----- Original Message -----
>
> Hi Dave,
>
> I hope I have captured everything you asked for here, if remote
> debugging over e-mail is too tedious, I can arrange to post a
> vmlinux/vmcore on our FTP site (roughly 600MB together).
Sure, you can do that if you'd like.
But anyway, the crash -d1 output is illuminating. You've got a 16-cpu
system, with all cpus online. But the compressed kdump header only
saw 1 cpu when it was created:
> header: 2cc1fe0
> signature: "KDUMP "
> header_version: 4
> utsname:
> sysname: Linux
> nodename: bahamut.mno.stratus.com
> release: 2.6.32-131.0.15.el6.exp10.bz16586.x86_64
> version: #1 SMP Thu Jun 16 13:13:45 EDT 2011
> machine: x86_64
> domainname: sraeng
> timestamp:
> tv_sec: 4e4fe6e3
> tv_usec: 0
> status: 0 ()
> block_size: 4096
> sub_hdr_size: 1
> bitmap_blocks: 288
> max_mapnr: 4718592
> total_ram_blocks: 0
> device_blocks: 0
> written_blocks: 0
> current_cpu: 0
> nr_cpus: 1 <== should be 16
> tasks[nr_cpus]: 0
and farther on, here is the array of note pointers that I
was asking about:
> num_prstatus_notes: 1
> notes_buf: 2cc4000
> notes[0]: 2cc4000
Since dumpfile header's nr_cpus was 1, the array has only one entry.
I cannot explain that. But the crash utility can only deal with what
it finds in the dumpfile.
Furthermore, these two error messages indicate that memory containing
per-cpu data was "excluded" are of prime importance here:
> crash: page excluded: kernel virtual address: ffffffff81bb3b00 type: "cpu number (per_cpu)"
> crash: page excluded: kernel virtual address: ffffffff81bb3b00 type: "cpu number (per_cpu)"
The fact that the page was specifically "page excluded" is troubling,
because it should *never* have been filtered by makedumpfile -d<level>.
But, since the dumpfile indicates that the crucial per-cpu page was filtered,
so there's nothing that the crash utility can do about it.
I'm guessing that, even though you are able to get to a prompt
with --no_elf_notes, any command that depends upon per-cpu data
would fail. Although, it might be interesting to know *which*
cpu was in play when those two error messages were displayed
in x86_64_per_cpu_init() and x86_64_get_smp_cpus(). There is
a loop in both functions -- can you dump out which cpu's
per-cpu data was inaccessible?
Thanks,
Dave
>
>
> *** Setup some breakpoints to watch bt->machdep:
>
> get_netdump_regs_x86_64(struct bt_info *bt, ulong *ripp, ulong *rspp)
> {
> ...
>
> if (((NETDUMP_DUMPFILE() || KDUMP_DUMPFILE()) &&
> VALID_STRUCT(user_regs_struct) && (bt->task == tt->panic_task)) ||
> (KDUMP_DUMPFILE() && (kt->flags & DWARF_UNWIND) &&
> (bt->flags & BT_DUMPFILE_SEARCH))) {
> ...
> 2287 bt->machdep = (void *)user_regs;
> ...
>
> if (ELF_NOTES_VALID() &&
> (bt->flags & BT_DUMPFILE_SEARCH) && DISKDUMP_DUMPFILE() &&
> (note = (Elf64_Nhdr *)
> diskdump_get_prstatus_percpu(bt->tc->processor))) {
> ...
> 2306 bt->machdep = (void *)user_regs;
> ...
>
> (gdb) break get_netdump_regs_x86_64
> Breakpoint 1 at 0x519740: file netdump.c, line 2238.
> (gdb) break netdump.c:2287
> Breakpoint 2 at 0x519970: file netdump.c, line 2287.
> (gdb) break netdump.c:2306
> Breakpoint 3 at 0x5199e7: file netdump.c, line 2306.
> (gdb) r
>
> please wait... (determining panic task)
> Breakpoint 1, get_netdump_regs_x86_64 (bt=0x7fffffffcd70,
> ripp=0x7fffffffcce0,
> rspp=0x7fffffffcce8) at netdump.c:2238
> 2238 {
> (gdb) c
> Continuing.
