Re: [PATCH] loongarch: irq: Move to generic_handle_arch_irq
From: Huacai Chen
Date: Sat Aug 13 2022 - 06:10:32 EST
Hi, Ren,
Thank you for your patch, but it seems wrong to me.
Please see c9b12b59e2ea4c3c7cedec7efb071b6 ("s390/entry: fix duplicate
tracking of irq nesting level"). ct_irq_enter() is named
rcu_irq_enter() in older kernels. And irq_enter = rcu_irq_enter +
irq_enter_rcu (for the latest kernel, irq_enter = ct_irq_enter +
irq_enter_rcu). If we use generic entry, irqentry_enter() has already
call rcu_irq_enter (or ct_irq_enter in the latest kernel), so the arch
code doesn't need to call it again. This has nothing to do with
HAVE_CONTEXT_TRACKING_USER, just avoid duplicated calls to
ct_irq_enter.
Huacai
On Sat, Aug 13, 2022 at 5:06 PM Guo Ren <guoren@xxxxxxxxxx> wrote:
>
> Abandon this patch because of the wrong commit log. Please see
> "[RESEND PATCH] loongarch: irq: Move to generic_handle_arch_irq".
>
> On Sat, Aug 13, 2022 at 4:58 PM <guoren@xxxxxxxxxx> wrote:
> >
> > From: Guo Ren <guoren@xxxxxxxxxxxxxxxxx>
> >
> > No reason to keep custom handle_loongarch_irq, and move to the generic
> > one. No reason to keep custom handle_loongarch_irq, and move to the
> > generic one. The patch also the fixup HAVE_CONTEXT_TRACKING_USER
> > feature, because handle_loongarch_irq missed ct_irq_enter/exit.
> >
> > Fixes: 0603839b18f4 ("LoongArch: Add exception/interrupt handling")
> > Signed-off-by: Guo Ren <guoren@xxxxxxxxxxxxxxxxx>
> > Signed-off-by: Guo Ren <guoren@xxxxxxxxxx>
> > ---
> > arch/loongarch/kernel/traps.c | 15 ++-------------
> > 1 file changed, 2 insertions(+), 13 deletions(-)
> >
> > diff --git a/arch/loongarch/kernel/traps.c b/arch/loongarch/kernel/traps.c
> > index aa1c95aaf595..06211640ba1f 100644
> > --- a/arch/loongarch/kernel/traps.c
> > +++ b/arch/loongarch/kernel/traps.c
> > @@ -588,17 +588,6 @@ asmlinkage void cache_parity_error(void)
> > panic("Can't handle the cache error!");
> > }
> >
> > -asmlinkage void noinstr handle_loongarch_irq(struct pt_regs *regs)
> > -{
> > - struct pt_regs *old_regs;
> > -
> > - irq_enter_rcu();
> > - old_regs = set_irq_regs(regs);
> > - handle_arch_irq(regs);
> > - set_irq_regs(old_regs);
> > - irq_exit_rcu();
> > -}
> > -
> > asmlinkage void noinstr do_vint(struct pt_regs *regs, unsigned long sp)
> > {
> > register int cpu;
> > @@ -608,7 +597,7 @@ asmlinkage void noinstr do_vint(struct pt_regs *regs, unsigned long sp)
> > cpu = smp_processor_id();
> >
> > if (on_irq_stack(cpu, sp))
> > - handle_loongarch_irq(regs);
> > + generic_handle_arch_irq(regs);
> > else {
> > stack = per_cpu(irq_stack, cpu) + IRQ_STACK_START;
> >
> > @@ -619,7 +608,7 @@ asmlinkage void noinstr do_vint(struct pt_regs *regs, unsigned long sp)
> > "move $s0, $sp \n" /* Preserve sp */
> > "move $sp, %[stk] \n" /* Switch stack */
> > "move $a0, %[regs] \n"
> > - "bl handle_loongarch_irq \n"
> > + "bl generic_handle_arch_irq \n"
> > "move $sp, $s0 \n" /* Restore sp */
> > : /* No outputs */
> > : [stk] "r" (stack), [regs] "r" (regs)
> > --
> > 2.36.1
> >
>
>
> --
> Best Regards
> Guo Ren
>