swsusp problem

swsusp problem

[curious]

hello.
i write because swsuspend don't work for me.
i try to echo disk > /sys/power/state
and just nothing happens, i have blinking cursor and machine freezes.

when i enabled debug i got :
stopping tasks: ========|
Shrinking memory... done (2684 pages freed)
swsusp: Need to copy 1454 pages
swsusp: critical section/: done (1454 pages copied)

.... and machine just sits there , doing nothing.
after reboot it boots like usual.

machine is Ts30M Viglen Dossier 486 SM
kernel is 2.6.18-rc5
here is config : http://zjeby.dyndns.org:8242/viglen.config



-- 

Re: swsusp problem

[Rafael J. Wysocki]

Hi,

On Sunday, 10 September 2006 02:13, curious wrote:
> hello.
> i write because swsuspend don't work for me.
> i try to echo disk > /sys/power/state
> and just nothing happens, i have blinking cursor and machine freezes.
> 
> when i enabled debug i got :
> stopping tasks: ========|
> Shrinking memory... done (2684 pages freed)
> swsusp: Need to copy 1454 pages
> swsusp: critical section/: done (1454 pages copied)
> 
> .... and machine just sits there , doing nothing.
> after reboot it boots like usual.
> 
> machine is Ts30M Viglen Dossier 486 SM
> kernel is 2.6.18-rc5
> here is config : http://zjeby.dyndns.org:8242/viglen.config

Could you boot the kernel with the init=/bin/bash command line argument
and do the following:

# mount /proc
# mount /sys
# echo 8 > /proc/sys/kernel/printk
# swapon -a
# echo disk > /sys/power/state

and see what happens?

Rafael


-- 
You never change things by fighting the existing reality.
		R. Buckminster Fuller

Re: swsusp problem

[curious]

On Sun, 10 Sep 2006, Rafael J. Wysocki wrote:

> Hi,
>
> On Sunday, 10 September 2006 02:13, curious wrote:
>> hello.
>> i write because swsuspend don't work for me.
>> i try to echo disk > /sys/power/state
>> and just nothing happens, i have blinking cursor and machine freezes.
>>
>> when i enabled debug i got :
>> stopping tasks: ========|
>> Shrinking memory... done (2684 pages freed)
>> swsusp: Need to copy 1454 pages
>> swsusp: critical section/: done (1454 pages copied)
>>
>> .... and machine just sits there , doing nothing.
>> after reboot it boots like usual.
>>
>> machine is Ts30M Viglen Dossier 486 SM
>> kernel is 2.6.18-rc5
>> here is config : http://zjeby.dyndns.org:8242/viglen.config
>
> Could you boot the kernel with the init=/bin/bash command line argument
> and do the following:
>
> # mount /proc
> # mount /sys
> # echo 8 > /proc/sys/kernel/printk
> # swapon -a
> # echo disk > /sys/power/state
>
> and see what happens?

same thing , except page count is different ofcourse.

Re: swsusp problem

[Pavel Machek]

On Sun 2006-09-10 21:24:15, curious wrote:
> 
> On Sun, 10 Sep 2006, Rafael J. Wysocki wrote:
> 
> >Hi,
> >
> >On Sunday, 10 September 2006 02:13, curious wrote:
> >>hello.
> >>i write because swsuspend don't work for me.
> >>i try to echo disk > /sys/power/state
> >>and just nothing happens, i have blinking cursor and machine freezes.
> >>
> >>when i enabled debug i got :
> >>stopping tasks: ========|
> >>Shrinking memory... done (2684 pages freed)
> >>swsusp: Need to copy 1454 pages
> >>swsusp: critical section/: done (1454 pages copied)
> >>
> >>.... and machine just sits there , doing nothing.
> >>after reboot it boots like usual.
> >>
> >>machine is Ts30M Viglen Dossier 486 SM
> >>kernel is 2.6.18-rc5
> >>here is config : http://zjeby.dyndns.org:8242/viglen.config
> >
> >Could you boot the kernel with the init=/bin/bash command line argument
> >and do the following:
> >
> ># mount /proc
> ># mount /sys
> ># echo 8 > /proc/sys/kernel/printk
> ># swapon -a
> ># echo disk > /sys/power/state
> >
> >and see what happens?
> 
> same thing , except page count is different ofcourse.

