<Entitlement>
tmlind - [ [PATCH] gpio: omap: Save and restore sysconfig ]
<siceloa>
I was just looking at that
* sicelo
doesn't understand this stuff so much, haha, but, is this for PM? does it affect us while we don't have PM working on N900 yet?
<sicelo>
i genuinely wish i could make PM work, even if it's only 50% efficients
<sicelo>
otherwise, gpio issue i noticed on N900 is with lp5523 driver (for keyboard LEDs and RGB) - the gpio for this chip is never set HIGH, so chip never gets enabled. i sent a patch to the LED maintainers and still waiting for feedback.
<tmlind>
sicelo: yeah this would be only with pm enabled
<tmlind>
sicelo: can happen easily during the boot while drivers are loading, so if a gpio bank is only partially working, weird things can happen
* sicelo
should do something about that serial for N900 ... otherwise we'll be left too far behind, and end up with lots of regressions, and useless device :)
<tmlind>
sicelo: presumably these gpio issues started happening with commit fb2c599f0566 ("ARM: omap3: enable off mode automatically")
<sicelo>
i see. but off mode isn't automatically enabled still? at least with 5.12-rc6, /sys/kernel/debug/pm_debug/enable_off_mode was initially 0
<tmlind>
sicelo: hmm it should be, you still need to idle the uarts though for any deeper idle modes to start triggering
<sicelo>
i'll test it again
<tmlind>
ok
<sicelo>
in fact i can boot 5.12-rc6 right away :)
<sicelo>
there was some bandgap/eocz stuff you discussed with fmg in the recent past, which i seem to think you mentioned could also have an effect on entering some of the idle levels. i'll check logs and see if any of that makes a difference/improvement for me
<tmlind>
yeah the bandgap fixes all are in v5.12-rc series already
<parazyd>
sicelo: btw thanks for investigating the dnsmasq configuration, will save me some time :)
<tmlind>
sicelo: but please apply that gpio patch as the gpio banks can get messed up during init
<sicelo>
tmlind: sure. i'll do that later today, with -rc7
<sicelo>
parazyd: yw. how does the pp name its cellular data interface? wwan as well, or?
<tmlind>
sicelo: ok thanks
<parazyd>
Yea I believe so
<sicelo>
tmlind: any suggested way for me to check if the patch works?
<tmlind>
sicelo: yeah see the droid4-pm script on detaching serial console and idling uarts
<tmlind>
sicelo: then disconnect usb and blank the lcd, and five seconds later do sudo cat /sys/kernel/debug/pm_debug/count and look at the core domain ret and off numbers
<sicelo>
sure
<tmlind>
unfortunately off number won't increase much with current kernels as there are way too many timers happening :(
<sicelo>
i said earlier that enable_off_mode was initially 0 ... so you're right, it's 1 as i've just booted up
uvos has joined #maemo-leste
<sicelo>
tmlind: i just briefly looked at the droi4-pm script. the are some registers/addresses read via devmem. i guess those come from the TRM? just tried one of them on N900 and naturally got core dumped :p
<sicelo>
"Address Hole seen by MPU at address 4a009540"