Skip to contentSkip to conversation

intr_remap: Simplify the code further

BUG In review Russ Anderson 4 participants 8 messages Oct 29, 2010 – Oct 30, 2010 x86

lore
Share & tools
lore

Russ Anderson

author Reviewed-by Ingo Molnar Acked-by Suresh Siddha
#0 lore raw

There is a regression that is causing a NULL pointer dereference in free_irte when shutting down xpc. git bisect narrowed it down to git commit d585d060b42bd36f6f0b23ff327d3b91f80c7139, which changed free_irte(). Reverse applying the patch fixes the problem.

git commit d585d060b42bd36f6f0b23ff327d3b91f80c7139
---------------------------------------------------------
commit d585d060b42bd36f6f0b23ff327d3b91f80c7139

Author: Thomas Gleixner <[email protected]>
Date:   Sun Oct 10 12:34:27 2010 +0200

intr_remap: Simplify the code further

Having irq_2_iommu in struct irq_cfg allows further simplifications.

---------------------------------------------------------

The failing output on real hardware:
----------------------------------------------------------------------------
Sending all processes the TERM signal... done
Sending all processes the KILL signal...
Please stand by while rebooting the system...

[ 4020.514342] BUG: unable to handle kernel NULL pointer dereference at 0000000000000080
[ 4020.523091] IP: [<ffffffff8121a791>] free_irte+0x46/0xe4
[ 4020.529024] PGD 3f5ac5067 PUD 3f6487067 PMD 0
[ 4020.534007] Oops: 0000 [#1] SMP
[ 4020.537626] last sysfs file:
/sys/devices/pci0000:00/0000:00:03.0/0000:04:00.0/host0/port-0:0/end_device-0:0/target0:0:0/0:0:0:0/vendor
[ 4020.551197] xpc : all partitions have deactivated
[ 4020.556437] CPU 25
[ 4020.558580] Modules linked in:
[ 4020.562199] 
[ 4020.563858] Pid: 13489, comm: reboot Not tainted 2.6.36-tip+ #5 /Stoutland Platform
[ 4020.572390] RIP: 0010:[<ffffffff8121a791>]  [<ffffffff8121a791>] free_irte+0x46/0xe4
[ 4020.581030] RSP: 0018:ffff88047cdb1c38  EFLAGS: 00010046 
[ 4020.586948] RAX: 0000000000000246 RBX: ffff8803f6b50b60 RCX: 0000000000000000
[ 4020.594900] RDX: 0000000000000113 RSI: 0000000000000000 RDI: 0000000000000000
[ 4020.602851] RBP: ffff88047cdb1c68 R08: ffff8803f6b50ac0 R09: 00000000ffffffff
[ 4020.610802] R10: ffff8803f78a2200 R11: 00000084f04c73c1 R12: 0000000000000000
[ 4020.618754] R13: 0000000000000246 R14: 0000000000000000 R15: 0000000000000000
[ 4020.626705] FS:  00007fec1bfd1700(0000) GS:ffff8803ffd20000(0000) knlGS:0000000000000000 
[ 4020.635722] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
[ 4020.642123] CR2: 0000000000000080 CR3: 00000003f5704000 CR4: 00000000000006e0
[ 4020.650074] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 4020.658026] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
[ 4020.665976] Process reboot (pid: 13489, threadinfo ffff88047cdb0000, task ffff88047a2058e0)
[ 4020.675282] Stack:
aster Resource Control: runlevel 6 has been                         reached
[ 4020.677520]  ffff88047cdb1c68
ffffffff8109172d ffff8803f6b50b80 ffff8803f6b50b80                                                         
INIT: no more processes left in this runlevel
[ 4020.685807]  0000000000000065 ffff8803f6b50b40             ffff88047cdb1c98 ffffffff8101cd32
[ 4020.694093]  ffff88047cdb1ca8 ffff8803f6b50b80 0000000000000065 0000000000000292
[ 4020.702379] Call Trace:
[ 4020.705109]  [<ffffffff8109172d>] ? irq_modify_status+0x55/0x5e 
[ 4020.711709]  [<ffffffff8101cd32>] destroy_irq+0x41/0x7e        
[ 4020.717534]  [<ffffffff81031fe2>] uv_teardown_irq+0xb6/0xc1  
[ 4020.723743]  [<ffffffff81090f92>] ? free_irq+0x50/0x59                                                  
[ 4020.729470]  [<ffffffff812e0da3>] ? xp_restrict_memprotect_uv+0x0/0x30
[ 4020.736744]  [<ffffffff812e3a12>] xpc_destroy_gru_mq_uv+0x54/0x98 
[ 4020.743533]  [<ffffffff812e3a66>] xpc_exit_uv+0x10/0x1e 
[ 4020.749355]  [<ffffffff812e1299>] xpc_do_exit+0x1e3/0x1f2
[ 4020.755371]  [<ffffffff812e12d3>] xpc_system_reboot+0x2b/0x2f
[ 4020.761778]  [<ffffffff815c0b19>] notifier_call_chain+0x33/0x5b
[ 4020.768378]  [<ffffffff810612a4>] __blocking_notifier_call_chain+0x4d/0x6a 
[ 4020.776038]  [<ffffffff810612d0>] blocking_notifier_call_chain+0xf/0x11   
[ 4020.783411]  [<ffffffff81055c1e>] kernel_restart_prepare+0x18/0x2e       
[ 4020.790289]  [<ffffffff81055c6d>] kernel_restart+0x11/0x43              
[ 4020.796401]  [<ffffffff81055de5>] sys_reboot+0x139/0x174               
[ 4020.802322]  [<ffffffff810ef23b>] ? fput+0x20d/0x21c                  
[ 4020.807853]  [<ffffffff810ec3a0>] ? filp_close+0x67/0x72             
[ 4020.813772]  [<ffffffff81002aab>] system_call_fastpath+0x16/0x1b    
[ 4020.820464] Code: c3 0f 84 b4 00 00 00 48 c7 c7 d4 d8 e5 81 45 31 e4 e8 95 2e 3a 00 66 83 7b 0a 00 49 89
c5 75 75 48 8b 33 0f b7 7b 08 0f b6 4b 0c <48> 8b 86 80 00 00 00 48 89 fa 48 c1 e2 04 48 03 10 b8 01 00 00
[ 4020.842190] RIP  [<ffffffff8121a791>] free_irte+0x46/0xe4
[ 4020.848217]  RSP <ffff88047cdb1c38>
[ 4020.852101] CR2: 0000000000000080 
[ 4021.091896] ---[ end trace 113a8c342207f0d1 ]---