What kind of machine is that? What cpu? Really 486?
								Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: swsusp problem

[Piotr Gluszenia Slawinski]

> What kind of machine is that? What cpu? Really 486?

yes, it's viglen dossier 486 based laptop .
cpuinfo detects it as 'GenuineIntel'
cpu family : 4
model : 3
model name : 486 DX/2
stepping : 6
cache_size : 0k
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 1
wp : yes
flags : fpu vme pse
bogomips : 31.55

Re: swsusp problem

[Pavel Machek]

Hi!

> >What kind of machine is that? What cpu? Really 486?
> 
> yes, it's viglen dossier 486 based laptop .

Forget it, this thing does not support 4MB pages, and swsusp currently
needs them. (We'll need to fix that, but not now -- fix is port of
page-table handling code from x86-64).

It should die with

        if (!cpu_has_pse) {
                printk(KERN_ERR "PSE is required for swsusp.\n");
                return -EPERM;
        }

...can you investigate why it does not?

							Pavel

> cpuinfo detects it as 'GenuineIntel'
> cpu family : 4
> model : 3
> model name : 486 DX/2
> stepping : 6
> cache_size : 0k
> fdiv_bug : no
> hlt_bug : no
> f00f_bug : no
> coma_bug : no
> fpu : yes
> fpu_exception : yes
> cpuid level : 1
> wp : yes
> flags : fpu vme pse
> bogomips : 31.55

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: swsusp problem

[Piotr Gluszenia Slawinski]

> It should die with
5B>
>        if (!cpu_has_pse) {
>                printk(KERN_ERR "PSE is required for swsusp.\n");
>                return -EPERM;
>        }
>
> ...can you investigate why it does not?

obviously because i disabled BUG() support :o

thanx for help

-- 

Re: swsusp problem

[Dmitry Torokhov]

On Sunday 10 September 2006 15:49, Pavel Machek wrote:
> Hi!
> 
> > >What kind of machine is that? What cpu? Really 486?
> > 
> > yes, it's viglen dossier 486 based laptop .
> 
> Forget it, this thing does not support 4MB pages, and swsusp currently
> needs them. (We'll need to fix that, but not now -- fix is port of
> page-table handling code from x86-64).
> 
> It should die with
> 
>         if (!cpu_has_pse) {
>                 printk(KERN_ERR "PSE is required for swsusp.\n");
>                 return -EPERM;
>         }
> 
> ...can you investigate why it does not?
> 

Probably because of this:

> > flags : fpu vme pse
                    ^^^

> > bogomips : 31.55
> 

-- 
Dmitry

Re: swsusp problem

[Pavel Machek]

Hi!

> > > >What kind of machine is that? What cpu? Really 486?
> > > 
> > > yes, it's viglen dossier 486 based laptop .
> > 
> > Forget it, this thing does not support 4MB pages, and swsusp currently
> > needs them. (We'll need to fix that, but not now -- fix is port of
> > page-table handling code from x86-64).
> > 
> > It should die with
> > 
> >         if (!cpu_has_pse) {
> >                 printk(KERN_ERR "PSE is required for swsusp.\n");
> >                 return -EPERM;
> >         }
> > 
> > ...can you investigate why it does not?
> > 
> 
> Probably because of this:

Oops, aha, really? I thought only pentium+ supported 4MB tables?

Does it also support cr4? Anyway, someone interested in history needs
to play with this one ;-).

> > > flags : fpu vme pse
>                     ^^^

								Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

thinkpad 360Cs keyboard problem

[Piotr Gluszenia Slawinski]

well, certainly 2.6.18 issue...
kernel boots up fine, but keyboard is totally messed up,
and locks up after some tries of use.

on 2.4.20 kernel keyboard works just fine

-- 

Re: thinkpad 360Cs keyboard problem

[Dmitry Torokhov]

On 9/11/06, Piotr Gluszenia Slawinski <curious@zjeby.dyndns.org> wrote:
> well, certainly 2.6.18 issue...

Are you saying that it works on 2.6.17 and is broken on 2.6.18?

