]> git.karo-electronics.de Git - karo-tx-linux.git/commitdiff
x86 cpu_hotplug: unmap cpu2node when the cpu is hotremoved
authorWen Congyang <wency@cn.fujitsu.com>
Thu, 25 Oct 2012 01:14:37 +0000 (12:14 +1100)
committerStephen Rothwell <sfr@canb.auug.org.au>
Thu, 1 Nov 2012 04:23:58 +0000 (15:23 +1100)
When a cpu is hotplugged, we call acpi_map_cpu2node() in
_acpi_map_lsapic() to store the cpu's node.  But we don't clear the cpu's
node in acpi_unmap_lsapic() when this cpu is hotremoved.  If the node is
also hotremoved, We will get the following messages:

[ 1646.771485] kernel BUG at include/linux/gfp.h:329!
[ 1646.828729] invalid opcode: 0000 [#1] SMP
[ 1646.877872] Modules linked in: ebtable_nat ebtables ipt_MASQUERADE iptable_nat nf_nat xt_CHECKSUM iptable_mangle bridge stp llc sunrpc ipt_REJECT nf_conntrack_ipv4 nf_defrag_ipv4 iptable_filter ip_tables ip6t_REJECT nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables binfmt_misc dm_mirror dm_region_hash dm_log dm_mod vhost_net macvtap macvlan tun uinput iTCO_wdt iTCO_vendor_support coretemp kvm_intel kvm crc32c_intel microcode pcspkr i2c_i801 i2c_core lpc_ich mfd_core ioatdma e1000e i7core_edac edac_core sg acpi_memhotplug igb dca sd_mod crc_t10dif megaraid_sas mptsas mptscsih mptbase scsi_transport_sas scsi_mod
[ 1647.588773] Pid: 3126, comm: init Not tainted 3.6.0-rc3-tangchen-hostbridge+ #13 FUJITSU-SV PRIMEQUEST 1800E/SB
[ 1647.711545] RIP: 0010:[<ffffffff811bc3fd>]  [<ffffffff811bc3fd>] allocate_slab+0x28d/0x300
[ 1647.810492] RSP: 0018:ffff88078a049cf8  EFLAGS: 00010246
[ 1647.874028] RAX: 0000000000000000 RBX: 0000000000000001 RCX: 0000000000000000
[ 1647.959339] RDX: 0000000000000001 RSI: 0000000000000001 RDI: 0000000000000246
[ 1648.044659] RBP: ffff88078a049d38 R08: 00000000000040d0 R09: 0000000000000001
[ 1648.129953] R10: 0000000000000000 R11: 0000000000000b5f R12: 00000000000052d0
[ 1648.215259] R13: ffff8807c1417300 R14: 0000000000030038 R15: 0000000000000003
[ 1648.300572] FS:  00007fa9b1b44700(0000) GS:ffff8807c3800000(0000) knlGS:0000000000000000
[ 1648.397272] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
[ 1648.465985] CR2: 00007fa9b09acca0 CR3: 000000078b855000 CR4: 00000000000007e0
[ 1648.551265] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 1648.636565] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
[ 1648.721838] Process init (pid: 3126, threadinfo ffff88078a048000, task ffff8807bb6f2650)
[ 1648.818534] Stack:
[ 1648.842548]  ffff8807c39d7fa0 ffffffff000040d0 00000000000000bb 00000000000080d0
[ 1648.931469]  ffff8807c1417300 ffff8807c39d7fa0 ffff8807c1417300 0000000000000001
[ 1649.020410]  ffff88078a049d88 ffffffff811bc4a0 ffff8807c1410c80 0000000000000000
[ 1649.109464] Call Trace:
[ 1649.138713]  [<ffffffff811bc4a0>] new_slab+0x30/0x1b0
[ 1649.199075]  [<ffffffff811bc978>] __slab_alloc+0x358/0x4c0
[ 1649.264683]  [<ffffffff810b71c0>] ? alloc_fair_sched_group+0xd0/0x1b0
[ 1649.341695]  [<ffffffff811be7d4>] kmem_cache_alloc_node_trace+0xb4/0x1e0
[ 1649.421824]  [<ffffffff8109d188>] ? hrtimer_init+0x48/0x100
[ 1649.488414]  [<ffffffff810b71c0>] ? alloc_fair_sched_group+0xd0/0x1b0
[ 1649.565402]  [<ffffffff810b71c0>] alloc_fair_sched_group+0xd0/0x1b0
[ 1649.640297]  [<ffffffff810a8bce>] sched_create_group+0x3e/0x110
[ 1649.711040]  [<ffffffff810bdbcd>] sched_autogroup_create_attach+0x4d/0x180
[ 1649.793260]  [<ffffffff81089614>] sys_setsid+0xd4/0xf0
[ 1649.854694]  [<ffffffff8167a029>] system_call_fastpath+0x16/0x1b
[ 1649.926483] Code: 89 c4 e9 73 fe ff ff 31 c0 89 de 48 c7 c7 45 de 9e 81 44 89 45 c8 e8 22 05 4b 00 85 db 44 8b 45 c8 0f 89 4f ff ff ff 0f 0b eb fe <0f> 0b 90 eb fd 0f 0b eb fe 89 de 48 c7 c7 45 de 9e 81 31 c0 44
[ 1650.161454] RIP  [<ffffffff811bc3fd>] allocate_slab+0x28d/0x300
[ 1650.232348]  RSP <ffff88078a049cf8>
[ 1650.274029] ---[ end trace adf84c90f3fea3e5 ]---

The reason is that: the cpu's node is not NUMA_NO_NODE, we will call
alloc_pages_exact_node() to alloc memory on the node, but the node is
offlined.

Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Pavel Machek <pavel@ucw.cz>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
arch/x86/kernel/acpi/boot.c

index bacf4b0d91f4e0e01c4b7bc7cbcfebe6e1e3a680..c9fcd8cc0e60e433d098c1322eea4f88a2a13fd4 100644 (file)
@@ -697,6 +697,11 @@ EXPORT_SYMBOL(acpi_map_lsapic);
 
 int acpi_unmap_lsapic(int cpu)
 {
+#ifdef CONFIG_ACPI_NUMA
+       set_apicid_to_node(per_cpu(x86_cpu_to_apicid, cpu), NUMA_NO_NODE);
+       numa_clear_node(cpu);
+#endif
+
        per_cpu(x86_cpu_to_apicid, cpu) = -1;
        set_cpu_present(cpu, false);
        num_processors--;