I have retried compiling the 3.4 Kernel this time in Squeeze. The Kernel compiles fine and will boot up to 255 cores; however, after that if fails out with the following Kernel oops when booting more than 255 cores (Kernel is compiled with 512). Here is the boot log with the option bootmem_debug=1. I have tried to shorten the boot log and leave what I think are the important parts; however if anyone needs the 35M boot log then I will gladly send as attachment. [ 0.000000] Initializing cgroup subsys cpuset [ 0.000000] Initializing cgroup subsys cpu [ 0.000000] Linux version 3.4.49 (beeij@debian-hpc) (gcc version 4.4.5 (Debian 4.4.5-8) ) #16 SMP Thu Aug 29 14:41:59 CDT 2013 [ 0.000000] EFI v1.10 by INTEL: SALsystab=0x1802c2d990 ACPI 2.0=0x1802c2da80 [ 0.000000] booting generic kernel on platform sn2 [ 0.000000] console [sn_sal0] enabled [ 0.000000] ACPI: RSDP 0000001802c2da80 00024 (v02 SGI) [ 0.000000] ACPI: XSDT 0000001802c38df0 00044 (v01 SGI XSDTSN2 00010001 ? 00000094) [ 0.000000] ACPI: APIC 0000001802c2f5a0 0152C (v01 SGI APICSN2 00010001 ? 00000001) [ 0.000000] ACPI: SRAT 0000001802c30ae0 02DB0 (v01 SGI SRATSN2 00010001 ? 00000001) [ 0.000000] ACPI: SLIT 0000001802c338a0 0312C (v01 SGI SLITSN2 00010001 ? 00000001) [ 0.000000] ACPI: FACP 0000001802c369e0 000F4 (v03 SGI FACPSN2 00030001 ? 00000001) [ 0.000000] ACPI Warning: 32/64X length mismatch in Pm1aEventBlock: 32/0 (20120320/tbfadt-548) [ 0.000000] ACPI Warning: 32/64X length mismatch in Pm1aControlBlock: 16/0 (20120320/tbfadt-548) [ 0.000000] ACPI Warning: 32/64X length mismatch in PmTimerBlock: 32/0 (20120320/tbfadt-548) [ 0.000000] ACPI Warning: 32/64X length mismatch in Gpe0Block: 64/0 (20120320/tbfadt-548) [ 0.000000] ACPI Warning: Invalid length for Pm1aEventBlock: 0, using default 32 (20120320/tbfadt-629) [ 0.000000] ACPI Warning: Invalid length for Pm1aControlBlock: 0, using default 16 (20120320/tbfadt-629) [ 0.000000] ACPI Warning: Invalid length for PmTimerBlock: 0, using default 32 (20120320/tbfadt-629) [ 0.000000] ACPI: DSDT 0000001802c3af20 00024 (v02 SGI DSDTSN2 00020001 ? 00002483) [ 0.000000] ACPI: FACS 0000001802c2e1e0 00040 [ 0.000000] ACPI: Local APIC address c0000000fee00000 [ 0.000000] 448 CPUs available, 448 CPUs total [ 0.000000] Number of logical nodes in system = 112 [ 0.000000] Number of memory chunks in system = 112 [ 0.000000] SMP: Allowing 448 CPUs, 0 hotplug CPUs [=================SNIP=======================] [ 0.000000] On node 63 totalpages: 504832 [ 0.000000] free_area_init_node: node 63, pgdat e0000fd8040c1f80, node_mem_map a0007ff57d62a000 [ 0.000000] DMA zone: 2650 pages used for memmap [ 0.000000] DMA zone: 0 pages reserved [ 0.000000] DMA zone: 502182 pages, LIFO batch:7 [ 0.000000] bootmem::alloc_bootmem_core nid=63 size=18 [1 pages] align=80 goal=4000000000000 limit=0 [ 0.000000] bootmem::__reserve nid=63 start=3f601318 end=3f601319 flags=1 [ 0.000000] bootmem::alloc_bootmem_core nid=63 size=18000 [6 pages] align=80 