[parisc-linux] K380 partial success, the state of the port?

[parisc-linux] K380 partial success, the state of the port?

[Heikki Hannikainen]

  Hi,

  I'm working on a K380 system with 4*240 MHz PA8200 CPUs, 2G of memory. I
initially installed it with the palinux-2.4.21-pa6-20030718-netinst.iso
netinstall CD images. I was amazed how smoothly it went! I've run the
alpha and sparc ports in the past on a bit less powerful hardware. I have
a couple of questions:

- the box seems awfully slow when compared to a similar HP-UX 11 box.
With 2.4.21-32 uniprocessor kernel 'time /bin/true' takes about 0.07s,
'time time /bin/true' takes 0.12s. 'time strace /bin/true' takes a
whopping 0.822s, almost all of which is spent in the system:

real    0m0.822s
user    0m0.010s
sys     0m0.790s

- with an SMP kernel it's much slower. 'time /bin/true' says:

real    0m0.420s
user    0m0.010s
sys     0m0.400s

'time time /bin/true' takes 0.65s, 'time strace /bin/true' takes 2.8
allowing me to read the output while it's running. I compiled the SMP
kernel myself, 2.4.24-pa0 from the CVS on 27.1.2004. On CPU bound tasks
it's ok, but fork/exec/ptrace seems very slow.

  Is the parisc port supposed to be this slow at it's current state, or is
this a problem which is specific to my setup?

- Should I go to the 2.6 tree? I compiled it from the CVS (to get the
tulip driver working) but the SCSI failed:

Linux version 2.6.1-pa10 (root@soul) (gcc version 3.0.4) #2 SMP Tue Jan 27

zalon_probe: Zalon vers field is 0x1, IRQ 37
ncr53c720-0: rev 0xf irq 37
ncr53c720-0: ID 7, Fast-10, Parity Checking, Differential
scsi0 : ncr53c8xx-3.4.3e
Using anticipatory io scheduler
scsi: Device offlined - not ready after error recovery: host 0 channel 0
id 0 l

  Is this a known problem?

  Thank you for your efforts!

  - Hessu

RE: [parisc-linux] K380 partial success, the state of the port?

[Joel Soete]

Hello,

>- Should I go to the 2.6 tree? I compiled it from the CVS (to get the
>tulip driver working but the SCSI failed:
>
>Linux version 2.6.1-pa10 (root@soul) (gcc version 3.0.4) #2 SMP Tue Jan
27


Afaik, 2.6 is not yet smp ready. Could you re-test it by just removing SMP
(there are already a lot of improvement in 2.6  :) )

hth,
    Joel

PS: I am happy that boot fine on the N I am testing 

-------------------------------------------------------------------------
Tiscali ADSL: 12 mois à 29,50 €/mois! L'Internet rapide, c'est pour tout
le monde.
http://reg.tiscali.be/default.asp?lg=fr

Re: [parisc-linux] K380 partial success, the state of the port?

[Randolph Chung]

>   Is the parisc port supposed to be this slow at it's current state, or is
> this a problem which is specific to my setup?

fork/exec is very slow on parisc right now. in other areas though
performance is not too shabby. Ryan posted lmbench results to the list
recently. Check the archive for details.

> - Should I go to the 2.6 tree? I compiled it from the CVS (to get the
> tulip driver working) but the SCSI failed:

Yes, you should go to 2.6, but as Joel pointed out SMP is not yet
working correctly.

We have a few scsi gurus on the parisc list so they can address the scsi
issues :)

randolph
-- 
Randolph Chung
Debian GNU/Linux Developer, hppa/ia64 ports
http://www.tausq.org/

Re: [parisc-linux] K380 partial success, the state of the port?

