<!doctype html public "-//w3c//dtd html 4.0 transitional//en"> <html> <tt>- Addressed a number of issues associated with CONFIG_SPARSEMEM</tt> <tt> kernels and kernels using updated manners for the linkage of</tt> <tt> their pglist_data structures, and pointers to their mem_map arrays.</tt> <tt> (anderson@redhat.com)</tt><tt></tt> <tt>- Implemented "kmem -n" for CONFIG_SPARSEMEM kernels; in addition</tt> <tt> to the pgdat- and zone-related data command output, it also</tt> <tt> displays a list of the SPARSEMEM mem_sections. Here is an</tt> <tt> example from an ia64:</tt> <tt> </tt> <tt> crash> kmem -n</tt> <tt> NODE SIZE PGLIST_DATA BOOTMEM_DATA NODE_ZONES</tt> <tt> 0 2359296 e000000008c00000 a000000100749b70 e000000008c00000</tt> <tt> e000000008c02400</tt> <tt> e000000008c04800</tt> <tt> e000000008c06c00</tt> <tt> MEM_MAP START_PADDR START_MAPNR</tt> <tt> e0000001040a3f00 0 0</tt> <tt> </tt> <tt> ZONE NAME SIZE MEM_MAP START_PADDR START_MAPNR</tt> <tt> 0 DMA 262144 e0000001040a3f00 0 0</tt> <tt> 1 DMA32 0 0 0 0</tt> <tt> 2 Normal 2097152 e0000001048a3f00 100000000 262144</tt> <tt> 3 HighMem 0 0 0 0</tt> <tt> </tt> <tt> -------------------------------------------------------------------</tt> <tt> </tt> <tt> NR SECTION CODED_MEM_MAP MEM_MAP PFN</tt> <tt> 0 e00000010409ff00 e0000001040a3f00 e0000001040a3f00 0</tt> <tt> 1 e00000010409ff08 e0000001040a3f00 e0000001044a3f00 65536</tt> <tt> 4 e00000010409ff20 e0000001038a3f00 e0000001048a3f00 262144</tt> <tt> 5 e00000010409ff28 e0000001038a3f00 e000000104ca3f00 327680</tt> <tt> 6 e00000010409ff30 e0000001038a3f00 e0000001050a3f00 393216</tt> <tt> 7 e00000010409ff38 e0000001038a3f00 e0000001054a3f00 458752</tt> <tt> 8 e00000010409ff40 e0000001038a3f00 e0000001058a3f00 524288</tt> <tt> 9 e00000010409ff48 e0000001038a3f00 e000000105ca3f00 589824</tt> <tt> 10 e00000010409ff50 e0000001038a3f00 e0000001060a3f00 655360</tt> <tt> 11 e00000010409ff58 e0000001038a3f00 e0000001064a3f00 720896</tt> <tt> 12 e00000010409ff60 e0000001038a3f00 e0000001068a3f00 786432</tt> <tt> 13 e00000010409ff68 e0000001038a3f00 e000000106ca3f00 851968</tt> <tt> 14 e00000010409ff70 e0000001038a3f00 e0000001070a3f00 917504</tt> <tt> 15 e00000010409ff78 e0000001038a3f00 e0000001074a3f00 983040</tt> <tt> 16 e00000010409ff80 e0000001038a3f00 e0000001078a3f00 1048576</tt> <tt> 17 e00000010409ff88 e0000001038a3f00 e000000107ca3f00 1114112</tt> <tt> 18 e00000010409ff90 e0000001038a3f00 e0000001080a3f00 1179648</tt> <tt> 19 e00000010409ff98 e0000001038a3f00 e0000001084a3f00 1245184</tt> <tt> 20 e00000010409ffa0 e0000001038a3f00 e0000001088a3f00 1310720</tt> <tt> 21 e00000010409ffa8 e0000001038a3f00 e000000108ca3f00 1376256</tt> <tt> 22 e00000010409ffb0 e0000001038a3f00 e0000001090a3f00 1441792</tt> <tt> 23 e00000010409ffb8 e0000001038a3f00 e0000001094a3f00 1507328</tt> <tt> 34 e0000001040a0010 e0000001010a3f00 e0000001098a3f00 2228224</tt> <tt> 35 e0000001040a0018 e0000001010a3f00 e000000109ca3f00 2293760</tt> <tt> crash></tt> <tt> (anderson@redhat.com)</tt><tt></tt> <tt>- Fix for "kmem -i" failure in CONFIG_SPARSEMEM kernels that would</tt> <tt> typically fail with the error message: "kmem: invalid kernel virtual</tt> <tt> address: 0 type: node_zones free_pages". (anderson@redhat.com)</tt><tt></tt> <tt>- Fix for "kmem -f" failure in CONFIG_SPARSEMEM kernels that would</tt> <tt> typically