@samitouri / QOSamiQemu / commits / 99bfc1c754

vhost-user-blk-server: fix opt_io_size=1 causing severe Windows I/O degradation

The QSD vhost-user-blk export sets opt_io_size=1 and min_io_size=1 in the virtio config. These values are reported to the guest through the VPD Block Limits page as OptimalTransferLength=1 block (512 bytes) and OptimalTransferLengthGranularity=1 block. Windows respects these hints and splits all I/O into ~512-byte requests, causing ~100x sequential throughput degradation (150 MB/s instead of 15+ GB/s). Linux is unaffected as its block layer ignores these values. Set both to 0 which means "not reported" per the SCSI Block Limits VPD spec, allowing Windows to use its own optimal I/O size defaults. Signed-off-by: Max Makarov <maxpain@linux.com> Message-ID: <20260330193451.76037-1-maxpain@linux.com> Reviewed-by: Kevin Wolf <kwolf@redhat.com> Signed-off-by: Kevin Wolf <kwolf@redhat.com>

Max Makarov committed Mar 30, 2026 at 22:34 UTC 99bfc1c7549894e849c41640f1c4dc4b3a1c963b

1 file changed +2 -2

block/export/vhost-user-blk-server.c

+2 -2

@@ -242,8 +242,8 @@ vu_blk_initialize_config(BlockDriverState *bs,

242 config->blk_size = cpu_to_le32(blk_size);

243 config->size_max = cpu_to_le32(0);

244 config->seg_max = cpu_to_le32(128 - 2);

245 - config->min_io_size = cpu_to_le16(1);

246 - config->opt_io_size = cpu_to_le32(1);

245 + config->min_io_size = cpu_to_le16(0);

246 + config->opt_io_size = cpu_to_le32(0);

247 config->num_queues = cpu_to_le16(num_queues);

248 config->max_discard_sectors =

249 cpu_to_le32(VIRTIO_BLK_MAX_DISCARD_SECTORS);