[Heikki Hannikainen]

On Wed, 28 Jan 2004, Randolph Chung wrote:

> > - Should I go to the 2.6 tree? I compiled it from the CVS (to get the
> > tulip driver working) but the SCSI failed:
>
> Yes, you should go to 2.6, but as Joel pointed out SMP is not yet
> working correctly.
>
> We have a few scsi gurus on the parisc list so they can address the scsi
> issues :)

  Ok! I compiled the same 2.6.1-pa10 from tuesday's CVS, uniprocessor
32bit, and it almost booted. The SCSI driver didn't barf, but it crashed
just before init started. Full console log at:
http://he.fi/misc/hppa/2.6.1-pa10-32-uni.crash1.consolelog
I'm not a kernel hacker, but I'd be happy to try any debugging or
testing code/patches if someone has an idea why this is failing.

EXT3-fs: mounted filesystem with ordered data mode.
VFS: Mounted root (ext3 filesystem) readonly.
Freeing unused kernel memory: 468k freed

Stack Dump:
...
Kernel addresses on the stack:
 [<10231f64>] elv_next_request+0x20/0x12c
 [<10103f38>] parisc_terminate+0x60/0xb0
 [<101041d4>] handle_interruption+0x24c/0x598
 [<10121d20>] io_schedule+0x40/0x6c
 [<1015f004>] __wait_on_buffer+0xbc/0x118
 [<10109088>] intr_check_sig+0x0/0xc
 [<102355a8>] submit_bio+0x80/0xd8
 [<102137a0>] init_dev+0x64/0x4d0
 [<10213798>] init_dev+0x5c/0x4d0
 [<10214328>] tty_open+0x88/0x3cc
 [<1016d3ac>] link_path_walk+0x788/0xa40
 [<10168008>] chrdev_open+0xd0/0x180
 [<1015e914>] get_empty_filp+0x68/0x118
 [<1015ceb4>] dentry_open+0x154/0x224
 [<1015cd40>] filp_open+0x4c/0x6c
 [<1015d288>] sys_open+0x54/0xa8
 [<10100204>] init+0x54/0xc8
 [<10108c5c>] ret_from_kernel_thread+0x1c/0x24


Kernel Fault: Code=15 regs=8ffa8380 (Addr=f305f306)

     YZrvWESTHLNXBCVMcbcbcbcbOGFRQPDI
PSW: 00000000000001001111111100001111 Not tainted
r00-03  00000000 10398010 10213798 10387178
r04-07  1039b42c 1039b42c 8fb7b840 00500001
r08-11  00000000 1045e99c 000f41fa 00000000
r12-15  1038c1c8 00000000 1038f810 1045e810
r16-19  f0001898 8ffa828c f000011c f305f306
r20-23  00000000 00000000 1010fdf0 00000002
r24-27  8ffa828c 00000000 10398730 10370010
r28-31  1039b42c 00000005 8ffa8380 102145d0
sr0-3   00000000 00000000 00000000 00000000
sr4-7   00000000 00000000 00000000 00000000

IASQ: 00000000 00000000 IAOQ: 1021379c 102137a0
 IIR: 0d130083    ISR: 00000000  IOR: f305f306
 CPU:        0   CR30: 8ffa8000 CR31: 103d4000
 ORIG_R28: 8ffa84c0
 IAOQ[0]: init_dev+0x60/0x4d0
 IAOQ[1]: init_dev+0x64/0x4d0
 RP(r2): init_dev+0x5c/0x4d0


  - Hessu

Re: [parisc-linux] K380 partial success, the state of the port?

[Ryan Bradetich]

I looked through your log because this crash looked very familar to me
:(

> Kernel command line: root=/dev/sda2 HOME=/ console=ttyB0
palo_kernel=2/boot/vmlinux

Good you have palo set for the Serial Mux.

But I do not see the Serial Mux in the console log.  Can you verify that
the CONFIG_SERIAL_MUX and CONFIG_SERLIAL_MUX_CONSOLE are compiled into
your kernel?

I would bet this is your problem just from the dump output ... and
futher confirmed by lack of serial mux in your boot log.

Should look something like:

Serial mux driver (8 ports) Revision: 0.2
ttyB0 at MMIO 0xfff78800 (polling) is a Mux
.
.
.

