From: Benjamin Herrenschmidt <benh@kernel.crashing.org>
To: Matthew Wilcox <matthew@wil.cx>
Cc: Linus Torvalds <torvalds@linux-foundation.org>,
David Miller <davem@davemloft.net>,
linux-arch@vger.kernel.org, scottwood@freescale.com,
linuxppc-dev@ozlabs.org, alan@lxorguk.ukuu.org.uk,
linux-kernel@vger.kernel.org, tpiepho@freescale.com
Subject: Re: MMIO and gcc re-ordering issue
Date: Wed, 28 May 2008 08:17:16 +1000 [thread overview]
Message-ID: <1211926636.3286.100.camel@pasglop> (raw)
In-Reply-To: <20080527214241.GA22636@parisc-linux.org>
On Tue, 2008-05-27 at 15:42 -0600, Matthew Wilcox wrote:
> On Wed, May 28, 2008 at 07:38:55AM +1000, Benjamin Herrenschmidt wrote:
> > A problem with __raw_ though is that they -also- don't do byteswap,
> > which is a pain in the neck as people use them for either one reason
> > (relaxed ordering) or the other (no byteswap) without always knowing the
> > consequences of doing so...
>
> That's why there's __readl() which does byteswap, but doesn't do
> ordering ...
Ah, that one is news to me. I don't think we ever had it on powerpc :-)
> > I'm happy to say that __raw is purely about ordering and make them
> > byteswap on powerpc tho (ie, make them little endian like the non-raw
> > counterpart).
>
> That would break a lot of drivers.
How many actually use __raw_ * ?
> > Some archs started providing writel_be etc... I added those to powerpc a
> > little while ago, and I tend to prefer that approach for the byteswap
> > issue.
> Those are for people who use big endian chips on little endian
> architectures.
Why limit them to LE architecture ? There is nothing fundamentally
speicifc to LE architectures here, and it's wrong to provide accessors
on some archs and not others. The endianness is a property of the device
registers. Current writel/readl are basically writel_le/readl_le. It
thus makes sense to have the opposite, ie, readl_be/writel_be, which
thus byteswaps on LE platforms and not on BE platforms, which is what I
provided on powerpc a while ago.
Ben.
WARNING: multiple messages have this Message-ID (diff)
From: Benjamin Herrenschmidt <benh@kernel.crashing.org>
To: Matthew Wilcox <matthew@wil.cx>
Cc: linux-arch@vger.kernel.org, linux-kernel@vger.kernel.org,
tpiepho@freescale.com, linuxppc-dev@ozlabs.org,
scottwood@freescale.com,
Linus Torvalds <torvalds@linux-foundation.org>,
David Miller <davem@davemloft.net>,
alan@lxorguk.ukuu.org.uk
Subject: Re: MMIO and gcc re-ordering issue
Date: Wed, 28 May 2008 08:17:16 +1000 [thread overview]
Message-ID: <1211926636.3286.100.camel@pasglop> (raw)
In-Reply-To: <20080527214241.GA22636@parisc-linux.org>
On Tue, 2008-05-27 at 15:42 -0600, Matthew Wilcox wrote:
> On Wed, May 28, 2008 at 07:38:55AM +1000, Benjamin Herrenschmidt wrote:
> > A problem with __raw_ though is that they -also- don't do byteswap,
> > which is a pain in the neck as people use them for either one reason
> > (relaxed ordering) or the other (no byteswap) without always knowing the
> > consequences of doing so...
>
> That's why there's __readl() which does byteswap, but doesn't do
> ordering ...
Ah, that one is news to me. I don't think we ever had it on powerpc :-)
> > I'm happy to say that __raw is purely about ordering and make them
> > byteswap on powerpc tho (ie, make them little endian like the non-raw
> > counterpart).
>
> That would break a lot of drivers.
How many actually use __raw_ * ?
> > Some archs started providing writel_be etc... I added those to powerpc a
> > little while ago, and I tend to prefer that approach for the byteswap
> > issue.
