[parisc-linux] c110 panic when try to boot 2.6.10-rc1-pa11?

[parisc-linux] c110 panic when try to boot 2.6.10-rc1-pa11?

[Joel Soete]

hello pa*,

2.6.10-rc1-pa10 works fine on this c110 (as soon as I don't try to use the narrow se crtlr)

unfortunately 2.6.10-rc1-pa11 failled to boot with following message:

SCSI device sda: drive cache: write back
  sda:Backtrace:
  [<10152bd8>] __blk_queue_bounce+0x210/0x290
  [<10244814>] __make_request+0x68/0x508
  [<10245064>] generic_make_request+0x20c/0x2bc
  [<10245180>] submit_bio+0x6c/0x158
  [<1016b710>] submit_bh+0x128/0x1d4
  [<1016a3f8>] block_read_full_page+0x1f8/0x340
  [<101451c0>] read_cache_page+0x90/0x320
  [<1019c374>] read_dev_sector+0x34/0x110
  [<1019cb20>] msdos_partition+0x98/0x5fc
  [<1019bd04>] check_partition+0xd0/0x164
  [<1019c174>] register_disk+0xd0/0x160
  [<10246e58>] add_disk+0x54/0x68
  [<10274764>] sd_probe+0x240/0x39c
  [<1023d87c>] bus_match+0x4c/0x84
  [<1023d9e8>] driver_attach+0x80/0xa8
  [<1023df48>] bus_add_driver+0xb0/0xd8


Kernel Fault: Code=26 regs=109949c0 (Addr=0000000c)
      YZrvWESTHLNXBCVMcbcbcbcbOGFRQPDI
PSW: 00000000000001001111111100001111 Not tainted
r00-03  00000000 10201000 10152bd8 00000000
r04-07  109a5790 00000200 00000000 00000010
r08-11  10994818 00000000 2fd092f4 1045d810
r12-15  00000000 10000000 1045d810 1045d810
r16-19  1016e9d8 f00000ac f00000a4 10994954
r20-23  10994954 00000000 10991958 10994958
r24-27  00000000 00000010 00000000 1038c010
r28-31  109a2120 10961bd4 109949c0 10146ee4
sr0-3   00000000 00000000 00000000 00000000
sr4-7   00000000 00000000 00000000 00000000

IASQ: 00000000 00000000 IAOQ: 10146ed0 10146ed4
  IIR: 0c781099    ISR: 00000000  IOR: 0000000c
  CPU:        0   CR30: 10994000 CR31: 103e4000
  ORIG_R28: 1045a810
  IAOQ[0]: mempool_alloc+0xa0/0x18c
  IAOQ[1]: mempool_alloc+0xa4/0x18c
  RP(r2): __blk_queue_bounce+0x210/0x290
Kernel panic - not syncing: Kernel Fault
  <0>Rebooting in 120 seconds..

Too bad :(

the diff between src trees is very small:
diff -NaurX dontdiff linux-2.6.10-rc1-pa10/Documentation/parisc/00-INDEX linux-2.6.10-rc1-pa11/Documentation/parisc/00-INDEX
--- linux-2.6.10-rc1-pa10/Documentation/parisc/00-INDEX 2004-08-23 10:52:25.000000000 +0200
+++ linux-2.6.10-rc1-pa11/Documentation/parisc/00-INDEX 2004-11-11 17:18:02.000000000 +0100
[snip]
diff -NaurX dontdiff linux-2.6.10-rc1-pa10/Documentation/parisc/todo linux-2.6.10-rc1-pa11/Documentation/parisc/todo
--- linux-2.6.10-rc1-pa10/Documentation/parisc/todo     2004-08-27 19:18:58.000000000 +0200
+++ linux-2.6.10-rc1-pa11/Documentation/parisc/todo     2004-11-11 17:18:02.000000000 +0100
[snip]
diff -NaurX dontdiff linux-2.6.10-rc1-pa10/Makefile linux-2.6.10-rc1-pa11/Makefile
--- linux-2.6.10-rc1-pa10/Makefile      2004-11-06 15:55:05.000000000 +0100
+++ linux-2.6.10-rc1-pa11/Makefile      2004-11-11 17:18:01.000000000 +0100
@@ -1,7 +1,7 @@
  VERSION = 2
  PATCHLEVEL = 6
  SUBLEVEL = 10
-EXTRAVERSION =-rc1-pa10
+EXTRAVERSION =-rc1-pa11
  NAME=Woozy Numbat

