[9fans] system crash during compile

[9fans] system crash during compile

[John Floren]

Hi everyone.
I've been trying to compile a new kernel (9pccpuf), but whenever it
gets to the "size 9pccpuf" point, I get the following error and a
system crash:

panic: executeio: page out I/O error
dumpstack disabled
cpu0: exiting

Admittedly, this system is not the most up to date (can't get dialup
working), so it could be a bug which is fixed now. However, with my
slow connection, I would rather not re-download the cd image. Has
anybody else seen this problem? It looks like it has something to do
with swap, but the swap had been working OK up until that point.
Thanks

John
--
TANSTAAFL! (There Ain't No Such Thing As A Free Lunch!)

Re: [9fans] system crash during compile

[Russ Cox]

> Hi everyone.
> I've been trying to compile a new kernel (9pccpuf), but whenever it
> gets to the "size 9pccpuf" point, I get the following error and a
> system crash:
>
> panic: executeio: page out I/O error
> dumpstack disabled
> cpu0: exiting
>
> Admittedly, this system is not the most up to date (can't get dialup
> working), so it could be a bug which is fixed now. However, with my
> slow connection, I would rather not re-download the cd image. Has
> anybody else seen this problem? It looks like it has something to do
> with swap, but the swap had been working OK up until that point.
> Thanks

You're running out of memory during the 8l command that
precedes the size command in the final linking recipe.
You might also be running out of swap.  You could try running
stats to watch your memory and swap usage during the mk
and see if swap fills.

Russ

Re: [9fans] system crash during compile

[John Floren]

On 7/9/06, Russ Cox <rsc@swtch.com> wrote:
> > Hi everyone.
> > I've been trying to compile a new kernel (9pccpuf), but whenever it
> > gets to the "size 9pccpuf" point, I get the following error and a
> > system crash:
> >
> > panic: executeio: page out I/O error
> > dumpstack disabled
> > cpu0: exiting
> >
> > Admittedly, this system is not the most up to date (can't get dialup
> > working), so it could be a bug which is fixed now. However, with my
> > slow connection, I would rather not re-download the cd image. Has
> > anybody else seen this problem? It looks like it has something to do
> > with swap, but the swap had been working OK up until that point.
> > Thanks
>
> You're running out of memory during the 8l command that
> precedes the size command in the final linking recipe.
> You might also be running out of swap.  You could try running
> stats to watch your memory and swap usage during the mk
> and see if swap fills.
>
> Russ
>

I've been running stats. Swap doesn't even reach 1/4 of capacity when
the crash occurs.

John
--
TANSTAAFL! (There Ain't No Such Thing As A Free Lunch!)

Re: [9fans] system crash during compile

[John Floren]

So far, nobody can shed any light on the problem? Bugger.

John
--
TANSTAAFL! (There Ain't No Such Thing As A Free Lunch!)

Re: Re: [9fans] system crash during compile

[andrey mirtchovski]

On 7/10/06, John Floren <slawmaster@gmail.com> wrote:
> So far, nobody can shed any light on the problem? Bugger.

didn't russ?

anyway, you haven't told us what architecture you have, how much ram,
how much hard disk space available and what time of day you're running
the command at (don't ask, sometimes it matters)... back in the day
ghostscript linking would panic a machine with no swap turned on and
only 64MB of ram, but those days are long gone, right?

right?!? i mean, i haven't had a need for a swap device since 2002!
what is that beastly node on which you must have a kernel compiled,
and why can't you just run plan9 in qemu on your main machine?

Re: Re: [9fans] system crash during compile

[John Floren]

