NetBSD on Sun's 64-bit Ultrasparc based machines (NetBSD/sparc64)

Jukka Salmi | 1 Feb 2008 02:00

Picon

Re: SATA controller recommendation for sparc64

Manuel Bouyer --> port-sparc64 (2007-11-10 16:14:32 +0100):
> Promise SATA adapters should work in a sparc64.

I'm just initialising a RAIDframe mirror on two SATA disks connected
to a Promise SATA300 TX4 controller in a Sun Blade 100, and there are
lots of lost interrupts...

Sun Blade 100 (UltraSPARC-IIe), No Keyboard
Copyright 1998-2002 Sun Microsystems, Inc.  All rights reserved.
OpenBoot 4.5, 512 MB memory installed, Serial #50874347.
[...]
NetBSD 4.99.52 (INSTALL) #0: Thu Jan 31 20:59:09 CET 2008
[...]
ppb0 at pci0 dev 5 function 0: Digital Equipment DC21152 PCI-PCI Bridge (rev. 0x03)
pci1 at ppb0 bus 1
pdcsata0 at pci1 dev 0 function 0
pdcsata0: Promise PDC40718 SATA300 controller (rev. 0x02)
pdcsata0: interrupting at ivec 19                        
atabus2 at pdcsata0 channel 0    
atabus3 at pdcsata0 channel 1
atabus4 at pdcsata0 channel 2
atabus5 at pdcsata0 channel 3
[...]
pdcsata0 port 0: device present, speed: 3.0Gb/s                       
pdcsata0 port 1: device present, speed: 3.0Gb/s
[...]
wd1 at atabus2 drive 0pdcsata0:0:0: lost interrupt         
	type: ata tc_bcount: 512 tc_skip: 0       
: <SAMSUNG HD321KJ>                        
wd1: 298 GB, 620181 cyl, 16 head, 63 sec, 512 bytes/sect x 625142448 sectors

(Continue reading)

Julian Coleman | 1 Feb 2008 09:30

Picon

NetBSD 4 crash in netbsd:ipsec_setspidx()

Hi,

Has anyone else seen this crash on NetBSD 4?

  kernel trap 34: mem address not aligned
  Stopped at      netbsd:ipsec_setspidx+0x20:     ld              [%g2 + 0x20], %g1

Backtrace is:

  db> bt
  ipsec_setspidx_mbuf(0, 18, 5aef210, 0, e00179d8, 1000000) at netbsd:ipsec_setspidx_mbuf+0x3c
  ipsec6_getpolicybyaddr(0, 1, 1, e0017bac, e0017ba8, 5) at netbsd:ipsec6_getpolicybyaddr+0x48
  ipsec6_in_reject_so(5aef210, 0, ffffffffffffffff, c000000, 28, 0) at
netbsd:ipsec6_in_reject_so+0x88 lo0, addresses 127.0.0.1/8 (loopback)
  ip6_input(5aef210, 5aef210, 0, 0, 1ff0020611c, 1) at netbsd:ip6_input+0x960
  ip6intr(5aef210, 10a00010, 1700, 1ff00200000, 1, 4000) at netbsd:ip6intr+0x28
  softnet(1, 0, e0017ed0, 0, 1403fa0, 13113c0) at netbsd:softnet+0x5c
  sparc_interrupt(0, 0, 14c0c98, 0, ffffffffffffffff, ffffffffffffc078) at netbsd:
  sparc_interrupt+0x1d4
  microtime(fb70120, 0, 160ecc0, 0, fda3570, 27) at netbsd:microtime+0x14
  mi_switch(fb70120, 0, 0, ef81f10, 2, 0) at netbsd:mi_switch+0x148

I'm not using IPSEC.  Kernel is 4.0_STABLE.

