[Qemu-devel] [PATCH] scsi-disk: provide "max write same length" for block limits VPD page

[Qemu-devel] [PATCH] scsi-disk: provide "max write same length" for block limits VPD page

[Ming Lei]

Since QEMU claims to support UNMAP, WRITE SAME and WRITE SAME 16
in the LBP VPD page, it is better to provide "max write same length"
in response for block limits VPD page because:

	- T10 SBC-3 doesn't describe priority explicitly when all three
	are enabled
	- host driver may prefer WRITE/WRITE 16, then try to parse
	"max write same length"

Signed-off-by: Ming Lei <ming.lei@canonical.com>
---
 hw/scsi/scsi-disk.c |   14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/hw/scsi/scsi-disk.c b/hw/scsi/scsi-disk.c
index 2f75d7d..b15bf4f 100644
--- a/hw/scsi/scsi-disk.c
+++ b/hw/scsi/scsi-disk.c
@@ -674,6 +674,20 @@ static int scsi_disk_emulate_inquiry(SCSIRequest *req, uint8_t *outbuf)
             outbuf[29] = (unmap_sectors >> 16) & 0xff;
             outbuf[30] = (unmap_sectors >> 8) & 0xff;
             outbuf[31] = unmap_sectors & 0xff;
+
+            /*
+             * maximum write same length, just borrow max unmap
+             * count because write same command need to support
+             * unmap
+             */
+            outbuf[36] = 0;
+            outbuf[37] = 0;
+            outbuf[38] = 0;
+            outbuf[39] = 0;
+            outbuf[40] = (max_unmap_sectors >> 24) & 0xff;
+            outbuf[41] = (max_unmap_sectors >> 16) & 0xff;
+            outbuf[42] = (max_unmap_sectors >> 8) & 0xff;
+            outbuf[43] = max_unmap_sectors & 0xff;
             break;
         }
         case 0xb2: /* thin provisioning */
-- 
1.7.9.5

Re: [Qemu-devel] [PATCH] scsi-disk: provide "max write same length" for block limits VPD page

[Paolo Bonzini]

On 05/12/2014 14:00, Ming Lei wrote:
> Since QEMU claims to support UNMAP, WRITE SAME and WRITE SAME 16
> in the LBP VPD page, it is better to provide "max write same length"
> in response for block limits VPD page because:
> 
> 	- T10 SBC-3 doesn't describe priority explicitly when all three
> 	are enabled
> 	- host driver may prefer WRITE/WRITE 16, then try to parse
> 	"max write same length"
> 
> Signed-off-by: Ming Lei <ming.lei@canonical.com>
> ---
>  hw/scsi/scsi-disk.c |   14 ++++++++++++++
>  1 file changed, 14 insertions(+)
> 
> diff --git a/hw/scsi/scsi-disk.c b/hw/scsi/scsi-disk.c
> index 2f75d7d..b15bf4f 100644
> --- a/hw/scsi/scsi-disk.c
> +++ b/hw/scsi/scsi-disk.c
> @@ -674,6 +674,20 @@ static int scsi_disk_emulate_inquiry(SCSIRequest *req, uint8_t *outbuf)
>              outbuf[29] = (unmap_sectors >> 16) & 0xff;
>              outbuf[30] = (unmap_sectors >> 8) & 0xff;
>              outbuf[31] = unmap_sectors & 0xff;
> +
> +            /*
> +             * maximum write same length, just borrow max unmap
> +             * count because write same command need to support
> +             * unmap
> +             */
> +            outbuf[36] = 0;
> +            outbuf[37] = 0;
> +            outbuf[38] = 0;
> +            outbuf[39] = 0;
> +            outbuf[40] = (max_unmap_sectors >> 24) & 0xff;
> +            outbuf[41] = (max_unmap_sectors >> 16) & 0xff;
> +            outbuf[42] = (max_unmap_sectors >> 8) & 0xff;
> +            outbuf[43] = max_unmap_sectors & 0xff;
>              break;
>          }
>          case 0xb2: /* thin provisioning */
> 

There is support for splitting write_zeroes requests in the block layer,
so this is a problem elsewhere in the stack.  It could be in QEMU block/
or in Linux.  Let's analyze it in the linux-scsi thread.

Paolo