Hi!
After getting _good_ power supply, I was able to untar and rm ltp tree
100 times so far. 2.6.38, ext3. (So metan no, I do not see any weird
ext3 problems).
Another strange thing is... I see lots of duplicated keystrokes --
like type "pavel" but machine understands "paavel". But it only seems
to happen when I'm _not_ trying to debug it?!
(
So... just because something needs 5V does not mean you should charge
it from USB.
Yes, I'm talking about zaurus. I noticed that it crashed rather
quickly when connected to USB power. (For example in tar/rm test).
So I took multimeter, and there was 4.15..4.5V on USB :-(.
With power from USB extension cable, it crashes in <5 minutes. It
seems to toggle charger a lot, as voltage on the cable is only
4.2..4.5V. I guess it is software problem in the charging code.
)
Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
Hi!
> After getting _good_ power supply, I was able to untar and rm ltp tree
> 100 times so far. 2.6.38, ext3. (So metan no, I do not see any weird
> ext3 problems).
Hmm, I'm using original adapter, but who knows if it's okay to use...
> Another strange thing is... I see lots of duplicated keystrokes --
> like type "pavel" but machine understands "paavel". But it only seems
> to happen when I'm _not_ trying to debug it?!
That seems to be side effect of machine under load (it has gpio keyboard).
> (
> So... just because something needs 5V does not mean you should charge
> it from USB.
>
> Yes, I'm talking about zaurus. I noticed that it crashed rather
> quickly when connected to USB power. (For example in tar/rm test).
> So I took multimeter, and there was 4.15..4.5V on USB :-(.
>
> With power from USB extension cable, it crashes in <5 minutes. It
> seems to toggle charger a lot, as voltage on the cable is only
> 4.2..4.5V. I guess it is software problem in the charging code.
> )
The zaurus actually is AFAIK unable to run from battery when connected to the
external power, so weak power line may explain some of the strange behaviour.
--
metan
On Friday, April 15, 2011 11:38:17 PM Cyril Hrubis wrote:
> Hi!
>
> > After getting _good_ power supply, I was able to untar and rm ltp tree
> > 100 times so far. 2.6.38, ext3. (So metan no, I do not see any weird
> > ext3 problems).
>
> Hmm, I'm using original adapter, but who knows if it's okay to use...
I can lend you Eric's, but promise you won't break it or do anything weird to it
;-)
>
> > Another strange thing is... I see lots of duplicated keystrokes --
> > like type "pavel" but machine understands "paavel". But it only seems
> > to happen when I'm _not_ trying to debug it?!
>
> That seems to be side effect of machine under load (it has gpio keyboard).
Yea ... it's that polled para-keyboard I guess.
>
> > (
> > So... just because something needs 5V does not mean you should charge
> > it from USB.
> >
> > Yes, I'm talking about zaurus. I noticed that it crashed rather
> > quickly when connected to USB power. (For example in tar/rm test).
> > So I took multimeter, and there was 4.15..4.5V on USB :-(.
> >
> > With power from USB extension cable, it crashes in <5 minutes. It
> > seems to toggle charger a lot, as voltage on the cable is only
> > 4.2..4.5V. I guess it is software problem in the charging code.
> > )
>
> The zaurus actually is AFAIK unable to run from battery when connected to
> the external power, so weak power line may explain some of the strange
> behaviour.
As for the charging code, do we even have anything even remotely resembling that
?
Hi!
> > After getting _good_ power supply, I was able to untar and rm ltp tree
> > 100 times so far. 2.6.38, ext3. (So metan no, I do not see any weird
> > ext3 problems).
>
> Hmm, I'm using original adapter, but who knows if it's okay to use...
It probably is...
> > Another strange thing is... I see lots of duplicated keystrokes --
> > like type "pavel" but machine understands "paavel". But it only seems
> > to happen when I'm _not_ trying to debug it?!
>
> That seems to be side effect of machine under load (it has gpio keyboard).
ouch. original corgi_kbd driver used timers, new one uses threads --
more affected by scheduling. I'm not even sure if the worker thread it
uses runs at elevated priority...
