* Re: [PATCH] irqchip/riscv-aplic: Register syscore operations only once
2026-03-09 8:11 [PATCH] irqchip/riscv-aplic: Register syscore operations only once liu.xuemei1
@ 2026-03-09 9:17 ` Thomas Gleixner
0 siblings, 0 replies; 2+ messages in thread
From: Thomas Gleixner @ 2026-03-09 9:17 UTC (permalink / raw)
To: liu.xuemei1, anup, pjw, palmer, aou, alex, liujingqi, cyan.yang,
nick.hu, yongxuan.wang
Cc: linux-riscv, linux-kernel
On Mon, Mar 09 2026 at 16:11, liu wrote:
> From: Jessica Liu <liu.xuemei1@zte.com.cn>
>
> We can have multiple APLIC instances so setup global state
s/We/A system/
Also which global state are you referring to?
> and resgister syscore operations only once.
register
You also fail to describe what the problem is caused by registering the
syscore ops more than once.
> Fixes: 95a8ddde3660 ("irqchip/riscv-aplic: Preserve APLIC states across suspend/resume")
> Signed-off-by: Jessica Liu <liu.xuemei1@zte.com.cn>
> ---
> drivers/irqchip/irq-riscv-aplic-main.c | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/irqchip/irq-riscv-aplic-main.c b/drivers/irqchip/irq-riscv-aplic-main.c
> index 4495ca26abf5..8299e0da6577 100644
> --- a/drivers/irqchip/irq-riscv-aplic-main.c
> +++ b/drivers/irqchip/irq-riscv-aplic-main.c
> @@ -20,6 +20,7 @@
> #include "irq-riscv-aplic-main.h"
>
> static LIST_HEAD(aplics);
> +static bool aplic_global_setup_done __ro_after_init;
Again. That global state reference is confusing at best.
> static void aplic_restore_states(struct aplic_priv *priv)
> {
> @@ -375,8 +376,10 @@ static int aplic_probe(struct platform_device *pdev)
> if (rc)
> dev_err_probe(dev, rc, "failed to setup APLIC in %s mode\n",
> msi_mode ? "MSI" : "direct");
> - else
> + else if (!aplic_global_setup_done) {
> register_syscore(&aplic_syscore);
> + aplic_global_setup_done = true;
> + }
Aside of violating the bracket rules, this code is broken already in a
very subtle way:
If the probe fails then it should not invoke
acpi_dev_clear_dependencies() either.
Something like the below makes it entirely clear what this is about.
The fix for the probe fail vs. acpi dependencies obviously wants to be
split out into a separate patch.
Thanks,
tglx
---
--- a/drivers/irqchip/irq-riscv-aplic-main.c
+++ b/drivers/irqchip/irq-riscv-aplic-main.c
@@ -116,6 +116,16 @@ static struct syscore aplic_syscore = {
.ops = &aplic_syscore_ops,
};
+static bool aplic_syscore_registered __ro_after_init;
+
+static void aplic_syscore_init(void);
+{
+ if (!aplic_syscore_registered) {
+ register_syscore(&aplic_syscore);
+ aplic_syscore_registered = true;
+ }
+}
+
static int aplic_pm_notifier(struct notifier_block *nb, unsigned long action, void *data)
{
struct aplic_priv *priv = container_of(nb, struct aplic_priv, genpd_nb);
@@ -372,18 +382,20 @@ static int aplic_probe(struct platform_d
rc = aplic_msi_setup(dev, regs);
else
rc = aplic_direct_setup(dev, regs);
- if (rc)
+
+ if (rc) {
dev_err_probe(dev, rc, "failed to setup APLIC in %s mode\n",
msi_mode ? "MSI" : "direct");
- else
- register_syscore(&aplic_syscore);
+ return rc;
+ }
+
+ aplic_syscore_register();
#ifdef CONFIG_ACPI
if (!acpi_disabled)
acpi_dev_clear_dependencies(ACPI_COMPANION(dev));
#endif
-
- return rc;
+ return 0;
}
static const struct of_device_id aplic_match[] = {
^ permalink raw reply [flat|nested] 2+ messages in thread