On 7/10/06, andrey mirtchovski <mirtchovski@gmail.com> wrote:
> On 7/10/06, John Floren <slawmaster@gmail.com> wrote:
> > So far, nobody can shed any light on the problem? Bugger.
>
> didn't russ?
>
> anyway, you haven't told us what architecture you have, how much ram,
> how much hard disk space available and what time of day you're running
> the command at (don't ask, sometimes it matters)... back in the day
> ghostscript linking would panic a machine with no swap turned on and
> only 64MB of ram, but those days are long gone, right?
>
> right?!? i mean, i haven't had a need for a swap device since 2002!
> what is that beastly node on which you must have a kernel compiled,
> and why can't you just run plan9 in qemu on your main machine?
>
Okay, the node is an IBM machine, Pentium 2 processor @ ~300 Mhz, 64
MB of RAM, ~3 GB of hard drive space. I have a fossil+venti
configuration set up; exact numbers, I can't tell you, because I'm
currently 200 miles away from the damn thing. I've tried running the
command at various times of day.
I can't run Plan 9 in qemu on my main machine because my main machine
is somewhere in Ohio at the moment (I think) and I won't have access
until fall. Besides, it just isn't the same when you're emulating it
;-)

John
--
TANSTAAFL! (There Ain't No Such Thing As A Free Lunch!)

Re: Re: [9fans] system crash during compile

[Francisco J Ballesteros]

You said your stats for memory usage did remain low. Was this the case?
Couldn´t it be running out of the % of memory used by the kernel?
You could try increasing kernelpcent if that was the case.

On 7/11/06, John Floren <slawmaster@gmail.com> wrote:
> On 7/10/06, andrey mirtchovski <mirtchovski@gmail.com> wrote:
> > On 7/10/06, John Floren <slawmaster@gmail.com> wrote:
> > > So far, nobody can shed any light on the problem? Bugger.
> >
> > didn't russ?
> >
> > anyway, you haven't told us what architecture you have, how much ram,
> > how much hard disk space available and what time of day you're running
> > the command at (don't ask, sometimes it matters)... back in the day
> > ghostscript linking would panic a machine with no swap turned on and
> > only 64MB of ram, but those days are long gone, right?
> >
> > right?!? i mean, i haven't had a need for a swap device since 2002!
> > what is that beastly node on which you must have a kernel compiled,
> > and why can't you just run plan9 in qemu on your main machine?
> >
> Okay, the node is an IBM machine, Pentium 2 processor @ ~300 Mhz, 64
> MB of RAM, ~3 GB of hard drive space. I have a fossil+venti
> configuration set up; exact numbers, I can't tell you, because I'm
> currently 200 miles away from the damn thing. I've tried running the
> command at various times of day.
> I can't run Plan 9 in qemu on my main machine because my main machine
> is somewhere in Ohio at the moment (I think) and I won't have access
> until fall. Besides, it just isn't the same when you're emulating it
> ;-)
>
> John
> --
> TANSTAAFL! (There Ain't No Such Thing As A Free Lunch!)
>
>

Re: Re: Re: [9fans] system crash during compile

[David Leimbach]

On 7/11/06, Francisco J Ballesteros <nemo@lsub.org> wrote:
> You said your stats for memory usage did remain low. Was this the case?
> Couldn´t it be running out of the % of memory used by the kernel?
> You could try increasing kernelpcent if that was the case.

Do the RAM dimms pass a memory test?  Memory errors are fun... I had
good DIMMs once and a bad slot, took about 1 year till I finally tried
that out and got back to 1GB on my PC.

Haven't turned that machine on much since I got my intel mac with
Parallels though. (especially since Plan 9 is working so nicely on it
now)

Dave

>
> On 7/11/06, John Floren <slawmaster@gmail.com> wrote:
> > On 7/10/06, andrey mirtchovski <mirtchovski@gmail.com> wrote:
> > > On 7/10/06, John Floren <slawmaster@gmail.com> wrote:
> > > > So far, nobody can shed any light on the problem? Bugger.
> > >
> > > didn't russ?
> > >
> > > anyway, you haven't told us what architecture you have, how much ram,
> > > how much hard disk space available and what time of day you're running
> > > the command at (don't ask, sometimes it matters)... back in the day
> > > ghostscript linking would panic a machine with no swap turned on and
> > > only 64MB of ram, but those days are long gone, right?
> > >
> > > right?!? i mean, i haven't had a need for a swap device since 2002!
> > > what is that beastly node on which you must have a kernel compiled,
> > > and why can't you just run plan9 in qemu on your main machine?
> > >
> > Okay, the node is an IBM machine, Pentium 2 processor @ ~300 Mhz, 64
> > MB of RAM, ~3 GB of hard drive space. I have a fossil+venti
> > configuration set up; exact numbers, I can't tell you, because I'm
> > currently 200 miles away from the damn thing. I've tried running the
> > command at various times of day.
> > I can't run Plan 9 in qemu on my main machine because my main machine
> > is somewhere in Ohio at the moment (I think) and I won't have access
> > until fall. Besides, it just isn't the same when you're emulating it
> > ;-)
> >
> > John
> > --
> > TANSTAAFL! (There Ain't No Such Thing As A Free Lunch!)
> >
> >
>