> kernel boots up fine, but keyboard is totally messed up,
> and locks up after some tries of use.

Could you try describing the exact issues with the keyboard? Missing
keypresses, wrong keys reported, etc?

Thanks.

-- 
Dmitry

Re: thinkpad 360Cs keyboard problem

[Piotr Gluszenia Slawinski]

On Tue, 12 Sep 2006, Dmitry Torokhov wrote:

> On 9/11/06, Piotr Gluszenia Slawinski <curious@zjeby.dyndns.org> wrote:
>>  well, certainly 2.6.18 issue...
>
> Are you saying that it works on 2.6.17 and is broken on 2.6.18?

no, it works on 2.4.20. i didn't tried yet with other 2.6.x versions.

btw. now i compiled 2.4.33.3 using gcc 3.4.6 ,
and even though i used 486 in cpu type resulting binary requires
TSC+ from cpu :o (now recompiling with gcc-2.95.3 , and
config_notsc and it seems to boot up)

however weird behaviour aswell, because stops just after init...
no matter wheter i use init=/bin/bash or regular init.
it reports mounting root fs (ext2), then freeing unused kernel
mem (52k) and stops .

when i press keys i see output like it should be , so at least
keyboard work :)

with 2.6.18-rc5 it boots just fine, but ofcourse keyboard is
broken.

>>  kernel boots up fine, but keyboard is totally messed up,
>>  and locks up after some tries of use.
>
> Could you try describing the exact issues with the keyboard? Missing
> keypresses, wrong keys reported, etc?

with prink enabled it prints series of 'unknown scancode'
and keys are randomly messed up, and it changes, so like pressing b
results with n, then space, then nothing at all.
after some tries keyboard locks up completely.

so wrong keys reported

Re: thinkpad 360Cs keyboard problem

[Dmitry Torokhov]

On 9/12/06, Piotr Gluszenia Slawinski <curious@zjeby.dyndns.org> wrote:
> On Tue, 12 Sep 2006, Dmitry Torokhov wrote:
>
> > On 9/11/06, Piotr Gluszenia Slawinski <curious@zjeby.dyndns.org> wrote:
>
> >>  kernel boots up fine, but keyboard is totally messed up,
> >>  and locks up after some tries of use.
> >
> > Could you try describing the exact issues with the keyboard? Missing
> > keypresses, wrong keys reported, etc?
>
> with prink enabled it prints series of 'unknown scancode'
> and keys are randomly messed up, and it changes, so like pressing b
> results with n, then space, then nothing at all.
> after some tries keyboard locks up completely.
>

Are you loading a custom keymap by any chance? Could I please see
dmesg with "i8042.debug log_buf_len=131072"?

-- 
Dmitry

Re: thinkpad 360Cs keyboard problem

[Piotr Gluszenia Slawinski]

On Tue, 12 Sep 2006, Dmitry Torokhov wrote:

> On 9/12/06, Piotr Gluszenia Slawinski <curious@zjeby.dyndns.org> wrote:
>>  On Tue, 12 Sep 2006, Dmitry Torokhov wrote:
>> 
>> >  On 9/11/06, Piotr Gluszenia Slawinski <curious@zjeby.dyndns.org> wrote:
>> 
>> > >   kernel boots up fine, but keyboard is totally messed up,
>> > >   and locks up after some tries of use.
>> > 
>> >  Could you try describing the exact issues with the keyboard? Missing
>> >  keypresses, wrong keys reported, etc?
>>
>>  with prink enabled it prints series of 'unknown scancode'
>>  and keys are randomly messed up, and it changes, so like pressing b
>>  results with n, then space, then nothing at all.
>>  after some tries keyboard locks up completely.
>> 
>
> Are you loading a custom keymap by any chance? Could I please see
> dmesg with "i8042.debug log_buf_len=131072"?

no custom keymaps . init=/bin/bash :d
uhm, i don't get what you mean by this dmesg syntax :o
i should probably attach serial conole and send you whole output,
as now (as keyboard is unuseable) i can't scroll screen.

btw. serio: i8042 KBD port at 0x60,0x64 irq 1
is found.
also
input: AT Raw Set 2 keyboard as /class/input/input1
is reported

ah, i use gcc-4.1.1 to compile kernel .