> Those are for people who use big endian chips on little endian
> architectures.
Why limit them to LE architecture ? There is nothing fundamentally
speicifc to LE architectures here, and it's wrong to provide accessors
on some archs and not others. The endianness is a property of the device
registers. Current writel/readl are basically writel_le/readl_le. It
thus makes sense to have the opposite, ie, readl_be/writel_be, which
thus byteswaps on LE platforms and not on BE platforms, which is what I
provided on powerpc a while ago.
Ben.
next prev parent reply other threads:[~2008-05-27 22:17 UTC|newest]
Thread overview: 319+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-05-20 20:40 [PATCH] [POWERPC] Improve (in|out)_beXX() asm code Trent Piepho
2008-05-20 20:40 ` Trent Piepho
2008-05-20 21:16 ` Benjamin Herrenschmidt
2008-05-20 21:16 ` Benjamin Herrenschmidt
2008-05-20 21:38 ` Scott Wood
2008-05-20 21:38 ` Scott Wood
2008-05-20 22:02 ` Benjamin Herrenschmidt
2008-05-20 22:02 ` Benjamin Herrenschmidt
2008-05-20 22:21 ` Trent Piepho
2008-05-20 22:15 ` Alan Cox
2008-05-20 22:15 ` Alan Cox
2008-05-20 22:35 ` Scott Wood
2008-05-20 22:35 ` Scott Wood
2008-05-20 22:39 ` David Miller
2008-05-20 22:39 ` David Miller
2008-05-20 22:43 ` Scott Wood
2008-05-20 22:43 ` Scott Wood
2008-05-20 22:53 ` David Miller
2008-05-20 22:53 ` David Miller
2008-05-23 4:24 ` Benjamin Herrenschmidt
2008-05-22 22:56 ` Trent Piepho
2008-05-22 22:56 ` Trent Piepho
2008-05-23 12:36 ` MMIO and gcc re-ordering (Was: [PATCH] [POWERPC] Improve (in|out)_beXX() asm code) Benjamin Herrenschmidt
2008-05-23 12:36 ` Benjamin Herrenschmidt
2008-05-23 12:50 ` Benjamin Herrenschmidt
2008-05-23 12:50 ` Benjamin Herrenschmidt
2008-05-23 21:14 ` Scott Wood
2008-05-23 21:14 ` Scott Wood
2008-05-23 22:47 ` Benjamin Herrenschmidt
2008-05-23 22:47 ` Benjamin Herrenschmidt
2008-05-27 1:33 ` MMIO and gcc re-ordering issue Benjamin Herrenschmidt
2008-05-27 1:33 ` Benjamin Herrenschmidt
2008-05-27 1:40 ` David Miller
2008-05-27 1:40 ` David Miller
2008-05-27 2:15 ` Benjamin Herrenschmidt
2008-05-27 2:15 ` Benjamin Herrenschmidt
2008-05-27 2:28 ` David Miller
2008-05-27 2:28 ` David Miller
2008-05-27 3:39 ` Benjamin Herrenschmidt
2008-05-27 3:39 ` Benjamin Herrenschmidt
2008-05-27 15:35 ` Linus Torvalds
2008-05-27 15:35 ` Linus Torvalds
2008-05-27 16:47 ` Linus Torvalds
2008-05-27 16:47 ` Linus Torvalds
2008-05-27 17:31 ` Linus Torvalds
2008-05-27 17:31 ` Linus Torvalds
