| Message ID | 20200312064817.19000-1-afzal.mohd.ma@gmail.com (mailing list archive) |
|---|---|
| State | Mainlined |
| Commit | 470cf1c28d2f601ea666a96d676c10b09b2321ab |
| Headers | show |
| Series | [v4] clocksource/drivers/timer-cs5535: request irq with non-NULL dev_id | expand |
subject was not supposed to have version info, happened by mistake, this is an incremental patch on top of v3 that has been applied to the timers/drivers/next branch. Regards afzal
On 12/03/2020 07:48, afzal mohammed wrote: > Recently all usages of setup_irq() was replaced by request_irq(). > request_irq() does a few sanity checks that were not done in > setup_irq(), if they fail irq registration will fail. One of the check > is to ensure that non-NULL dev_id is passed in the case of shared irq. > > Fix it by passing non-NULL dev_id while registering the shared irq. > > Fixes: cc2550b421aa ("clocksource: Replace setup_irq() by request_irq()") > Signed-off-by: afzal mohammed <afzal.mohd.ma@gmail.com> > --- > > Hi Daniel, > > i have assumed that you would not rebase your drivers/next tree & put the > SHA-id in Fixes tag. Applied, thanks
Hello: This patch was applied to khilman/linux-amlogic.git (refs/heads/for-next). On Thu, 12 Mar 2020 12:18:17 +0530 you wrote: > Recently all usages of setup_irq() was replaced by request_irq(). > request_irq() does a few sanity checks that were not done in > setup_irq(), if they fail irq registration will fail. One of the check > is to ensure that non-NULL dev_id is passed in the case of shared irq. > > Fix it by passing non-NULL dev_id while registering the shared irq. > > [...] Here is a summary with links: - [v4] clocksource/drivers/timer-cs5535: request irq with non-NULL dev_id https://git.kernel.org/khilman/linux-amlogic/c/470cf1c28d2f601ea666a96d676c10b09b2321ab You are awesome, thank you!
diff --git a/drivers/clocksource/timer-cs5535.c b/drivers/clocksource/timer-cs5535.c index 51ea0509fb25..d47acfe848ae 100644 --- a/drivers/clocksource/timer-cs5535.c +++ b/drivers/clocksource/timer-cs5535.c @@ -133,6 +133,7 @@ static irqreturn_t mfgpt_tick(int irq, void *dev_id) static int __init cs5535_mfgpt_init(void) { + unsigned long flags = IRQF_NOBALANCING | IRQF_TIMER | IRQF_SHARED; struct cs5535_mfgpt_timer *timer; int ret; uint16_t val; @@ -152,9 +153,7 @@ static int __init cs5535_mfgpt_init(void) } /* And register it with the kernel */ - ret = request_irq(timer_irq, mfgpt_tick, - IRQF_NOBALANCING | IRQF_TIMER | IRQF_SHARED, - DRV_NAME, NULL); + ret = request_irq(timer_irq, mfgpt_tick, flags, DRV_NAME, timer); if (ret) { printk(KERN_ERR DRV_NAME ": Unable to set up the interrupt.\n"); goto err_irq;
Recently all usages of setup_irq() was replaced by request_irq(). request_irq() does a few sanity checks that were not done in setup_irq(), if they fail irq registration will fail. One of the check is to ensure that non-NULL dev_id is passed in the case of shared irq. Fix it by passing non-NULL dev_id while registering the shared irq. Fixes: cc2550b421aa ("clocksource: Replace setup_irq() by request_irq()") Signed-off-by: afzal mohammed <afzal.mohd.ma@gmail.com> --- Hi Daniel, i have assumed that you would not rebase your drivers/next tree & put the SHA-id in Fixes tag. Regards afzal drivers/clocksource/timer-cs5535.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)