commit a71b092 "ARM: Convert handle_IRQ to use __handle_domain_irq" breaks DNS-323

[Marc Zyngier]

On 14/08/15 16:43, Detlef Vollmann wrote:
> On 08/14/15 17:16, Marc Zyngier wrote:
>> On 14/08/15 15:13, Gregory CLEMENT wrote:
>>> Hi Detlef,
>>>
>>> On 14/08/2015 16:05, Detlef Vollmann wrote:
>>>> Hello,
>>>>
>>>> this commit:
>>>>   > commit a71b092a9c68685a270ebdde7b5986ba8787e575
>>>>   > Author: Marc Zyngier <marc.zyngier at arm.com>
>>>>   > Date:   Tue Aug 26 11:03:18 2014 +0100
>>>>   >
>>>>   >     ARM: Convert handle_IRQ to use __handle_domain_irq
>>>> [...]
>>>> causes the kernel not to boot anymore on my DNS-323.
>>>> Soc: MV88F5182 Rev 2
>>>> CPU: ARM926 (Rev 0) running @ 500Mhz
>>>> CPU: Feroceon [41069260] revision 0 (ARMv5TEJ), cr=a005317f
>>>> Orion ID: MV88F5182-A2. TCLK=166666667.
>>>> DNS-323: Identified HW revision B1
>>>>
>>>> Boot stops after "Uncompressing Linux... done, booting the kernel."
>>>> Reverting the commit works fine for me.
>>>> Tested 4.1.2 and 4.2.0-rc6 with orion5x_defconfig.
>>>
>>> Could you try the following patch ?
>>>
>>> http://lists.infradead.org/pipermail/linux-arm-kernel/2015-July/356501.html
>>
>> As a side note, having to revert that patch to get your system up and
>> running is a reliable indication that something is wrong with your
>> interrupt mapping, and that you are using 0 as a valid interrupt number
>> (which has been declared invalid a long time ago by the chief penguin).
> Well, in this case it wasn't me ;-)

Guilty by association! ;-)

> I wanted to avoid looking at the sources myself as the DNS-323 is still
> a widely used hardware.

You're missing out! Think of all these sleepless nights you could spend
staring at all kind of weird and fascinating code! :D

>> Hopefully Gregory has put you in the right direction.
> Yes.  Now I'll use that patch until some fix is accepted in mainline.

Hopefully soon.

Thanks,

	M.
-- 
Jazz is not dead. It just smells funny...