On Tue, 03/06 12:44, Daniel Henrique Barboza wrote:
> The calculation of the max_transfer atribute of BlockDriverState
> makes considerations such as max_segments and transfer_length via
> the BLKSECTGET ioctl (if available).
>
> However, bl->max_transfer isn't considered when emulating the INQUIRY
> 'Block Limit' response to the scsi-hd devices. This leads to situations
> where the declared max_sectors from the INQUIRY response is inconsistent
> with the block limits, which isn't ideal. It can also be misleading to the
> user that sets /sys/block/<dev>/queue/max_sectors_kb to a certain
> value, then finds a different value in the guest OS for the same disk.
>
> Following the same logic scsi_read_complete from scsi-generic.c does
> when patching the response of the Block Limits VPD back to the guest,
> change the max_io_sectors value of the emulated Block Limits VPD
> response by considering the blk_get_max_transfer of the related
> BlockDriverState. Use MIN_NOT_ZERO to be sure that the minimal
> value is chosen.
>
> Given that we're changing max_io_sectors, consider that min_io_sectors
> and opt_io_sectors can't be greater than the new calculated value.
>
> Signed-off-by: Daniel Henrique Barboza <danie...@linux.vnet.ibm.com>
> ---
> hw/scsi/scsi-disk.c | 15 +++++++++++++++
> 1 file changed, 15 insertions(+)
>
> diff --git a/hw/scsi/scsi-disk.c b/hw/scsi/scsi-disk.c
> index 49d2559d93..c65c1ce56d 100644
> --- a/hw/scsi/scsi-disk.c
> +++ b/hw/scsi/scsi-disk.c
> @@ -704,6 +704,21 @@ static int scsi_disk_emulate_inquiry(SCSIRequest *req,
> uint8_t *outbuf)
> page_code);
> return -1;
> }
> + if (s->qdev.type == TYPE_DISK) {
> + int max_transfer_blk =
> blk_get_max_transfer(s->qdev.conf.blk);
> + int max_io_sectors_blk =
> + max_transfer_blk / s->qdev.blocksize;
> +
> + max_io_sectors =
> + MIN_NON_ZERO(max_io_sectors_blk, max_io_sectors);
> +
> + /* min_io_size and opt_io_size can't be greater than
> + * max_io_sectors */
> + min_io_size =
> + MIN_NON_ZERO(min_io_size, max_io_sectors);
> + opt_io_size =
> + MIN_NON_ZERO(opt_io_size, max_io_sectors);
> + }
> /* required VPD size with unmap support */
> buflen = 0x40;
> memset(outbuf + 4, 0, buflen - 4);
> --
> 2.14.3
>
Reviewed-by: Fam Zheng <f...@redhat.com>
