On Wed, Nov 30, 2005 at 02:29:10PM -0500, Salyzyn, Mark wrote: > Christoph Hellwig [mailto:hch@xxxxxxxxxxxxx] writes: > > NACK on the write filter. If we can agree on it beeing > > useful (and I must > > admit I tend to the contrary opinion) it should be done in > > higher layers. > > Our customers won't stand still if we do not write (and read) protect > the raid components. So you propose the introduction of an > sdev->io_protect flag? write a patch to do it and argue for it. in the driver it's in the wrong place for certain and not acceptable by any means. > > Also why do we need the !sdev->no_uld_attach && !sdev->channel checks? > > This is additional insurance, the fact that channel 0 is the array > channel. We do not want the no_uld_attach to be part of the array > channel. But again the invariant is that no_uld_attach is never going to be set for channel 0. Just checking for channel 0 is fine with me, although a #define AAC_RAID_CHANNEL 0 would be even better to explain what's going on. Similar issue is the tagged_supported check, it's totally magic to me and at least needs a very good comment. - : send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
