On Wed, 2023-10-18 at 08:28 +0300, Tony Lindgren wrote:
> * Richard Purdie <richard.purdie@xxxxxxxxxxxxxxxxxxx> [231017 22:15]:
> > On Tue, 2023-10-17 at 09:56 +0300, Tony Lindgren wrote:
> > > * Richard Purdie <richard.purdie@xxxxxxxxxxxxxxxxxxx> [231016 08:10]:
> > > > The port sometimes doesn't come up properly at boot.
> > > >
> > > > To be clear, the "\n\n" from the qemu side into the port doesn't seem
> > > > to help. The "echo helloB > /dev/ttyS1" inside the image does seem to
> > > > wake it up.
> > >
> > > So if I understand correctly, this issue still happens with kernel patched
> > > with commit 81a61051e0ce ("serial: core: Fix checks for tx runtime PM
> > > state"), and the issue now only happens sometimes.
> >
> > The issue has always been intermittent and it appeared to happen less
> > frequently with 81a61051e0ce added but it was hard to know if I was
> > imagining that.
>
> Oh OK.
>
> > > I wonder if the following additional change might help?
> >
> > I've added it into testing and have not reproduced the failure with it
> > applied yet, locally or on our autobuilder. We need to sort some
> > release pieces which have been delayed by these issues and we're going
> > with a workaround for that. Once that is built I can get back to
> > testing this change more extensively, see if we can still provoke the
> > issue or not. It may take a day or two of testing before we know with
> > any certainty if the issue is resolved or not.
>
> Thanks for the update, let's wait a few days and see then.
Our release build failed with our workaround, probably as this piece is
missing.
I've not seen the failure occur in any build with it applied so far. As
such I think we'll move over to this patch as it seems to be better.
Whether it is fixed or not I'm still not 100% sure but it is looking
more likely. Thanks for the patch/fix!
Cheers,
Richard
