Re: [PATCH] gpio: omap: Fix double trigger for level interrupts

[Markus Schneider-Pargmann <msp@xxxxxxxxxxxx>]

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@xxxxxx>
> 
> 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@xxxxxxxxxxxx/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