[PATCH] powerpc/irq: Allow softirq to hardirq stack transition

[PATCH] powerpc/irq: Allow softirq to hardirq stack transition

[Michael Ellerman]

Allow a transition from the softirq stack to the hardirq stack when
handling a hardirq. Doing so means a hardirq received while deep in
softirq processing is less likely to cause a stack overflow of the
softirq stack.

Previously it wasn't safe to do so because irq_exit() (which initiates
softirq processing) was called on the hardirq stack.

That was changed in commit 1b1b6a6f4cc0 ("powerpc: handle irq_enter/
irq_exit in interrupt handler wrappers") and 1346d00e1bdf ("powerpc:
Don't select HAVE_IRQ_EXIT_ON_IRQ_STACK").

The allowed transitions are now:
 - process stack -> hardirq stack
 - process stack -> softirq stack
 - process stack -> softirq stack -> hardirq stack

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
---
 arch/powerpc/kernel/irq.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/arch/powerpc/kernel/irq.c b/arch/powerpc/kernel/irq.c
index 6f7d4edaa0bc..7504ceec5c58 100644
--- a/arch/powerpc/kernel/irq.c
+++ b/arch/powerpc/kernel/irq.c
@@ -284,15 +284,14 @@ static __always_inline void call_do_irq(struct pt_regs *regs, void *sp)
 void __do_IRQ(struct pt_regs *regs)
 {
 	struct pt_regs *old_regs = set_irq_regs(regs);
-	void *cursp, *irqsp, *sirqsp;
+	void *cursp, *irqsp;
 
 	/* Switch to the irq stack to handle this */
 	cursp = (void *)(current_stack_pointer & ~(THREAD_SIZE - 1));
 	irqsp = hardirq_ctx[raw_smp_processor_id()];
-	sirqsp = softirq_ctx[raw_smp_processor_id()];
 
 	/* Already there ? If not switch stack and call */
-	if (unlikely(cursp == irqsp || cursp == sirqsp))
+	if (unlikely(cursp == irqsp))
 		__do_irq(regs, current_stack_pointer);
 	else
 		call_do_irq(regs, irqsp);
-- 
2.41.0

Re: [PATCH] powerpc/irq: Allow softirq to hardirq stack transition

[Christophe Leroy]

Le 30/11/2023 à 13:50, Michael Ellerman a écrit :
> Allow a transition from the softirq stack to the hardirq stack when
> handling a hardirq. Doing so means a hardirq received while deep in
> softirq processing is less likely to cause a stack overflow of the
> softirq stack.
> 
> Previously it wasn't safe to do so because irq_exit() (which initiates
> softirq processing) was called on the hardirq stack.
> 
> That was changed in commit 1b1b6a6f4cc0 ("powerpc: handle irq_enter/
> irq_exit in interrupt handler wrappers") and 1346d00e1bdf ("powerpc:
> Don't select HAVE_IRQ_EXIT_ON_IRQ_STACK").
> 
> The allowed transitions are now:
>   - process stack -> hardirq stack
>   - process stack -> softirq stack
>   - process stack -> softirq stack -> hardirq stack

It means you don't like my patch 
https://patchwork.ozlabs.org/project/linuxppc-dev/patch/6cd9d8bb2258d8b51999c2584eac74423d2b5e29.1657203774.git.christophe.leroy@csgroup.eu/ 
?

I never got any feedback.


> 
> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
> ---
>   arch/powerpc/kernel/irq.c | 5 ++---
>   1 file changed, 2 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/powerpc/kernel/irq.c b/arch/powerpc/kernel/irq.c
> index 6f7d4edaa0bc..7504ceec5c58 100644
> --- a/arch/powerpc/kernel/irq.c
> +++ b/arch/powerpc/kernel/irq.c
> @@ -284,15 +284,14 @@ static __always_inline void call_do_irq(struct pt_regs *regs, void *sp)
>   void __do_IRQ(struct pt_regs *regs)
>   {
>   	struct pt_regs *old_regs = set_irq_regs(regs);
> -	void *cursp, *irqsp, *sirqsp;
> +	void *cursp, *irqsp;
>   
>   	/* Switch to the irq stack to handle this */
>   	cursp = (void *)(current_stack_pointer & ~(THREAD_SIZE - 1));
>   	irqsp = hardirq_ctx[raw_smp_processor_id()];
> -	sirqsp = softirq_ctx[raw_smp_processor_id()];
>   
>   	/* Already there ? If not switch stack and call */
> -	if (unlikely(cursp == irqsp || cursp == sirqsp))
> +	if (unlikely(cursp == irqsp))
>   		__do_irq(regs, current_stack_pointer);
>   	else
>   		call_do_irq(regs, irqsp);

Re: [PATCH] powerpc/irq: Allow softirq to hardirq stack transition

[Michael Ellerman]

Christophe Leroy <christophe.leroy@csgroup.eu> writes:
> Le 30/11/2023 à 13:50, Michael Ellerman a écrit :
>> Allow a transition from the softirq stack to the hardirq stack when
>> handling a hardirq. Doing so means a hardirq received while deep in
>> softirq processing is less likely to cause a stack overflow of the
>> softirq stack.
>> 
>> Previously it wasn't safe to do so because irq_exit() (which initiates
>> softirq processing) was called on the hardirq stack.
>> 
>> That was changed in commit 1b1b6a6f4cc0 ("powerpc: handle irq_enter/
>> irq_exit in interrupt handler wrappers") and 1346d00e1bdf ("powerpc:
>> Don't select HAVE_IRQ_EXIT_ON_IRQ_STACK").
>> 
>> The allowed transitions are now:
>>   - process stack -> hardirq stack
>>   - process stack -> softirq stack
>>   - process stack -> softirq stack -> hardirq stack
>
> It means you don't like my patch 
> https://patchwork.ozlabs.org/project/linuxppc-dev/patch/6cd9d8bb2258d8b51999c2584eac74423d2b5e29.1657203774.git.christophe.leroy@csgroup.eu/ 
> ?