> > With power from USB extension cable, it crashes in <5 minutes. It
> > seems to toggle charger a lot, as voltage on the cable is only
> > 4.2..4.5V. I guess it is software problem in the charging code.
> > )
>
> The zaurus actually is AFAIK unable to run from battery when connected to the
> external power, so weak power line may explain some of the strange behaviour.
?? IIRC Stanislav told me that hardware will always run off whatever
is available...?
Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
On Saturday, April 16, 2011 09:41:34 AM Pavel Machek wrote:
> Hi!
>
> > > After getting _good_ power supply, I was able to untar and rm ltp tree
> > > 100 times so far. 2.6.38, ext3. (So metan no, I do not see any weird
> > > ext3 problems).
> >
> > Hmm, I'm using original adapter, but who knows if it's okay to use...
>
> It probably is...
>
> > > Another strange thing is... I see lots of duplicated keystrokes --
> > > like type "pavel" but machine understands "paavel". But it only seems
> > > to happen when I'm _not_ trying to debug it?!
> >
> > That seems to be side effect of machine under load (it has gpio
> > keyboard).
>
> ouch. original corgi_kbd driver used timers, new one uses threads --
> more affected by scheduling. I'm not even sure if the worker thread it
> uses runs at elevated priority...
>
> > > With power from USB extension cable, it crashes in <5 minutes. It
> > > seems to toggle charger a lot, as voltage on the cable is only
> > > 4.2..4.5V. I guess it is software problem in the charging code.
> > > )
> >
> > The zaurus actually is AFAIK unable to run from battery when connected to
> > the external power, so weak power line may explain some of the strange
> > behaviour.
>
> ?? IIRC Stanislav told me that hardware will always run off whatever
> is available...?
> Pavel
Backtrace from QEMU ... I think this is what Cyril observed. But heck, it
happens in QEMU too ;-)
Unable to handle kernel NULL pointer dereference at virtual address 00000000
pgd = c0004000
[00000000] *pgd=00000000
Internal error: Oops: 17 [#1] PREEMPT
last sysfs file:
CPU: 0 Not tainted (2.6.39-rc5+ #81)
PC is at complete+0x30/0x90
LR is at giveback+0x134/0x150
pc : [<c00eca34>] lr : [<c023c984>] psr: a0000193
sp : c782ff08 ip : c782fec8 fp : c782ff1c
r10: c02b78c4 r9 : c7828720 r8 : c880c000
r7 : c7865828 r6 : c7865894 r5 : c7865828 r4 : a0000113
r3 : c782e000 r2 : 00000102 r1 : f2e00000 r0 : 00000000
Flags: NzCv IRQs off FIQs on Mode SVC_32 ISA ARM Segment kernel
Control: 00007977 Table: a5774000 DAC: 00000017
Process ksoftirqd/0 (pid: 3, stack limit = 0xc782e278)
Stack: (0xc782ff08 to 0xc7830000)
ff00: 60000113 c010497c c786f4d8 c786f4d8 00000000 c023c984
ff20: c786f4d8 c7868760 c7865878 c023cc28 c786f4d8 60000113 00000000 00000000
ff40: c02e2980 c02b7a3c 00000000 00000000 00000101 00000008 c02b78c4 c00f4bd8
ff60: c782e000 00000001 00000018 c02e2980 00000101 c00f5128 c7821660 c782e000
ff80: c782ffac 00000006 00000000 c782e000 c02e2980 00000000 00000001 00000000