/etc/init.d/rc: line 317: 13489 Killed $link start
----------------------------------------------------------------------------

Output from on the simulator:
----------------------------------------------------------------------------

[    0.205968] xpc : can't setup our reserved page

Breakpoint reached at <0xffffffff811fdeb8> on cpu 0

All cpus stopped because one or more cpus hit breakpoint(s); the "stat" cmd will show which cpu(s) hit breakpoints and which were still running.
<199934998> <bust_spinlocks> 55 push %rbp
mdb:/> lastct 100

199933847: ret  <__phys_addr+0x33> -> <virt_to_head_page+0x9>
199933857: ret  <virt_to_head_page+0x2f> -> <kfree+0x4d>
199933885: ret  <kfree+0xb3> -> <pde_put+0x68>
199933889: ret  <pde_put+0x6c> -> <remove_proc_entry+0x230>
199933897: ret  <remove_proc_entry+0x23e> -> <unregister_handler_proc+0x2c>
199933901: ret  <unregister_handler_proc+0x30> -> <__free_irq+0x134>

199933903: call <__free_irq+0x137> -> <synchronize_irq> 199933910: call <synchronize_irq+0xd> -> <irq_to_desc> 199933915: call <irq_to_desc+0xd> -> <radix_tree_lookup> 199933943: ret <radix_tree_lookup+0x5a> -> <irq_to_desc+0x12> 199933945: ret <irq_to_desc+0x13> -> <synchronize_irq+0x12> 199933953: call <synchronize_irq+0x30> -> <_raw_spin_lock_irqsave> 199933966: ret <_raw_spin_lock_irqsave+0x22> -> <synchronize_irq+0x35> 199933970: call <synchronize_irq+0x40> -> <_raw_spin_unlock_irqrestore>

