Messages in this thread | | | Date | Thu, 15 Feb 2007 15:42:59 -0800 | From | Zachary Amsden <> | Subject | Re: [PATCH 9/11] Panic delay fix |
| |
Pavel Machek wrote: > > You know it is ugly. Alan demonstrated it even hurts performance, but > being ugly is the main problem. >
No argument with that. Well, we're ok with dropping it. Actually, reverting the entire set of udelay changes now seems wise. The same bug that happened with i8042 can happen with any hardware device driven by the VM in passthrough mode - potentially USB or IDE CDROM or direct SCSI.
Since that is per-device and not global, a global udelay disable really is broken in that case, and recompiling individual C files or modules for passthrough vs. non-passthrough is not the answer.
So Rusty, Chris, Jeremy, any objections to killing udelay() and friends in paravirt-ops? It would simplify things a bit. The only thing we lose is a slightly faster boot time in the 100% emulated device case. I'm ok with losing that. Even the PIT fast paths don't use udelay, so I don't think we care for runtime performance at all.
Zach - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |