[stgt] bs_rdwr_request

[ronnie sahlberg]

On Mon, Aug 27, 2012 at 10:16 PM, FUJITA Tomonori
<fujita.tomonori at lab.ntt.co.jp> wrote:
> On Mon, 27 Aug 2012 22:10:22 -0700
> ronnie sahlberg <ronniesahlberg at gmail.com> wrote:
>
>> On Sun, Aug 26, 2012 at 11:51 PM, FUJITA Tomonori
>> <fujita.tomonori at lab.ntt.co.jp> wrote:
>>> On Mon, 27 Aug 2012 09:42:02 +0300
>>> Roi Dayan <roi.dayan at gmail.com> wrote:
>>>
>>>> On Mon, Aug 27, 2012 at 3:39 AM, FUJITA Tomonori
>>>> <fujita.tomonori at lab.ntt.co.jp> wrote:
>>>> > On Sun, 26 Aug 2012 15:58:02 +0300
>>>> > Roi Dayan <roi.dayan at gmail.com> wrote:
>>>> >
>>>> >> On Sun, Aug 26, 2012 at 3:32 PM, Roi Dayan <roi.dayan at gmail.com> wrote:
>>>> >> > On Fri, Aug 24, 2012 at 1:06 PM, <frederik.vos at linvirt.nl> wrote:
>>>> >> >>
>>>> >> >> I tested also version 1.0.28: no problem
>>>> >> >> version 1.0.29: no problem
>>>> >> >> version 1.0.30: there is the problem
>>>> >> >>
>>>> >> >>
>>>> >> >
>>>> >> >
>>>> >> >
>>>> >> > Hi,
>>>> >> >
>>>> >> > I tested as well with ESXi 5.1 and I could not login with ESXi as initiator
>>>> >> > The error from tgtd was as follows:
>>>> >> >
>>>> >> > tgtd: add_mode_page(1648) Mode Page 10 (0x01): param_count 6 != MODE
>>>> >> > PAGE size : 31
>>>> >> >
>>>> >> >
>>>> >> > I checked for commits between 1.0.29 - 1.0.30 that modify or call this
>>>> >> > function and found
>>>> >> > that commit 9a95b4431ccc01b82cb4febc735485cd06cd5ea4 added a new call
>>>> >> > to add_mode_page()
>>>> >> > and the error is the result of that new call:
>>>> >> >
>>>> >> > 518     +  /* Control Extensions mode page:  TCMOS:1 */
>>>> >> > 519     +  add_mode_page(lu, "0x0a:1:0x1c:0x04:0x00:0x00");
>>>> >> >
>>>> >> >
>>>> >> > After removing this call ESXi logged in fine and without any problems.
>>>> >> >
>>>> >> > frederik,
>>>> >> > mind trying it as well?
>>>> >
>>>> > I guess that it's due to compare-and-write command but I might be
>>>> > wrong.
>>>> >
>>>>
>>>>
>>>> Hi,
>>>>
>>>> The failure is because add_mode_page expect size of data according to
>>>> the size specified in the page string
>>>> i.e. for the call: add_mode_page(lu, "0x0a:1:0x1c:0x04:0x00:0x00");
>>>> the size is 0x1c (28) and the data count is 3 (0x04 0x00:0x00) so
>>>> add_mode_page fails.
>>>>
>>>> The patch I sent accepts it and it means rest of the bytes in the page
>>>> data will be 0 because of how the page is allocated.
>>>
>>> Ah, I see. Thanks for the explanation.
>>>
>>>
>>>> Do you still want a patch to update the callers instead?
>>>
>>> No. Your patch sounds ok. Ronnie, any opinion?
>>>
>>> But I don't think that your patch fixes Frederik's problem.
>>
>> Ouch,
>>
>> That is definitely a bug for the 0x0a:0x01 modepage as Roi points out.
>> I have been travelling and my mailbox is a bit 'unstructured' right
>> now so I can not see Roi's patch
>> but as SPC states this modepage must be 0x1c in size, (spc4 7.5.8)
>> we should pad the mode page from the current 6 bytes to the proper 31
>> bytes byt adding zeroes to it.
>
> Yeah, but COMPARE_AND_WRITE bug is not related with this?

No it is not I think.

Can we get a network trace of the failure?
Then I can try to fix it over the next few days

regards
ronnie sahlberg
--
To unsubscribe from this list: send the line "unsubscribe stgt" in
the body of a message to majordomo at vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html