On 7/11/06 12:19, "Isaku Yamahata" <[EMAIL PROTECTED]> wrote:
> Hmm.. since you don't believe any race, then please revert > this mb patch except NULL check chunk. > I certainly observed that tapdisk failed to get new request on IA64. > If it is really caused by the race, I will sent a patch again. > (or I'll sent another patch which fixes another issues) I agreed the other barriers were needed, just not that one. If it were required, we'd have the same race in blkback.c. > BTW, > why is wmb() in write_rsp_to_ring() of tools/blktap/drivers/tapdisk.c > necessary? > RING_PUSH_RESPONSES() of kick_responses() issues wmb() so that > wmb() in write_rsp_to_ring() isn't needed, I think. I agree. I'll remove it. -- Keir _______________________________________________ Xen-ia64-devel mailing list [email protected] http://lists.xensource.com/xen-ia64-devel