199933977: ret  <_raw_spin_unlock_irqrestore+0xa> -> <synchronize_irq+0x45>
199933990: ret  <synchronize_irq+0xd4> -> <__free_irq+0x13c>
199934002: ret  <__free_irq+0x178> -> <free_irq+0x40>
199934004: call <free_irq+0x43> -> <kfree>
199934018: call <kfree+0x48> -> <virt_to_head_page>
199934021: call <virt_to_head_page+0x4> -> <__phys_addr>
199934030: ret  <__phys_addr+0x33> -> <virt_to_head_page+0x9>
199934040: ret  <virt_to_head_page+0x2f> -> <kfree+0x4d>
199934068: ret  <kfree+0xb3> -> <free_irq+0x48>

199934070: call <free_irq+0x4b> -> <chip_bus_sync_unlock> 199934078: ret <chip_bus_sync_unlock+0x17> -> <free_irq+0x50> 199934084: ret <free_irq+0x58> -> <xpc_destroy_gru_mq_uv+0x4a> 199934086: call <xpc_destroy_gru_mq_uv+0x4f> -> <uv_teardown_irq> 199934095: call <uv_teardown_irq+0x17> -> <_raw_spin_lock_irqsave> 199934108: ret <_raw_spin_lock_irqsave+0x22> -> <uv_teardown_irq+0x1c> 199934148: call <uv_teardown_irq+0x7c> -> <rb_erase> 199934180: ret <rb_erase+0x27f> -> <uv_teardown_irq+0x81> 199934182: call <uv_teardown_irq+0x84> -> <kfree>
199934196: call <kfree+0x48> -> <virt_to_head_page> 199934199: call <virt_to_head_page+0x4> -> <__phys_addr>

199934208: ret  <__phys_addr+0x33> -> <virt_to_head_page+0x9>
199934218: ret  <virt_to_head_page+0x2f> -> <kfree+0x4d>
199934246: ret  <kfree+0xb3> -> <uv_teardown_irq+0x89>

199934250: call <uv_teardown_irq+0xa9> -> <_raw_spin_unlock_irqrestore> 199934257: ret <_raw_spin_unlock_irqrestore+0xa> -> <uv_teardown_irq+0xae> 199934259: call <uv_teardown_irq+0xb1> -> <destroy_irq> 199934268: call <destroy_irq+0x13> -> <irq_get_irq_data> 199934271: call <irq_get_irq_data+0x4> -> <irq_to_desc> 199934276: call <irq_to_desc+0xd> -> <radix_tree_lookup>

199934304: ret  <radix_tree_lookup+0x5a> -> <irq_to_desc+0x12>
199934306: ret  <irq_to_desc+0x13> -> <irq_get_irq_data+0x9>
199934308: ret  <irq_get_irq_data+0xa> -> <destroy_irq+0x18>

199934315: call <destroy_irq+0x2b> -> <irq_modify_status> 199934324: call <irq_modify_status+0x11> -> <irq_to_desc> 199934329: call <irq_to_desc+0xd> -> <radix_tree_lookup> 199934357: ret <radix_tree_lookup+0x5a> -> <irq_to_desc+0x12> 199934359: ret <irq_to_desc+0x13> -> <irq_modify_status+0x16> 199934367: call <irq_modify_status+0x33> -> <_raw_spin_lock_irqsave> 199934380: ret <_raw_spin_lock_irqsave+0x22> -> <irq_modify_status+0x38> 199934388: call <irq_modify_status+0x50> -> <_raw_spin_unlock_irqrestore> 199934395: ret <_raw_spin_unlock_irqrestore+0xa> -> <irq_modify_status+0x55> 199934401: ret <irq_modify_status+0x5d> -> <destroy_irq+0x30> 199934405: call <destroy_irq+0x3c> -> <free_irte>
199934413: call <free_irte+0x11> -> <irq_2_iommu>
199934416: call <irq_2_iommu+0x4> -> <irq_get_irq_data> 199934419: call <irq_get_irq_data+0x4> -> <irq_to_desc> 199934424: call <irq_to_desc+0xd> -> <radix_tree_lookup>

199934452: ret  <radix_tree_lookup+0x5a> -> <irq_to_desc+0x12>
199934454: ret  <irq_to_desc+0x13> -> <irq_get_irq_data+0x9>
199934456: ret  <irq_get_irq_data+0xa> -> <irq_2_iommu+0x9>
199934464: ret  <irq_2_iommu+0x20> -> <free_irte+0x16>