>
> Breakpoint 3, get_netdump_regs_x86_64 (bt=0x7fffffffcd70,
> ripp=0x7fffffffcce0,
> rspp=0x7fffffffcce8) at netdump.c:2306
> 2306 bt->machdep = (void *)user_regs;
> (gdb) p user_regs
> $1 = 0xd14084 ""
> (gdb) c
> Continuing.
>
> Breakpoint 1, get_netdump_regs_x86_64 (bt=0x7fffffffcd70,
> ripp=0x7fffffffcce0,
> rspp=0x7fffffffcce8) at netdump.c:2238
> 2238 {
> (gdb) c
> Continuing.
>
> Program received signal SIGSEGV, Segmentation fault.
> x86_64_get_dumpfile_stack_frame (rsp=0x7fffffffcce8,
> rip=0x7fffffffcce0,
> bt_in=0x7fffffffcd70) at x86_64.c:4183
> 4183 ur_rip = ULONG(user_regs +
>
>
> *** So in its second invocation, get_netdump_regs_x86_64() never sets
> bt->machdep (only breakpoint 1 fired)
>
> *** Let's see what diskdump_get_prstatus_percpu() is returning
>
> (gdb) break diskdump_get_prstatus_percpu
> Breakpoint 1 at 0x526070: file diskdump.c, line 1451.
> (gdb) r
> please wait... (determining panic task)
> Breakpoint 1, diskdump_get_prstatus_percpu (cpu=0) at diskdump.c:1451
> 1451 return dd->nt_prstatus_percpu[cpu];
> (gdb) display dd->nt_prstatus_percpu[0]@16
> 1: dd->nt_prstatus_percpu[0]@16 = {0xd1c000, 0x0, 0x0, 0xd26472,
> 0xbf35ab2,
> 0xd26472, 0x200000012, 0xd1c850, 0xd1c600, 0x1010000012b,
> 0xffffffff814e4fa0, 0x14e4fa0, 0x4270, 0x0, 0x0, 0x0}
> (gdb) c
> Continuing.
>
> Breakpoint 1, diskdump_get_prstatus_percpu (cpu=1) at diskdump.c:1451
> 1451 return dd->nt_prstatus_percpu[cpu];
> 1: dd->nt_prstatus_percpu[0]@16 = {0xd1c000, 0x0, 0x0, 0xd26472,
> 0xbf35ab2,
> 0xd26472, 0x200000012, 0xd1c850, 0xd1c600, 0x1010000012b,
> 0xffffffff814e4fa0, 0x14e4fa0, 0x4270, 0x0, 0x0, 0x0}
>
>
> *** See crash -d1 vmlinux vmcore output at the bottom of the mail,
> particularly the part that says...
>
> crash: page excluded: kernel virtual address: ffffffff81bb3b00 type:
> "cpu number (per_cpu)"
> crash: get_cpus_present: present: 16
>
>
>
> *** Bogus note->n_descsz value
> *** Apply first patch to get us further into ELF Note processing
>
> >From inside netdump.c :: get_regs_from_note() at the point of the
> >fault, I don't see dd->nt_prstatus[], for dd is now type
> >*diskdump_data... The *note passed in can be found in
> >dd->nt_prstatus_percpu[] however...
>
> please wait... (determining panic task)
> Program received signal SIGSEGV, Segmentation fault.
> get_regs_from_note (note=0xd26472 "\b", ip=0x7fffffffc590,
> sp=0x7fffffffc598)
> at netdump.c:2221
> 2221 *sp = ULONG(user_regs + offset_sp);
> (gdb) p/x *((Elf64_Nhdr *)note)
> $1 = {n_namesz = 0x8, n_descsz = 0xccf80000, n_type = 0x8}
> (gdb) p dd->nt_prstatus_percpu[0]@16
> $2 = {0xd1c000, 0x0, 0x0, 0xd26472, 0xbf35ab2, 0xd26472, 0x200000012,
> 0xd1c850, 0xd1c600, 0x1010000012b, 0xffffffff814e4fa0, 0x14e4fa0,
> 0x4270,
> 0x0, 0x0, 0x0}
> (gdb) ptype dd
> type = struct diskdump_data {
> char *filename;
> ulong flags;
> int dfd;
> FILE *ofp;
> int machine_type;
> struct disk_dump_header *header;
> struct disk_dump_sub_header *sub_header;
> struct kdump_sub_header *sub_header_kdump;
> size_t data_offset;
> int block_size;
> int block_shift;
> char *bitmap;
> int bitmap_len;
> char *dumpable_bitmap;
> int byte;
> int bit;
> char *compressed_page;
> char *curbufptr;
> unsigned char *notes_buf;
> void **nt_prstatus_percpu;
> uint num_prstatus_notes;
> struct page_cache_hdr page_cache_hdr[16];
> char *page_cache_buf;
> int evict_index;
> ulong evictions;
> ulong cached_reads;
> ulong *valid_pages;
> ulong accesses;
> } *
>
>
>
> *** Unpatched crash -d1 vmlinux vmcore output:
>
> crash 5.1.8
> Copyright (C) 2002-2011 Red Hat, Inc.
> Copyright (C) 2004, 2005, 2006 IBM Corporation
> Copyright (C) 1999-2006 Hewlett-Packard Co
> Copyright (C) 2005, 2006 Fujitsu Limited
> Copyright (C) 2006, 2007 VA Linux Systems Japan K.K.
> Copyright (C) 2005 NEC Corporation
> Copyright (C) 1999, 2002, 2007 Silicon Graphics, Inc.
> Copyright (C) 1999, 2000, 2001, 2002 Mission Critical Linux, Inc.
> This program is free software, covered by the GNU General Public
> License,
> and you are welcome to change it and/or distribute copies of it under
> certain conditions. Enter "help copying" to see the conditions.
> This program has absolutely no warranty. Enter "help warranty" for
> details.
>
> compressed kdump: header->utsname.machine: x86_64
> diskdump_data:
> filename: vmcore
> flags: 6 (KDUMP_CMPRS_LOCAL|ERROR_EXCLUDED)
> dfd: 3
> ofp: 0
> machine_type: 62 (EM_X86_64)
>
> header: 2cc1fe0
> signature: "KDUMP "
> header_version: 4
> utsname:
> sysname: Linux
> nodename: bahamut.mno.stratus.com
> release: 2.6.32-131.0.15.el6.exp10.bz16586.x86_64
> version: #1 SMP Thu Jun 16 13:13:45 EDT 2011
> machine: x86_64
> domainname: sraeng
> timestamp:
> tv_sec: 4e4fe6e3
> tv_usec: 0
> status: 0 ()
> block_size: 4096
> sub_hdr_size: 1
> bitmap_blocks: 288
> max_mapnr: 4718592
> total_ram_blocks: 0
> device_blocks: 0
> written_blocks: 0
> current_cpu: 0
> nr_cpus: 1
> tasks[nr_cpus]: 0
>
> sub_header: 0 (n/a)
>
> sub_header_kdump: 2cc2ff0
> phys_base: 0
> dump_level: 31 (0x1f)
> (DUMP_EXCLUDE_ZERO|DUMP_EXCLUDE_CACHE|DUMP_EXCLUDE_CACHE_PRI|DUMP_EXCLUDE_USER_DATA|DUMP_EXCLUDE_FREE)
> offset_vmcoreinfo: 11bc
> size_vmcoreinfo: 1392
> OSRELEASE=2.6.32-131.0.15.el6.exp10.bz16586.x86_64
> PAGESIZE=4096
> SYMBOL(init_uts_ns)=ffffffff81a2e8c0
> SYMBOL(node_online_map)=ffffffff81ba0860
> SYMBOL(swapper_pg_dir)=ffffffff81a25000
> SYMBOL(_stext)=ffffffff81000198
> SYMBOL(vmlist)=ffffffff81ee60b8
> SYMBOL(mem_section)=ffffffff81ef03c0
> LENGTH(mem_section)=4096
> SIZE(mem_section)=32
> OFFSET(mem_section.section_mem_map)=0
> SIZE(page)=56
> SIZE(pglist_data)=212416
> SIZE(zone)=34496
> SIZE(free_area)=88
> SIZE(list_head)=16
> SIZE(nodemask_t)=64
> OFFSET(page.flags)=0
> OFFSET(page._count)=8
> OFFSET(page.mapping)=24
> OFFSET(page.lru)=40
> OFFSET(pglist_data.node_zones)=0
> OFFSET(pglist_data.nr_zones)=212288
> OFFSET(pglist_data.node_start_pfn)=212312
> OFFSET(pglist_data.node_spanned_pages)=212328
> OFFSET(pglist_data.node_id)=212336
> OFFSET(zone.free_area)=32864
> OFFSET(zone.vm_stat)=34032
> OFFSET(zone.spanned_pages)=34344
> OFFSET(free_area.free_list)=0
> OFFSET(list_head.next)=0
