From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Date: Tue, 21 Aug 2007 20:55:29 +0000 Subject: Re: [PATCH 2/3] dma: override "dma_flags_set_dmaflush" for sn-ia64 Message-Id: <1187729729.18410.48.camel@localhost.localdomain> List-Id: References: <20070818002746.GU1813@sgi.com> <46C94FD5.6000006@sgi.com> <20070821193522.GD5592@sgi.com> <20070821130515.6e745b17.randy.dunlap@oracle.com> In-Reply-To: <20070821130515.6e745b17.randy.dunlap@oracle.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Randy Dunlap Cc: akepner@sgi.com, Jes Sorensen , linux-kernel , rdreier@cisco.com, linux-ia64 On Tue, 2007-08-21 at 13:05 -0700, Randy Dunlap wrote: > > diff --git a/Documentation/DMA-API.txt b/Documentation/DMA-API.txt > > index cc7a8c3..e117b72 100644 > > --- a/Documentation/DMA-API.txt > > +++ b/Documentation/DMA-API.txt > > @@ -392,6 +392,28 @@ Notes: You must do this: > > > > See also dma_map_single(). > > > > +int > > +dma_flags_set_dmaflush(int dir) > > + > > +Amend dir (one of the enum dma_data_direction values), with a platform- > > no comma. > > > +specific "dmaflush" attribute. Unless the platform supports "posted DMA" > > add comma after "posted DMA" and drop lots of trailing spaces. > > > +this is a no-op Almost every platform supports posted DMA ... its a property of most PCI bridge chips. > > +On platforms that support posted DMA, dma_flags_set_dmaflush() causes > > +device writes to the associated memory region to flush in-flight DMA. > > +This can be important, for example, when (DMA) writes to the memory > > +region indicate that DMA of data is complete. If DMA of data and DMA of > > +the completion indication race, as they can do when the platform supports > > +posted DMA, then the completion indication may arrive in host memory > > +ahead of some data. This isn't possible on most platforms. PCI write posting can only be flushed by a read transaction on the device (or sometimes any device on the bridge). Either this interface is misnamed and misdescribed, or it can't work for most systems. > > +To prevent this, you might map the memory region used for completion > > +indications as follows: > > + > > + int count, flags = dma_flags_set_dmaflush(DMA_BIDIRECTIONAL); > > + ..... > > + count = dma_map_sg(dev, sglist, nents, flags); James