  # *DOCUMENTATION*
diff -NaurX dontdiff linux-2.6.10-rc1-pa10/arch/parisc/configs/a500_defconfig linux-2.6.10-rc1-pa11/arch/parisc/configs/a500_defconfig
--- linux-2.6.10-rc1-pa10/arch/parisc/configs/a500_defconfig    2004-08-23 10:52:33.000000000 +0200
+++ linux-2.6.10-rc1-pa11/arch/parisc/configs/a500_defconfig    2004-11-11 17:18:05.000000000 +0100
[snip]
diff -NaurX dontdiff linux-2.6.10-rc1-pa10/arch/parisc/lib/memcpy.c linux-2.6.10-rc1-pa11/arch/parisc/lib/memcpy.c
--- linux-2.6.10-rc1-pa10/arch/parisc/lib/memcpy.c      2004-10-16 18:08:06.000000000 +0200
+++ linux-2.6.10-rc1-pa11/arch/parisc/lib/memcpy.c      2004-11-13 17:15:17.000000000 +0100
@@ -110,8 +110,8 @@
         "\t" EXC_WORD "\t1b\n"                          \
         "\t" EXC_WORD "\t" #_e "\n"                     \
         "\t.previous\n"                                 \
-       : _tt(_t), "+r"(_a)                             \
-       : "1"(_a)                                       \
+       : _tt(_t)                                       \
+       : "r"(_a)                                       \
         : "r8")

  #define def_store_ai_insn(_insn,_sz,_tt,_s,_a,_t,_e)   \
@@ -121,16 +121,16 @@
         "\t" EXC_WORD "\t1b\n"                          \
         "\t" EXC_WORD "\t" #_e "\n"                     \
         "\t.previous\n"                                 \
-       : "+r"(_a)                                      \
-       : _tt(_t), "0"(_a)                              \
+       :                                               \
+       : "r"(_a), _tt(_t)                              \
         : "r8")

  #define ldbma(_s, _a, _t, _e) def_load_ai_insn(ldbs,1,"=r",_s,_a,_t,_e)
  #define stbma(_s, _t, _a, _e) def_store_ai_insn(stbs,1,"r",_s,_a,_t,_e)
  #define ldwma(_s, _a, _t, _e) def_load_ai_insn(ldw,4,"=r",_s,_a,_t,_e)
  #define stwma(_s, _t, _a, _e) def_store_ai_insn(stw,4,"r",_s,_a,_t,_e)
-#define flddma(_s, _a, _t, _e) def_load_ai_insn(fldd,8,"=f",_s,_a,_t,_e)
-#define fstdma(_s, _t, _a, _e) def_store_ai_insn(fstd,8,"f",_s,_a,_t,_e)
+#define flddma(_s, _a, _t, _e) def_load_ai_insn(fldd,8,"=Z",_s,_a,_t,_e)
+#define fstdma(_s, _t, _a, _e) def_store_ai_insn(fstd,8,"Z",_s,_a,_t,_e)

