Re: [Qemu-devel] [PULL] xen patches
On Tue, Apr 24, 2012 at 12:28:35PM +0100, Stefano Stabellini wrote: xen_disk: use bdrv_aio_flush instead of bdrv_flush This one seems completely broken, as it just queues up the flushes and writes without any ordering between them. Linux filesystems absolutely rely on a REQ_FUA request which gets mapped to BLKIF_OP_FLUSH_DISKCACHE to flush the data it actually sent. I'm not sure who actually expects the preflush, but the only point in ever doing it would be to sequence it properly vs the write.
Re: [Qemu-devel] [PULL] xen patches
Anthony, please ignore this pull request, I'll rework and resubmit later. On Wed, 25 Apr 2012, Christoph Hellwig wrote: On Tue, Apr 24, 2012 at 12:28:35PM +0100, Stefano Stabellini wrote: xen_disk: use bdrv_aio_flush instead of bdrv_flush This one seems completely broken, as it just queues up the flushes and writes without any ordering between them. Linux filesystems absolutely rely on a REQ_FUA request which gets mapped to BLKIF_OP_FLUSH_DISKCACHE to flush the data it actually sent. I'm not sure who actually expects the preflush, but the only point in ever doing it would be to sequence it properly vs the write. I see.. I need to make sure that the bdrv_aio_flush is fully completed before issuing any reads or writes, not just ordered correctly: calling bdrv_aio_flush before bdrv_aio_readv/writev is not enough.
[Qemu-devel] [PULL] xen patches
Hi Anthony, please pull: git://xenbits.xen.org/people/sstabellini/qemu-dm.git for_anthony it contains a couple of xen_disk improvements and Xen HVM reboot support. John V. Baboval (1): xen: Support guest reboots Stefano Stabellini (2): xen_disk: use bdrv_aio_flush instead of bdrv_flush xen_disk: implement BLKIF_OP_FLUSH_DISKCACHE, remove BLKIF_OP_WRITE_BARRIER hw/xen_common.h |2 +- hw/xen_disk.c | 21 ++--- xen-all.c | 18 +++--- 3 files changed, 22 insertions(+), 19 deletions(-) Thanks, Stefano