vpack270 and sleep mode

[marek.vasut@gmail.com (Marek Vasut)]

Dne ?t 25. kv?tna 2010 15:26:54 237 Rumjantsev Egor (PROG) napsal(a):
> 25.05.2010 18:39, Marek Vasut ?????:
> > Dne ?t 25. kv?tna 2010 14:38:09 Marek Vasut napsal(a):
> >> Dne ?t 25. kv?tna 2010 14:13:01 237 Rumjantsev Egor (PROG) napsal(a):
> >>> Hello, Marek
> >> 
> >> Hi,
> >> 
> >>> Could you help to solve the following problem:
> >>> I try to move my board based on Voipac PXA270 module to sleep state.
> >>> I'm running kernel 2.6.28.10 on it.
> >> 
> >> CCed Voipac and linux-arm-kernel. Btw. that's not Voipac official
> >> kernel, is it?
> 
> Yes it isn't. I patched vanilla kernel to get basic Voipac module
> support and add some my code for our board.

Please, use what's in mainline then. (git.kernel.org, search for ycmiao, use -
devel branch).
> 
> > My bad, CCed wrong linux-arm-kernel, fixed now :)
> > 
> >>> When i try echo mem>  /sys/power/state i see messages that system goes
> >>> to sleep state but after pushing wake-up button (GPIO0) i don't see
> >>> that a bootloader is running (i set up some GPIOs as flags in
> >>> bootloader and see on them whith oscilloscope).
> >> 
> >> The bootloader checks RCSR once you wake it up and it sets-up SDRAM. If
> >> there's a sleep-wakeup bit, it branches to location in PSPR and there's
> >> linux kernel already.
> 
> i know that as i have to use U-boot 1.3.0 for my board. It's also not
> Voipac's u-boot version.

Can't you use u-boot-pxa from git.denx.de ?

