On Mon, 2021-10-11 at 23:20 +0000, Ward, David - 0665 - MITLL wrote:
On 9/29/21 3:29 PM, Adam Williamson wrote:
> On Wed, 2021-09-29 at 16:04 +0000, Patrick McAfee wrote:
> > Hey all,
> > => First time posting here. When I tried to install the beta on
> > my
> > laptop while in EFI mode the installer would crash after installing
> > the bootloader and transition to configuring software and go back
> > to
> > GDM. When I switch my laptop to Legacy it doesn't happen, which
> > leads
> > me to believe something is up with the EFI shim or something along
> > those lines. Is this a known bug that anyone knows of? Also where
> > do
> > I report this one exactly? I know this isn't a very detailed
> > report
> > but I wanted to check before making a fully detailed report out.
> > Thanks!
>
> Hi Patrick! We'd need to see some logs to know what's going on, I
> think. This
> isn't a general bug, I don't think, as we do test an EFI install from
> live media in
> our automated tests, and those tests are passing. Does the installer
> show
> you any kind of crash report interface, or does it just flat out
> disappear? If it
> just goes away, we'll probably need to see what's in the various
> installer log
> files in /tmp , plus the system journal. Thanks!
Adam,
I can consistently reproduce this exactly as described here, on a brand
new laptop (but not on an older one). There is a SIGSEGV occurring in
gnome-shell, which would appear from the backtrace to be caused by a
recent upstream commit. Please see bug 2009637.
Should this be considered for a release blocker? I'm honestly not sure
whether this issue is quite hardware constrained, or perhaps if we're
just the first ones reporting it.
Thanks for the update - as you probably saw, I got Ray to take a look
at the bug and he saw the problem right away, a fix is now heading to
updates-testing. We're not 100% sure why it sometimes happens and
sometimes doesn't, though it's got to have *something* to do with a
password or authentication process somehow happening along the way.
Given that, I proposed it as a freeze exception issue, which is easier
to vote on and will let us get the fix in for the release.
On that note: because ABRT is disabled when the live media boots, it
does not capture this crash - unless I manually start the ABRT services
before the installer. Is that intended? One of the basic release
criteria is that installer failures should be reportable to Bugzilla.
The nuance here is that it's not Anaconda which is crashing, but the
fact that gnome-shell is crashing is essentially having the same
impact.
The criterion is really covering the installer's own mechanism
specifically. abrt crash reporting is intentionally disabled on lives
since 2018:
https://pagure.io/fedora-kickstarts/c/4f4fe63125e76ef53056e10b5266ea99396...
--
Adam Williamson
Fedora QA
IRC: adamw | Twitter: adamw_ha
https://www.happyassassin.net