Thanks,

J

--

-- 
  My other computer also runs NetBSD    /        Sailing at Newbiggin

(Continue reading)

Matthias Scheler | 1 Feb 2008 14:23

Picon

4.99.51 kernel crashes while detecting a PCI card


	Hello,

when I tried out a 4.99.51 kernel on my ULTRA60 to check whether
PR port-sparc64/3034 is fixed it paniced when it detected the PCI-X card:

bge0 at pci1 dev 1 function 0: Broadcom BCM5701 Gigabit Ethernet
cpu0: data fault: pc=12eac80 addr=0
kernel trap 30: data access exception
Stopped in pid 0.1 (system) at  netbsd:pci_conf_read:   ldx             [%o0 + %g0], %g1
db> bt
config_attach_loc(3aacd00, 1808be8, 1c053bc, 1c05358, 1, 12c6600) at netbsd:config_attach_loc+0x158
pci_probe_device(0, f0088f1400800800, 0, 0, 0, 0) at netbsd:pci_probe_device+0x254
sparc64_pci_enumerate_bus(3ab9000, 13a2ca8, 0, 42, 40, 181cc00) at netbsd:sparc64_pci_enumerate_bus+0x324
pcirescan(3ab9000, 1448278, 13a2ca8, 3ab9037, 0, 3ab9037) at netbsd:pcirescan+0x10
pciattach(0, 3ab9000, 1c057d8, 1810800, 1813000, 2) at netbsd:pciattach+0x14c
config_attach_loc(3ab9000, 1808ac8, 0, 1c057d8, 1, 12e90c0) at netbsd:config_attach_loc+0x158
psycho_attach(3aadf00, 3aaa200, 1c059f0, 1810800, 1813000, 2) at netbsd:psycho_attach+0x4f8
config_attach_loc(3aaa200, 1808a98, 0, 1c059f0, 1, 13142a0) at netbsd:config_attach_loc+0x158
mainbus_attach(0, 3aadf00, 0, 1810800, 1464800, 1314000) at netbsd:mainbus_attach+0x474
config_attach_loc(3aadf00, 1808858, 0, 0, 1, 7) at netbsd:config_attach_loc+0x158
cpu_configure(1469000, 1208ae0, 0, 1810800, 1814c00, 146a9d8) at netbsd:cpu_configure+0x1d0
configure(185f0b8, 2, 187c400, 12526c0, 1468ca0, 1884c00) at netbsd:configure+0x3c         
main(0, 11ebe40, ffffffffffffe000, 1fff, 1889800, 1c02000) at netbsd:main+0x1dc
cpu_initialize(f0067260, 0, 10fc00, fff83d00, 10ea78, fff83e00) at netbsd:cpu_initialize+0xf8

Any ideas? The machines boots fine with a very old 4.99.18 kernel.

	Kind regards

(Continue reading)

Martin Husemann | 1 Feb 2008 15:17

Picon

Re: 4.99.51 kernel crashes while detecting a PCI card

On Fri, Feb 01, 2008 at 01:23:51PM +0000, Matthias Scheler wrote:
> Any ideas? The machines boots fine with a very old 4.99.18 kernel.

This is probably an unaligned pci_conf_read() call - can you build a 
netbsd.gdb kernel and locate the caller of it?

Martin

matthew green | 1 Feb 2008 16:58

Picon

re: 4.99.51 kernel crashes while detecting a PCI card


   when I tried out a 4.99.51 kernel on my ULTRA60 to check whether
   PR port-sparc64/3034 is fixed it paniced when it detected the PCI-X card:

   bge0 at pci1 dev 1 function 0: Broadcom BCM5701 Gigabit Ethernet
   cpu0: data fault: pc=12eac80 addr=0

null pointer access.

this seem very strange, and i do not think it has to do with
pci_conf_read().

   kernel trap 30: data access exception
   Stopped in pid 0.1 (system) at  netbsd:pci_conf_read:   ldx             [%o0 + %g0], %g1
   db> bt
   config_attach_loc(3aacd00, 1808be8, 1c053bc, 1c05358, 1, 12c6600) at netbsd:config_attach_loc+0x158
   pci_probe_device(0, f0088f1400800800, 0, 0, 0, 0) at netbsd:pci_probe_device+0x254
   sparc64_pci_enumerate_bus(3ab9000, 13a2ca8, 0, 42, 40, 181cc00) at netbsd:sparc64_pci_enumerate_bus+0x324

can you instrument this function to see where it dies?

thanks.

.mrg.