2008-06-02 10:36 ` Ingo Molnar
2008-06-02 10:36 ` Ingo Molnar
2008-06-02 21:53 ` Benjamin Herrenschmidt
2008-06-02 21:53 ` Benjamin Herrenschmidt
2008-05-27 21:12 ` Benjamin Herrenschmidt
2008-05-27 21:12 ` Benjamin Herrenschmidt
2008-05-27 18:23 ` Trent Piepho
2008-05-27 18:23 ` Trent Piepho
2008-05-27 18:33 ` Scott Wood
2008-05-27 18:33 ` Scott Wood
2008-05-27 21:10 ` Benjamin Herrenschmidt
2008-05-27 21:10 ` Benjamin Herrenschmidt
2008-05-27 21:30 ` Linus Torvalds
2008-05-27 21:30 ` Linus Torvalds
2008-05-27 21:38 ` Alan Cox
2008-05-27 21:38 ` Alan Cox
2008-05-27 21:53 ` Matthew Wilcox
2008-05-27 21:53 ` Matthew Wilcox
2008-05-27 21:46 ` Alan Cox
2008-05-27 21:46 ` Alan Cox
2008-05-27 22:02 ` Linus Torvalds
2008-05-27 22:02 ` Linus Torvalds
2008-05-27 21:59 ` Linus Torvalds
2008-05-27 21:59 ` Linus Torvalds
2008-05-27 21:38 ` Benjamin Herrenschmidt
2008-05-27 21:38 ` Benjamin Herrenschmidt
2008-05-27 21:42 ` Matthew Wilcox
2008-05-27 21:42 ` Matthew Wilcox
2008-05-27 22:17 ` Benjamin Herrenschmidt [this message]
2008-05-27 22:17 ` Benjamin Herrenschmidt
2008-05-28 8:36 ` Haavard Skinnemoen
2008-05-28 8:36 ` Haavard Skinnemoen
2008-05-29 11:05 ` Pantelis Antoniou
2008-05-29 11:05 ` Pantelis Antoniou
2008-05-30 1:13 ` Benjamin Herrenschmidt
2008-05-30 1:13 ` Benjamin Herrenschmidt
2008-05-30 6:07 ` Haavard Skinnemoen
2008-05-30 6:07 ` Haavard Skinnemoen
2008-05-30 7:24 ` Benjamin Herrenschmidt
2008-05-30 7:24 ` Benjamin Herrenschmidt
2008-05-30 8:27 ` Haavard Skinnemoen
2008-05-30 8:27 ` Haavard Skinnemoen
2008-05-30 9:22 ` Geert Uytterhoeven
2008-05-30 9:22 ` Geert Uytterhoeven
2008-06-02 8:11 ` Haavard Skinnemoen
2008-06-02 8:11 ` Haavard Skinnemoen
2008-06-02 15:48 ` Scott Wood
2008-06-02 15:48 ` Scott Wood
2008-06-03 7:46 ` Haavard Skinnemoen
2008-06-03 7:46 ` Haavard Skinnemoen
2008-06-04 15:31 ` Linus Torvalds
2008-06-04 15:31 ` Linus Torvalds
2008-05-27 21:55 ` Linus Torvalds
2008-05-27 21:55 ` Linus Torvalds
2008-05-27 22:19 ` Benjamin Herrenschmidt
2008-05-27 22:19 ` Benjamin Herrenschmidt
2008-05-29 7:10 ` Arnd Bergmann
2008-05-29 7:10 ` Arnd Bergmann
2008-05-29 10:46 ` Alan Cox
2008-05-29 10:46 ` Alan Cox
2008-06-02 7:24 ` Russell King
2008-06-02 7:24 ` Russell King
2008-06-03 4:16 ` Nick Piggin
2008-06-03 4:16 ` Nick Piggin
2008-06-03 4:32 ` Benjamin Herrenschmidt
2008-06-03 4:32 ` Benjamin Herrenschmidt
2008-06-03 6:11 ` Nick Piggin
2008-06-03 6:11 ` Nick Piggin
2008-06-03 6:48 ` Benjamin Herrenschmidt
