Code review comment for ~xnox/casper:papercuts

On Wed, 27 May 2020 at 03:03, Michael Hudson-Doyle
<email address hidden> wrote:
>
> This all looks good, just one question.
>
> Diff comments:
>
> > diff --git a/scripts/casper b/scripts/casper
> > index 6e9ff35..6cecb79 100644
> > --- a/scripts/casper
> > +++ b/scripts/casper
> > @@ -885,6 +890,10 @@ mountroot() {
> > if [ "${livefs_root}" ]; then
> > break
> > fi
> > + # No more udev events, no more devices, nothing to wait for
>
> Are we just waiting for udev here? I thought the issue was that sometimes it takes some time for hardware to show up at all. If I'm mistaken, +1 for sure.
>

I don't know how to make a distinction between hardware that is slow
and we are enumerating 1000s of hard drives looking for our CDROM,
versus interactive netboot where there will never be a drive matching
for us.

My assumption here is that, if we are enumerating 1000s of hard
drives, udev queue will be full of unprocessed events.

On the s390x LPAR, it appears that the installer is hung up, one hits
enter a few times, and then none of the `read -p` work, cause stdin is
full of things.

Whereas on s390x QEMU, the CDROM should appear there quite quickly.

Regards,

Dimitri.