Re: Debian 7.1 on QNAP TS-412 mdadm always reassembles without drive in 4th bay
Am 03.07.2013 14:48, schrieb Arnaud Patard (Rtp):
>> Thanks for your analysis! Can you please file a bug report on the
>> linux package in the Debian BTS, so this can be forwarded upstream to
>> look for a good solution.
>
> I don't want to look to pessimistic but adding 5s delays has few chances
> to be accepted upstream. It's a very huge delay and in this patch, it's
> added for all chipsets supported by sata_mv. Moreover, there's no clue if
> it's an issue with the controller, the ts-412 hw or whatever.
True. Especially since running eg. report-hw on both my TS-410 and my
TS-412 gives the *exact* same result as far as hardware IDs and revision
numbers are concerned - including those of the Kirkwood SoC (88F6281 rev
03) and used SATA chip (88SX7042 rev 02).
The only difference seems to be the clock speed of the core (800Mhz for
the 410 vs. 1200MHz for the 412).
That, however, is weird in itself, because the part number of the SoC
88F6281 is only listed at www.marvell.com as having clock speeds of >=
1GHz (http://www.marvell.com/embedded-processors/kirkwood)
> I fear that upstream will ask us to justify in depth this and we
> won't be able to do so. We can try to send it anyway and try to deal
> with upstream's comments but don't hope too much (I would even guess
> that the patch in its current form as no chances to be accepted, sorry).
How about checking the kirkwood subtype selected in the kernel config
and only apply the delay if the TS-41X option was set? It would not be
much of an improvement, but it would be better than the version I posted.
I guess if there were more people with this problem, one could actually
try to narrow the issue down better? Hint: In case you read this, own a
QNAP TS-412 and faced the same trouble installing debian, NOW would be
the time to speak up ;-)
Martin
Reply to: