Suspend to disk panicked in -test9.

Suspend to disk panicked in -test9.

[Rob Landley]

Unfortunately, while I was writing down the panic on a piece of paper, the 
screen blanking code kicked in while I was still copying down the register 
values.  I remember that the call trace mentioned some variant of a 
write_stuff_to_disk call, but that's not that useful...

When is the last time that the screen blanking code actually accomplished 
something useful?  These days it seems to exist for the purpose of destroying 
panic call traces and annoying people.  (I seem to remember that pressing a 
key used to make it come back, but now we're forced to use the input core 
that no longer seems to be the case...)

I also seem to remember a patch floating by on the list that would make 
console screen blanking go away.  I really think console screen blanking NOT 
being enabled should be the default these days.  Or at the very least, when 
there's a panic it should get shut off.  I'll add looking into that to my 
to-do list, but will probably get to it somewhere around 2009...

Rob

Re: Suspend to disk panicked in -test9.

[Steven Cole]

On Wednesday 29 October 2003 05:57 pm, Rob Landley wrote:
> Unfortunately, while I was writing down the panic on a piece of paper, the
> screen blanking code kicked in while I was still copying down the register
> values.  I remember that the call trace mentioned some variant of a
> write_stuff_to_disk call, but that's not that useful...
>
> When is the last time that the screen blanking code actually accomplished
> something useful?  These days it seems to exist for the purpose of
> destroying panic call traces and annoying people.  (I seem to remember that
> pressing a key used to make it come back, but now we're forced to use the
> input core that no longer seems to be the case...)
>
> I also seem to remember a patch floating by on the list that would make
> console screen blanking go away.  I really think console screen blanking
> NOT being enabled should be the default these days.  Or at the very least,
> when there's a panic it should get shut off.  I'll add looking into that to
> my to-do list, but will probably get to it somewhere around 2009...
>
> Rob

In the meantime, keeping a digital camera close by when testing is a
low tech/high tech solution to this. 

Steven

Re: Suspend to disk panicked in -test9.

[Zwane Mwaikambo]

On Wed, 29 Oct 2003, Steven Cole wrote:

> In the meantime, keeping a digital camera close by when testing is a
> low tech/high tech solution to this. 
> 
> Steven

Or you could just disable screen blanking entirely; setterm(1)

Re: Suspend to disk panicked in -test9.

[Rob Landley]

On Wednesday 29 October 2003 20:35, Steven Cole wrote:
> On Wednesday 29 October 2003 05:57 pm, Rob Landley wrote:
> > Unfortunately, while I was writing down the panic on a piece of paper,
> > the screen blanking code kicked in while I was still copying down the
> > register values.  I remember that the call trace mentioned some variant
> > of a write_stuff_to_disk call, but that's not that useful...
> >
> > When is the last time that the screen blanking code actually accomplished
> > something useful?  These days it seems to exist for the purpose of
> > destroying panic call traces and annoying people.  (I seem to remember
> > that pressing a key used to make it come back, but now we're forced to
> > use the input core that no longer seems to be the case...)
> >
> > I also seem to remember a patch floating by on the list that would make
> > console screen blanking go away.  I really think console screen blanking
> > NOT being enabled should be the default these days.  Or at the very
> > least, when there's a panic it should get shut off.  I'll add looking
> > into that to my to-do list, but will probably get to it somewhere around
> > 2009...
> >
> > Rob
>
> In the meantime, keeping a digital camera close by when testing is a
> low tech/high tech solution to this.

Very few McDonalds have a digital camera behind the register to loan out.  I 
was lucky they printed out some blank cash register paper for me to write the 
panic down on.  (Ordinarily, I take notes on my laptop...)

If this was easily reproducible, I'd recreate it at home under a serial 
console.  (Well, this being a "modern" laptop with no serial port, maybe I 
could I could rig up a parallel port console or something.  But the 
principle's the same.  No, don't ask me why this thing has no serial port but 
does have a parallel port.  Ask IBM.)

Does netconsole handle panics?  (Would it work through a wireless card on an 
internal cardbus?  I also have a pcmcia 10baseT card around here somewhere I 
could plug a some cat5 into, if I can get any untangled from the big ball of 
miscelanous obsolete computer stuf in the "old parts" box.  It's been a 
couple months since I needed to reclaim anything that box, but I think I know 
where I left it.  I vaguely remember a friend asking me what an 
unrecognizable component was...  For the record, it was a 5 1/4 to 3.5 floppy 
cable adapter...)

> Steven

Rob

Re: Suspend to disk panicked in -test9.

[Rob Landley]

On Wednesday 29 October 2003 21:24, Zwane Mwaikambo wrote:
> On Wed, 29 Oct 2003, Steven Cole wrote:
> > In the meantime, keeping a digital camera close by when testing is a
> > low tech/high tech solution to this.
> >
> > Steven
>
> Or you could just disable screen blanking entirely; setterm(1)

Ah, cool.  Thanks.  (It's now in my init script.)

Do you know if this affects x's screensaver timeout?  (Probably not, but I'm 
not sure how to look it up.  Maybe X gets the value from console 7, somehow?  
Then again, if this was a per-console thing setting it in the initscripts 
when the console is /dev/console rather than virtual wouldn't necessarily 
work...  Huh.)

