[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <56AB29C4.8060304@rempel-privat.de>
Date: Fri, 29 Jan 2016 09:58:44 +0100
From: Oleksij Rempel <linux@...pel-privat.de>
To: Thomas Gleixner <tglx@...utronix.de>
Cc: kernel@...gutronix.de, linux-kernel@...r.kernel.org,
jason@...edaemon.net, marc.zyngier@....com
Subject: Re: [PATCH] irqchip: mxs: add missing set_handle_irq()
Am 29.01.2016 um 09:54 schrieb Thomas Gleixner:
> On Fri, 29 Jan 2016, Oleksij Rempel wrote:
>> Am 29.01.2016 um 09:33 schrieb Thomas Gleixner:
>>> On Fri, 29 Jan 2016, Oleksij Rempel wrote:
>>>
>>>> this will fix IRQ related issues on Alphascale ASM9260
>>>
>>> This really does not qualify as a changelog.
>>>
>>> 1) What's the problem and what consequences does it have
>>>
>>> 2) What's the fix.
>>>
>>> Thanks,
>>>
>>> tglx
>>>
>>
>> The driver provide this function:
>> asmlinkage void __exception_irq_entry icoll_handle_irq(struct pt_regs *regs)
>> {
>> u32 irqnr;
>>
>> irqnr = __raw_readl(icoll_priv.stat);
>> __raw_writel(irqnr, icoll_priv.vector);
>> handle_domain_irq(icoll_domain, irqnr, regs);
>> }
>>
>> i assume at some late review point instead of move it to asm9260_of_init
>> i removed it completely. As result, the driver was compilable but not
>> working.
>> Should i put this explanation to changelog?
>
> So the problem is:
>
> The rework of the driver missed to move the call to set_handle_irq() into
> asm9260_of_init(). As a consequence no interrupt entry point is installed and
> no interrupts are delivered (or whatever the outcome is, crash, explode ...)
>
> Solution is:
>
> Install the interrupt entry handler.
>
> Right?
Right.
Should i resend a patch with new comment?
> And that wants a tag:
>
> Fixes: nnnnnnnnnnnn ("irqchip/......")
>
> So we can mark it for stable.
>
> Thanks,
>
> tglx
>
--
Regards,
Oleksij
Download attachment "signature.asc" of type "application/pgp-signature" (214 bytes)
Powered by blists - more mailing lists