2020-03-04 00:42:02

by afzal mohammed

Subject: [PATCH v3] xtensa: replace setup_irq() by request_irq()

request_irq() is preferred over setup_irq(). Invocations of setup_irq()
occur after memory allocators are ready.

Per tglx[1], setup_irq() existed in olden days when allocators were not
ready by the time early interrupts were initialized.

Hence replace setup_irq() by request_irq().

[1] https://lkml.kernel.org/r/alpine.DEB.2.20.1710191609480.1971@nanos

Signed-off-by: afzal mohammed <[email protected]>
---
Hi Max Filippov,

i believe you are the maintainer of xtensa, if you are okay w/ this change,
please consider taking it thr' your tree, else please let me know.

Regards
afzal

Link to v2 & v1,
[v2] https://lkml.kernel.org/r/[email protected]
[v1] https://lkml.kernel.org/r/[email protected]

v3:
* Split out from tree wide series, as Thomas suggested to get it thr'
respective maintainers
* Modify pr_err displayed in case of error
* Re-arrange code & choose pr_err args as required to improve readability
* Remove irrelevant parts from commit message & improve

v2:
* Replace pr_err("request_irq() on %s failed" by
pr_err("%s: request_irq() failed"
* Commit message massage

arch/xtensa/kernel/smp.c | 8 ++------
arch/xtensa/kernel/time.c | 12 +++++-------
2 files changed, 7 insertions(+), 13 deletions(-)

diff --git a/arch/xtensa/kernel/smp.c b/arch/xtensa/kernel/smp.c
index 83b244ce61ee..cd85a7a2722b 100644
--- a/arch/xtensa/kernel/smp.c
+++ b/arch/xtensa/kernel/smp.c
@@ -53,16 +53,12 @@ static void system_flush_invalidate_dcache_range(unsigned long start,
#define IPI_IRQ 0

static irqreturn_t ipi_interrupt(int irq, void *dev_id);
-static struct irqaction ipi_irqaction = {
- .handler = ipi_interrupt,
- .flags = IRQF_PERCPU,
- .name = "ipi",
-};

void ipi_init(void)
{
unsigned irq = irq_create_mapping(NULL, IPI_IRQ);
- setup_irq(irq, &ipi_irqaction);
+ if (request_irq(irq, ipi_interrupt, IRQF_PERCPU, "ipi", NULL))
+ pr_err("Failed to request irq %u (ipi)\n", irq);
}

static inline unsigned int get_core_count(void)
diff --git a/arch/xtensa/kernel/time.c b/arch/xtensa/kernel/time.c
index 69db8c93c1f9..77971fe4cc95 100644
--- a/arch/xtensa/kernel/time.c
+++ b/arch/xtensa/kernel/time.c
@@ -128,12 +128,6 @@ static irqreturn_t timer_interrupt(int irq, void *dev_id)
return IRQ_HANDLED;
}

-static struct irqaction timer_irqaction = {
- .handler = timer_interrupt,
- .flags = IRQF_TIMER,
- .name = "timer",
-};
-
void local_timer_setup(unsigned cpu)
{
struct ccount_timer *timer = &per_cpu(ccount_timer, cpu);
@@ -184,6 +178,8 @@ static inline void calibrate_ccount(void)

void __init time_init(void)
{
+ int irq;
+
of_clk_init(NULL);
#ifdef CONFIG_XTENSA_CALIBRATE_CCOUNT
pr_info("Calibrating CPU frequency ");
@@ -199,7 +195,9 @@ void __init time_init(void)
__func__);
clocksource_register_hz(&ccount_clocksource, ccount_freq);
local_timer_setup(0);
- setup_irq(this_cpu_ptr(&ccount_timer)->evt.irq, &timer_irqaction);
+ irq = this_cpu_ptr(&ccount_timer)->evt.irq;
+ if (request_irq(irq, timer_interrupt, IRQF_TIMER, "timer", NULL))
+ pr_err("Failed to request irq %d (timer)\n", irq);
sched_clock_register(ccount_sched_clock_read, 32, ccount_freq);
timer_probe();
}
--
2.25.1

2020-03-04 02:42:24

by Max Filippov

Subject: Re: [PATCH v3] xtensa: replace setup_irq() by request_irq()

On Tue, Mar 3, 2020 at 4:41 PM afzal mohammed <[email protected]> wrote:
>
> request_irq() is preferred over setup_irq(). Invocations of setup_irq()
> occur after memory allocators are ready.
>
> Per tglx[1], setup_irq() existed in olden days when allocators were not
> ready by the time early interrupts were initialized.
>
> Hence replace setup_irq() by request_irq().
>
> [1] https://lkml.kernel.org/r/alpine.DEB.2.20.1710191609480.1971@nanos
>
> Signed-off-by: afzal mohammed <[email protected]>
> ---
> Hi Max Filippov,
>
> i believe you are the maintainer of xtensa, if you are okay w/ this change,
> please consider taking it thr' your tree, else please let me know.

Sure. Applied to my xtensa tree.

--
Thanks.
-- Max