Rob

Re: Suspend to disk panicked in -test9.

[Zwane Mwaikambo]

On Wed, 29 Oct 2003, Rob Landley wrote:

> Ah, cool.  Thanks.  (It's now in my init script.)
> 
> Do you know if this affects x's screensaver timeout?  (Probably not, but I'm 
> not sure how to look it up.  Maybe X gets the value from console 7, somehow?  
> Then again, if this was a per-console thing setting it in the initscripts 
> when the console is /dev/console rather than virtual wouldn't necessarily 
> work...  Huh.)

You can set xscreensaver's timeout with xscreensaver-demo.

Re: Suspend to disk panicked in -test9.

[Felipe Alfaro Solana]

On Thu, 2003-10-30 at 01:57, Rob Landley wrote:

> When is the last time that the screen blanking code actually accomplished 
> something useful?  These days it seems to exist for the purpose of destroying 
> panic call traces and annoying people.  (I seem to remember that pressing a 
> key used to make it come back, but now we're forced to use the input core 
> that no longer seems to be the case...)

I always do set "blankinterval" to 0 in drivers/char/vt.c before
compiling :-)

Re: Suspend to disk panicked in -test9.

[bill davidsen]

In article <200310291857.40722.rob@landley.net>,
Rob Landley  <rob@landley.net> wrote:
| Unfortunately, while I was writing down the panic on a piece of paper, the 
| screen blanking code kicked in while I was still copying down the register 
| values.  I remember that the call trace mentioned some variant of a 
| write_stuff_to_disk call, but that's not that useful...
| 
| When is the last time that the screen blanking code actually accomplished 
| something useful?  These days it seems to exist for the purpose of destroying 
| panic call traces and annoying people.  (I seem to remember that pressing a 
| key used to make it come back, but now we're forced to use the input core 
| that no longer seems to be the case...)
| 
| I also seem to remember a patch floating by on the list that would make 
| console screen blanking go away.  I really think console screen blanking NOT 
| being enabled should be the default these days.  Or at the very least, when 
| there's a panic it should get shut off.  I'll add looking into that to my 
| to-do list, but will probably get to it somewhere around 2009...

Or people who want it that way could put the setterm call in their
rc.local, of course. No patches required and the rest of the world
doesn't have to turn it on.
-- 
bill davidsen <davidsen@tmr.com>
  CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.

Re: Suspend to disk panicked in -test9.

[bill davidsen]

In article <200310292333.06470.rob@landley.net>,
Rob Landley  <rob@landley.net> wrote:

| > In the meantime, keeping a digital camera close by when testing is a
| > low tech/high tech solution to this.
| 
| Very few McDonalds have a digital camera behind the register to loan out.  I 
| was lucky they printed out some blank cash register paper for me to write the 
| panic down on.  (Ordinarily, I take notes on my laptop...)

If you don't carry a camera then bring a spare laptop ;-)

| If this was easily reproducible, I'd recreate it at home under a serial 
| console.  (Well, this being a "modern" laptop with no serial port, maybe I 
| could I could rig up a parallel port console or something.  But the 
| principle's the same.  No, don't ask me why this thing has no serial port but 
| does have a parallel port.  Ask IBM.)

I assume because serial mice and printers are rare, parallel printers
are not. That said, I would not buy a laptop w/o serial, too useful for
other things.
-- 
bill davidsen <davidsen@tmr.com>
  CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.

Re: Suspend to disk panicked in -test9.

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 680 bytes --]

On Fri, 07 Nov 2003 15:14:06 GMT, davidsen@tmr.com (bill davidsen)  said:

> Or people who want it that way could put the setterm call in their
> rc.local, of course. No patches required and the rest of the world
> doesn't have to turn it on.

Of course, this means that you have to know beforehand that your
machine is going to panic.  And of course, since the screen is  blanked,
you don't get a clue what caused it so you don't get any information
on how to reproduce it....  We're supposed to be designing a quality
operating system, not something that's a set-up for a Dilbert strip.

Rob Landley is right - there should be a patch to make it go away
if the kernel panics.  

[-- Attachment #2: Type: application/pgp-signature, Size: 226 bytes --]

Re: Suspend to disk panicked in -test9.

[Bill Davidsen]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 997 bytes --]

On Fri, 7 Nov 2003 Valdis.Kletnieks@vt.edu wrote:

> On Fri, 07 Nov 2003 15:14:06 GMT, davidsen@tmr.com (bill davidsen)  said:
> 
> > Or people who want it that way could put the setterm call in their
> > rc.local, of course. No patches required and the rest of the world
> > doesn't have to turn it on.
> 
> Of course, this means that you have to know beforehand that your
> machine is going to panic.

Unless it panics in the boot and never loads the first startup file you
can execute the blank disable there.

> Rob Landley is right - there should be a patch to make it go away
> if the kernel panics.  

There should be a lot of things, including writing the dump to a disk
partition like AIX, Solaris, etc, etc. That never got in, either. I'd
rather see the blank disable as a configurable boot time option, so people
who don't want it don't get it, and vice-versa. 

-- 
bill davidsen <davidsen@tmr.com>
  CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.

[-- Attachment #2: Type: APPLICATION/PGP-SIGNATURE, Size: 226 bytes --]