From: Russell King - ARM Linux <linux@arm.linux.org.uk>
To: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Robin Holt <holt@sgi.com>,
Laurent Desnogues <laurent.desnogues@gmail.com>,
Jamie Lokier <jamie@shareable.org>,
David Xiao <dxiao@broadcom.com>, Ben Dooks <ben-linux@fluff.org>,
Hugh Dickins <hugh.dickins@tiscali.co.uk>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
v4l2_linux <linux-media@vger.kernel.org>,
"linux-arm-kernel@lists.arm.linux.org.uk"
<linux-arm-kernel@lists.arm.linux.org.uk>
Subject: Re: How to efficiently handle DMA and cache on ARMv7 ? (was "Is get_user_pages() enough to prevent pages from being swapped out ?")
Date: Fri, 7 Aug 2009 21:28:29 +0100 [thread overview]
Message-ID: <20090807202829.GF31543@n2100.arm.linux.org.uk> (raw)
In-Reply-To: <200908072211.45283.laurent.pinchart@ideasonboard.com>
On Fri, Aug 07, 2009 at 10:11:40PM +0200, Laurent Pinchart wrote:
> Ok. Maybe the kernel mapping from L_PTE_MT_UNCACHED to strongly ordered for
> ARMv6 and up (not sure about how it worked for previous versions) brought some
> confusion. I'll try to be more precise now.
It's something we should correct.
> Does that mean that, in theory, all DMA transfers in the DMA_FROM_DEVICE
> direction are currently broken on ARMv7 ?
Technically, yes. I haven't had a stream of bug reports which tends
to suggest that either the speculation isn't that aggressive in current
silicon, or we're just lucky so far.
> The ARM Architecture Reference Manual (ARM DDI 0100I) states that
Bear in mind that DDI0100 is out of date now. There's a different
document number for it (I forget what it is.)
> "• If the same memory locations are marked as having different memory types
> (Normal, Device, or Strongly Ordered), for example by the use of synonyms in a
> virtual to physical address mapping, UNPREDICTABLE behavior results.
>
> • If the same memory locations are marked as having different cacheable
> attributes, for example by the use of synonyms in a virtual to physical
> address mapping, UNPREDICTABLE behavior results."
Both of these we end up doing. The current position is "yes, umm, we're
not sure what we can do about that"... which also happens to be mine as
well. Currently, my best solution is to go for minimal lowmem and
maximal highmem - so _everything_ gets mapped in on an as required
basis.
> This would be broken if a fully cached Normal mapping already existed for
> those physical pages. You seem to imply that's the case, but I'm not sure to
> understand why.
The kernel direct mapping maps all system (low) memory with normal
memory cacheable attributes.
So using vmalloc, dma_alloc_coherent, using pages in userspace all
create duplicate mappings of pages.
next prev parent reply other threads:[~2009-08-07 20:28 UTC|newest]
Thread overview: 42+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-08-06 10:08 How to efficiently handle DMA and cache on ARMv7 ? (was "Is get_user_pages() enough to prevent pages from being swapped out ?") Laurent Pinchart
2009-08-06 11:46 ` Ben Dooks
2009-08-06 13:06 ` Laurent Pinchart
2009-08-06 18:46 ` David Xiao
2009-08-06 19:16 ` Chetan.Loke
2009-08-06 20:15 ` Jamie Lokier
2009-08-06 22:25 ` Russell King - ARM Linux
2009-08-07 5:59 ` David Xiao
2009-08-07 7:58 ` Laurent Pinchart
2009-08-07 8:10 ` Russell King - ARM Linux
2009-08-07 9:54 ` Jamie Lokier
2009-08-07 9:59 ` Russell King - ARM Linux
2009-08-07 12:07 ` Laurent Desnogues
2009-08-07 13:15 ` Robin Holt
2009-08-07 19:01 ` Russell King - ARM Linux
2009-08-07 20:11 ` Laurent Pinchart
2009-08-07 20:28 ` Russell King - ARM Linux [this message]
2009-08-07 22:25 ` David Xiao
2009-08-10 13:49 ` Laurent Pinchart
2009-08-07 8:08 ` Russell King - ARM Linux
2009-08-07 10:23 ` Jamie Lokier
2009-08-07 19:03 ` Russell King - ARM Linux
2009-08-11 9:31 ` Catalin Marinas
2009-08-11 18:23 ` David Xiao
2009-08-07 7:48 ` Laurent Pinchart
2009-08-25 12:53 ` Steven Walter
2009-08-25 22:02 ` David Xiao
2009-08-25 23:17 ` Laurent Pinchart
2009-08-26 17:22 ` David Xiao
2009-09-01 13:31 ` Russell King - ARM Linux
2009-09-01 18:08 ` David Xiao
2009-09-01 13:28 ` Russell King - ARM Linux
2009-09-01 13:43 ` Laurent Pinchart
2009-09-01 14:18 ` Russell King - ARM Linux
2009-09-01 16:53 ` Hugh Dickins
2009-09-02 15:10 ` Imre Deak
2009-09-03 7:31 ` Imre Deak
2009-09-03 8:36 ` Russell King - ARM Linux
2009-09-08 13:05 ` Steven Walter
2009-08-07 7:29 ` Laurent Pinchart
2009-08-07 8:12 ` Matthieu CASTET
2009-08-07 10:13 ` How to efficiently handle DMA and cache on ARMv7 ? (was " Is " Laurent Pinchart
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=20090807202829.GF31543@n2100.arm.linux.org.uk \
--to=linux@arm.linux.org.uk \
--cc=ben-linux@fluff.org \
--cc=dxiao@broadcom.com \
--cc=holt@sgi.com \
--cc=hugh.dickins@tiscali.co.uk \
--cc=jamie@shareable.org \
--cc=laurent.desnogues@gmail.com \
--cc=laurent.pinchart@ideasonboard.com \
--cc=linux-arm-kernel@lists.arm.linux.org.uk \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-media@vger.kernel.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