<Wizzup>
(that number is real and it's my test sim, so feel free to spam it)
xmn has quit [Ping timeout: 252 seconds]
<Wizzup>
that is my n900 db btw
_inky has joined #maemo-leste
<freemangordon>
Wizzup: I think I can debug this oops, by using hints provided by sre
<Wizzup>
freemangordon: I didn't mean the ssi one, but fixing that is helpful too
<freemangordon>
a
<freemangordon>
ah
<freemangordon>
not sure I can help with idle one
* freemangordon
looks at the backtrace
<Wizzup>
the idle one is the real painful one
<Wizzup>
I think the next steps for me are: build 5.15 without our patches and just try to boot it 10x, see if it's stable
<Wizzup>
if not, go to (say) 5.1 and try to boot it without our patches
<Wizzup>
if that is stable, try to bisect
<Wizzup>
the problem is that (afaik) the n900 is broken at many points in between 5.1 and 5.15 so a bisect is not at all easy
<freemangordon>
hmm, but 5.1 won;t idle
<Wizzup>
If by idle you mean RET or OFF increases, then I don't think this happens in 5.15 either and it still crashes
<freemangordon>
ah
<freemangordon>
ok
<Wizzup>
I mean, yes, I can make it go into RET or OFF, but it crashes even when it doesn't hit RET or OFF
<freemangordon>
I see
<Wizzup>
(or it hits RET for the first time somehow and then crashes, but I cannot see that, and it seems unlikely)
<freemangordon>
it makes sense then to bisect
<Wizzup>
also the problem is that it's semi random, so you need to repeat every boot say 10x
<Wizzup>
just to make sure the problem is actually gone
<freemangordon>
ugh
<Wizzup>
I have no way to repeat the problem other than "well let's try a few more times"
<freemangordon>
yeah
<Wizzup>
booting to leste with full desktop usually makes it crash pretty reliably about 3 minutes in (uptime), but getting a kernel all the way to that point is a whole other nightmare
<freemangordon>
my favorite type of bugs :(
<Wizzup>
the absolute worst yeah
<Wizzup>
that said I don't think tmlind has been able to reproduce yet (per his remarks) so maybe it's just in our patches on top of 5.15
<Wizzup>
if we're lucky
<freemangordon>
could you run "scripts/decode_stacktrace.sh" for both oops-es and send results on the ML?
<freemangordon>
I can;t do that as I need the matching vmlinuz
<Wizzup>
yes, but I probably (much) later in the evening
<freemangordon>
ok
<Wizzup>
but only probably*
<freemangordon>
yeah
<Wizzup>
freemangordon: that said if you don't see these resets on your n900/kernel that's also a data point
<freemangordon>
but, I am rarely booting to full desktop
<freemangordon>
dthe difference I guess is in the config though
<freemangordon>
Wizzup: what about pulling 5.15.6? there is a pile of fixes on top of 5.15.2
<Wizzup>
I mean, yeah, but if tmlind is not aware of any changes in that domain it doesn't seem that worthwhile likely