Julian Coleman | 1 Feb 2008 17:28

Picon

Re: NetBSD 4 crash in netbsd:ipsec_setspidx()

> I'm not using IPSEC.  Kernel is 4.0_STABLE.

Maybe the problem isn't with IPSEC at all.  I've just had another crash on
a kernel built without IPSEC:

  panic: lockmgr: no context

Backtrace is:

  lockmgr(1874e20, 1, 0, 57020312000, 1, 4000) at netbsd:lockmgr+0xa9c
  uvm_fault_internal(1874e18, 1874e18, 1, 0, 4, 4) at netbsd:uvm_fault_internal+0xd4
  data_access_fault(e0017700, 30, 12d55f4, 57020312000, 57020313160, 400) at netbsd:data_access_fault+0xd4
  ?(1899d68, 5aedb30, ffffffffffffffff, 20, 201, 5a77400) at 0x1008708
  m_freem(726f757020313130, 815c, 94, 5c, 35, 90) at netbsd:m_freem+0x4c
  tcp_input(5aedb30, 20, 4, 4, ed35838, 0) at netbsd:tcp_input+0x1b14
  tcp6_input(101, e0017c84, 6, c000000, 28, 0) at netbsd:tcp6_input+0x70
  ip6_input(5aedb30, 5aedb30, 0, 0, 1ff0020611c, 1) at netbsd:ip6_input+0x8d4
  ip6intr(5aedb30, 1080010, 1700, 1ff00200000, 1, 4000) at netbsd:ip6intr+0x28
  softnet(1, 0, e0017ed0, ffffffffffffc940, 13e6260, 0) at netbsd:softnet+0x5c
  sparc_interrupt(0, 0, 14a2f58, 0, ffffffffffffffff, 0) at netbsd:sparc_interrupt+0x1d4
  microtime(fb4b880, 0, 0, 0, 0, 0) at netbsd:microtime+0x14
  mi_switch(fb4b880, 0, 78, 0, 0, 1816420) at netbsd:mi_switch+0x148

This crash happened whilst I was playing with ipf (trying to get it not to
block IPv6 ICMP).  From the db> prompt, `show mbuf` gave:

  MBUF 0x13f1fa4
  trap type 0x34: cpu 0, pc=12d4060 npc=12bdf40 pstate=ffffffff991d0006<PRIV,IE>
  kernel trap 34: mem address not aligned

(Continue reading)

Matthias Scheler | 2 Feb 2008 01:18

Picon

Re: 4.99.51 kernel crashes while detecting a PCI card

On Fri, Feb 01, 2008 at 03:17:02PM +0100, Martin Husemann wrote:
> On Fri, Feb 01, 2008 at 01:23:51PM +0000, Matthias Scheler wrote:
> > Any ideas? The machines boots fine with a very old 4.99.18 kernel.
> 
> This is probably an unaligned pci_conf_read() call - can you build a 
> netbsd.gdb kernel and locate the caller of it?

How what I do that?

In the past I've used a crash dump and "netbsd.gdb" to figure out such
things. But I cannot can a crash dump in this case because the machine
doesn't even finish auto configuration

	Kind regards

--

-- 
Matthias Scheler                                  http://zhadum.org.uk/

Martin Husemann | 2 Feb 2008 01:52

Picon

Re: 4.99.51 kernel crashes while detecting a PCI card

On Sat, Feb 02, 2008 at 12:18:27AM +0000, Matthias Scheler wrote:
> How what I do that?

Matt is right, this is not an unaligned access but a null pointer deref.
Just build the same kernel with makeoptions DEBUG=-g, then do

  gdb netbsd.gdb

and check the pc that the crash reported, like "list *(0xXXXXX)".

Martin

Matthias Scheler | 2 Feb 2008 11:55

Picon

Re: 4.99.51 kernel crashes while detecting a PCI card

On Sat, Feb 02, 2008 at 01:52:25AM +0100, Martin Husemann wrote:
> On Sat, Feb 02, 2008 at 12:18:27AM +0000, Matthias Scheler wrote:
> > How what I do that?
> 
> Matt is right, this is not an unaligned access but a null pointer deref.
> Just build the same kernel with makeoptions DEBUG=-g, then do
> 
>   gdb netbsd.gdb
> 
> and check the pc that the crash reported, like "list *(0xXXXXX)".

