Re: [Fastboot] Ia64 kdump patch

[Jay Lan <jlan@sgi.com>]

Hi Nanhai and horms,

Zou, Nanhai wrote:
> 
[snip]
>  Forget to mention,
>  If you are using same kernel as first and crash kernel, you'd better pass an additional 
>  "maxcpus=1" kernel parameter to the second kernel.

The "maxpus=1' did help. Without it, the kernel went into double panic.
However, with "maxcpus=1" i saw crash dump and Call Trace...

holism.engr.sgi.com login: SysRq : Trigger a crashdump 
         kernel BUG at kernel/irq/migration.c:39! 
                  bash[3213]: bugcheck! 0 [1] 
                           Modules linked in: radeon drm agpgart nfs 
lockd sunrpc binfmt_misc dm_mirror dmi 
                                               Pid: 3213, CPU 1, comm: 
                bash                                    psr : 
00001010085a2010 ifs : 800000000000038b ip  : [<a0000001000e1c00>] 
Notdip is at move_native_irq+0x1a0/0x360 
             unat: 0000000000000000 pfs : 000000000000038b rsc : 
0000000000000000            rnat: d6304e388f73eebc bsps: 
9d40fde11cbbf11b pr  : 0000000000596599            ldrs: 
0000000000000000 ccv : 0000000000000000 fpsr: 0009804c8a70033f 
   csd : 0000000000000000 ssd : 0000000000000000 
            b0  : a0000001000e1c00 b6  : a000000100036060 b7  : 
a000000100010150            f6  : 1003e0000006c1c8a11a7 f7  : 
1003e0000000000000514                         f8  : 
1003e0000006c1c8a0c93 f9  : 1003e0000000000000001 
   f10 : 0fffd9999999996900000 f11 : 1003e0000000000000000 
            r1  : a000000100b66020 r2  : 000000000000048d r3  : 
a0000001009665b8            r8  : 000000000000002c r9  : 
0000000000003446 r10 : a00000010097d1f8            r11 : 
0000000000000000 r12 : e00000001a62f7e0 r13 : e00000001a628000 
   r14 : 0000000000004000 r15 : a00000010097d200 r16 : 0000000000000000 
            r17 : 0000000000004000 r18 : 0000000000000060 r19 : 
0000000000004000            r20 : a00000010097c430 r21 : 
0000000000000000 r22 : a00000010096a4b8            r23 : 
a00000010097d210 r24 : a00000010097d1f0 r25 : a000000100910c88 
   r26 : a000000100910c88 r27 : a000000100966270 r28 : 0000000000000034 
            r29 : 0000000000000034 r30 : 0000000000000000 r31 : 
a00000010097d1cc 
                                     Call Trace: 
                                               [<a000000100013880>] 
show_stack+0x40/0xa0 
                   spà0000001a62f370 bspà0000001a6294b8 
[<a0000001000144e0>] show_regs+0x840/0x880 
                                        spà0000001a62f540 
bspà0000001a629460         [<a0000001000363a0>] die+0x1c0/0x2e0 
 
spà0000001a62f540 bspà0000001a629418         [<a000000100036510>] 
die_if_kernel+0x50/0x80 
                   spà0000001a62f560 bspà0000001a6293e0 
[<a000000100037c10>] ia64_bad_break+0x270/0x4a0 
                                        spà0000001a62f560 
bspà0000001a6293b8         [<a00000010000c320>] 
ia64_leave_kernel+0x0/0x280 
                   spà0000001a62f610 bspà0000001a6293b8 
[<a0000001000e1c00>] move_native_irq+0x1a0/0x360 
                                        spà0000001a62f7e0 
bspà0000001a629360         [<a00000010004ecb0>] 
iosapic_end_level_irq+0x30/0xe0 
                   spà0000001a62f7e0 bspà0000001a629340 
[<a00000010005d620>] machine_crash_shutdown+0x5c0/0x680 
                                        spà0000001a62f7e0 
bspà0000001a629300         [<a0000001000d4fd0>] crash_kexec+0x70/0xc0 
 
spà0000001a62fc60 bspà0000001a6292e0         [<a00000010045dfc0>] 
sysrq_handle_crashdump+0x20/0x40 
                   spà0000001a62fe20 bspà0000001a6292b8 
[<a00000010045e1a0>] __handle_sysrq+0x160/0x300 
                                        spà0000001a62fe20 
bspà0000001a629268         [<a0000001001b9770>] 
write_sysrq_trigger+0xb0/0xe0 
                   spà0000001a62fe20 bspà0000001a629238 
[<a00000010012f1d0>] vfs_write+0x1b0/0x340 
                                        spà0000001a62fe20 
bspà0000001a6291e0         [<a00000010012fe50>] sys_write+0x70/0xe0 
 
spà0000001a62fe20 bspà0000001a629168         [<a00000010000c180>] 
ia64_ret_from_syscall+0x0/0x20 
                   spà0000001a62fe30 bspà0000001a629168 
[<a000000000010620>] __kernel_syscall_via_break+0x0/0x20 
                                        spà0000001a630000 
bspà0000001a629168 
                                     Fedora Core release 5 (Bordeaux) 
                                              Kernel 2.6.18-rc2 on an 
ia64 
 
holism.engr.sgi.com login:

However, the second kernel was not booted and system remained
alive.

The /proc/{sysrq-trigger,iomem} after the trigger were as below:
(holism,9) ls -l /proc/{sysrq-trigger,iomem}
-r--r--r-- 1 root root 0 Aug  9 08:10 /proc/iomem
--w------- 1 root root 0 Aug  9 08:10 /proc/sysrq-trigger
(holism,10)

I built the kernel based on 2.6.18-rc2 with Nanhai's
kexec-kdump-ia64-2.6.16.patch and my fixes to
   - replace  irq_descp(dev->irq) with irq_desc + dev->irq
   - replace  desc->handle with desc-chip
for compilation.

As to kexec-tools, i use kdump9 with Nanhai's
kexec-tools-kdump9-ia64-zou.patch.

What did i miss here? Do i miss any patch(es)?
My machine is a HP zx6000 loaded with FC5.

Thanks!
  - jay

> 
> Thanks
> Zou Nan hai