On Tuesday, September 23, 2008 11:59 am Dave Jones wrote: > On Tue, Sep 23, 2008 at 11:46:36AM -0700, Jesse Barnes wrote: > > On Friday, September 19, 2008 1:31 pm David Miller wrote: > > > From: Roland Dreier <rdreier@xxxxxxxxx> > > > Date: Fri, 19 Sep 2008 10:10:20 -0700 > > > > > > > I investigated a little and discovered that current Fedora kernels > > > > are built with CONFIG_PCI_MSI=y but have an extra patch added that > > > > defaults MSI to off (and requires booting with "pci=msi" to enable > > > > MSI). > > > > > > That makes me feel like a real idiot for trying to make > > > sure all the in-kernel MSI quirks are there and sane. > > > > Chuck and/or Dave (Jones), can one of you summarize the list of bugs you > > ran into last time you enabled MSI unconditionally on Fedora? Maybe we > > can convert the errors you saw into new blacklist entries... If things > > break horribly everytime you try MSI, I can't really blame you for > > disabling it, but it would be nice if we could make some progress here. > > I don't have a list of bug reports handy, but here's one such example > I found searching for pci=nomsi in fedora bugzilla.. > https://bugzilla.redhat.com/show_bug.cgi?id=438621 > Where the SATA controller wouldn't boot without that option. Thanks, assuming there aren't other SATA issues in that bug it sounds like we need to blacklist that chipset. > It was nowhere near as bad as the MMCONFIG option, which seemed to be > utterly busted everywhere, but it was still something that came up > often enough at one point, that we flipped the default state. I still have to sneak in a fix for the real root cause of that problem (we're just papering over it atm), but yeah it was pretty annoying. Thanks, -- Jesse Barnes, Intel Open Source Technology Center -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