I've built a new kernel with debugging symbols the stack trace looks
like this now:

config_attach_loc(3aacd00, 1808be8, 1c053bc, 1c05358, 1, 12c65c0) at netbsd:config_attach_loc+0x158
pci_probe_device(0, f0088f1400800800, 0, 0, 0, 0) at netbsd:pci_probe_device+0x254
sparc64_pci_enumerate_bus(3ab9000, 13a2ce8, 0, 42, 40, 181cc00) at netbsd:sparc64_pci_enumerate_bus+0x324
pcirescan(3ab9000, 14482c0, 13a2ce8, 3ab9037, 0, 3ab9037) at netbsd:pcirescan+0x10
pciattach(0, 3ab9000, 1c057d8, 1810800, 1813000, 2) at netbsd:pciattach+0x14c
config_attach_loc(3ab9000, 1808ac8, 0, 1c057d8, 1, 12e9080) at netbsd:config_attach_loc+0x158
psycho_attach(3aadf00, 3aaa200, 1c059f0, 1810800, 1813000, 2) at netbsd:psycho_attach+0x4f8
config_attach_loc(3aaa200, 1808a98, 0, 1c059f0, 1, 1314260) at netbsd:config_attach_loc+0x158
mainbus_attach(0, 3aadf00, 0, 1810800, 1464800, 1314000) at netbsd:mainbus_attach+0x474
config_attach_loc(3aadf00, 1808858, 0, 0, 1, 7) at netbsd:config_attach_loc+0x158
cpu_configure(1469000, 1208ac0, 0, 1810800, 1814c00, 146a9c8) at netbsd:cpu_configure+0x1d0
configure(185f0b8, 2, 187c400, 1252680, 1468c90, 1884c00) at netbsd:configure+0x3c         
main(0, 11ebe00, ffffffffffffe000, 1fff, 1889800, 1c02000) at netbsd:main+0x1dc
cpu_initialize(f0067260, 0, 10fc00, fff83d00, 10ea78, fff83e00) at netbsd:cpu_initialize+0xf8

And here is what "gdb" found:

(Continue reading)

Matthias Scheler | 2 Feb 2008 12:58

Picon

Re: 4.99.51 kernel crashes while detecting a PCI card

On Sat, Feb 02, 2008 at 10:55:02AM +0000, Matthias Scheler wrote:
> And here is what "gdb" found:
> 
> (gdb) list *(config_attach_loc+0x158)
> 0x1221798 is in config_attach_loc (/src/NetBSD-current/src/sys/kern/subr_autoconf.c:1276).
> 1271
> 1272    #if defined(SPLASHSCREEN) && defined(SPLASHSCREEN_PROGRESS)
> 1273            if (splash_progress_state)
> 1274                    splash_progress_update(splash_progress_state);
> 1275    #endif
> 1276            (*dev->dv_cfattach->ca_attach)(parent, dev, aux);
> 1277    #if defined(SPLASHSCREEN) && defined(SPLASHSCREEN_PROGRESS)
> 1278            if (splash_progress_state)
> 1279                    splash_progress_update(splash_progress_state);
> 1280    #endif

I realized that the backtrace didn't give the right function. Here is
what disassembling at the report instruction pointer prints out:

0x12eac60 is in pci_conf_read (/src/NetBSD-current/src/sys/arch/sparc64/dev/pci_machdep.c:379).
374     /* assume we are mapped little-endian/side-effect */
375     pcireg_t
376     pci_conf_read(pci_chipset_tag_t pc, pcitag_t tag, int reg)
377     {
378             struct psycho_pbm *pp = pc->cookie;
379             struct psycho_softc *sc = pp->pp_sc;
380             pcireg_t val = (pcireg_t)~0;
381     
382             DPRINTF(SPDB_CONF, ("pci_conf_read: tag %lx reg %x ", 
383                     (long)tag, reg));

(Continue reading)

Group

gmane.os.netbsd.ports.sparc64.