I did like your patch :)

But then we got reports of folks hitting stack overflow in some distro
kernels, and in at least some cases it was a hardirq coming in during
softirq handling and overflowing the softirq stack.

> I never got any feedback.

Sorry, not enough hours in the day.

cheers

Re: [PATCH] powerpc/irq: Allow softirq to hardirq stack transition

[Christophe Leroy]

Le 01/12/2023 à 11:05, Michael Ellerman a écrit :
> Christophe Leroy <christophe.leroy@csgroup.eu> writes:
>> Le 30/11/2023 à 13:50, Michael Ellerman a écrit :
>>> Allow a transition from the softirq stack to the hardirq stack when
>>> handling a hardirq. Doing so means a hardirq received while deep in
>>> softirq processing is less likely to cause a stack overflow of the
>>> softirq stack.
>>>
>>> Previously it wasn't safe to do so because irq_exit() (which initiates
>>> softirq processing) was called on the hardirq stack.
>>>
>>> That was changed in commit 1b1b6a6f4cc0 ("powerpc: handle irq_enter/
>>> irq_exit in interrupt handler wrappers") and 1346d00e1bdf ("powerpc:
>>> Don't select HAVE_IRQ_EXIT_ON_IRQ_STACK").
>>>
>>> The allowed transitions are now:
>>>    - process stack -> hardirq stack
>>>    - process stack -> softirq stack
>>>    - process stack -> softirq stack -> hardirq stack
>>
>> It means you don't like my patch
>> https://patchwork.ozlabs.org/project/linuxppc-dev/patch/6cd9d8bb2258d8b51999c2584eac74423d2b5e29.1657203774.git.christophe.leroy@csgroup.eu/
>> ?
> 
> I did like your patch :)
> 
> But then we got reports of folks hitting stack overflow in some distro
> kernels, and in at least some cases it was a hardirq coming in during
> softirq handling and overflowing the softirq stack.

Fair enough, I'll discard it.

> 
>> I never got any feedback.
> 
> Sorry, not enough hours in the day.
> 

Yes same problem here :)

Re: [PATCH] powerpc/irq: Allow softirq to hardirq stack transition

[Christophe Leroy]

Le 30/11/2023 à 13:50, Michael Ellerman a écrit :
> Allow a transition from the softirq stack to the hardirq stack when
> handling a hardirq. Doing so means a hardirq received while deep in
> softirq processing is less likely to cause a stack overflow of the
> softirq stack.
> 
> Previously it wasn't safe to do so because irq_exit() (which initiates
> softirq processing) was called on the hardirq stack.
> 
> That was changed in commit 1b1b6a6f4cc0 ("powerpc: handle irq_enter/
> irq_exit in interrupt handler wrappers") and 1346d00e1bdf ("powerpc:
> Don't select HAVE_IRQ_EXIT_ON_IRQ_STACK").
> 
> The allowed transitions are now:
>   - process stack -> hardirq stack
>   - process stack -> softirq stack
>   - process stack -> softirq stack -> hardirq stack
> 
> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>


Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu>



> ---
>   arch/powerpc/kernel/irq.c | 5 ++---
>   1 file changed, 2 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/powerpc/kernel/irq.c b/arch/powerpc/kernel/irq.c
> index 6f7d4edaa0bc..7504ceec5c58 100644
> --- a/arch/powerpc/kernel/irq.c
> +++ b/arch/powerpc/kernel/irq.c
> @@ -284,15 +284,14 @@ static __always_inline void call_do_irq(struct pt_regs *regs, void *sp)
>   void __do_IRQ(struct pt_regs *regs)
>   {
>   	struct pt_regs *old_regs = set_irq_regs(regs);
> -	void *cursp, *irqsp, *sirqsp;
> +	void *cursp, *irqsp;
>   
>   	/* Switch to the irq stack to handle this */
>   	cursp = (void *)(current_stack_pointer & ~(THREAD_SIZE - 1));
>   	irqsp = hardirq_ctx[raw_smp_processor_id()];
> -	sirqsp = softirq_ctx[raw_smp_processor_id()];
>   
>   	/* Already there ? If not switch stack and call */
> -	if (unlikely(cursp == irqsp || cursp == sirqsp))
> +	if (unlikely(cursp == irqsp))
>   		__do_irq(regs, current_stack_pointer);
>   	else
>   		call_do_irq(regs, irqsp);

Re: [PATCH] powerpc/irq: Allow softirq to hardirq stack transition

[Michael Ellerman]

On Thu, 30 Nov 2023 23:50:45 +1100, Michael Ellerman wrote:
> Allow a transition from the softirq stack to the hardirq stack when
> handling a hardirq. Doing so means a hardirq received while deep in
> softirq processing is less likely to cause a stack overflow of the
> softirq stack.
> 
> Previously it wasn't safe to do so because irq_exit() (which initiates
> softirq processing) was called on the hardirq stack.
> 
> [...]

Applied to powerpc/next.

[1/1] powerpc/irq: Allow softirq to hardirq stack transition
      https://git.kernel.org/powerpc/c/4eb20bf34ea296f648971a8528e32cd80efcbe89

cheers