199934470: call <free_irte+0x2c> -> <_raw_spin_lock_irqsave> 199934483: ret <_raw_spin_lock_irqsave+0x22> -> <free_irte+0x31> 199934490: int <free_irte+0x46> -> <page_fault> 199934492: call <page_fault+0x4> -> <error_entry>

----------------------------------------------------------------------------

Signature · 2 lines
Russ Anderson, OS RAS/Partitioning Project Lead  
SGI - Silicon Graphics Inc          [email protected]
#1 lore raw
Quoted 5 lines · Russ Anderson

There is a regression that is causing a NULL pointer dereference in free_irte when shutting down xpc. git bisect narrowed it down to git commit d585d060b42bd36f6f0b23ff327d3b91f80c7139, which changed free_irte(). Reverse applying the patch fixes the problem.

Russ, Is this with !CONFIG_SPARSE_IRQ?

thanks,
suresh

Quoted 169 lines

git commit d585d060b42bd36f6f0b23ff327d3b91f80c7139
---------------------------------------------------------
commit d585d060b42bd36f6f0b23ff327d3b91f80c7139

Author: Thomas Gleixner <[email protected]>
Date:   Sun Oct 10 12:34:27 2010 +0200

intr_remap: Simplify the code further

Having irq_2_iommu in struct irq_cfg allows further simplifications.

---------------------------------------------------------

The failing output on real hardware:
----------------------------------------------------------------------------
Sending all processes the TERM signal... done
Sending all processes the KILL signal...
Please stand by while rebooting the system...

[ 4020.514342] BUG: unable to handle kernel NULL pointer dereference at 0000000000000080
[ 4020.523091] IP: [<ffffffff8121a791>] free_irte+0x46/0xe4
[ 4020.529024] PGD 3f5ac5067 PUD 3f6487067 PMD 0
[ 4020.534007] Oops: 0000 [#1] SMP
[ 4020.537626] last sysfs file:
/sys/devices/pci0000:00/0000:00:03.0/0000:04:00.0/host0/port-0:0/end_device-0:0/target0:0:0/0:0:0:0/vendor
[ 4020.551197] xpc : all partitions have deactivated
[ 4020.556437] CPU 25
[ 4020.558580] Modules linked in:
[ 4020.562199] 
[ 4020.563858] Pid: 13489, comm: reboot Not tainted 2.6.36-tip+ #5 /Stoutland Platform
[ 4020.572390] RIP: 0010:[<ffffffff8121a791>]  [<ffffffff8121a791>] free_irte+0x46/0xe4
[ 4020.581030] RSP: 0018:ffff88047cdb1c38  EFLAGS: 00010046 
[ 4020.586948] RAX: 0000000000000246 RBX: ffff8803f6b50b60 RCX: 0000000000000000
[ 4020.594900] RDX: 0000000000000113 RSI: 0000000000000000 RDI: 0000000000000000
[ 4020.602851] RBP: ffff88047cdb1c68 R08: ffff8803f6b50ac0 R09: 00000000ffffffff
[ 4020.610802] R10: ffff8803f78a2200 R11: 00000084f04c73c1 R12: 0000000000000000
[ 4020.618754] R13: 0000000000000246 R14: 0000000000000000 R15: 0000000000000000
[ 4020.626705] FS:  00007fec1bfd1700(0000) GS:ffff8803ffd20000(0000) knlGS:0000000000000000 
[ 4020.635722] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
[ 4020.642123] CR2: 0000000000000080 CR3: 00000003f5704000 CR4: 00000000000006e0
[ 4020.650074] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 4020.658026] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
[ 4020.665976] Process reboot (pid: 13489, threadinfo ffff88047cdb0000, task ffff88047a2058e0)
[ 4020.675282] Stack:
aster Resource Control: runlevel 6 has been                         reached
[ 4020.677520]  ffff88047cdb1c68
ffffffff8109172d ffff8803f6b50b80 ffff8803f6b50b80                                                         
INIT: no more processes left in this runlevel
[ 4020.685807]  0000000000000065 ffff8803f6b50b40             ffff88047cdb1c98 ffffffff8101cd32
[ 4020.694093]  ffff88047cdb1ca8 ffff8803f6b50b80 0000000000000065 0000000000000292
[ 4020.702379] Call Trace:
[ 4020.705109]  [<ffffffff8109172d>] ? irq_modify_status+0x55/0x5e 
[ 4020.711709]  [<ffffffff8101cd32>] destroy_irq+0x41/0x7e        
[ 4020.717534]  [<ffffffff81031fe2>] uv_teardown_irq+0xb6/0xc1  
[ 4020.723743]  [<ffffffff81090f92>] ? free_irq+0x50/0x59                                                  
[ 4020.729470]  [<ffffffff812e0da3>] ? xp_restrict_memprotect_uv+0x0/0x30
[ 4020.736744]  [<ffffffff812e3a12>] xpc_destroy_gru_mq_uv+0x54/0x98 
[ 4020.743533]  [<ffffffff812e3a66>] xpc_exit_uv+0x10/0x1e 
[ 4020.749355]  [<ffffffff812e1299>] xpc_do_exit+0x1e3/0x1f2
[ 4020.755371]  [<ffffffff812e12d3>] xpc_system_reboot+0x2b/0x2f
[ 4020.761778]  [<ffffffff815c0b19>] notifier_call_chain+0x33/0x5b
[ 4020.768378]  [<ffffffff810612a4>] __blocking_notifier_call_chain+0x4d/0x6a 
[ 4020.776038]  [<ffffffff810612d0>] blocking_notifier_call_chain+0xf/0x11   
[ 4020.783411]  [<ffffffff81055c1e>] kernel_restart_prepare+0x18/0x2e       
[ 4020.790289]  [<ffffffff81055c6d>] kernel_restart+0x11/0x43              
[ 4020.796401]  [<ffffffff81055de5>] sys_reboot+0x139/0x174               
[ 4020.802322]  [<ffffffff810ef23b>] ? fput+0x20d/0x21c                  
[ 4020.807853]  [<ffffffff810ec3a0>] ? filp_close+0x67/0x72             
[ 4020.813772]  [<ffffffff81002aab>] system_call_fastpath+0x16/0x1b    
[ 4020.820464] Code: c3 0f 84 b4 00 00 00 48 c7 c7 d4 d8 e5 81 45 31 e4 e8 95 2e 3a 00 66 83 7b 0a 00 49 89
c5 75 75 48 8b 33 0f b7 7b 08 0f b6 4b 0c <48> 8b 86 80 00 00 00 48 89 fa 48 c1 e2 04 48 03 10 b8 01 00 00
[ 4020.842190] RIP  [<ffffffff8121a791>] free_irte+0x46/0xe4
[ 4020.848217]  RSP <ffff88047cdb1c38>
[ 4020.852101] CR2: 0000000000000080 
[ 4021.091896] ---[ end trace 113a8c342207f0d1 ]---

