Re: [PATCH] irq, x86: Remove IRQ_DISABLED check in process context IRQ move

[ebiederm@xmission.com (Eric W. Biederman)]

"Pallipadi, Venkatesh" <venkatesh.pallipadi@intel.com> writes:

> As discussed in the thread here
> http://marc.info/?l=linux-kernel&m=123964468521142&w=2
>
> On Fri, Apr 10, 2009 at 3:02 PM, Eric W. Biederman
> <ebiederm@xmission.com> wrote:
>>
>> It looks like some additional bugs have slipped in since last I looked.
>>
>> set_irq_affinity does this:
>> ifdef CONFIG_GENERIC_PENDING_IRQ
>>        if (desc->status & IRQ_MOVE_PCNTXT || desc->status & IRQ_DISABLED) {
>>                cpumask_copy(desc->affinity, cpumask);
>>                desc->chip->set_affinity(irq, cpumask);
>>        } else {
>>                desc->status |= IRQ_MOVE_PENDING;
>>                cpumask_copy(desc->pending_mask, cpumask);
>>        }
>> #else
>>
>> That IRQ_DISABLED case is a software state and as such it has nothing to
>> do with how safe it is to move an irq in process context.
>>
>
> "Pallipadi, Venkatesh" <venkatesh.pallipadi@intel.com> writes:
>> On Sat, 2009-04-11 at 04:01 -0700, Eric W. Biederman wrote:
>> >
>> > If the goal is moving MSIs, we should modify the msi code to be safe
>> > in process context and to set IRQ_MOVE_PCNTXT.
>> >
>> > The only reason we migrate MSIs in interrupt context today is that there
>> > wasn't infrastructure for support migration both in interrupt context
>> > and outside of it.
>>
>> Yes. The idea here was to force the MSI migration to happen in process
>> context. One of the patches in the series did
>>
>>         disable_irq(dev->irq);
>>         irq_set_affinity(dev->irq, cpumask_of(dev->cpu));
>>         enable_irq(dev->irq);
>>
>> with the above patch adding irq/manage code check for interrupt disabled
>> and moving the interrupt in process context.
>>
>> IIRC, there was no IRQ_MOVE_PCNTXT when we were developing this HPET
>> code and we ended up having this ugly hack. IRQ_MOVE_PCNTXT was there
>> when we eventually submitted the patch upstream. But, looks like I did a
>> blind rebasing instead of using IRQ_MOVE_PCNTXT in hpet MSI code. That
>> was my fault. Will send a patch to fix this ugliness.
>
> Below patch fixes this. i.e., revert
> commit 932775a4ab622e3c99bd59f14cc7d96722f79501
> and add PCNTXT to HPET MSI setup. Also removes copying of desc->affinity
> in generic code as set_affinity routines are doing it internally.

Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>

This looks good.

Do you think you could take this one step farther, place a read
after the hpet_msi_write to flush the write to the interrupt source, 
and then finish up the work to change the irq reception setup?

Roughly like ir_set_msi_irq_affinity?

That way we really do get everything done in process context.

> Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
> ---
>  arch/x86/kernel/apic/io_apic.c |    2 ++
>  kernel/irq/manage.c            |    5 ++---
>  2 files changed, 4 insertions(+), 3 deletions(-)
>
> diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
> index 767fe7e..aaf8212 100644
> --- a/arch/x86/kernel/apic/io_apic.c
> +++ b/arch/x86/kernel/apic/io_apic.c
> @@ -3667,12 +3667,14 @@ int arch_setup_hpet_msi(unsigned int irq)
>  {
>  	int ret;
>  	struct msi_msg msg;
> +	struct irq_desc *desc = irq_to_desc(irq);
>  
>  	ret = msi_compose_msg(NULL, irq, &msg);
>  	if (ret < 0)
>  		return ret;
>  
>  	hpet_msi_write(irq, &msg);
> +	desc->status |= IRQ_MOVE_PCNTXT;
>  	set_irq_chip_and_handler_name(irq, &hpet_msi_type, handle_edge_irq,
>  		"edge");
>  
> diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c
> index 7e2e7dd..2734eca 100644
> --- a/kernel/irq/manage.c
> +++ b/kernel/irq/manage.c
> @@ -109,10 +109,9 @@ int irq_set_affinity(unsigned int irq, const struct cpumask *cpumask)
>  	spin_lock_irqsave(&desc->lock, flags);
>  
>  #ifdef CONFIG_GENERIC_PENDING_IRQ
> -	if (desc->status & IRQ_MOVE_PCNTXT || desc->status & IRQ_DISABLED) {
> -		cpumask_copy(desc->affinity, cpumask);
> +	if (desc->status & IRQ_MOVE_PCNTXT)
>  		desc->chip->set_affinity(irq, cpumask);
> -	} else {
> +	else {
>  		desc->status |= IRQ_MOVE_PENDING;
>  		cpumask_copy(desc->pending_mask, cpumask);
>  	}
> -- 
> 1.6.0.6