Re: x86_86 SMP megaraid_mbox hangups and panics.

[dormando <dormando@rydia.net>]

Ju, Seokmann wrote:
> Hi,
> > Most of the time the server hits: "megaraid: probe new device" - with 
> > the device information, then hangs and starts the 180 second 
> > countdown: 
> > "megaraid: wait for FW to boot [blah]"
> > After which I get a VFS panic for not having a root disk.
> This means the controller is NOT taking any commands from the driver at that time.
> In other words, the F/W is NOT ready to take any command, yet.
> It sounds like that the controller is NOT in good condition for some reason and needs to check sanity of it.
> You may want to check with LSI logic SE team.
>
> Thank you,

Can you confirm that you read my whole message? I might be a complete 
idiot, but let me reiterate some highlights and add a few more details 
from my last mail:

If I build a 32-bit SMP OR 64-bit UP kernel, the cards will boot and 
work *every time*.

Most of the time this is a *kernel panic* in megaraid_ack_sequence, 
somewhere through megaraid_isr in megaraid_mbox.c
*Sometimes* it results in the firmware hanging like you mentioned above.
If I compile the drivers as modules instead of statically, this 
*sometimes* results in the machine booting all the way. Once the machine 
boots up, I can give it a good 'ole I/O beatdown and it does not flinch.

As in: I boot it once, it hangs. I reboot, it panics. I reboot, it 
panics. I reboot, it works. I have 16 cards in 16 systems that all 
exhibit the same behavior. Looks like an x86_64 SMP timing issue of some 
kind to me, and less of a card issue. Please correct me if I am wrong!

-Dormando
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/