Skip to contentSkip to conversation

xen: fix hypervisor crash with xentrace

PATCH In review Christoph Egger 3 participants 5 messages Apr 6, 2011

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 xen: fix hypervisor crash with xentrace

Hi,

Attached patch fixes hypervisor crashes when xentrace buffer gets initialized.

The xen backtraces I have seen are:

(XEN) Xen call trace:

(XEN)    [<ffff82c48025cf70>] early_stack+0x0/0x48
(XEN)    [<ffff82c4801278d0>] tb_control+0x182/0x278
(XEN)    [<ffff82c480124aac>] do_sysctl+0x1fc/0xa40
(XEN)    [<ffff82c480213248>] syscall_enter+0xc8/0x122
(XEN)
(XEN) Pagetable walk from ffff82c48025cf70:
(XEN)  L4[0x105] = 00000000cfcaa027 5555555555555555
(XEN)  L3[0x112] = 00000000cfcab027 5555555555555555
(XEN)  L2[0x001] = 000000022f27b063 5555555555555555
(XEN)  L1[0x05c] = 00000000cfc5c262 5555555555555555

(XEN)
(XEN) **************************************** (XEN) Panic on CPU 0:
(XEN) FATAL PAGE FAULT
(XEN) [error_code=0010]
(XEN) Faulting linear address: ffff82c48025cf70 (XEN) ****************************************

(XEN) Xen call trace:

(XEN)    [<ffff82c48025d852>] notifier_chain_register+0x0/0x35
(XEN)    [<ffff82c4801275b2>] alloc_trace_bufs+0x5cd/0x7a9
(XEN)    [<ffff82c480127910>] tb_control+0x182/0x278
(XEN)    [<ffff82c480124aec>] do_sysctl+0x1fc/0xa40
(XEN)    [<ffff82c480213288>] syscall_enter+0xc8/0x122
(XEN)
(XEN) Pagetable walk from ffff82c48025d852:
(XEN)  L4[0x105] = 00000000cfcaa027 5555555555555555
(XEN)  L3[0x112] = 00000000cfcab027 5555555555555555
(XEN)  L2[0x001] = 000000022f27b063 5555555555555555
(XEN)  L1[0x05d] = 00000000cfc5d262 5555555555555555

(XEN)
(XEN) **************************************** (XEN) Panic on CPU 1:
(XEN) FATAL PAGE FAULT
(XEN) [error_code=0010]
(XEN) Faulting linear address: ffff82c48025d852 (XEN) ****************************************

Signature · 6 lines
---to satisfy European Law for business letters:
Advanced Micro Devices GmbH
Einsteinring 24, 85609 Dornach b. Muenchen
Geschaeftsfuehrer: Alberto Bozzo, Andrew Bowd
Sitz: Dornach, Gemeinde Aschheim, Landkreis Muenchen
Registergericht Muenchen, HRB Nr. 43632
xen/common/cpu.c +1 -1
diff -r 3fd604b77a5e xen/common/cpu.c
@@ -54,7 +54,7 @@ void cpu_hotplug_done(void)
 
 static NOTIFIER_HEAD(cpu_chain);
 
-void __init register_cpu_notifier(struct notifier_block *nb)
+void register_cpu_notifier(struct notifier_block *nb)
 {
     if ( !spin_trylock(&cpu_add_remove_lock) )
         BUG(); /* Should never fail as we are called only during boot. */
xen/common/notifier.c +1 -1
diff -r 3fd604b77a5e xen/common/notifier.c
@@ -19,7 +19,7 @@
  * Adds a notifier to a raw notifier chain.
  * All locking must be provided by the caller.
  */
-void __init notifier_chain_register(
+void notifier_chain_register(
     struct notifier_head *nh, struct notifier_block *n)
 {
     struct list_head *chain = &nh->head.chain;

_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xensource.com/xen-devel

#1 lore raw
Quoted 4 lines · Christoph Egger

Hi,

Attached patch fixes hypervisor crashes when xentrace buffer gets initialized.

Forgot to mention this regression was introduced in c/s 23134:f5d33782a57b by Jan Beulich.

Quoted 47 lines

The xen backtraces I have seen are:

(XEN) Xen call trace:

(XEN)    [<ffff82c48025cf70>] early_stack+0x0/0x48
(XEN)    [<ffff82c4801278d0>] tb_control+0x182/0x278
(XEN)    [<ffff82c480124aac>] do_sysctl+0x1fc/0xa40
(XEN)    [<ffff82c480213248>] syscall_enter+0xc8/0x122
(XEN)
(XEN) Pagetable walk from ffff82c48025cf70:
(XEN)  L4[0x105] = 00000000cfcaa027 5555555555555555
(XEN)  L3[0x112] = 00000000cfcab027 5555555555555555
(XEN)  L2[0x001] = 000000022f27b063 5555555555555555
(XEN)  L1[0x05c] = 00000000cfc5c262 5555555555555555

(XEN)
(XEN) **************************************** (XEN) Panic on CPU 0:
(XEN) FATAL PAGE FAULT
(XEN) [error_code=0010]
(XEN) Faulting linear address: ffff82c48025cf70 (XEN) ****************************************

(XEN) Xen call trace:

(XEN)    [<ffff82c48025d852>] notifier_chain_register+0x0/0x35
(XEN)    [<ffff82c4801275b2>] alloc_trace_bufs+0x5cd/0x7a9
(XEN)    [<ffff82c480127910>] tb_control+0x182/0x278
(XEN)    [<ffff82c480124aec>] do_sysctl+0x1fc/0xa40
(XEN)    [<ffff82c480213288>] syscall_enter+0xc8/0x122
(XEN)
(XEN) Pagetable walk from ffff82c48025d852:
(XEN)  L4[0x105] = 00000000cfcaa027 5555555555555555
(XEN)  L3[0x112] = 00000000cfcab027 5555555555555555
(XEN)  L2[0x001] = 000000022f27b063 5555555555555555
(XEN)  L1[0x05d] = 00000000cfc5d262 5555555555555555

(XEN)
(XEN) **************************************** (XEN) Panic on CPU 1:
(XEN) FATAL PAGE FAULT
(XEN) [error_code=0010]
(XEN) Faulting linear address: ffff82c48025d852 (XEN) ****************************************

Signature · 6 lines
---to satisfy European Law for business letters:
Advanced Micro Devices GmbH
Einsteinring 24, 85689 Dornach b. Muenchen
Geschaeftsfuehrer: Alberto Bozzo, Andrew Bowd
Sitz: Dornach, Gemeinde Aschheim, Landkreis Muenchen
Registergericht Muenchen, HRB Nr. 43632
#3 lore raw
In reply to Christoph Egger
Quoted 4 lines · Christoph Egger

Hi,

Attached patch fixes hypervisor crashes when xentrace buffer gets initialized.

Should be fixed by c/s 23173. Thanks!

-- Keir

Quoted 45 lines

The xen backtraces I have seen are:

(XEN) Xen call trace:

(XEN)    [<ffff82c48025cf70>] early_stack+0x0/0x48
(XEN)    [<ffff82c4801278d0>] tb_control+0x182/0x278
(XEN)    [<ffff82c480124aac>] do_sysctl+0x1fc/0xa40
(XEN)    [<ffff82c480213248>] syscall_enter+0xc8/0x122
(XEN)
(XEN) Pagetable walk from ffff82c48025cf70:
(XEN)  L4[0x105] = 00000000cfcaa027 5555555555555555
(XEN)  L3[0x112] = 00000000cfcab027 5555555555555555
(XEN)  L2[0x001] = 000000022f27b063 5555555555555555
(XEN)  L1[0x05c] = 00000000cfc5c262 5555555555555555

(XEN)
(XEN) **************************************** (XEN) Panic on CPU 0:
(XEN) FATAL PAGE FAULT
(XEN) [error_code=0010]
(XEN) Faulting linear address: ffff82c48025cf70 (XEN) ****************************************

(XEN) Xen call trace:

(XEN)    [<ffff82c48025d852>] notifier_chain_register+0x0/0x35
(XEN)    [<ffff82c4801275b2>] alloc_trace_bufs+0x5cd/0x7a9
(XEN)    [<ffff82c480127910>] tb_control+0x182/0x278
(XEN)    [<ffff82c480124aec>] do_sysctl+0x1fc/0xa40
(XEN)    [<ffff82c480213288>] syscall_enter+0xc8/0x122
(XEN)
(XEN) Pagetable walk from ffff82c48025d852:
(XEN)  L4[0x105] = 00000000cfcaa027 5555555555555555
(XEN)  L3[0x112] = 00000000cfcab027 5555555555555555
(XEN)  L2[0x001] = 000000022f27b063 5555555555555555
(XEN)  L1[0x05d] = 00000000cfc5d262 5555555555555555

(XEN)
(XEN) **************************************** (XEN) Panic on CPU 1:
(XEN) FATAL PAGE FAULT
(XEN) [error_code=0010]
(XEN) Faulting linear address: ffff82c48025d852 (XEN) ****************************************

#2 lore raw
In reply to Christoph Egger
Quoted 10 lines
Quoted 1 line
Quoted 1 line · Christoph Egger
Quoted 4 lines · Christoph Egger

Hi,

Attached patch fixes hypervisor crashes when xentrace buffer gets

initialized.

Forgot to mention this regression was introduced in c/s 23134:f5d33782a57b by Jan Beulich.

But just removing the __init-s isn't the right way to fix this (Did you notice the comment in register_cpu_notifier(), which I also pointed out in the changeset description). Apparently I overlooked a call to register_cpu_notifier(), and it's that which ought to get moved into an __init function.

Jan

#4 lore raw
In reply to Christoph Egger
Quoted 3 lines
Quoted 1 line
Quoted 1 line · Christoph Egger

Attached patch fixes hypervisor crashes when xentrace buffer gets initialized.

Just saw that Keir already fixed this the proper way as c/s 23173:94cef9aaf0cd. Thanks, Keir!

Jan

Quoted 45 lines

The xen backtraces I have seen are:

(XEN) Xen call trace:

(XEN)    [<ffff82c48025cf70>] early_stack+0x0/0x48
(XEN)    [<ffff82c4801278d0>] tb_control+0x182/0x278
(XEN)    [<ffff82c480124aac>] do_sysctl+0x1fc/0xa40
(XEN)    [<ffff82c480213248>] syscall_enter+0xc8/0x122
(XEN)
(XEN) Pagetable walk from ffff82c48025cf70:
(XEN)  L4[0x105] = 00000000cfcaa027 5555555555555555
(XEN)  L3[0x112] = 00000000cfcab027 5555555555555555
(XEN)  L2[0x001] = 000000022f27b063 5555555555555555
(XEN)  L1[0x05c] = 00000000cfc5c262 5555555555555555

(XEN)
(XEN) **************************************** (XEN) Panic on CPU 0:
(XEN) FATAL PAGE FAULT
(XEN) [error_code=0010]
(XEN) Faulting linear address: ffff82c48025cf70 (XEN) ****************************************

(XEN) Xen call trace:

(XEN)    [<ffff82c48025d852>] notifier_chain_register+0x0/0x35
(XEN)    [<ffff82c4801275b2>] alloc_trace_bufs+0x5cd/0x7a9
(XEN)    [<ffff82c480127910>] tb_control+0x182/0x278
(XEN)    [<ffff82c480124aec>] do_sysctl+0x1fc/0xa40
(XEN)    [<ffff82c480213288>] syscall_enter+0xc8/0x122
(XEN)
(XEN) Pagetable walk from ffff82c48025d852:
(XEN)  L4[0x105] = 00000000cfcaa027 5555555555555555
(XEN)  L3[0x112] = 00000000cfcab027 5555555555555555
(XEN)  L2[0x001] = 000000022f27b063 5555555555555555
(XEN)  L1[0x05d] = 00000000cfc5d262 5555555555555555

(XEN)
(XEN) **************************************** (XEN) Panic on CPU 1:
(XEN) FATAL PAGE FAULT
(XEN) [error_code=0010]
(XEN) Faulting linear address: ffff82c48025d852 (XEN) ****************************************