Thanks,

- Ryan


On Thu, 2004-01-29 at 03:03, Heikki Hannikainen wrote:
> On Wed, 28 Jan 2004, Randolph Chung wrote:
> 
> > > - Should I go to the 2.6 tree? I compiled it from the CVS (to get the
> > > tulip driver working) but the SCSI failed:
> >
> > Yes, you should go to 2.6, but as Joel pointed out SMP is not yet
> > working correctly.
> >
> > We have a few scsi gurus on the parisc list so they can address the scsi
> > issues :)
> 
>   Ok! I compiled the same 2.6.1-pa10 from tuesday's CVS, uniprocessor
> 32bit, and it almost booted. The SCSI driver didn't barf, but it crashed
> just before init started. Full console log at:
> http://he.fi/misc/hppa/2.6.1-pa10-32-uni.crash1.consolelog
> I'm not a kernel hacker, but I'd be happy to try any debugging or
> testing code/patches if someone has an idea why this is failing.
> 
> EXT3-fs: mounted filesystem with ordered data mode.
> VFS: Mounted root (ext3 filesystem) readonly.
> Freeing unused kernel memory: 468k freed
> 
> Stack Dump:
> ...
> Kernel addresses on the stack:
>  [<10231f64>] elv_next_request+0x20/0x12c
>  [<10103f38>] parisc_terminate+0x60/0xb0
>  [<101041d4>] handle_interruption+0x24c/0x598
>  [<10121d20>] io_schedule+0x40/0x6c
>  [<1015f004>] __wait_on_buffer+0xbc/0x118
>  [<10109088>] intr_check_sig+0x0/0xc
>  [<102355a8>] submit_bio+0x80/0xd8
>  [<102137a0>] init_dev+0x64/0x4d0
>  [<10213798>] init_dev+0x5c/0x4d0
>  [<10214328>] tty_open+0x88/0x3cc
>  [<1016d3ac>] link_path_walk+0x788/0xa40
>  [<10168008>] chrdev_open+0xd0/0x180
>  [<1015e914>] get_empty_filp+0x68/0x118
>  [<1015ceb4>] dentry_open+0x154/0x224
>  [<1015cd40>] filp_open+0x4c/0x6c
>  [<1015d288>] sys_open+0x54/0xa8
>  [<10100204>] init+0x54/0xc8
>  [<10108c5c>] ret_from_kernel_thread+0x1c/0x24
> 
> 
> Kernel Fault: Code=15 regs=8ffa8380 (Addr=f305f306)
> 
>      YZrvWESTHLNXBCVMcbcbcbcbOGFRQPDI
> PSW: 00000000000001001111111100001111 Not tainted
> r00-03  00000000 10398010 10213798 10387178
> r04-07  1039b42c 1039b42c 8fb7b840 00500001
> r08-11  00000000 1045e99c 000f41fa 00000000
> r12-15  1038c1c8 00000000 1038f810 1045e810
> r16-19  f0001898 8ffa828c f000011c f305f306
> r20-23  00000000 00000000 1010fdf0 00000002
> r24-27  8ffa828c 00000000 10398730 10370010
> r28-31  1039b42c 00000005 8ffa8380 102145d0
> sr0-3   00000000 00000000 00000000 00000000
> sr4-7   00000000 00000000 00000000 00000000
> 
> IASQ: 00000000 00000000 IAOQ: 1021379c 102137a0
>  IIR: 0d130083    ISR: 00000000  IOR: f305f306
>  CPU:        0   CR30: 8ffa8000 CR31: 103d4000
>  ORIG_R28: 8ffa84c0
>  IAOQ[0]: init_dev+0x60/0x4d0
>  IAOQ[1]: init_dev+0x64/0x4d0
>  RP(r2): init_dev+0x5c/0x4d0
> 
> 
>   - Hessu
> 
> _______________________________________________
> parisc-linux mailing list
> parisc-linux@lists.parisc-linux.org
> http://lists.parisc-linux.org/mailman/listinfo/parisc-linux
>