  #define def_load_insn(_insn,_tt,_s,_o,_a,_t,_e)        \
         __asm__ __volatile__ (                          \
@@ -297,7 +297,7 @@
  unsigned long pa_memcpy(void *dstp, const void *srcp, unsigned long len)
  {
         register unsigned long src, dst, t1, t2, t3;
-       register char *pcs, *pcd;
+       register unsigned char *pcs, *pcd;
         register unsigned int *pws, *pwd;
         register double *pds, *pdd;
         unsigned long ret = 0;
diff -NaurX dontdiff linux-2.6.10-rc1-pa10/arch/parisc/mm/init.c linux-2.6.10-rc1-pa11/arch/parisc/mm/init.c
--- linux-2.6.10-rc1-pa10/arch/parisc/mm/init.c 2004-10-30 17:37:52.000000000 +0200
+++ linux-2.6.10-rc1-pa11/arch/parisc/mm/init.c 2004-11-13 19:52:50.425203881 +0100
@@ -59,8 +59,6 @@

  static struct resource sysram_resources[MAX_PHYSMEM_RANGES];

-static unsigned long max_pfn;
-
  /* The following array is initialized from the firmware specific
   * information retrieved in kernel/inventory.c.
   */
diff -NaurX dontdiff linux-2.6.10-rc1-pa10/drivers/parisc/sba_iommu.c linux-2.6.10-rc1-pa11/drivers/parisc/sba_iommu.c
--- linux-2.6.10-rc1-pa10/drivers/parisc/sba_iommu.c    2004-11-12 21:42:51.000000000 +0100
+++ linux-2.6.10-rc1-pa11/drivers/parisc/sba_iommu.c    2004-11-11 17:18:13.000000000 +0100
[snip]
=========><=========
(this last one doesn't affect c110)

I first suspected the more complex i.e. memcpy.c and revert it first but it doesn't help :O

OTC reverting the second one (i.e. init.c) help a bit in the way the system boot again but wrongly as you can judge by your self 
with following init errors:

Cleaning up ifupdown...done.
Starting raid devices:
sed: -e expression #1, char 87: unterminated `s' command
sed: -e expression #1, char 15: unterminated `s' command
sed: -e expression #1, char 15: unterminated `s' command
sed: -e expression #1, char 15: unterminated `s' command
sed: -e expression #1, char 15: unterminated `s' command
sed: -e expression #1, char 15: unterminated `s' command
sed: -e expression #1, char 15: unterminated `s' command
done.
Setting up LVM Volume Groups...

[...]
Discovering hardware: discover: ../../lib/device-xml.c:180: create_device: Assertion `vendor_name != ((void *)0)' f.
Running 0dns-down to make sure resolv.conf is ok..awk: cmd. line:1: BEGIN {FS = "." } /resolv\.conf\.bak/ {print $Ntawk: cmd. line:1:
                                                                           ^ unexpected newline or end of string
done.
/dev/shm/network/...Initializing: /etc/network/ifstate.
grep: /etc/network/option: No such file or directory
Setting up IP spoofing protection: rp_filter.
grep: /etc/network/option: No such file or directory
grep: /etc/network/option: No such file or directory
Configuring network interfaces...run-parts: failed to open directory /etc/network/if-pre-up.: No such file or direcy
Disabled Privacy Extensions on device 103ac490(lo)
run-parts: failed to open directory /etc/network/if-up.: No such file or directory
run-parts: failed to open directory /etc/network/if-pre-up.: No such file or directory
eth0: link ok.
run-parts: failed to open directory /etc/network/if-up.: No such file or directory
done.
Usage: /etc/init.d/portmap {start|stop|reload|restart}
Setting the System Clock using the Hardware Clock as reference...
System Clock set. Local time: Sat Nov 13 19:20:14 CET 2004
Usage: /etc/init.d/ntpdate {start|stop|restart|reload|force-reload}
cat: /proc/sys/kernel/random/poolsiz: No such file or directory
Initializing random number generator...cat: /var/lib/urandom/random-see: No such file or directory
urandom start: failed.
done.
[...]
Starting portmap daemon: portmap.
grep: /etc/inetd.con: No such file or directory
Starting MTA: update-exim4.co: invalid option -- l
Terminating...
grep: /etc/inetd.con: No such file or directory
Starting file alteration monitor: FAM.
Starting mouse interface server: gpm.
Starting printer spooler: lpd/sbin/start-stop-daemo: option `--exec' requires an argument
Try `/sbin/start-stop-daemo --help' for more information.
.
Starting printer spooler: lpd/sbin/start-stop-daemo: option `--exec' requires an argument
Try `/sbin/start-stop-daemo --help' for more information.
.
Starting network benchmark server: netserver.
rsync daemon not enabled in /etc/default/rsync
Starting OpenBSD Secure Shell server: sshd.
Starting the system activity data collector: sadc.
Setting up X font server socket directory /tmp/.font-unix...done.
Starting X font server: xfsusage: /usr/bin/X11/xf [-config config_file] [-daemon] [-droppriv] [-ls listen_socket] []
  already running.
Usage: /etc/init.d/xinetd {start|stop|reload|force-reload|restart}
Usage: nfs-common {start|stop|restart}
Usage: /etc/init.d/ntp-server {start|stop|restart|force-reload}
Starting RAID monitor daemon: mdadm -F.
Starting proxy server: squid.
Starting anac(h)ronistic cron: anacron.
Starting deferred execution scheduler: atd.
Starting periodic command scheduler: cron.
cat: /etc/X11/default-display-manage: No such file or directory
Not starting GNOME Display Manager (gdm); it is not the default display manager.