-------------------- stuff below might be result of my ignorance

i just checked 2.2.27 kernel. it behaves quite identical to 2.4.33.3,
just after 'freeing unused kernel memory' it touches disk , and just
sits there. kepresses are echoed, and keyboard output is 
ok. i can reboot it by ctrl-alt-delete too .
when i boot it with 'fosh' shell commandline appears, but trying
to use any tool like bash, ls , etc results with signal 11.
tools are statically compiled against glibc-2.4 (compiled for 486 cpu
on gentoo)
busybox works fine.

on 2.4.33.3 kernel Xvesa (from kdrive package) complains
'set_thread_area failed when setting up thread-local storage'

cardmgr works fine aswell...(though was linked against different glibc)
  i guess there is some problem with math emulation and glibc,
as on 2.6.18 everything works just fine.

i recall few years ago i tried to make 386SX work on 2.4.20 , and 
there were problems with math emulation (some apps worked, some not)

Re: thinkpad 360Cs keyboard problem

[Dmitry Torokhov]

On Tuesday 12 September 2006 17:21, Piotr Gluszenia Slawinski wrote:
> On Tue, 12 Sep 2006, Dmitry Torokhov wrote:
> 
> > On 9/12/06, Piotr Gluszenia Slawinski <curious@zjeby.dyndns.org> wrote:
> >>  On Tue, 12 Sep 2006, Dmitry Torokhov wrote:
> >> 
> >> >  On 9/11/06, Piotr Gluszenia Slawinski <curious@zjeby.dyndns.org> wrote:
> >> 
> >> > >   kernel boots up fine, but keyboard is totally messed up,
> >> > >   and locks up after some tries of use.
> >> > 
> >> >  Could you try describing the exact issues with the keyboard? Missing
> >> >  keypresses, wrong keys reported, etc?
> >>
> >>  with prink enabled it prints series of 'unknown scancode'
> >>  and keys are randomly messed up, and it changes, so like pressing b
> >>  results with n, then space, then nothing at all.
> >>  after some tries keyboard locks up completely.
> >> 
> >
> > Are you loading a custom keymap by any chance? Could I please see
> > dmesg with "i8042.debug log_buf_len=131072"?
> 
> no custom keymaps . init=/bin/bash :d
> uhm, i don't get what you mean by this dmesg syntax :o

These were kernel boot options. If you let the system boot normally
so that syslog is running I'd expect debug messages end somewhere
(like in /var/log/messages) so you could reboot and send me that file.

> i should probably attach serial conole and send you whole output,
> as now (as keyboard is unuseable) i can't scroll screen.
> 
> btw. serio: i8042 KBD port at 0x60,0x64 irq 1
> is found.
> also
> input: AT Raw Set 2 keyboard as /class/input/input1
> is reported
> 

It could be that it lies about being in raw mode and actually is in
translated mode.

-- 
Dmitry

Re: swsusp problem

[Pavel Machek]

On Sun 10-09-06 02:13:13, curious wrote:

Can we get some real name? I guess my spam filters ate this...


> i write because swsuspend don't work for me.
> i try to echo disk > /sys/power/state
> and just nothing happens, i have blinking cursor and 
> machine freezes.
> 
> when i enabled debug i got :
> stopping tasks: ========|
> Shrinking memory... done (2684 pages freed)
> swsusp: Need to copy 1454 pages
> swsusp: critical section/: done (1454 pages copied)
> 
> .... and machine just sits there , doing nothing.
> after reboot it boots like usual.

> 
> machine is Ts30M Viglen Dossier 486 SM

intel 486 cpu?

Can you try with minimal config?
							Pavel

-- 
Thanks for all the (sleeping) penguins.

Re: swsusp problem

[Piotr Gluszenia Slawinski]

> Can we get some real name? I guess my spam filters ate this...
yes, excuse me (fixed)

> intel 486 cpu?
uhm. yes.
i see cpudetection fail... but as i checked just on
this 486 and pentium 133 - on both 'cache_size' field
in cpuinfo shows 0k .

on pentium2 machines (pentium II deschutes) cache is detected ok .


> Can you try with minimal config?
what do you mean by minimal config ?

--