[Gta04-owner] various powermanagement issues

[Andreas Kemnade]

On Mon, 26 Oct 2015 08:25:46 +0100
"H. Nikolaus Schaller" <hns at goldelico.com> wrote:

> Hi,
> 
> Am 25.10.2015 um 17:32 schrieb H. Nikolaus Schaller <hns at goldelico.com>:
> 
> > Hi Andreas,
> > 
> > Am 25.10.2015 um 14:10 schrieb Andreas Kemnade <andreas at kemnade.info>:
> > 
> >> On Sun, 25 Oct 2015 08:41:40 +0100
> >> "H. Nikolaus Schaller" <hns at goldelico.com> wrote:
> >> 
> >>>>> When I find time (sigh... I assume people are still waiting for GTA04A5
> >>>>> news...), I can look into the U-Boot systest approach.
> >>>>> 
> >>>> I managed to hang uboot with systest ch...
> >>> 
> >>> Yes, for me it hangs as well. So there might be another bug.
> >>> 
> >> PHY_DPLL clock does not come up for some reason. And systest ch waits
> >> forever for it.
> > 
> > I have fixed the
> > 
> > #define BCIAUTOWEN       (TWL4030_BASEADD_PM_MASTER + 0x020)
> > 
> > to become
> > 
> > #define BCIAUTOWEN       0x020
> > 
> > (TWL4030_BASEADD_PM_MASTER is 0x036 so summing things up
> > makes 0x056 which is a really wrong bit pattern).
> > 
> > and now it appears to work (PHY_DPLL fixed as well):
> > 
> > GTA04 # systest cha     
> > USB charger detected
> > Main battery charge:    3557 mV
> > Battery temperature:    39 C
> > Backup battery voltage: 3206 mV
> > AC charger voltage:     0 mV
> > USB charger voltage:    5124 mV
> > Charging current:       66 mA
> > GTA04 # 
> > GTA04 # 
> > GTA04 # 
> > GTA04 # 
> > GTA04 # systest cha
> > USB charger detected
> > Main battery charge:    4215 mV
> > Battery temperature:    39 C
> > Backup battery voltage: 3210 mV
> > AC charger voltage:     0 mV
> > USB charger voltage:    5028 mV
> > Charging current:       173 mA
> > GTA04 # 
> > GTA04 # 
> > 
> > Only the charging current doesn't appear to be reasonable, but I can
> > now take out the battery while connected to USB. So there is more
> > than 60 mA or 173 mA available...
> > 
> > So I will make a patch for our letux-u-boot and provide new SD card
> > images so that everyone who wants can upgrade the U-Boot in flash.
> > 
> > The binary of this u-boot is already here:
> > 
> > 	http://download.goldelico.com/letux-u-boot/GTA04/latest/
> > 
> > Thanks for identifying the root cause of this issue!
> > 
> > BR,
> > Nikolaus
> > 
> >> Found that out by doing everything via i2c mw/md commands
> >> Linear charging works with less things enabled. 
> >> 
> >>>> I tried to insert these lines into boot.scr
> >>>> 
> >>>> i2c dev 0 
> >>>> i2c mw 0x4b 0x3d 0x30 1
> >>>> i2c mw 0x4a 0x75 0x26 1
> >>>> i2c mw 0x4a 0x95 0xf3 1
> >>>> 
> >>>> That starts charging for me.
> >>> 
> >>> Great!
> >>> 
> >> #enable madc
> >> i2c mw 0x4a 0x00 0x01 1
> >> #disable charging it bciboot 
> >> i2c mw 0x4b 0x3d 0x30 1
> >> #enable linear charging
> >> i2c mw 0x4a 0x75 0x26 1
> >> i2c mw 0x4a 0x95 0xf3 1
> >> #current monitoring
> >> i2c mw 0x4a 0x85 0x9c 1
> >> i2c mw 0x4a 0x88 0xf0 1
> >> 
> >> gives a bit better behavior. Drawback is that some kernels do not
> >> behave well when linear charging is enabled before bootup.
> >> And I would prefer if the kernel switches back to automagic charging.
> >> 
> >>>> I found this line in board/goldelico/gta04/twl4030-additions.h:
> >>>> #define BCIAUTOWEN       (TWL4030_BASEADD_PM_MASTER + 0x020)
> >>>> 
> >>>> That looks wrong for a *bit*  definition.
> >>> 
> >>> Indeed!
> >>> 
> >>> I have looked around and it appears to come from some TI / WindRiver source
> >>> code from 2009 or so. E.g.
> >>> 
> >>> https://github.com/CMNookTablet/acclaim_cyanoboot/blob/master/include/twl4030.h
> >>> 
> >>> This may indeed be buggy.
> >>> 
> >>> In Linux it was suggested (today's code is very different) as a bit value:
> >>> 
> >>> https://lwn.net/Articles/364310/
> >>> 
> >>> #define BCIAUTOWEN		0x20
> >>> 
> >> The uboot stuff was an exercise to refresh my knowledge about charging
> >> to be able to deal with current kernel code.
> >> I have stolen i2c 0x4a and 0x4b (via echo 0x4a >unbind)from the twl4030
> >> driver. Via i2cget I could find out that
> >> bciiref was set to 0! I could enable charging with i2cset. And when the
> >> kernel notices that theft. It became angry of course!
> 
> Some more tests have revealed another strange issue:
> 
> * if I boot MLO + U-Boot from MMC (AUX button pressed), the systest charge command now works
> * if I boot X-Loader + U-Boot from NAND, the systest charge still hangs (after detecting USB cable present)
> 
my i2c mw stuff works in both cases.
I think there should at least be a timeout for the DPLL check.

> I have not identified the reason why, but it may have to do with the boot sequence.
> 
> It is (http://projects.goldelico.com/p/gta04-main/downloads/54/):
> 
> * with AUX: USB -> UART -> MMC -> NAND
> * w/o AUX: NAND -> USB -> UART -> MMC
> 
> So the USB (twl4030-phy) may be differently initialized since the BootROM
> stops this search as soon as it finds a boot loader. And therefore the same command
> sequence in U-Boot may work or fail.
> 
What about regulators? In the kernel there is one regulator enabled in twl4030_charger.c

Regards,
Andreas
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 490 bytes
Desc: not available
URL: <http://lists.goldelico.com/pipermail/gta04-owner/attachments/20151030/c44db957/attachment.asc>