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.le...@csgroup.eu/
?
I never got any feedback.
>
> Signed-off-by: Michael Ellerman <m...@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);