2008-06-03 6:48 ` Benjamin Herrenschmidt
2008-06-03 6:53 ` Paul Mackerras
2008-06-03 6:53 ` Paul Mackerras
2008-06-03 7:18 ` Nick Piggin
2008-06-03 7:18 ` Nick Piggin
2008-06-03 14:47 ` Linus Torvalds
2008-06-03 14:47 ` Linus Torvalds
2008-06-03 18:47 ` Trent Piepho
2008-06-03 18:47 ` Trent Piepho
2008-06-03 18:55 ` Matthew Wilcox
2008-06-03 18:55 ` Matthew Wilcox
2008-06-03 19:57 ` Trent Piepho
2008-06-03 19:57 ` Trent Piepho
2008-06-03 21:35 ` Matthew Wilcox
2008-06-03 21:35 ` Matthew Wilcox
2008-06-03 21:58 ` Trent Piepho
2008-06-03 21:58 ` Trent Piepho
2008-06-04 2:00 ` Nick Piggin
2008-06-04 2:00 ` Nick Piggin
2008-06-03 19:07 ` Linus Torvalds
2008-06-03 19:07 ` Linus Torvalds
2008-06-04 2:05 ` Nick Piggin
2008-06-04 2:05 ` Nick Piggin
2008-06-04 2:46 ` Linus Torvalds
2008-06-04 2:46 ` Linus Torvalds
2008-06-04 11:47 ` Alan Cox
2008-06-04 11:47 ` Alan Cox
2008-06-10 6:56 ` Nick Piggin
2008-06-10 6:56 ` Nick Piggin
2008-06-10 17:41 ` Jesse Barnes
2008-06-10 17:41 ` Jesse Barnes
2008-06-10 18:10 ` James Bottomley
2008-06-10 18:10 ` James Bottomley
2008-06-10 19:05 ` Roland Dreier
2008-06-10 19:05 ` Roland Dreier
2008-06-10 19:19 ` Jesse Barnes
2008-06-10 19:19 ` Jesse Barnes
2008-06-11 3:29 ` Nick Piggin
2008-06-11 3:29 ` Nick Piggin
2008-06-11 3:29 ` Nick Piggin
2008-06-11 3:29 ` Nick Piggin
2008-06-11 3:40 ` Benjamin Herrenschmidt
2008-06-11 3:40 ` Benjamin Herrenschmidt
2008-06-11 4:06 ` Nick Piggin
2008-06-11 4:06 ` Nick Piggin
2008-06-11 16:07 ` Jesse Barnes
2008-06-11 16:07 ` Jesse Barnes
2008-06-12 11:27 ` Nick Piggin
2008-06-12 11:27 ` Nick Piggin
2008-06-11 4:18 ` Paul Mackerras
2008-06-11 4:18 ` Paul Mackerras
2008-06-11 5:00 ` Nick Piggin
2008-06-11 5:00 ` Nick Piggin
2008-06-11 5:13 ` Paul Mackerras
2008-06-11 5:13 ` Paul Mackerras
2008-06-11 5:35 ` Nick Piggin
2008-06-11 5:35 ` Nick Piggin
2008-06-11 6:02 ` Nick Piggin
2008-06-11 6:02 ` Nick Piggin
2008-06-12 12:14 ` Paul Mackerras
2008-06-12 12:14 ` Paul Mackerras
2008-06-12 13:08 ` Nick Piggin
2008-06-12 13:08 ` Nick Piggin
2008-06-11 14:46 ` Linus Torvalds
2008-06-11 14:46 ` Linus Torvalds
2008-06-11 5:20 ` Paul Mackerras
2008-06-11 5:20 ` Paul Mackerras
2008-06-04 2:19 ` Nick Piggin
2008-06-04 2:19 ` Nick Piggin
2008-06-03 19:43 ` Trent Piepho
2008-06-03 19:43 ` Trent Piepho
2008-06-03 21:33 ` Matthew Wilcox
2008-06-03 21:33 ` Matthew Wilcox
2008-06-03 21:44 ` Trent Piepho
2008-06-03 21:44 ` Trent Piepho
2008-06-04 2:25 ` Nick Piggin
