On Wed, May 04, 2011 at 04:18:01AM -0400, Jean Delvare wrote:
> On Wed, 04 May 2011 10:08:12 +0300, Niclas Knuts wrote:
> > On 3.5.2011 15:25, Jean Delvare wrote:
> > > On Tue, 03 May 2011 14:59:37 +0300, Niclas Knuts wrote:
> > >> On 3.5.2011 14:42, Jean Delvare wrote:
> > >>> On Tue, 03 May 2011 14:29:27 +0300, Niclas Knuts wrote:
> > >>>> I am trying to setup lm-sensors for use with the Texas Instruments
> > >>>> TWL4030 Chip (the main processor is a OMAP3503). But I cannot seem to
> > >>>> get it to detect the chip. I have added the hwmon drivers and I am using
> > >>>> up to date MADC drivers that I have backported to kernel 2.6.32 (I am
> > >>>> running Angstrom).
> > >>> Disclaimer: I don't know a thing about this chip nor its drivers.
> > >>>
> > >>> There is no detection support for this device in sensors-detect, if
> > >>> this was your question. I don't think this device can be detected from
> > >>> user-space.
> > >> I am aware that the automatic detect doesn't work, I have read through
> > >> the sensors-detect script.
> > >>> My understanding is that the drivers for the TWL4030 should all get
> > >>> loaded automatically thanks to module aliases, as long as the TWL4030
> > >>> itself is properly registered during platform setup. Check in lsmod
> > >>> whether twl4030-madc and twl4030-madc-hwmon are actually loaded. If
> > >>> not, try loading them manually
> > >> lsmod doesn't show the drivers, but they should have been compiled into
> > >> the kernel, not as modules. Maybe
> > >> that is the problem?
> > > No, this shouldn't make a difference. Check under /sys/devices/platform
> > > if the platform devices have been properly created
> > Ok, they are missing from there. I guess there is some problem with
> > compiling in the drivers then.
>
> Again, no. Modules or built-in should make no difference.
>
> > Is it enough to compile them in or do I
> > initialize them in the board specific files in any way?
>
> Please take a look at:
> drivers/mfd/twl-core.c
>
> As I understand it, the board code has to instantiate one of the I2C
> devices listed in twl_ids. This will call twl_probe() which in turn
> should instantiate all the relevant platform devices.
> arch/arm/mach-omap2 contains many examples:
>
> $ grep -rI '"twl4030"' arch/
> arch/arm/mach-omap2/board-omap3stalker.c: I2C_BOARD_INFO("twl4030", 0x48),
> arch/arm/mach-omap2/board-igep0020.c: I2C_BOARD_INFO("twl4030", 0x48),
> arch/arm/mach-omap2/board-omap3evm.c: I2C_BOARD_INFO("twl4030", 0x48),
> arch/arm/mach-omap2/board-2430sdp.c: I2C_BOARD_INFO("twl4030", 0x48),
> arch/arm/mach-omap2/board-omap3beagle.c: I2C_BOARD_INFO("twl4030", 0x48),
> arch/arm/mach-omap2/board-3430sdp.c: I2C_BOARD_INFO("twl4030", 0x48),
> arch/arm/mach-omap2/board-igep0030.c: I2C_BOARD_INFO("twl4030", 0x48),
> arch/arm/mach-omap2/board-ldp.c: I2C_BOARD_INFO("twl4030", 0x48),
> arch/arm/mach-omap2/board-omap3touchbook.c: I2C_BOARD_INFO("twl4030", 0x48),
> arch/arm/mach-omap2/board-omap3logic.c: I2C_BOARD_INFO("twl4030", 0x48),
>
> That being said, I can't find any code to instantiate
> twl4030_madc_hwmon, so it might as well be that this part is missing
> from the upstream kernel. Cc'ing Keerthy and Axel Lin who probably know
> what's going on.
>
There was an exchange about that last night. Keerthy mentioned that he would submit a patch.
Guenter
_______________________________________________
lm-sensors mailing list
lm-sensors@xxxxxxxxxxxxxx
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