> I set or clear some of GPIO depending of reason we enter on bootloader.
> And after pushing wake-up button my flags state doen't change.
> 
> >>> This is a part of kernel output when the system goes to sleep mode:
> >>> 
> >>> dm9000 dm9000.0: LATE suspend
> >>> platform pxa27x-pwm.1: LATE suspend
> >>> platform pxa27x-pwm.0: LATE suspend
> >>> platform pxa27x-ssp.2: LATE suspend
> >>> platform pxa27x-ssp.1: LATE suspend
> >>> platform pxa27x-ssp.0: LATE suspend
> >>> pxa2xx-i2c pxa2xx-i2c.1: LATE suspend
> >>> platform sa1100-rtc: LATE suspend
> >>> platform pxa2xx-i2s: LATE suspend
> >>> pxa2xx-uart pxa2xx-uart.2: LATE suspend
> >>> pxa2xx-uart pxa2xx-uart.1: LATE suspend
> >>> pxa2xx-uart pxa2xx-uart.0: LATE suspend
> >>> platform pxa27x-udc: LATE suspend
> >>> pxa_pm_enter(): RCSR value: 0x00000000
> >>> pxa_pm_enter(): state: 3
> >>> pxa_pm_enter(): Power regs:
> >>> pxa_pm_enter(): GPDR0: 0x0070E200
> >>> pxa_pm_enter(): PWER: 0x00000001
> >>> pxa_pm_enter(): PFER: 0x00000000
> >>> pxa_pm_enter(): PRER: 0x00000001
> >>> pxa_pm_enter(): PGSR0: 0x00010000
> >>> pxa_pm_enter(): PGSR1: 0x000F2000
> >>> pxa_pm_enter(): PGSR2: 0x0404C000
> >>> pxa_pm_enter(): PGSR3: 0x00000020
> >>> pxa27x_cpu_pm_enter(): go to suspend : 3
> >>> pxa27x_cpu_pm_enter(): PCFR : 0x61
> >>> pxa27x_cpu_pm_enter(): PSSR : 0x0
> >>> pxa27x_cpu_pm_enter(): PSLR : 0xCC100004
> >>> pxa27x_cpu_pm_enter(): OSCC : 0x3
> >>> 
> >>> but if i type command echo standby>  /sys/power/state
> >>> the system reacts on wake-up button but hangs after some messages about
> >>> null dereferencing pointers.
> >>> 
> >>> is there any possibility to trace such problem?
> >> 
> >> Try CONFIG_PM_ADVANCED_DEBUG.
> 
> i've looked through kernel sources and didn't such identifer
> 
Maybe you need newer kernel, see above.
> 
>   Enable that and check if the tests work, one
> 
> >> by one. What are the messages you get once the kernel crashes?
> 
> i've tryied to run tests.
> echo devices > /sys/power/pm_test works fine
> echo core > /sys/power/pm_test hangs with:
> 
> suspend debug: Waiting for 5 seconds.
> platform pxa27x-udc: EARLY resume
> pxa2xx-uart pxa2xx-uart.0: EARLY resume
> pxa2xx-uart pxa2xx-uart.1: EARLY resume
> pxa2xx-uart pxa2xx-uart.2: EARLY resume
> platform pxa2xx-i2s: EARLY resume
> platform sa1100-rtc: EARLY resume
> platform pxa2xx-i2c.1: EARLY resume
> platform pxa27x-ssp.0: EARLY resume
> platform pxa27x-ssp.1: EARLY resume
> platform pxa27x-ssp.2: EARLY resume
> platform pxa27x-pwm.0: EARLY resume
> platform pxa27x-pwm.1: EARLY resume
> dm9000 dm9000.0: EARLY resume
> vpac270-ssp vpac270-ssp.0: EARLY resume
> platform vpac270-pm.0: EARLY resume
> platform pxa_g64_driver.0: EARLY resume
> platform pxa2xx_dir_keypad.0: EARLY resume
> platform pxa_capture.0: EARLY resume
> platform pxa2xx-ac97: EARLY resume
> platform mfrc531-ssp.0: EARLY resume
> pxa2xx-flash pxa2xx-flash.0: EARLY resume
> Unable to handle kernel paging request at virtual address a03ab26f
> pgd = c7b10000
> [a03ab26f] *pgd=8020040e(bad)
> Internal error: Oops: 803 [#1] PREEMPT
> Modules linked in: ipv6 pxa_reg
> Unable to handle kernel paging request at virtual address a03ab26f
> pgd = c7b10000
> [a03ab26f] *pgd=8020040e(bad)
> Internal error: Oops: 803 [#1] PREEMPT
> Modules linked in: ipv6 pxa_reg
> CPU: 0    Not tainted  (2.6.28.10-vpac3-splash #90)
> PC is at bad_channel+0x80166/0x91f28
> LR is at platform_pm_resume_noirq+0x48/0x54
