Messages in this thread | | | Date | Tue, 28 Feb 2006 14:34:59 -0500 | From | Benjamin LaHaise <> | Subject | Re: [PATCH] Define wc_wmb, a write barrier for PCI write combining |
| |
On Tue, Feb 28, 2006 at 11:20:24AM -0800, Bryan O'Sullivan wrote: > So if we just use wmb(), we incur a 16-cycle penalty on every packet we > send. This has a deleterious and measurable effect on performance.
sfence doesn't guarantee that the write will be flushed, especially if the chipset gets involved. The only way to do that is the same way any pci write can be flushed, which is to read from a register on the device in question.
-ben -- "Ladies and gentlemen, I'm sorry to interrupt, but the police are here and they've asked us to stop the party." Don't Email: <dont@kvack.org>. - 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/
| |