| Message ID | 20240326145439.1293412-1-msp@baylibre.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | gpio: omap: Fix double trigger for level interrupts | expand |
In the bug report email thread Markus Mirevik said "The interrupt associated with the GPIO module still fires twice" so while this patch is an improvement, it might not be a complete solution? > This patch was originally proposed by Grygorii Strashko. The way to give authorship credit is you make the first line of your patch: From: Grygorii Strashko <grygorii.strashko@ti.com> When the patch is applied then git will assign authorship credit but remove that line from the git log. On Tue, Mar 26, 2024 at 03:50:14PM +0100, Markus Schneider-Pargmann wrote: > Set gpio trigger before clearing the irq status. > This commit message needs some work. When you're reviewing on email, it's kind of common to read the commit message without reading the subject. See how the patch looks like on lore: https://lore.kernel.org/linux-gpio/20240326145439.1293412-1-msp@baylibre.com/T/#u The subject is up on the first line, but it's mixed in with the headers so it's easy to skip. Go ahead a restate the subject but in different words. But also copy and paste more of the problem from the bug report. To me if I were a user the important bit is that the bug ends up hogging the CPU. "The problem is that the interrupt handler was is run twice for each frame. It hogs a lot of CPU time. Fix this by setting the GPIO trigger before clearing the IRQ status." regards, dan carpenter
Hi Dan, On Tue, Mar 26, 2024 at 06:29:45PM +0300, Dan Carpenter wrote: > In the bug report email thread Markus Mirevik said "The interrupt > associated with the GPIO module still fires twice" so while this patch > is an improvement, it might not be a complete solution? Yes, that is possible. I actually only had the level interrupt double trigger bug which is fixed by this change. > > > This patch was originally proposed by Grygorii Strashko. > > The way to give authorship credit is you make the first line of your > patch: > > From: Grygorii Strashko <grygorii.strashko@ti.com> > > When the patch is applied then git will assign authorship credit but > remove that line from the git log. Thanks, yes I am aware, though I didn't want to add that author tag without someone agreeing with that. As it wasn't a formal patch, there was no From: already present or a SoB by Grygorii. Also there was no reaction on a ping from me on the original thread. That's why I opted for the comment in the commit, the explanation and as I said in the letter part I don't care about author, just that it gets upstream. > > On Tue, Mar 26, 2024 at 03:50:14PM +0100, Markus Schneider-Pargmann wrote: > > Set gpio trigger before clearing the irq status. > > > > This commit message needs some work. When you're reviewing on email, > it's kind of common to read the commit message without reading the > subject. See how the patch looks like on lore: > > https://lore.kernel.org/linux-gpio/20240326145439.1293412-1-msp@baylibre.com/T/#u > > The subject is up on the first line, but it's mixed in with the headers > so it's easy to skip. Go ahead a restate the subject but in different > words. > > But also copy and paste more of the problem from the bug report. To me > if I were a user the important bit is that the bug ends up hogging the > CPU. > > "The problem is that the interrupt handler was is run twice for each > frame. It hogs a lot of CPU time. Fix this by setting the GPIO trigger > before clearing the IRQ status." Yes, thanks, I will work on that commit message. Best Markus
diff --git a/drivers/gpio/gpio-omap.c b/drivers/gpio/gpio-omap.c index 76d5d87e9681..74b8fe2995e1 100644 --- a/drivers/gpio/gpio-omap.c +++ b/drivers/gpio/gpio-omap.c @@ -696,6 +696,9 @@ static void omap_gpio_unmask_irq(struct irq_data *d) raw_spin_lock_irqsave(&bank->lock, flags); omap_set_gpio_irqenable(bank, offset, 1); + if (trigger) + omap_set_gpio_triggering(bank, offset, trigger); + /* * For level-triggered GPIOs, clearing must be done after the source * is cleared, thus after the handler has run. OMAP4 needs this done @@ -705,9 +708,6 @@ static void omap_gpio_unmask_irq(struct irq_data *d) trigger & (IRQ_TYPE_LEVEL_HIGH | IRQ_TYPE_LEVEL_LOW)) omap_clear_gpio_irqstatus(bank, offset); - if (trigger) - omap_set_gpio_triggering(bank, offset, trigger); - raw_spin_unlock_irqrestore(&bank->lock, flags); }
Set gpio trigger before clearing the irq status. This patch was originally proposed by Grygorii Strashko. Cc: Grygorii Strashko <grygorii.strashko@ti.com> Reported-by: Markus Mirevik <markus.mirevik@dpsolutions.se> Closes: https://lore.kernel.org/all/20220122235959.GA10737@sol/T/ Signed-off-by: Markus Schneider-Pargmann <msp@baylibre.com> --- Hi everyone, this patch helped me on the beagleboneblack to remove the mentioned double trigger of level interrupts. This diff was proposed by Grygorii in the thread linked in the commit message. I am not sure why this never made it into the kernel, that's why I sending this patch. I did not create the diff just made a patch out of it, I don't care about being the author but I would be happy if this would get merged or some other solution to the problem. Thanks! Best Markus drivers/gpio/gpio-omap.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)