goal=4000000000000 limit=0 [ 0.000000] bootmem::__reserve nid=63 start=3f601319 end=3f60131f flags=1 [ 0.000000] Could not find start_pfn for node 64 [ 0.000000] On node 64 totalpages: 0 [ 0.000000] free_area_init_node: node 64, pgdat e000101804102000, node_mem_map a0007ff5b562a000 [ 0.000000] Could not find start_pfn for node 65 [ 0.000000] On node 65 totalpages: 0 [ 0.000000] free_area_init_node: node 65, pgdat e000105804142080, node_mem_map a0007ff5ed62a000 [ 0.000000] Could not find start_pfn for node 66 [ 0.000000] On node 66 totalpages: 0 [=================SNIP=======================] [ 0.000000] BUG: Bad page state in process swapper pfn:40601318 [ 0.000000] page:a0007ff5b5642d40 count:0 mapcount:1 mapping: (null) index:0x0 [ 0.000000] page flags: 0x0() [ 0.000000] Modules linked in: [ 0.000000] Unable to handle kernel NULL pointer dereference (address 0000000000000018) [ 0.000000] swapper[0]: Oops 11003706212352 [1] [ 0.000000] Modules linked in: [ 0.000000] [ 0.000000] Pid: 0, CPU 0, comm: swapper [ 0.000000] psr : 00001210084a2018 ifs : 800000000000cc18 ip : [<a0000001003ea1b1>] Not tainted (3.4.49) [ 0.000000] ip is at __copy_user+0x891/0x950 [ 0.000000] unat: 0000000000000000 pfs : 0000000000000792 rsc : 0000000000000003 [ 0.000000] rnat: 0000000000000000 bsps: 0000000000000000 pr : 0bad0bad0baa55a9 [ 0.000000] ldrs: 0000000000000000 ccv : 0000000000000000 fpsr: 0009804c8a70433f [ 0.000000] csd : 0000000000000000 ssd : 0000000000000000 [ 0.000000] b0 : a000000100043430 b6 : a000000100043660 b7 : a00000010000c3b0 [ 0.000000] f6 : 000000000000000000000 f7 : 1003e9e3779b97f4a7c16 [ 0.000000] f8 : 1003e0a00000010001577 f9 : 10006c7fffffffd73ea5c [ 0.000000] f10 : 1003e0000000000000000 f11 : 1003e0044b82fa09b5a53 [ 0.000000] r1 : a000000100dfa9e0 r2 : a000000100ac75f0 r3 : a000000100ac75f8 [ 0.000000] r8 : 0000000000000298 r9 : 0000000000000013 r10 : 0000000000000000 [ 0.000000] r11 : 0bad0bad0baa11e9 r12 : a000000100ac7550 r13 : a000000100ac0000 [ 0.000000] r14 : a000000100e44080 r15 : a000000100e44030 r16 : 0000000000000298 [ 0.000000] r17 : 0000000000000010 r18 : 0000000000000018 r19 : a000000100ac7850 [ 0.000000] r20 : 0000000000000290 r21 : a000000100ac75b4 r22 : a000000100c12f20 [ 0.000000] r23 : a000000100ac75b0 r24 : 0000000000000000 r25 : a000000100e44030 [ 0.000000] r26 : a0000001007ea718 r27 : 0000000000018869 r28 : a000000100ac4000 [ 0.000000] r29 : 0000000000000014 r30 : 0000000000000000 r31 : 0000000000000792 It looks like after Node 64 which would be cores 256 and up, it can not find start_pfn and it shows 0 total pages. The instruction pointer is at [__copy_user+0x891/0x950]. In the meantime I have compiled the 2.6.35 Kernel with support for 1024 CPUs that works as a hold over. Anyone have any ideas as to why it is failing out at this point? Thanks, Beeij -- To unsubscribe from this list: send the line "unsubscribe linux-ia64" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
|