Anders Hjalmarsson | 7 Jan 00:40 2008
X-Face
Picon

Some alpha problems in current

I tried current (from todays sources) on alpha, but there were some
problems building a kernel.
I do not think they are worth a pr since they should be trivial to fix.

Builing a MP kernel:
make depend fails for several files with messages similar to this:

In file included from /usr/src/sys/sys/simplelock.h:87,
                 from ./machine/pcb.h:37,
                 from /usr/src/sys/sys/user.h:37,
                 from /usr/src/sys/net/bpf.c:56:
./machine/lock.h:168:1: error: "SPINLOCK_SPIN_HOOK" redefined
In file included from /usr/src/sys/uvm/uvm_param.h:75,
                 from /usr/src/sys/sys/mbuf.h:91,
                 from /usr/src/sys/net/bpf.c:52:
/usr/src/sys/sys/lock.h:222:1: error: this is the location of the previous defin
ition
In file included from /usr/src/sys/sys/simplelock.h:87,
                 from ./machine/pcb.h:37,
                 from /usr/src/sys/sys/user.h:37,
                 from /usr/src/sys/net/bpf.c:56:
./machine/lock.h:181:1: error: "SPINLOCK_BACKOFF_HOOK" redefined
In file included from /usr/src/sys/uvm/uvm_param.h:75,
                 from /usr/src/sys/sys/mbuf.h:91,
                 from /usr/src/sys/net/bpf.c:52:
/usr/src/sys/sys/lock.h:225:1: error: this is the location of the previous defin
ition
mkdep: compile failed.
*** Error code 1

(Continue reading)

Andrew Doran | 7 Jan 01:44 2008
Picon

Re: Some alpha problems in current

Hi Anders,

On Mon, Jan 07, 2008 at 12:40:44AM +0100, Anders Hjalmarsson wrote:

> I tried current (from todays sources) on alpha, but there were some
> problems building a kernel.
> I do not think they are worth a pr since they should be trivial to fix.

Thanks, I have applied the fixes.

> The following patch got me far enough that it hanged late in the boot while
> starting local daemons.

Can you get into the debugger? Can you get a backtrace, and what does ps/l
show?

Thanks,
Andrew

Anders Hjalmarsson | 7 Jan 02:27 2008
X-Face
Picon

Re: Some alpha problems in current

On Mon, 07 Jan 2008 00:44:39 GMT, Andrew Doran wrote:

> > The following patch got me far enough that it hanged late in the boot while
> > starting local daemons.
> 
> Can you get into the debugger? Can you get a backtrace, and what does ps/l
> show?
> 
Unfortunately I have not found out how to get into ddb voluntarily. It seems
sending a break does not put me in ddb (or my terminal server is misconfigured)
The machine did not repond to pings, nor to keypresses like ^T

The last part of the boot messages:
Starting ntpd.
Starting rwhod.
FATAL: Could not determine fully qualified hostname.  Please set 'visible_hostna

Then I gave up and rebooted it.

It seems it was trying to start squid, which unfortunately could not
determine its hostname and probably called abort().

[the name lookup problem was unrelated to this problem,
I have had that problem before during startup, starting it manually after boot
works]

With a uni-processor kernel i get the following:

FATAL: Could not determine fully qualified hostname.  Please set 'visible_hostname'

(Continue reading)

Anders Hjalmarsson | 8 Jan 00:50 2008
X-Face
Picon

Re: Some alpha problems in current

On Mon, 07 Jan 2008 02:27:25 +0100, Anders Hjalmarsson wrote:

> On Mon, 07 Jan 2008 00:44:39 GMT, Andrew Doran wrote:
> 
> > > The following patch got me far enough that it hanged late in the boot whi
le
> > > starting local daemons.
> > 
> > Can you get into the debugger? Can you get a backtrace, and what does ps/l
> > show?
> > 
> Unfortunately I have not found out how to get into ddb voluntarily. It seems
> sending a break does not put me in ddb (or my terminal server is misconfigure
d)

