From mboxrd@z Thu Jan 1 00:00:00 1970 From: jbarnes@sgi.com (Jesse Barnes) Date: Fri, 16 Jan 2004 17:21:00 +0000 Subject: Re: [PATCH] readX_relaxed interface Message-Id: <20040116172100.GA13553@sgi.com> List-Id: References: <20040115204913.GA8172@sgi.com> <20040116003224.GF23253@kroah.com> In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Linus Torvalds Cc: Greg KH , linux-pci@atrey.karlin.mff.cuni.cz, linux-kernel@vger.kernel.org, linux-ia64@vger.kernel.org, jeremy@sgi.com On Thu, Jan 15, 2004 at 09:00:10PM -0800, Linus Torvalds wrote: > If you care about machine check errors, use a special interface for that. > A _really_ special one. Especially as on many systems you'll likely have > to read status registers etc (and clear them before doing the IO) to see > the errors. > > So that way you can get errors working, AND it won't actually make normal > code any uglier. How about one that allows you to register an error handling function for a given address range and/or device? That would cover both read() and write() cases, and would be optional so drivers wouldn't be forced to become more complicated. Jesse