[parisc-linux] Watchdog on parisc?

[parisc-linux] Watchdog on parisc?

[Jan-Benedict Glaw]

[-- Attachment #1: Type: text/plain, Size: 898 bytes --]

Hi!

I'm currently playing a bit with my B132L and 715. They're working quite
nice, except that kernels aren't too stable (gcc-3.2, I'll do some more
test compiles with gcc-3.0, too).

I've looked through the docs available if a watchdog is mentioned. In
./docs/platforms/b1xx-service.pdf there's a watchdog timeout error
declared (during selftest) so I *think* there is a watchdog, isn't
there? I've not yet seen a driver for it, nor have I seen any
documentation. Is there a watchdog (and possibly some docu or an hpux
vmunix to disassemble) so I could possibly write something better than
softdog.c?

MfG, JBG

-- 
   Jan-Benedict Glaw       jbglaw@lug-owl.de    . +49-172-7608481
   "Eine Freie Meinung in  einem Freien Kopf    | Gegen Zensur
    fuer einen Freien Staat voll Freier Bürger" | im Internet!
   Shell Script APT-Proxy: http://lug-owl.de/~jbglaw/software/ap2/

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [parisc-linux] Watchdog on parisc?

[Grant Grundler]

On Fri, Feb 14, 2003 at 07:11:24PM +0100, Jan-Benedict Glaw wrote:
> Hi!
> 
> I'm currently playing a bit with my B132L and 715. They're working quite
> nice, except that kernels aren't too stable (gcc-3.2, I'll do some more
> test compiles with gcc-3.0, too).

gcc-3.2.1 has a few bugs that have bitten us already (search mail
archive on "pty.c"). I'm not sure if it's better to stick with
gcc 3.0 or grab something that's bleeding edge (gcc 3.3.x).

> I've looked through the docs available if a watchdog is mentioned. In
> ./docs/platforms/b1xx-service.pdf there's a watchdog timeout error
> declared (during selftest) so I *think* there is a watchdog, isn't
> there?

I'm not sure what this is refering to. The CPU has some HW checking
that is a sort of watchdog but it's CPU internal and only the firmware
touches those.

> I've not yet seen a driver for it, nor have I seen any
> documentation. Is there a watchdog (and possibly some docu or an hpux
> vmunix to disassemble) so I could possibly write something better than
> softdog.c?

not if it's what I'm thinking of (transaction timeout, ie causes an
HPMC). This is a built-in HW feature. 

grant

Re: [parisc-linux] Watchdog on parisc?

[Jan-Benedict Glaw]

[-- Attachment #1: Type: text/plain, Size: 2319 bytes --]

On Sun, 2003-02-16 01:31:10 -0700, Grant Grundler <grundler@dsl2.external.hp.com>
wrote in message <20030216083110.GC18888@dsl2.external.hp.com>:
> On Fri, Feb 14, 2003 at 07:11:24PM +0100, Jan-Benedict Glaw wrote:
> > Hi!
> > 
> > I'm currently playing a bit with my B132L and 715. They're working quite
> > nice, except that kernels aren't too stable (gcc-3.2, I'll do some more
> > test compiles with gcc-3.0, too).
> 
> gcc-3.2.1 has a few bugs that have bitten us already (search mail
> archive on "pty.c"). I'm not sure if it's better to stick with
> gcc 3.0 or grab something that's bleeding edge (gcc 3.3.x).

Yeah, I know. I followed that thread. gcc-3.0 seems to produce a more
stable kernel, but even there are some gotchas (eg. strace'ing programs
often leads to stuck userland while I still can ping the box or complete
the three-way-handshake of a starting TCP connection...).

I'd possibly spend some time in invensting this...

> > I've looked through the docs available if a watchdog is mentioned. In
> > ./docs/platforms/b1xx-service.pdf there's a watchdog timeout error
> > declared (during selftest) so I *think* there is a watchdog, isn't
> > there?
> 
> I'm not sure what this is refering to. The CPU has some HW checking
> that is a sort of watchdog but it's CPU internal and only the firmware
> touches those.

So there's nothing what's like a watchdog as you know it from some
machines (SGI Indys for example) or as add-on cards in machines (PCI/ISA
cards).

> > I've not yet seen a driver for it, nor have I seen any
> > documentation. Is there a watchdog (and possibly some docu or an hpux
> > vmunix to disassemble) so I could possibly write something better than
> > softdog.c?
> 
> not if it's what I'm thinking of (transaction timeout, ie causes an
> HPMC). This is a built-in HW feature. 

Hmmm... Triggering a HPMC would be a neat feature. So we could think
about a way to initiate something that, if not handled, would trigger a
HPMC some 60 seconds later:-?

MfG, JBG

-- 
   Jan-Benedict Glaw       jbglaw@lug-owl.de    . +49-172-7608481
   "Eine Freie Meinung in  einem Freien Kopf    | Gegen Zensur
    fuer einen Freien Staat voll Freier Bürger" | im Internet!
   Shell Script APT-Proxy: http://lug-owl.de/~jbglaw/software/ap2/

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [parisc-linux] Watchdog on parisc?

[Grant Grundler]

On Sun, Feb 16, 2003 at 10:42:35AM +0100, Jan-Benedict Glaw wrote:
> So there's nothing what's like a watchdog as you know it from some
> machines (SGI Indys for example) or as add-on cards in machines (PCI/ISA
> cards).

Offhand, I think all the HP9000/8xx servers have either built-in
or add-on "management ports" for remote console. And that always
includes ability to reset or cycle power on the machine (or generate TOC).

> > not if it's what I'm thinking of (transaction timeout, ie causes an
> > HPMC). This is a built-in HW feature. 
> 
> Hmmm... Triggering a HPMC would be a neat feature. So we could think
> about a way to initiate something that, if not handled, would trigger a
> HPMC some 60 seconds later:-?

Generating an HPMC is easy. reference an address in MMIO space that
isn't maintained by a device.  Figuring out a failsafe way of knowing
when to trigger the HPMC is the hardpart.
 
Besides, Lamont proposed a patch to reset the machine after HPMC/panic.
I thought it was a great idea but several people I trust tell me it
needs a bit more work.

grant