I made a new attempt with a slightly newer kernel (sources from this evening)
and disabled start of squid.
It hanged at approximately the same point, after starting rwhod.
I made a "halt in" which paniced the kernel and I got the following:

Updating motd.
Starting ntpd.
Starting rwhod.
RMC>halt in    

Returning to COM port    

halted CPU 0  
CPU 1 is not halted      
CPU 2 is not halted      
(Continue reading)

Andrew Doran | 12 Jan 18:03 2008
Picon

Re: Some alpha problems in current

On Tue, Jan 08, 2008 at 12:50:35AM +0100, Anders Hjalmarsson wrote:

> cpu_Debugger() at netbsd:cpu_Debugger+0x4
> panic() at netbsd:panic+0x1c8
> console_restart() at netbsd:console_restart+0x78
> XentRestart() at netbsd:XentRestart+0x90
> --- console restart (from ipl 5) ---
> nullop() at netbsd:nullop
> mutex_vector_enter() at netbsd:mutex_vector_enter+0x320
> pool_put() at netbsd:pool_put+0x40
> pmap_tlb_shootdown_job_put() at netbsd:pmap_tlb_shootdown_job_put+0x30
> pmap_do_tlb_shootdown() at netbsd:pmap_do_tlb_shootdown+0x16c
> alpha_ipi_process() at netbsd:alpha_ipi_process+0xb8
> interrupt() at netbsd:interrupt+0x80
> XentInt() at netbsd:XentInt+0x1c
> --- interrupt (from ipl 0) ---
> idle_loop() at netbsd:idle_loop+0x164
> exception_return() at netbsd:exception_return

I made some changes to the pmap that could help the situation (rev 1.230 of
arch/alpha/alpha/pmap.c). Can you try updating and see if the problem still
persists? If it happens again, it would be really useful if you could get
a backtrace from all the CPUs. For example:

mach cpu 0
bt
mach cpu 1
bt 
mach cpu 2
bt
(Continue reading)

Anders Hjalmarsson | 13 Jan 00:33 2008
X-Face
Picon

Re: Some alpha problems in current

On Sat, 12 Jan 2008 17:03:34 GMT, Andrew Doran wrote:

> I made some changes to the pmap that could help the situation (rev 1.230 of
> arch/alpha/alpha/pmap.c). Can you try updating and see if the problem still
> persists? If it happens again, it would be really useful if you could get
> a backtrace from all the CPUs. For example:
> 
> mach cpu 0
> bt
> mach cpu 1
> bt 
> mach cpu 2
> bt
> ....
> 

Somewhat better, it stayed up for more than 10 minutes (multiuser, with
quite a few daemons started, some xterms, top, xosview etc).
Unfortunately I did not get a good backtrace on the other cpus,
but it looks like there is still some kind of pmap-related deadlock.

This is an ES40 with 4 cpus, in case that matters.

Here is what I managed to extract from ddb:

RMC>halt in

Returning to COM port

halted CPU 0
(Continue reading)

mel kravitz | 16 Jan 17:10 2008

netbsd-4.0 boots on alpha ES45 ?

Fresh boot cdrom and tried floppies, (Version-4.0) both fail booting on 
alpha ES45, both work on LX164.
Any one have any better luck perhaps with 2.0 or 1.6?
Mel

Havard Eidnes | 29 Jan 22:33 2008
Picon

Package binaries for NetBSD/alpha 2.1 / pkgsrc-2007Q3

Hi,

I've uploaded the results of a bulk rebuild for NetBSD/alpha
2.1 to

   ftp://ftp.NetBSD.org/pub/pkgsrc/packages/NetBSD/alpha/2.1_2007Q3/

A total of 5.3GB of packages were uploaded.

Following our relatively recent policy, the 2007Q1 packages for
this platform will be removed; we keep only at most two sets of
binaries for each os-version / architecture.

The source tree this was built from was updated Dec 12 2007.

The next update for NetBSD/alpha 2.1 will be based on the 2007Q4
pkgsrc branch.

Regards,

- Håvard


Gmane