/etc/init.d/rc: line 317: 13489 Killed $link start
----------------------------------------------------------------------------

Output from on the simulator:
----------------------------------------------------------------------------

[    0.205968] xpc : can't setup our reserved page

Breakpoint reached at <0xffffffff811fdeb8> on cpu 0

All cpus stopped because one or more cpus hit breakpoint(s); the "stat" cmd will show which cpu(s) hit breakpoints and which were still running.
<199934998> <bust_spinlocks> 55 push %rbp
mdb:/> lastct 100

199933847: ret  <__phys_addr+0x33> -> <virt_to_head_page+0x9>
199933857: ret  <virt_to_head_page+0x2f> -> <kfree+0x4d>
199933885: ret  <kfree+0xb3> -> <pde_put+0x68>
199933889: ret  <pde_put+0x6c> -> <remove_proc_entry+0x230>
199933897: ret  <remove_proc_entry+0x23e> -> <unregister_handler_proc+0x2c>
199933901: ret  <unregister_handler_proc+0x30> -> <__free_irq+0x134>

199933903: call <__free_irq+0x137> -> <synchronize_irq> 199933910: call <synchronize_irq+0xd> -> <irq_to_desc> 199933915: call <irq_to_desc+0xd> -> <radix_tree_lookup> 199933943: ret <radix_tree_lookup+0x5a> -> <irq_to_desc+0x12> 199933945: ret <irq_to_desc+0x13> -> <synchronize_irq+0x12> 199933953: call <synchronize_irq+0x30> -> <_raw_spin_lock_irqsave> 199933966: ret <_raw_spin_lock_irqsave+0x22> -> <synchronize_irq+0x35> 199933970: call <synchronize_irq+0x40> -> <_raw_spin_unlock_irqrestore>