Re: Re: Re: [9fans] system crash during compile

[John Floren]

On 7/11/06, David Leimbach <leimy2k@gmail.com> wrote:
> On 7/11/06, Francisco J Ballesteros <nemo@lsub.org> wrote:
> > You said your stats for memory usage did remain low. Was this the case?
> > Couldn´t it be running out of the % of memory used by the kernel?
> > You could try increasing kernelpcent if that was the case.
>
> Do the RAM dimms pass a memory test?  Memory errors are fun... I had
> good DIMMs once and a bad slot, took about 1 year till I finally tried
> that out and got back to 1GB on my PC.
>
> Haven't turned that machine on much since I got my intel mac with
> Parallels though. (especially since Plan 9 is working so nicely on it
> now)
>
> Dave
I'm pretty sure the memory is okay. Francisco Ballesteros seems to
have misunderstood me, or perhaps I wrote poorly; my RAM usage goes to
100%. *Swap* usage stays low the whole time, though. There's only one
DIMM in the computer, and it seems to be working okay.

John
-- 
TANSTAAFL! (There Ain't No Such Thing As A Free Lunch!)

Re: Re: Re: Re: [9fans] system crash during compile

[andrey mirtchovski]

> 100%. *Swap* usage stays low the whole time, though. There's only one
> DIMM in the computer, and it seems to be working okay.

low but above zero? you've made sure you have turned swap on, right?
see swap(8).

perhaps it would be best to use /386/9pccpuf, send the list your
pccpuf config file and have someone compile a kernel for you, or cpu
to a 9grid machine and compile it there?

Re: Re: Re: Re: [9fans] system crash during compile

[John Floren]

On 7/11/06, andrey mirtchovski <mirtchovski@gmail.com> wrote:
> > 100%. *Swap* usage stays low the whole time, though. There's only one
> > DIMM in the computer, and it seems to be working okay.
>
> low but above zero? you've made sure you have turned swap on, right?
> see swap(8).
>
> perhaps it would be best to use /386/9pccpuf, send the list your
> pccpuf config file and have someone compile a kernel for you, or cpu
> to a 9grid machine and compile it there?
>
Yes, low but non-zero. I would use /386/9pccpuf, but I don't appear to
*have* it. My install disc is from last October, and I'm not sure it
includes all the precompiled kernels. I'd try one of the others, but I
haven't yet gotten dialup working on the machine, so it can't connect
to the Internet.

John
--
TANSTAAFL! (There Ain't No Such Thing As A Free Lunch!)

Re: Re: Re: Re: [9fans] system crash during compile

[Russ Cox]

Swap doesn't work.  Find a machine with more memory
or remove some stuff from your kernel.

Russ

Re: Re: Re: Re: [9fans] system crash during compile

[John Floren]

On 7/12/06, Russ Cox <rsc@swtch.com> wrote:
> Swap doesn't work.  Find a machine with more memory
> or remove some stuff from your kernel.
>
> Russ
Why doesn't swap work? Are you talking about on my machine,
specifically, or just in the distribution as a whole?


John
--
TANSTAAFL! (There Ain't No Such Thing As A Free Lunch!)

Re: Re: Re: Re: [9fans] system crash during compile

[Francisco J Ballesteros]

I've seen a machine with available swap space block due to lack
of memory. It seems that under certain circumstances the machine
is not able to get more memory by paging out things.