> OFFSET(list_head.prev)=8
> OFFSET(vm_struct.addr)=8
> LENGTH(zone.free_area)=11
> SYMBOL(log_buf)=ffffffff81a37210
> SYMBOL(log_end)=ffffffff81d5b820
> SYMBOL(log_buf_len)=ffffffff81a37208
> SYMBOL(logged_chars)=ffffffff81ddb920
> LENGTH(free_area.free_list)=5
> NUMBER(NR_FREE_PAGES)=0
> NUMBER(PG_lru)=5
> NUMBER(PG_private)=11
> NUMBER(PG_swapcache)=16
> SYMBOL(phys_base)=ffffffff81a2d010
> SYMBOL(init_level4_pgt)=ffffffff81a25000
> SYMBOL(node_data)=ffffffff81b9cda0
> LENGTH(node_data)=512
> CRASHTIME=1313859299
> offset_note: 1040
> size_note: 1780
> num_prstatus_notes: 1
> notes_buf: 2cc4000
> notes[0]: 2cc4000
> NT_PRSTATUS_offset: 1040
>
> data_offset: 122000
> block_size: 4096
> block_shift: 12
> bitmap: 7fa5296fc010
> bitmap_len: 1179648
> dumpable_bitmap: 7fa528890010
> byte: 0
> bit: 0
> compressed_page: 2cdeb30
> curbufptr: 0
>
> page_cache_hdr[0]:
> pg_flags: 0 ()
> pg_addr: 0
> pg_bufptr: 2cceb20
> pg_hit_count: 0
> page_cache_hdr[1]:
> pg_flags: 0 ()
> pg_addr: 0
> pg_bufptr: 2ccfb20
> pg_hit_count: 0
> page_cache_hdr[2]:
> pg_flags: 0 ()
> pg_addr: 0
> pg_bufptr: 2cd0b20
> pg_hit_count: 0
> page_cache_hdr[3]:
> pg_flags: 0 ()
> pg_addr: 0
> pg_bufptr: 2cd1b20
> pg_hit_count: 0
> page_cache_hdr[4]:
> pg_flags: 0 ()
> pg_addr: 0
> pg_bufptr: 2cd2b20
> pg_hit_count: 0
> page_cache_hdr[5]:
> pg_flags: 0 ()
> pg_addr: 0
> pg_bufptr: 2cd3b20
> pg_hit_count: 0
> page_cache_hdr[6]:
> pg_flags: 0 ()
> pg_addr: 0
> pg_bufptr: 2cd4b20
> pg_hit_count: 0
> page_cache_hdr[7]:
> pg_flags: 0 ()
> pg_addr: 0
> pg_bufptr: 2cd5b20
> pg_hit_count: 0
> page_cache_hdr[8]:
> pg_flags: 0 ()
> pg_addr: 0
> pg_bufptr: 2cd6b20
> pg_hit_count: 0
> page_cache_hdr[9]:
> pg_flags: 0 ()
> pg_addr: 0
> pg_bufptr: 2cd7b20
> pg_hit_count: 0
> page_cache_hdr[10]:
> pg_flags: 0 ()
> pg_addr: 0
> pg_bufptr: 2cd8b20
> pg_hit_count: 0
> page_cache_hdr[11]:
> pg_flags: 0 ()
> pg_addr: 0
> pg_bufptr: 2cd9b20
> pg_hit_count: 0
> page_cache_hdr[12]:
> pg_flags: 0 ()
> pg_addr: 0
> pg_bufptr: 2cdab20
> pg_hit_count: 0
> page_cache_hdr[13]:
> pg_flags: 0 ()
> pg_addr: 0
> pg_bufptr: 2cdbb20
> pg_hit_count: 0
> page_cache_hdr[14]:
> pg_flags: 0 ()
> pg_addr: 0
> pg_bufptr: 2cdcb20
> pg_hit_count: 0
> page_cache_hdr[15]:
> pg_flags: 0 ()
> pg_addr: 0
> pg_bufptr: 2cddb20
> pg_hit_count: 0
>
> page_cache_buf: 2cceb20
> evict_index: 0
> evictions: 0
> accesses: 0
> cached_reads: 0
> valid_pages: 2ccc710
> crash: pv_init_ops exists: ARCH_PVOPS
> compressed kdump: phys_base: 0
> gdb vmlinux
> GNU gdb (GDB) 7.0
> Copyright (C) 2009 Free Software Foundation, Inc.