fail with the error message: "kmem: invalid kernel virtual</tt> <tt> address: ab8 type: node_zones name". (anderson@redhat.com)</tt><tt></tt> <tt>- Fix for "kmem -f" failure in 2.6.17 kernels (possibly earlier) that</tt> <tt> would fail with the error message: "kmem: invalid structure member</tt> <tt> offset: zone_zone_mem_map". (anderson@redhat.com)</tt><tt></tt> <tt>- Fix for "kmem [address]" failure in 2.6.17 kernels (possibly earlier)</tt> <tt> that would fail with the error message: "kmem: invalid structure</tt> <tt> member offset: zone_zone_mem_map". (anderson@redhat.com)</tt><tt></tt> <tt>- Fix for "kmem -i" that resulted in a bogus "CACHED" page count</tt> <tt> value. (anderson@redhat.com)</tt><tt></tt> <tt>- As an result of the last "kmem -i" fix, I've added a new "kmem -V"</tt> <tt> option that dumps the kernel's new vm_stat[] array contents by</tt> <tt> their enum values:</tt><tt></tt> <tt> crash> kmem -V</tt> <tt> NR_ANON_PAGES: 38656</tt> <tt> NR_FILE_MAPPED: 3116</tt> <tt> NR_FILE_PAGES: 141106</tt> <tt> NR_SLAB: 58605</tt> <tt> NR_PAGETABLE: 1059</tt> <tt> NR_FILE_DIRTY: 7</tt> <tt> NR_WRITEBACK: 0</tt> <tt> NR_UNSTABLE_NFS: 0</tt> <tt> NR_BOUNCE: 0</tt> <tt> NUMA_HIT: 86475467</tt> <tt> NUMA_MISS: 0</tt> <tt> NUMA_FOREIGN: 0</tt> <tt> NUMA_INTERLEAVE_HIT: 31523</tt> <tt> NUMA_LOCAL: 86475467</tt> <tt> NUMA_OTHER: 0</tt> <tt> crash></tt><tt></tt> <tt> Interally, a new dump_vm_stat() function has been added to access</tt> <tt> any of the items in the list. (anderson@redhat.com)</tt><tt></tt> <tt>- Implemented support for relocatable x86_64 live kernels and kdump</tt> <tt> generated vmcores. Without this patch, attempts to analyze those</tt> <tt> kernels would fail during initialization with the error message:</tt> <tt> "crash: vmlinux and vmcore do not match!" (anderson@redhat.com)</tt><tt></tt> <tt>- Support for recognizing real-time signals in the "sig" command.</tt> <tt> (olivier.daudel@u-paris10.fr)</tt><tt></tt> <tt>- Fix for "sys -c" display of "sys_ni_syscall" entries that showed</tt> <tt> different system call names that have the same (W) symbol value</tt> <tt> as the (T) symbol "sys_ni_syscall". For example:</tt><tt></tt> <tt> crash> sym -l | grep ffffffff802a38b6</tt> <tt> ffffffff802a38b6 (W) compat_sys_ipc</tt> <tt> ffffffff802a38b6 (W) compat_sys_keyctl</tt> <tt> ffffffff802a38b6 (W) compat_sys_sysctl</tt> <tt> ffffffff802a38b6 (W) ppc_rtas</tt> <tt> ffffffff802a38b6 (T) sys_ni_syscall</tt> <tt> ffffffff802a38b6 (W) sys_pciconfig_iobase</tt> <tt> ffffffff802a38b6 (W) sys_pciconfig_read</tt> <tt> ffffffff802a38b6 (W) sys_pciconfig_write</tt> <tt> ffffffff802a38b6 (W) sys_spu_create</tt> <tt> ffffffff802a38b6 (W) sys_spu_run</tt> <tt> ffffffff802a38b6 (W) sys_vm86</tt> <tt> ffffffff802a38b6 (W) sys_vm86old</tt> <tt> crash></tt><tt></tt> <tt>- Depending upon the kernel, one of those symbols would be displayed</tt> <tt> instead of sys_ni_syscall. (olivier.daudel@u-paris10.fr)</tt><tt></tt> <tt>- Fix for "sig" command where in later 2.6 kernels, the queued signal</tt> <tt> list at the end of the display would loop back on itself, repeatedly</tt> <tt> displaying the same queued signal(s). (olivier.daudel@u-paris10.fr)</tt> <tt></tt> <tt></tt> <tt>Download from: <A HREF="http://people.redhat.com/anderson">http://people.redhat.com/anderson</A></tt> <tt> </tt> <tt></tt> </html>