199933977: ret  <_raw_spin_unlock_irqrestore+0xa> -> <synchronize_irq+0x45>
199933990: ret  <synchronize_irq+0xd4> -> <__free_irq+0x13c>
199934002: ret  <__free_irq+0x178> -> <free_irq+0x40>
199934004: call <free_irq+0x43> -> <kfree>
199934018: call <kfree+0x48> -> <virt_to_head_page>
199934021: call <virt_to_head_page+0x4> -> <__phys_addr>
199934030: ret  <__phys_addr+0x33> -> <virt_to_head_page+0x9>
199934040: ret  <virt_to_head_page+0x2f> -> <kfree+0x4d>
199934068: ret  <kfree+0xb3> -> <free_irq+0x48>

199934070: call <free_irq+0x4b> -> <chip_bus_sync_unlock> 199934078: ret <chip_bus_sync_unlock+0x17> -> <free_irq+0x50> 199934084: ret <free_irq+0x58> -> <xpc_destroy_gru_mq_uv+0x4a> 199934086: call <xpc_destroy_gru_mq_uv+0x4f> -> <uv_teardown_irq> 199934095: call <uv_teardown_irq+0x17> -> <_raw_spin_lock_irqsave> 199934108: ret <_raw_spin_lock_irqsave+0x22> -> <uv_teardown_irq+0x1c> 199934148: call <uv_teardown_irq+0x7c> -> <rb_erase> 199934180: ret <rb_erase+0x27f> -> <uv_teardown_irq+0x81> 199934182: call <uv_teardown_irq+0x84> -> <kfree>
199934196: call <kfree+0x48> -> <virt_to_head_page> 199934199: call <virt_to_head_page+0x4> -> <__phys_addr>

199934208: ret  <__phys_addr+0x33> -> <virt_to_head_page+0x9>
199934218: ret  <virt_to_head_page+0x2f> -> <kfree+0x4d>
199934246: ret  <kfree+0xb3> -> <uv_teardown_irq+0x89>

199934250: call <uv_teardown_irq+0xa9> -> <_raw_spin_unlock_irqrestore> 199934257: ret <_raw_spin_unlock_irqrestore+0xa> -> <uv_teardown_irq+0xae> 199934259: call <uv_teardown_irq+0xb1> -> <destroy_irq> 199934268: call <destroy_irq+0x13> -> <irq_get_irq_data> 199934271: call <irq_get_irq_data+0x4> -> <irq_to_desc> 199934276: call <irq_to_desc+0xd> -> <radix_tree_lookup>

199934304: ret  <radix_tree_lookup+0x5a> -> <irq_to_desc+0x12>
199934306: ret  <irq_to_desc+0x13> -> <irq_get_irq_data+0x9>
199934308: ret  <irq_get_irq_data+0xa> -> <destroy_irq+0x18>

199934315: call <destroy_irq+0x2b> -> <irq_modify_status> 199934324: call <irq_modify_status+0x11> -> <irq_to_desc> 199934329: call <irq_to_desc+0xd> -> <radix_tree_lookup> 199934357: ret <radix_tree_lookup+0x5a> -> <irq_to_desc+0x12> 199934359: ret <irq_to_desc+0x13> -> <irq_modify_status+0x16> 199934367: call <irq_modify_status+0x33> -> <_raw_spin_lock_irqsave> 199934380: ret <_raw_spin_lock_irqsave+0x22> -> <irq_modify_status+0x38> 199934388: call <irq_modify_status+0x50> -> <_raw_spin_unlock_irqrestore> 199934395: ret <_raw_spin_unlock_irqrestore+0xa> -> <irq_modify_status+0x55> 199934401: ret <irq_modify_status+0x5d> -> <destroy_irq+0x30> 199934405: call <destroy_irq+0x3c> -> <free_irte>
199934413: call <free_irte+0x11> -> <irq_2_iommu>
199934416: call <irq_2_iommu+0x4> -> <irq_get_irq_data> 199934419: call <irq_get_irq_data+0x4> -> <irq_to_desc> 199934424: call <irq_to_desc+0xd> -> <radix_tree_lookup>

