From: Arnd Bergmann <arnd@arndb.de>
To: Roland Dreier <rdreier@cisco.com>
Cc: Mark Nelson <markn@au1.ibm.com>,
linuxppc-dev@ozlabs.org, cbe-oss-dev@ozlabs.org
Subject: Re: [patch 9/9] powerpc/cell: Add DMA_ATTR_STRONG_ORDERING dma attribute and use in IOMMU code
Date: Wed, 16 Jul 2008 09:54:03 +0200 [thread overview]
Message-ID: <200807160954.03633.arnd@arndb.de> (raw)
In-Reply-To: <ada7ibmd115.fsf@cisco.com>
On Wednesday 16 July 2008, Roland Dreier wrote:
> =A0> Strong ordering is only active when both the bridge and the IOMMU en=
able
> =A0> it, but for correctly written drivers, this only results in a slowdo=
wn.
>=20
> So when would someone use this dma attribute? =A0As a hack to fix drivers
> where the real fix is too complicated?
This is used in the Axon PCIe endpoint drivers, e.g. in the Roadrunner
machine. The reason was to improve roundtrip latency by doing only
mmio stores, not loads, on each side of the PCIe connection, which
turn into posted DMA operations on the other end. With relaxed ordering,
the posted writes may be observed out of order. Strong ordering makes
sure they arrive in-order without having to do a non-posted mmio read
or eieio operation on the receiver side.
Arnd <><
next prev parent reply other threads:[~2008-07-16 7:54 UTC|newest]
Thread overview: 25+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-07-15 19:51 [patch 0/9] Cell patches for 2.6.27, version 2 arnd
2008-07-15 19:51 ` [patch 1/9] powerpc/cell/edac: log a syndrome code in case of correctable error arnd
2008-07-17 5:59 ` Benjamin Herrenschmidt
2008-07-17 18:35 ` Doug Thompson
2008-07-15 19:51 ` [patch 2/9] powerpc/axonram: use only one block device major number arnd
2008-07-15 19:51 ` [patch 3/9] powerpc/axonram: enable partitioning of the Axons DDR2 DIMMs arnd
2008-07-15 19:51 ` [patch 4/9] powerpc/cell/cpufreq: add spu aware cpufreq governor arnd
2008-07-15 19:51 ` [patch 5/9] powerpc/cell: cleanup sysreset_hack for IBM cell blades arnd
2008-07-15 19:51 ` [patch 6/9] powerpc/cell: add support for power button of future " arnd
2008-07-15 19:51 ` [patch 7/9] azfs: initial submit of azfs, a non-buffered filesystem arnd
2008-07-17 6:13 ` Benjamin Herrenschmidt
2008-07-22 9:49 ` [Cbe-oss-dev] " Christoph Hellwig
2008-07-15 19:51 ` [patch 8/9] powerpc/dma: use the struct dma_attrs in iommu code arnd
2008-07-15 19:51 ` [patch 9/9] powerpc/cell: Add DMA_ATTR_STRONG_ORDERING dma attribute and use in IOMMU code arnd
2008-07-15 20:34 ` Roland Dreier
2008-07-15 21:27 ` Arnd Bergmann
2008-07-16 2:18 ` Roland Dreier
2008-07-16 7:54 ` Arnd Bergmann [this message]
2008-07-17 6:20 ` [Cbe-oss-dev] " Benjamin Herrenschmidt
2008-07-17 14:53 ` Arnd Bergmann
2008-07-17 20:10 ` Benjamin Herrenschmidt
2008-07-17 20:10 ` Benjamin Herrenschmidt
2008-07-18 13:03 ` [PATCH] Add DMA_ATTR_WEAK_ORDERING dma attribute and use in Cell " Arnd Bergmann
2008-07-19 7:29 ` [Cbe-oss-dev] " Jeremy Kerr
2008-07-19 8:36 ` Arnd Bergmann
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=200807160954.03633.arnd@arndb.de \
--to=arnd@arndb.de \
--cc=cbe-oss-dev@ozlabs.org \
--cc=linuxppc-dev@ozlabs.org \
--cc=markn@au1.ibm.com \
--cc=rdreier@cisco.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.