Search Archive

Page

1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9

Articles in period: 85

February 2008

Language

Change language

Options

Current view: Threads only / Showing only 20 lines / Not hiding cited text.
Change to All messages, whole messages, or hide cited text.

Post a message
NNTP Newsgroup
Classic Gmane web interface
XML

XML

RSS Feed
List Information

About Gmane

Recent entries

New firefox on sparc64 (13 May 2013)
Problem building perl5 on NetBSD 6 / Sparc64 (4 May 2013)
Apply for loan @ 2%... (10 Apr 2013)
power failure interrupt on Netra X1 (25 Feb 2013)
DIMM population information (25 Feb 2013)
seeprom(4) support (8 Feb 2013)
U80 console trouble (6 Feb 2013)
SB1000/SB2000 now quieter (3 Feb 2013)
pkgsrc/emulators/tme + tap(4) (29 Jan 2013)
RE Archetech Last Minute A4 Page Offer (27 Jan 2013)
V240 Status? (20 Jan 2013)
T1 support? (21 Dec 2012)
npf with 6.0 failing to load module (14 Dec 2012)
SunFire v240 hardware question (2 Dec 2012)
Installing Perl5 not working (24 Nov 2012)
Using %stick where available (6 Nov 2012)
ultra45 port - mostly works! (30 Oct 2012)
Early crash on SB2000 (panic: init: no memory (29 Oct 2012)
Why is checksumming so much slower on a sparc64? (25 Oct 2012)
bge fail on netboot of NetBSD-6 on v215 and v240 (17 Oct 2012)
Returning to firmware broken on USii? (9 Oct 2012)

Archives

12	May 2013
1	April 2013
1	March 2013
13	February 2013
29	January 2013
16	December 2012
39	November 2012
24	October 2012
3	September 2012
7	August 2012
14	July 2012
3	June 2012
13	May 2012
2	April 2012
18	March 2012
31	February 2012
1	January 2012
4	December 2011
26	November 2011
1	October 2011
19	September 2011
17	August 2011
11	July 2011
24	June 2011
37	May 2011
23	April 2011
2	March 2011
59	February 2011
38	January 2011
17	December 2010
14	November 2010
18	October 2010
5	September 2010
9	July 2010
30	June 2010
88	May 2010
22	April 2010
54	March 2010
13	February 2010
72	January 2010
7	December 2009
21	November 2009
9	October 2009
14	September 2009
79	August 2009
35	July 2009
13	June 2009
40	May 2009
15	April 2009
26	March 2009
80	February 2009
51	January 2009
29	December 2008
17	November 2008
57	October 2008
38	September 2008
46	August 2008
40	July 2008
62	June 2008
57	May 2008
60	April 2008
137	March 2008
82	February 2008
60	January 2008
49	December 2007
87	November 2007
30	October 2007
115	September 2007
72	August 2007
46	July 2007
33	June 2007
58	May 2007
83	April 2007
42	March 2007
50	February 2007
24	January 2007
66	December 2006
29	November 2006
108	October 2006
38	September 2006
72	August 2006
109	July 2006
36	June 2006
34	May 2006
108	April 2006
32	March 2006
70	February 2006
89	January 2006
88	December 2005
97	November 2005
102	October 2005
107	September 2005
86	August 2005
142	July 2005
196	June 2005
159	May 2005
165	April 2005
193	March 2005
218	February 2005
152	January 2005
74	December 2004
91	November 2004
204	October 2004
62	September 2004
56	August 2004
81	July 2004
141	June 2004
56	May 2004
2	April 2004
55	March 2004
68	February 2004
80	January 2004
64	December 2003
56	November 2003
48	October 2003
49	September 2003
78	August 2003
70	July 2003
66	June 2003
72	May 2003
43	April 2003
65	March 2003
95	February 2003
79	January 2003
62	December 2002
97	November 2002
122	October 2002
67	September 2002
89	August 2002
69	July 2002
99	June 2002
88	May 2002
54	April 2002
22	March 2002

Design

Zawodny | Right Menu | Left Menu

Your Own Design

Paste the URL of your CSS below. Download CSS template