I didn't analyse in much detail but eg /etc/X11/default-display-manage exists but it's name seems to be longer:
# ll /etc/X11/default*                                                                                 ls: 
/etc/X11/default-display-manageß: No such file or directory

Randolph: any idea?

Thanks in advance for attention,
	Joel
_______________________________________________
parisc-linux mailing list
parisc-linux@lists.parisc-linux.org
http://lists.parisc-linux.org/mailman/listinfo/parisc-linux

Re: 2.6.10-rc1-pa16 still panic on c110 [WAS: [parisc-linux] c110 panic when try to boot 2.6.10-rc1-pa11?]

[Grant Grundler]

On Sat, Nov 20, 2004 at 08:37:08PM +0000, Joel Soete wrote:
> > IAOQ[1]: mempool_alloc+0xa4/0x18c
> > RP(r2): __blk_queue_bounce+0x210/0x290

My guess is this is because the default value is wrong...hrm..
but ccio driver should be clearing parisc_bus_is_phys.
Let me commit the pending change i have and then see how it goes.


> and I don't understand why reverting this patch:
> diff -u -r1.15 -r1.16
> --- linux-2.6/arch/parisc/mm/init.c	2004/10/24 05:47:27	1.15
> +++ linux-2.6/arch/parisc/mm/init.c	2004/11/10 19:55:37	1.16
> @@ -59,8 +59,6 @@
> 
>  static struct resource sysram_resources[MAX_PHYSMEM_RANGES];
> 
> -static unsigned long max_pfn;
> -
>  /* The following array is initialized from the firmware specific
>   * information retrieved in kernel/inventory.c.
>   */
> 
> solve the pb.
> 
> That doen't make sence to me as far as I test with success this kernel on 
> b180, b2k (32 and 64 bit) and n4k (smp) without pb
> (btw the same success with my rc2 merge on b2k [32 and 64 bit] and b180 :) )

hrm. We bounce when we think an address is greater than the number
of bits a device can DMA directly to. So either the max_pfn or
the dma_mask is being set wrong.

> Well the c110 is the only model I use with NCR 53C720 & LASI ASIC interface 
> (i.e. NCR 53C710)

Either of these drivers may not be setting the corresponding dma_mask.
I need to check.

grant
_______________________________________________
parisc-linux mailing list
parisc-linux@lists.parisc-linux.org
http://lists.parisc-linux.org/mailman/listinfo/parisc-linux

Re: 2.6.10-rc1-pa16 still panic on c110 [WAS: [parisc-linux] c110 panic when try to boot 2.6.10-rc1-pa11?]