> pc : [<a03ab23e>]    lr : [<a01cb758>]    psr: a00000b3
> sp : c6b3be64  ip : c6b3be74  fp : c6b3be70
> r10: 00000004  r9 : c79711b4  r8 : c7a1e000
> r7 : 00000005  r6 : a03da1b0  r5 : a03f22e4  r4 : a03dd328
> r3 : a03ab23b  r2 : 00000000  r1 : a03f22e4  r0 : a03dd320
> Flags: NzCv  IRQs off  FIQs on  Mode SVC_32  ISA Thumb  Segment user
> Control: 0000397f  Table: a7b10000  DAC: 00000015
> Process sh (pid: 309, stack limit = 0xc6b3a268)
> Stack: (0xc6b3be64 to 0xc6b3c000)
> be60:          c6b3be88 c6b3be74 a01cd9b0 a01cb71c a03dd328 00000010
> c6b3bea8
> be80: c6b3be8c a01ce248 a01cd968 00000010 00000000 00000003 a030d660
> c6b3bebc
> bea0: c6b3beac a01ce8c8 a01ce1e4 00000000 c6b3bed8 c6b3bec0 a006fd10
> a01ce8b8
> bec0: 0000db96 a038c7db 00000003 c6b3bef4 c6b3bedc a006ff20 a006fc10
> a038c7db
> bee0: 00000003 00000003 c6b3bf1c c6b3bef8 a0070040 a006fdfc c780dc14
> 00000004
> bf00: c79711a0 c781ada0 a03eeea4 c6b3bf80 c6b3bf2c c6b3bf20 a018afe8
> a006ffb0
> bf20: c6b3bf58 c6b3bf30 a00e1da8 a018afd0 c7ab1b20 4001d000 c6b3bf80
> 00000004
> bf40: a00300a4 c6b3a000 00000000 c6b3bf7c c6b3bf5c a0099b58 a00e1ca8
> c6b3bf80
> bf60: 00000000 00000000 c7ab1b20 00000004 c6b3bfa4 c6b3bf80 a0099cb8
> a0099ab0
> bf80: 00000000 00000000 00000000 00000004 4001d000 401565f8 00000000
> c6b3bfa8
> bfa0: a002ff00 a0099c78 00000004 4001d000 00000001 4001d000 00000004
> 00000000
> bfc0: 00000004 4001d000 401565f8 00000004 00000001 9ef94b58 00000000
> 9ef94a44
> bfe0: 00000000 9ef9498c 4008df64 400e28b4 60000010 00000001 80936021
> 80936421
> Backtrace:
> [<a01cb710>] (platform_pm_resume_noirq+0x0/0x54) from [<a01cd9b0>]
> (pm_noirq_op+0x54/0x88)
> [<a01cd95c>] (pm_noirq_op+0x0/0x88) from [<a01ce248>]
> (dpm_power_up+0x70/0xec)
>   r5:00000010 r4:a03dd328
> [<a01ce1d8>] (dpm_power_up+0x0/0xec) from [<a01ce8c8>]
> (device_power_up+0x1c/0x20)
>   r7:a030d660 r6:00000003 r5:00000000 r4:00000010
> [<a01ce8ac>] (device_power_up+0x0/0x20) from [<a006fd10>]
> (suspend_devices_and_enter+0x10c/0x1b4)
>   r4:00000000
> [<a006fc04>] (suspend_devices_and_enter+0x0/0x1b4) from [<a006ff20>]
> (enter_state+0x130/0x1b4)
>   r6:00000003 r5:a038c7db r4:0000db96
> [<a006fdf0>] (enter_state+0x0/0x1b4) from [<a0070040>]
> (state_store+0x9c/0xc4)
>   r6:00000003 r5:00000003 r4:a038c7db
> [<a006ffa4>] (state_store+0x0/0xc4) from [<a018afe8>]
> (kobj_attr_store+0x24/0x28)
> [<a018afc4>] (kobj_attr_store+0x0/0x28) from [<a00e1da8>]
> (sysfs_write_file+0x10c/0x144)
> [<a00e1c9c>] (sysfs_write_file+0x0/0x144) from [<a0099b58>]
> (vfs_write+0xb4/0x140)
> [<a0099aa4>] (vfs_write+0x0/0x140) from [<a0099cb8>] (sys_write+0x4c/0x80)
>   r7:00000004 r6:c7ab1b20 r5:00000000 r4:00000000
> [<a0099c6c>] (sys_write+0x0/0x80) from [<a002ff00>]
> (ret_fast_syscall+0x0/0x2c)
>   r6:401565f8 r5:4001d000 r4:00000004
> Code: 74732820 20657461 0a296425 69666300 (646d635f)
> ---[ end trace 213b55b29148afbf ]---
> eth0: link down

What's pxa_reg module? Does it have proper suspend/resume functions?
> 
> >> There was also a way to keep the console enabled until the kernel didn't
> >> totally suspend, but I can't find that option now, maybe someone will be
> >> more successful. And maybe I'm mistaking this with an option to don't
> >> blank consoles until then, it's been some time since I needed that.
> >> 
> >>> P.S.: sorry if i disturb you, but you the only man in ARM linux mailing
> >>> list who posted messages about Voipac modules.
> >> 
> >> You're welcome.
> >> 
> >>> Best regards !!!