199934452: ret  <radix_tree_lookup+0x5a> -> <irq_to_desc+0x12>
199934454: ret  <irq_to_desc+0x13> -> <irq_get_irq_data+0x9>
199934456: ret  <irq_get_irq_data+0xa> -> <irq_2_iommu+0x9>
199934464: ret  <irq_2_iommu+0x20> -> <free_irte+0x16>

199934470: call <free_irte+0x2c> -> <_raw_spin_lock_irqsave> 199934483: ret <_raw_spin_lock_irqsave+0x22> -> <free_irte+0x31> 199934490: int <free_irte+0x46> -> <page_fault> 199934492: call <page_fault+0x4> -> <error_entry>

----------------------------------------------------------------------------

#2 lore raw
Quoted 8 lines · Suresh Siddha
Quoted 5 lines · Russ Anderson

There is a regression that is causing a NULL pointer dereference in free_irte when shutting down xpc. git bisect narrowed it down to git commit d585d060b42bd36f6f0b23ff327d3b91f80c7139, which changed free_irte(). Reverse applying the patch fixes the problem.

Russ, Is this with !CONFIG_SPARSE_IRQ?

CONFIG_SPARSE_IRQ=y

Signature · 2 lines
Russ Anderson, OS RAS/Partitioning Project Lead  
SGI - Silicon Graphics Inc          [email protected]
#3 lore raw
Quoted 8 lines · Russ Anderson
Quoted 7 lines · Suresh Siddha
Quoted 5 lines · Russ Anderson

There is a regression that is causing a NULL pointer dereference in free_irte when shutting down xpc.  git bisect narrowed it down to git commit d585d060b42bd36f6f0b23ff327d3b91f80c7139, which changed free_irte().  Reverse applying the patch fixes the problem.

Quoted 5 lines
Quoted 1 line

Russ, Is this with !CONFIG_SPARSE_IRQ?

CONFIG_SPARSE_IRQ=y

can you try attached patch?

Yinghai

arch/x86/kernel/apic/io_apic.c +1 -1
index 5e43502..383f4bb 100644
@@ -3109,7 +3109,7 @@ void destroy_irq(unsigned int irq)
 
 	irq_set_status_flags(irq, IRQ_NOREQUEST|IRQ_NOPROBE);
 
-	if (intr_remapping_enabled)
+	if (irq_remapped(get_irq_chip_data(irq)))
 		free_irte(irq);
 	raw_spin_lock_irqsave(&vector_lock, flags);
 	__clear_irq_vector(irq, cfg);
#4 lore raw
Quoted 9 lines · Yinghai Lu
Quoted 8 lines · Russ Anderson
Quoted 7 lines · Suresh Siddha
Quoted 5 lines · Russ Anderson

There is a regression that is causing a NULL pointer dereference in free_irte when shutting down xpc.  git bisect narrowed it down to git commit d585d060b42bd36f6f0b23ff327d3b91f80c7139, which changed free_irte().  Reverse applying the patch fixes the problem.

Quoted 7 lines
Quoted 5 lines
Quoted 1 line

Russ, Is this with !CONFIG_SPARSE_IRQ?

CONFIG_SPARSE_IRQ=y

can you try attached patch?

That fixes it.

Quoted 1 line

Yinghai

Thanks!

Signature · 2 lines
Russ Anderson, OS RAS/Partitioning Project Lead  
SGI - Silicon Graphics Inc          [email protected]
#5 lore raw

PATCH 1/1 x86, uv: Fix uv with destroy_irq()

Russ found that:

| There is a regression that is causing a NULL pointer dereference
| in free_irte when shutting down xpc.  git bisect narrowed it down
| to git commit d585d060b42bd36f6f0b23ff327d3b91f80c7139, which
| changed free_irte().  Reverse applying the patch fixes the problem.

and he bisected to

| commit d585d060b42bd36f6f0b23ff327d3b91f80c7139
| Author: Thomas Gleixner <[email protected]>
|Date:   Sun Oct 10 12:34:27 2010 +0200
|
|   intr_remap: Simplify the code further
|
|   Having irq_2_iommu in struct irq_cfg allows further simplifications.