2008-06-04 2:25 ` Nick Piggin
2008-06-04 6:39 ` Trent Piepho
2008-06-04 6:39 ` Trent Piepho
2008-06-03 22:26 ` Benjamin Herrenschmidt
2008-06-03 22:26 ` Benjamin Herrenschmidt
2008-05-27 3:42 ` Arjan van de Ven
2008-05-27 3:42 ` Arjan van de Ven
2008-05-27 4:08 ` Roland Dreier
2008-05-27 4:08 ` Roland Dreier
2008-05-27 4:20 ` Arjan van de Ven
2008-05-27 4:20 ` Arjan van de Ven
2008-05-27 7:08 ` Benjamin Herrenschmidt
2008-05-27 7:08 ` Benjamin Herrenschmidt
2008-05-27 15:50 ` Roland Dreier
2008-05-27 15:50 ` Roland Dreier
2008-05-27 16:37 ` James Bottomley
2008-05-27 16:37 ` James Bottomley
2008-05-27 17:38 ` Roland Dreier
2008-05-27 17:38 ` Roland Dreier
2008-05-27 17:53 ` James Bottomley
2008-05-27 17:53 ` James Bottomley
2008-05-27 18:07 ` Roland Dreier
2008-05-27 18:07 ` Roland Dreier
2008-05-27 18:17 ` Roland Dreier
2008-05-27 18:17 ` Roland Dreier
2008-05-27 21:23 ` Chris Friesen
2008-05-27 21:23 ` Chris Friesen
2008-05-27 21:29 ` Roland Dreier
2008-05-27 21:29 ` Roland Dreier
2008-05-27 23:04 ` Paul Mackerras
2008-05-27 23:04 ` Paul Mackerras
2008-05-27 21:11 ` Benjamin Herrenschmidt
2008-05-27 21:11 ` Benjamin Herrenschmidt
2008-05-27 21:33 ` Roland Dreier
2008-05-27 21:33 ` Roland Dreier
2008-05-27 22:13 ` Benjamin Herrenschmidt
2008-05-27 22:13 ` Benjamin Herrenschmidt
2008-05-27 22:39 ` Roland Dreier
2008-05-27 22:39 ` Roland Dreier
2008-05-29 14:47 ` Jes Sorensen
2008-05-29 14:47 ` Jes Sorensen
2008-05-29 15:01 ` James Bottomley
2008-05-29 15:01 ` James Bottomley
2008-05-30 9:36 ` Jes Sorensen
2008-05-30 9:36 ` Jes Sorensen
2008-05-30 17:21 ` Jesse Barnes
2008-05-30 17:21 ` Jesse Barnes
2008-05-31 7:57 ` Jeremy Higdon
2008-05-31 7:57 ` Jeremy Higdon
2008-05-29 21:40 ` Benjamin Herrenschmidt
2008-05-29 21:40 ` Benjamin Herrenschmidt
2008-05-29 21:48 ` Trent Piepho
2008-05-29 21:48 ` Trent Piepho
2008-05-29 22:05 ` Benjamin Herrenschmidt
2008-05-29 22:05 ` Benjamin Herrenschmidt
2008-05-30 1:53 ` Trent Piepho
2008-05-30 1:53 ` Trent Piepho
2008-05-29 21:53 ` Jesse Barnes
2008-05-29 21:53 ` Jesse Barnes
2008-05-30 9:39 ` Jes Sorensen
2008-05-30 9:39 ` Jes Sorensen
2008-05-30 9:48 ` Jes Sorensen
2008-05-30 9:48 ` Jes Sorensen
2008-05-31 8:14 ` Pavel Machek
2008-05-31 8:14 ` Pavel Machek
2008-06-02 9:48 ` Jes Sorensen
2008-06-02 9:48 ` Jes Sorensen
2008-05-29 22:06 ` Roland Dreier
2008-05-29 22:06 ` Roland Dreier
2008-05-29 22:25 ` Trent Piepho
2008-05-29 22:25 ` Trent Piepho
