Re: [Scst-devel] WRITE_SAME16 issue

[Vladislav B. <vs...@vl...>]

Hi Bart,

Bart Van Assche, on 06/18/2012 09:00 AM wrote:
> On 06/18/12 06:23, Sietse van Zanen wrote:
>>> -----Original Message-----
>
>>> From: Bart Van Assche [mailto:bva...@ac...]
>>> Sent: Sunday, June 17, 2012 18:43
>>> To: Sietse van Zanen
>>> Cc: scs...@li...
>>> Subject: Re: [Scst-devel] WRITE_SAME16 issue
>>>
>>> On 06/16/12 06:41, Sietse van Zanen wrote:
>>>
>>>> Jun 11 09:03:09 rdsan01 kernel: [52557.668695] [4377]: scst:
>>>> scst_parse_cmd:780:Warning: expected transfer length 512 for opcode 0x93 (handler vdisk_fileio, target iscsi) doesn't match decoded value 1048576 Jun 11 09:03:09 rdsan01 kernel: [52557.668704] [4377]: scst_parse_cmd:782:Suspicious CDB:
>>>> Jun 11 09:03:09 rdsan01 kernel: [52557.668707]  (h)___0__1__2__3__4__5__6__7__8__9__A__B__C__D__E__F
>>>> Jun 11 09:03:09 rdsan01 kernel: [52557.668713]    0: 93 00 00 00 00 00 5d cd 18 00 00 00 08 00 00 00   ......].........
>>>>
>>>> Is this a known issue? I am using 2x ESXi5.0u1 ->  iscsi_scst ->
>>>> vdisk_fileio ->  zfs (on linux) ->  dm-crypt ->  sata
>>>
>>>
>>> The expected transfer length value looks fine but the "decoded value" not. With which SCST version did this occur ?
>
>>
>> 2.2.0 running on debian squeeze 6.0.5 with a 3.2.18 kernel (patched).
>>
>> System is an intel p35 chipset (ich9) with a core2 quad q6600 cpu an 8GB mem.
>
> As far as I can see the warning is innocent and can be ignored safely. Anyway, this
> patch should make the warning go away:
>
> [PATCH] scst 2.2: Fix WRITE SAME(16) parsing
>
> Signed-off-by: Bart Van Assche<bva...@ac...>
> ---
>   scst/src/scst_lib.c |    2 +-
>   1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/scst/src/scst_lib.c b/scst/src/scst_lib.c
> index 9e23029..4a10df6 100644
> --- a/scst/src/scst_lib.c
> +++ b/scst/src/scst_lib.c
> @@ -558,7 +558,7 @@ static const struct scst_sdbops scst_scsi_op_table[] = {
>   	 0, get_trans_len_none},
>   	{0x93, "O    O          ", "WRITE SAME(16)",
>   	 SCST_DATA_WRITE, SCST_TRANSFER_LEN_TYPE_FIXED|SCST_WRITE_MEDIUM,
> -	 10, get_trans_len_4},
> +	 10, get_trans_len_single},
>   	{0x93, " M              ", "ERASE(16)",
>   	 SCST_DATA_NONE, SCST_LONG_TIMEOUT|SCST_WRITE_MEDIUM,
>   	 0, get_trans_len_none},

The patch is OK for me. You can commit it, if you'd like.

Thanks,
Vlad