We need to use irq_mapped() for every irq instead of intr_remapping_enabled overall

---
 arch/x86/kernel/apic/io_apic.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/kernel/apic/io_apic.c +1 -1
Index: linux-2.6/arch/x86/kernel/apic/io_apic.c
===================================================================
@@ -3109,7 +3109,7 @@ void destroy_irq(unsigned int irq)
 
 	irq_set_status_flags(irq, IRQ_NOREQUEST|IRQ_NOPROBE);
 
-	if (intr_remapping_enabled)
+	if (irq_remapped(get_irq_chip_data(irq)))
 		free_irte(irq);
 	raw_spin_lock_irqsave(&vector_lock, flags);
 	__clear_irq_vector(irq, cfg);
#6 lore raw

Re: [PATCH] x86, uv: Fix uv with destroy_irq()

Quoted 35 lines · Yinghai Lu

Russ found that:

| There is a regression that is causing a NULL pointer dereference
| in free_irte when shutting down xpc.  git bisect narrowed it down
| to git commit d585d060b42bd36f6f0b23ff327d3b91f80c7139, which
| changed free_irte().  Reverse applying the patch fixes the problem.

and he bisected to

| commit d585d060b42bd36f6f0b23ff327d3b91f80c7139
| Author: Thomas Gleixner <[email protected]>
|Date:   Sun Oct 10 12:34:27 2010 +0200
|
|   intr_remap: Simplify the code further
|
|   Having irq_2_iommu in struct irq_cfg allows further simplifications.

We need to use irq_mapped() for every irq instead of intr_remapping_enabled overall

---
 arch/x86/kernel/apic/io_apic.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Index: linux-2.6/arch/x86/kernel/apic/io_apic.c
===================================================================

arch/x86/kernel/apic/io_apic.c void destroy_irq(unsigned int irq) · line 3112

Quoted 5 lines
arch/x86/kernel/apic/io_apic.c +1 -1
@@ -3109,7 +3109,7 @@ void destroy_irq(unsigned int irq)
 
 	irq_set_status_flags(irq, IRQ_NOREQUEST|IRQ_NOPROBE);
 
-	if (intr_remapping_enabled)
+	if (irq_remapped(get_irq_chip_data(irq)))

No need for get_irq_chip_data(). We have cfg already. Will fix.

Thanks,

tglx

Quoted 4 lines
arch/x86/kernel/apic/io_apic.c +0 -0
 		free_irte(irq);
 	raw_spin_lock_irqsave(&vector_lock, flags);
 	__clear_irq_vector(irq, cfg);
#7 lore raw

[tip:x86/urgent] x86: Check irq_remapped instead of remapping_enabled in destroy_irq()

In reply to Yinghai Lu
Commit-ID:  7b79462a20826a7269322113c68ca78d5f67c0bd
Gitweb:     http://git.kernel.org/tip/7b79462a20826a7269322113c68ca78d5f67c0bd
Author:     Yinghai Lu <[email protected]>
AuthorDate: Sat, 30 Oct 2010 01:19:29 -0700
Committer:  Thomas Gleixner <[email protected]>
CommitDate: Sat, 30 Oct 2010 10:28:31 +0200

x86: Check irq_remapped instead of remapping_enabled in destroy_irq()

Russ Anderson reported:

| There is a regression that is causing a NULL pointer dereference
| in free_irte when shutting down xpc. git bisect narrowed it down
| to git commit d585d06(intr_remap: Simplify the code further), which
| changed free_irte(). Reverse applying the patch fixes the problem.

We need to use irq_remapped() for each irq instead of checking only intr_remapping_enabled as there might be non remapped irqs even when remapping is enabled.

[ tglx: use cfg instead of retrieving it again. Massaged changelog ]

Reported-bisected-and-tested-by Russ Anderson <[email protected]>
Cc Suresh Siddha <[email protected]>

LKML-Reference: <[email protected]>

---
 arch/x86/kernel/apic/io_apic.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
arch/x86/kernel/apic/io_apic.c +1 -1
index 0929191..7cc0a72 100644
@@ -3109,7 +3109,7 @@ void destroy_irq(unsigned int irq)
 
 	irq_set_status_flags(irq, IRQ_NOREQUEST|IRQ_NOPROBE);
 
-	if (intr_remapping_enabled)
+	if (irq_remapped(cfg))
 		free_irte(irq);
 	raw_spin_lock_irqsave(&vector_lock, flags);
 	__clear_irq_vector(irq, cfg);