2008-05-30 3:56 ` Paul Mackerras
2008-05-30 3:56 ` Paul Mackerras
2008-05-31 7:52 ` Jeremy Higdon
2008-05-31 7:52 ` Jeremy Higdon
2008-06-02 9:56 ` Jes Sorensen
2008-06-02 9:56 ` Jes Sorensen
2008-06-02 21:02 ` Jeremy Higdon
2008-06-02 21:02 ` Jeremy Higdon
2008-06-03 4:33 ` Nick Piggin
2008-06-03 4:33 ` Nick Piggin
2008-06-03 8:15 ` Jeremy Higdon
2008-06-03 8:15 ` Jeremy Higdon
2008-06-03 8:19 ` Nick Piggin
2008-06-03 8:19 ` Nick Piggin
2008-06-03 8:45 ` Jeremy Higdon
2008-06-03 8:45 ` Jeremy Higdon
2008-06-03 16:52 ` Jesse Barnes
2008-06-03 16:52 ` Jesse Barnes
2008-06-05 8:40 ` Jes Sorensen
2008-06-05 8:40 ` Jes Sorensen
2008-06-05 8:43 ` Benjamin Herrenschmidt
2008-06-05 8:43 ` Benjamin Herrenschmidt
2008-06-12 15:07 ` Matthew Wilcox
2008-06-12 15:07 ` Matthew Wilcox
2008-06-13 0:07 ` Benjamin Herrenschmidt
2008-06-13 0:07 ` Benjamin Herrenschmidt
2008-05-31 8:04 ` Pavel Machek
2008-05-31 8:04 ` Pavel Machek
2008-05-27 8:24 ` Alan Cox
2008-05-27 8:24 ` Alan Cox
2008-05-27 15:28 ` Jonathan Corbet
2008-05-27 15:28 ` Jonathan Corbet
2008-05-20 22:55 ` [PATCH] [POWERPC] Improve (in|out)_beXX() asm code Trent Piepho
2008-05-20 22:55 ` Trent Piepho
2008-05-21 14:01 ` Benjamin Herrenschmidt
2008-05-21 14:01 ` Benjamin Herrenschmidt
2008-05-20 22:00 ` Trent Piepho
2008-05-20 22:00 ` Trent Piepho
2008-05-21 14:00 ` Benjamin Herrenschmidt
2008-05-21 14:00 ` Benjamin Herrenschmidt
2008-05-21 19:44 ` Trent Piepho
2008-05-21 19:44 ` Trent Piepho
2008-05-21 20:41 ` Benjamin Herrenschmidt
2008-05-21 20:41 ` Benjamin Herrenschmidt
2008-05-20 22:00 ` Andreas Schwab
2008-05-20 22:00 ` Andreas Schwab
2008-05-20 22:11 ` Trent Piepho
2008-05-20 22:11 ` Trent Piepho
2008-05-20 22:47 ` Andreas Schwab
2008-05-20 22:47 ` Andreas Schwab
2008-05-20 23:14 ` Trent Piepho
2008-05-20 23:14 ` Trent Piepho
2008-05-21 8:03 ` Andreas Schwab
2008-05-21 8:03 ` Andreas Schwab
2008-05-21 20:25 ` Trent Piepho
2008-05-21 20:25 ` Trent Piepho
2008-05-27 23:48 ` [PATCH V2] [POWERPC] Improve (in|out)_[bl]eXX() " Trent Piepho
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=1211926636.3286.100.camel@pasglop \
--to=benh@kernel.crashing.org \
--cc=alan@lxorguk.ukuu.org.uk \
--cc=davem@davemloft.net \
--cc=linux-arch@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linuxppc-dev@ozlabs.org \
--cc=matthew@wil.cx \
--cc=scottwood@freescale.com \
--cc=torvalds@linux-foundation.org \
--cc=tpiepho@freescale.com \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.