                   ` [PATCH 1/2] FRV: Implement atomic64_t [ver #2] David Howells
2009-07-01 23:46                   ` [PATCH 2/2] FRV: Add basic performance counter support " David Howells
2009-07-02 21:10                   ` [PATCH 1/2] FRV: Implement atomic64_t Eric Dumazet
2009-07-02 21:28                     ` Linus Torvalds
2009-07-02 22:08                       ` [PATCH] x86: atomic64_t should be 8 bytes aligned Eric Dumazet
2009-07-02 23:53                         ` Linus Torvalds
2009-07-03  6:14                           ` Ingo Molnar
2009-07-03 12:42                           ` [tip:perfcounters/urgent] x86: atomic64: The atomic64_t data type should be 8 bytes aligned on 32-bit too tip-bot for Eric Dumazet
2009-07-03 16:58                             ` Linus Torvalds
2009-07-03 17:49                               ` H. Peter Anvin
2009-07-03 12:42                           ` [tip:perfcounters/urgent] x86: atomic64: Move the 32-bit atomic64_t implementation to a .c file tip-bot for Ingo Molnar
2009-07-03 16:47                             ` Linus Torvalds
2009-07-03 18:31                               ` [tip:perfcounters/urgent] x86: atomic64: Clean up atomic64_sub_and_test() and atomic64_add_negative() tip-bot for Ingo Molnar
2009-07-03 19:18                               ` tip-bot for Ingo Molnar
2009-07-04  0:05                             ` [tip:perfcounters/urgent] x86: atomic64: Move the 32-bit atomic64_t implementation to a .c file Paul Mackerras
2009-07-05 11:25                               ` Ingo Molnar
2009-07-03 12:43                           ` [tip:perfcounters/urgent] x86: atomic64: Improve atomic64_read() tip-bot for Eric Dumazet
2009-07-03 12:43                           ` [tip:perfcounters/urgent] x86: atomic64: Improve cmpxchg8b() tip-bot for Eric Dumazet
2009-07-03 12:43                           ` [tip:perfcounters/urgent] x86: atomic64: Improve atomic64_add_return() tip-bot for Ingo Molnar
2009-07-03 12:43                           ` [tip:perfcounters/urgent] x86: atomic64: Reduce size of functions tip-bot for Ingo Molnar
2009-07-03 12:44                           ` [tip:perfcounters/urgent] x86: atomic64: Fix unclean type use in atomic64_xchg() tip-bot for Ingo Molnar
2009-07-03 17:02                             ` Linus Torvalds
2009-07-03 18:00                               ` Ingo Molnar
2009-07-03 12:44                           ` [tip:perfcounters/urgent] x86: atomic64: Improve atomic64_read() tip-bot for Eric Dumazet
2009-07-03 14:50                             ` [PATCH -tip] x86: atomic64: inline atomic64_read() Eric Dumazet
2009-07-03 18:04                               ` Ingo Molnar
2009-07-03 18:10                                 ` Arjan van de Ven
2009-07-03 18:18                                   ` Ingo Molnar
2009-07-03 18:25                                     ` Andi Kleen
2009-07-03 18:30                                     ` Arjan van de Ven
2009-07-03 18:43                                       ` Ingo Molnar
2009-07-03 18:24                                   ` Andi Kleen
2009-07-03 18:31                                   ` [tip:perfcounters/urgent] x86: atomic64: Optimize CMPXCHG8B sequences to not use the LOCK prefix tip-bot for Ingo Molnar
2009-07-03 18:45                                     ` Ingo Molnar
2009-07-03 19:10                                 ` [PATCH -tip] x86: atomic64: inline atomic64_read() Linus Torvalds
2009-07-03 19:17                                   ` Ingo Molnar
2009-07-03 19:38                                     ` Linus Torvalds
2009-07-03 21:40                                       ` Ingo Molnar
2009-07-03 18:31                               ` [tip:perfcounters/urgent] x86: atomic64: Inline atomic64_read() again tip-bot for Eric Dumazet
2009-07-03 19:18                               ` tip-bot for Eric Dumazet
2009-07-04  9:49                               ` tip-bot for Eric Dumazet
2009-07-03 12:44                           ` [tip:perfcounters/urgent] x86: atomic64: Code atomic(64)_read and atomic(64)_set in C not CPP tip-bot for Paul Mackerras
2009-07-03 12:48                           ` tip-bot for Paul Mackerras
2009-07-03 12:48                           ` [tip:perfcounters/urgent] x86: atomic64: Improve atomic64_read() tip-bot for Eric Dumazet
2009-07-03 15:33                           ` [tip:perfcounters/urgent] x86: atomic64: Export APIs to modules tip-bot for Ingo Molnar
2009-07-03 18:30                             ` tip-bot for Ingo Molnar
2009-07-03 18:30                             ` [tip:perfcounters/urgent] x86: atomic64: Improve atomic64_xchg() tip-bot for Ingo Molnar
2009-07-03 12:01                       ` [patch] x86: atomic64_t: Improve atomic64_add_return() Ingo Molnar
2009-07-03 12:26                         ` [PATCH] x86: atomic64_t: _cmpxchg() & _read() optimizations Eric Dumazet
2009-07-03 12:40                           ` Ingo Molnar
2009-07-03 17:38                         ` Linus Torvalds [this message]
2009-07-03  6:05                     ` [PATCH 1/2] FRV: Implement atomic64_t Eric Dumazet
2009-07-03 12:27                       ` Ingo Molnar
2009-07-03 12:39                         ` Eric Dumazet
2009-07-03 11:17                     ` Ingo Molnar
2009-07-03 11:26                       ` Ingo Molnar
2009-07-01 17:33                 ` David Howells
2009-07-01 23:48                 ` David Howells
2009-07-01 16:47               ` [PATCH 2/2] FRV: Add basic performance counter support David Howells
2009-07-01 21:10                 ` Ingo Molnar
2009-07-01 15:19             ` [PATCH] FRV: Wire up new syscalls David Howells

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=alpine.LFD.2.01.0907031035340.3210@localhost.localdomain \
    --to=torvalds@linux-foundation.org \
    --cc=akpm@linux-foundation.org \
    --cc=arnd@arndb.de \
    --cc=dhowells@redhat.com \
    --cc=eric.dumazet@gmail.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=paulus@samba.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

Powered by JetHome