ffa0: 00000000 00000000 00000000 c00f524c c7823f60 00000000 c782ffd4 c00f51bc
ffc0: 00000000 c0108d4c c00de570 00000000 00000000 00000000 c782ffd8 c782ffd8
ffe0: 00000000 c7823f60 c0108cc8 c00de570 00000013 c00de570 00000000 00000000
[<c00eca34>] (complete+0x30/0x90) from [<c023c984>] (giveback+0x134/0x150)
[<c023c984>] (giveback+0x134/0x150) from [<c023cc28>]
(pump_transfers+0x130/0x6f0)
[<c023cc28>] (pump_transfers+0x130/0x6f0) from [<c00f4bd8>]
(tasklet_action+0x84/0xdc)
[<c00f4bd8>] (tasklet_action+0x84/0xdc) from [<c00f5128>]
(__do_softirq+0x74/0x108)
[<c00f5128>] (__do_softirq+0x74/0x108) from [<c00f524c>]
(run_ksoftirqd+0x90/0x180)
[<c00f524c>] (run_ksoftirqd+0x90/0x180) from [<c0108d4c>] (kthread+0x84/0x8c)
[<c0108d4c>] (kthread+0x84/0x8c) from [<c00de570>] (kernel_thread_exit+0x0/0x8)
Code: e3c3303f e5932004 e2822001 e5832004 (e5903000)
---[ end trace 989a8f1146a65d49 ]---
Kernel panic - not syncing: Fatal exception in interrupt
[<c00e2200>] (unwind_backtrace+0x0/0xf0) from [<c024e13c>] (panic+0x70/0x198)
[<c024e13c>] (panic+0x70/0x198) from [<c00e0c1c>] (die+0x2c8/0x31c)
[<c00e0c1c>] (die+0x2c8/0x31c) from [<c00e3350>] (__do_kernel_fault+0x64/0x88)
[<c00e3350>] (__do_kernel_fault+0x64/0x88) from [<c00e3524>]
(do_page_fault+0x1b0/0x1c4)
[<c00e3524>] (do_page_fault+0x1b0/0x1c4) from [<c00d72cc>]
(do_DataAbort+0x34/0x94)
[<c00d72cc>] (do_DataAbort+0x34/0x94) from [<c00dd1ac>] (__dabt_svc+0x4c/0x60)
Exception stack(0xc782fec0 to 0xc782ff08)
fec0: 00000000 f2e00000 00000102 c782e000 a0000113 c7865828 c7865894 c7865828
fee0: c880c000 c7828720 c02b78c4 c782ff1c c782fec8 c782ff08 c023c984 c00eca34
ff00: a0000193 ffffffff
[<c00dd1ac>] (__dabt_svc+0x4c/0x60) from [<c00eca34>] (complete+0x30/0x90)
[<c00eca34>] (complete+0x30/0x90) from [<c023c984>] (giveback+0x134/0x150)
[<c023c984>] (giveback+0x134/0x150) from [<c023cc28>]
(pump_transfers+0x130/0x6f0)
[<c023cc28>] (pump_transfers+0x130/0x6f0) from [<c00f4bd8>]
(tasklet_action+0x84/0xdc)
[<c00f4bd8>] (tasklet_action+0x84/0xdc) from [<c00f5128>]
(__do_softirq+0x74/0x108)
[<c00f5128>] (__do_softirq+0x74/0x108) from [<c00f524c>]
(run_ksoftirqd+0x90/0x180)
[<c00f524c>] (run_ksoftirqd+0x90/0x180) from [<c0108d4c>] (kthread+0x84/0x8c)
[<c0108d4c>] (kthread+0x84/0x8c) from [<c00de570>] (kernel_thread_exit+0x0/0x8)
Hi!
> Backtrace from QEMU ... I think this is what Cyril observed. But heck, it
> happens in QEMU too ;-)
Yes, that is pretty much it.
It happens randomly, eg. adding printk() line into kernel code (somewhere in
mmc stack). Or sometimes crashes in the pccard initalization when two cards are
inserted at the boot time.
--
metan
> Hi!
> > Backtrace from QEMU ... I think this is what Cyril observed. But heck,
> > it happens in QEMU too ;-)
>
> Yes, that is pretty much it.
>
> It happens randomly, eg. adding printk() line into kernel code
> (somewhere in mmc stack). Or sometimes crashes in the pccard
> initalization when two cards are inserted at the boot time.
>
It happens randomly in qemu too.
> --
> metan