[Grant Grundler]

On Sun, Nov 21, 2004 at 04:18:26PM +0000, Joel Soete wrote:
> That works far better: no more panic at boot time :))

good. I think the other problems you are have are either real HW problems,
bugs in the driver or bugs in the hard disk firmware.
You can rule out some of the hard disk firmware bugs by checking for
newer versions of the firmware either at HP's site (URL
is in the parisc-linux FAQ) or disk vendor's site.

Real HW might be "fixed" by a "low level format".
See scsiformat in scsitools package.

grant
_______________________________________________
parisc-linux mailing list
parisc-linux@lists.parisc-linux.org
http://lists.parisc-linux.org/mailman/listinfo/parisc-linux

Re: 2.6.10-rc1-pa16 still panic on c110 [WAS: [parisc-linux] c110 panic when try to boot 2.6.10-rc1-pa11?]

[Joel Soete]

Hello Grant,

Just a short followup:

Grant Grundler wrote:
> On Sun, Nov 21, 2004 at 04:18:26PM +0000, Joel Soete wrote:
> 
>>That works far better: no more panic at boot time :))
> 
> 
> good. I think the other problems you are have are either real HW problems,
I used one of the latest support+ cd I had (hpux 11.0 March 2003) and ODE DISKUTIL (I am not a hp engineer so I can get passwd for 
DISKEXPT, as well as LASIDIAG) (so that I don't have change system) but it doesn't detect any read error :(

> bugs in the driver or bugs in the hard disk firmware.

> You can rule out some of the hard disk firmware bugs by checking for
> newer versions of the firmware either at HP's site (URL
> is in the parisc-linux FAQ) 
Unfortunately I didn't find any fw upgrade available for this 'SEAGATE ST34573W' (presently fw revision HP11) into 
<ftp://us-ffs.external.hp.com/firmware_patches/hp/disk/catalog> and http://itrc.hp.com/ is unavailable right now :(

> or disk vendor's site.
> 
It's a hp disk supply so I would prefer a hp 'patch'

> Real HW might be "fixed" by a "low level format".
> See scsiformat in scsitools package.
> 
I will see as ultimate solution ;)

Thanks,
	Joel
_______________________________________________
parisc-linux mailing list
parisc-linux@lists.parisc-linux.org
http://lists.parisc-linux.org/mailman/listinfo/parisc-linux

Re: 2.6.10-rc1-pa16 still panic on c110 [WAS: [parisc-linux] c110 panic when try to boot 2.6.10-rc1-pa11?]

[Joel Soete]

Hello all,

Having far less time to spend to linux :_(, I come back shortly with those investigation:


Grant Grundler wrote:
> On Sun, Nov 21, 2004 at 04:18:26PM +0000, Joel Soete wrote:
> 
>>That works far better: no more panic at boot time :))
> 
> 
> good. I think the other problems you are have are either real HW problems,
> bugs in the driver or bugs in the hard disk firmware.
> You can rule out some of the hard disk firmware bugs by checking for
> newer versions of the firmware either at HP's site (URL
> is in the parisc-linux FAQ) or disk vendor's site.
> 
There are effectively a new fw revision for this disk but only available for hp engineer :-)

But getting back this disk to office, I had the idea to connect this disk to a b180 using the same lasi700 driver and drive exactely 
the same stress test (http://lists.parisc-linux.org/pipermail/parisc-linux/2004-November/025277.html) and no pb occured.

so could it be a hw pb with the scsi ctrlr ncr53c710 (the narrow se)?

Thanks again,
	Joel
_______________________________________________
parisc-linux mailing list
parisc-linux@lists.parisc-linux.org
http://lists.parisc-linux.org/mailman/listinfo/parisc-linux