genii has quit [Quit: Morning comes early.... GO LEAFS GO!]
mszyprow|home has joined #linux-exynos
mszyprow|home has quit [Ping timeout: 240 seconds]
<krzk>
wwilly_: hey, do the bisect, that's the easiest
<krzk>
wwilly_: you start bisect with known good and known bad revision and test each bisect-point
<krzk>
wwilly_: since issue is reproducible, bisect should be able to point it fast (you will need around 13-14 tries)
<krzk>
wwilly_: just be sure that tested biection-point is really tested, so either the revision is good or bad. The worse is when issue is not easy reproducible and then you might mark wrong-revision-but-not-reproduced-now with "bisect good"
<krzk>
it will break entire process and bisect will go to wrong direction
<krzk>
when you find offending commit, send a bug report to lkml to folks responsible for it (get_maintainers.pl) and author and any signed-offs and for example us + samsung-soc
<krzk>
Sometimes the real commit causing issue is not the one pointed by bisect . Bisect-found-commit just discovered the issue. Brought to light.
<krzk>
Ah, and, if problem is reproducible on v5.6 then of course use v5.6, bnot v5.6.3
<wwilly_>
what is bisect exactly? dichotomic search on git log revision from the last working to the first faulty?
<wwilly_>
I found a rpi3b that uses a smsc95xx as well, I'm trying on this one currently
<wwilly_>
arf ok it's a git feature
<wwilly_>
never used that ... :)
<wwilly_>
thanks
<wwilly_>
don't know all features from git, just play with stash ... :) know that I pushing my phd in kernel instead of user space...
<wwilly_>
a pain in the ash :)
<krzk>
git help biect
<krzk>
git bisect start
<krzk>
git bisect good v5.5 (assuming it is good)
<krzk>
git bisect bad v5.6 (assumign it is bad)
<krzk>
and now compile + testing + git bisect good/bad
<wwilly_>
yeah got it
<wwilly_>
but would start with v5.5.16 not? as this one is working perfectly well on both boards
<krzk>
no, I mentioned, do not use stables because: a. it will be longer (e.g. 13 -> 15 tests), b. it might point you to stable backport commit and then what?
<krzk>
unless issue does not appear on v5.6... then it is already hint narrowing bisect and pointing where to search
<wwilly_>
ok thansk
<wwilly_>
so rpi3b is working with v5.6.3, will try bissect for the xu3 over the week-end
<wwilly_>
have a nice one krzk
<krzk>
You too, bye
LiquidAcid has joined #linux-exynos
wwilly_ has quit [Ping timeout: 258 seconds]
wwilly has joined #linux-exynos
LiquidAcid has quit [Quit: Leaving]
<wwilly>
krzk, it's a bit a pain in the ass, I have some kernel panic with different revision