But with memory so cheap...
My video card has a lot more memory than the first machine I used to
run plan 9 at urjc :-)

On 7/13/06, John Floren <slawmaster@gmail.com> wrote:
> On 7/12/06, Russ Cox <rsc@swtch.com> wrote:
> > Swap doesn't work.  Find a machine with more memory
> > or remove some stuff from your kernel.
> >
> > Russ
> Why doesn't swap work? Are you talking about on my machine,
> specifically, or just in the distribution as a whole?
>
>
> John
> --
> TANSTAAFL! (There Ain't No Such Thing As A Free Lunch!)
>
>

Re: Re: Re: Re: [9fans] system crash during compile

[John Floren]

On 7/12/06, Francisco J Ballesteros <nemo@lsub.org> wrote:
> I've seen a machine with available swap space block due to lack
> of memory. It seems that under certain circumstances the machine
> is not able to get more memory by paging out things.
>
> But with memory so cheap...
> My video card has a lot more memory than the first machine I used to
> run plan 9 at urjc :-)
>
I got this machine for free, so I'm rather reluctant to put any money
into it :-)
I guess I'll just have to download the newest ISO, which apparently
has 9pccpuf already compiled and in /386/

John
--
TANSTAAFL! (There Ain't No Such Thing As A Free Lunch!)

Re: Re: Re: Re: [9fans] system crash during compile

[Russ Cox]

You may also be able to
hget http://plan9.bell-labs.com/sources/plan9/386/9pccpuf.gz >/386/9pccpuf.gz

Swap has never worked.  It mostly works, but there have
always been hard-to-reproduce and then even harder-to-debug
little issues, and I just don't trust it at all.  With memory so
cheap, there is little motivation to fix it.

Russ

Re: Re: Re: Re: [9fans] system crash during compile

[Russ Cox]

> You may also be able to
> hget http://plan9.bell-labs.com/sources/plan9/386/9pccpuf.gz >/386/9pccpuf.gz

Or 9fs sources; cp /n/sources/plan9/386/9pccpuf.gz /386

but I assume that you don't have 9P connectivity to
the machine or else you'd have done that already.

Russ

Re: Re: Re: Re: [9fans] system crash during compile

[Brantley Coile]

> Swap has never worked.  It mostly works, but there have
> always been hard-to-reproduce and then even harder-to-debug
> little issues, and I just don't trust it at all.  With memory so
> cheap, there is little motivation to fix it.

That explains our mail processing problems that we had a few years ago
when the incoming load was so high that it exceeded local store and we
turned on swap, but things were still funky.  (wound up limiting the number
of connections allowed to port 25 at a given time.)

Might I suggest we remove swap from the system, at least remove it
enough so people don't fall prey to it?

I would argue backing store is a thing of the past and that Sandy
Fraser's objections to the Plan 9 team early on, as I understand his
objections to be, were correct.  Let's just use paging to manage the
store in the box and to perhaps demand load text.  We're a long way
from tiny, expensive local stores and slow drums.

Let's just take it out.

Re: Re: Re: Re: [9fans] system crash during compile

[John Floren]

On 7/13/06, Brantley Coile <brantley@coraid.com> wrote:
> > Swap has never worked.  It mostly works, but there have
> > always been hard-to-reproduce and then even harder-to-debug
> > little issues, and I just don't trust it at all.  With memory so
> > cheap, there is little motivation to fix it.
>
> That explains our mail processing problems that we had a few years ago
> when the incoming load was so high that it exceeded local store and we
> turned on swap, but things were still funky.  (wound up limiting the number
> of connections allowed to port 25 at a given time.)
>
> Might I suggest we remove swap from the system, at least remove it
> enough so people don't fall prey to it?
>
> I would argue backing store is a thing of the past and that Sandy
> Fraser's objections to the Plan 9 team early on, as I understand his
> objections to be, were correct.  Let's just use paging to manage the
> store in the box and to perhaps demand load text.  We're a long way
> from tiny, expensive local stores and slow drums.
>
> Let's just take it out.
>
Swap is handy. When you're running on 32 MB of RAM on an old laptop
that you don't want to spend on to upgrade, slow virtual memory is
better than nothing at all. I would volunteer to make swap work, but
it seems like it would be one of those things that's more deeply tied
to the rest of the system (which I don't know well); also, I don't
really know C.

John
--
TANSTAAFL! (There Ain't No Such Thing As A Free Lunch!)