> License GPLv3+: GNU GPL version 3 or later <
> http://gnu.org/licenses/gpl.html >
> This is free software: you are free to change and redistribute it.
> There is NO WARRANTY, to the extent permitted by law. Type "show
> copying"
> and "show warranty" for details.
> This GDB was configured as "x86_64-unknown-linux-gnu"...
>
> cpu_possible_map: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
> cpu_present_map: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
> cpu_online_map: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
> base kernel version: 2.6.32
> verify_namelist:
> dumpfile /proc/version:
> Linux version 2.6.32-131.0.15.el6.exp10.bz16586.x86_64
> (root@xxxxxxxxxxxxxxxxxxxx) (gcc version 4.4.5 20110214 (Red Hat
> 4.4.5-6) (GCC) ) #1 SMP Thu Jun 16 13:13:45 EDT 2011
> vmlinux:
> Linux version 2.6.32-131.0.15.el6.exp10.bz16586.x86_64
> (root@xxxxxxxxxxxxxxxxxxxx) (gcc version 4.4.5 20110214 (Red Hat
> 4.4.5-6) (GCC) ) #1 SMP Thu Jun 16 13:13:45 EDT 2011
>
> crash: page excluded: kernel virtual address: ffffffff81bb3b00 type:
> "cpu number (per_cpu)"
> crash: get_cpus_present: present: 16
> crash: page excluded: kernel virtual address: ffffffff81bb3b00 type:
> "cpu number (per_cpu)"
> crash: get_cpus_present: present: 16
> IRQ stack link register: undetermined
> PAGESIZE=4096
> mem_section_size = 32768
> NR_SECTION_ROOTS = 4096
> NR_MEM_SECTIONS = 524288
> SECTIONS_PER_ROOT = 128
> SECTION_ROOT_MASK = 0x7f
> PAGES_PER_SECTION = 32768
> node_online_map: [3, 0, 0, 0, 0, 0, 0, 0] -> nodes online: 2
> node_table[0]:
> id: 0
> pgdat: ffff880000020040
> size: 0
> present: 0
> mem_map: ffffea0000000000
> start_paddr: 0
> start_mapnr: 0
> WARNING: sparsemem: invalid section number: 137438888923
> WARNING: sparsemem: invalid section number: 137438888923
> crash: invalid kernel virtual address: 0 type: "readstring
> characters"
> crash: invalid kernel virtual address: 0 type: "readstring
> characters"
> node_table[1]:
> id: 1
> pgdat: ffff880280000040
> size: 2097152
> present: 2097152
> mem_map: ffffea0008c00000
> start_paddr: 280000000
> start_mapnr: 2621440
> NOTE: page_hash_table does not exist in this kernel
> ^Mplease wait... (gathering kmem slab cache data)
> kmem_cache_downsize: SIZE(kmem_cache_s): 36968
> cache_cache.buffer_size: 32896
> kmem_cache_downsize: nr_node_ids: 2
> ^M ^MNOTE: unwind_table structure has changed, or does not exist in
> this kernel
> init_unwind_table: DWARF_UNWIND_EH_FRAME
> ^Mplease wait... (gathering module symbol data)^M ^M^Mplease wait...
> (gathering task table data)^M ^Mcrash: get_cpus_online: online: 16
> ^Mplease wait... (determining panic task)
> crash: get_active_set_panic_task: failed
>
>
> Thanks,
>
> -- Joe Lawrence
> --
> Crash-utility mailing list
> Crash-utility@xxxxxxxxxx
> https://www.redhat.com/mailman/listinfo/crash-utility
>
--
Crash-utility mailing list
Crash-utility@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/crash-utility
