Skip to contentSkip to conversation

WARNING: at mm/bootmem.c:535 alloc_arch_preferred_bootmem

DISCUSSION Christian Kujau 4 participants 6 messages Jun 30, 2009 – Jul 1, 2009 Xen hypervisor interface

lore
Share & tools
lore
#0 lore raw

Hi there,

this topic has been discussed [0] already and a bug for powerpc[1] has been opened, I'm still seeing this with the latest -git in a x86_64 Xen DomU. The domain continues to boot however, no oopses or crashes so far, it's just this warning during bootup. The thread about the "early boot SLAB" came to a halt, but the warning is still there - is a fix for this still in the works?

Full dmesg/.config: http://nerdbynature.de/bits/2.6.31-rc1/

Thanks,
Christian.

[0] http://bugzilla.kernel.org/show_bug.cgi?id=13650 / Badness on the Warp
[1] http://lkml.org/lkml/2009/6/10/396 / Early boot SLAB for 2.6.31

Signature · 3 lines
BOFH excuse #36:

dynamic software linking table corrupted
#1 lore raw

Hi Christian,

Quoted 9 lines · Christian Kujau

Hi there,

this topic has been discussed [0] already and a bug for powerpc[1] has been opened, I'm still seeing this with the latest -git in a x86_64 Xen DomU. The domain continues to boot however, no oopses or crashes so far, it's just this warning during bootup. The thread about the "early boot SLAB" came to a halt, but the warning is still there - is a fix for this still in the works?

Quoted 7 lines

Full dmesg/.config: http://nerdbynature.de/bits/2.6.31-rc1/

Thanks,
Christian.

[0] http://bugzilla.kernel.org/show_bug.cgi?id=13650 / Badness on the Warp
[1] http://lkml.org/lkml/2009/6/10/396 / Early boot SLAB for 2.6.31

I think you're the first one to report this. The warning is harmless but the following patch should fix things up. Thanks!

Pekka

Quoted 1 line

From dc231c59194a973b80e6932bb4c162f2076635cb Mon Sep 17 00:00:00 2001

From: Pekka Enberg <[email protected]>
Date: Tue, 30 Jun 2009 11:35:19 +0300
Subject: [PATCH] xen: Use kcalloc() in xen_init_IRQ()

The init_IRQ() function is now called with slab allocator initialized. Therefore, we must not use the bootmem allocator in xen_init_IRQ(). Fixes the following boot-time warning:

  [    0.000000] Linux version 2.6.30 (dummy@sid) (gcc version 4.4.0 (Debian 4.4.0-8) ) #1 Tue Jun 23 02:13:42 CEST 2009
  [    0.000000] Command line: root=/dev/xvda1 console=hvc0 init=/sbin/init
  [    0.000000] KERNEL supported cpus:
  [    0.000000]   Intel GenuineIntel
  [    0.000000]   AMD AuthenticAMD
  [    0.000000]   Centaur CentaurHauls
  [    0.000000] BIOS-provided physical RAM map:
  [    0.000000]  Xen: 0000000000000000 - 00000000000a0000 (usable)
  [    0.000000]  Xen: 00000000000a0000 - 0000000000100000 (reserved)
  [    0.000000]  Xen: 0000000000100000 - 0000000008000000 (usable)
  [    0.000000] DMI not present or invalid.
  [    0.000000] last_pfn = 0x8000 max_arch_pfn = 0x400000000
  [    0.000000] initial memory mapped : 0 - 20000000
---
 drivers/xen/events.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)
drivers/xen/events.c +2 -2
index 891d2e9..abad71b 100644
@@ -927,9 +927,9 @@ static struct irq_chip xen_dynamic_chip __read_mostly = {
 void __init xen_init_IRQ(void)
 {
 	int i;
-	size_t size = nr_cpu_ids * sizeof(struct cpu_evtchn_s);
 
-	cpu_evtchn_mask_p = alloc_bootmem(size);
+	cpu_evtchn_mask_p = kcalloc(nr_cpu_ids, sizeof(struct cpu_evtchn_s),
+				    GFP_KERNEL);
 	BUG_ON(cpu_evtchn_mask_p == NULL);
 
 	init_evtchn_cpu_bindings();
Signature · 1 line
1.5.6.3
#2 lore raw
Quoted 2 lines · Pekka Enberg

I think you're the first one to report this. The warning is harmless but the following patch should fix things up. Thanks!

Yes, your patch makes the warning go away, Thanks!

But, I think the patch description should read:

Quoted 14 lines

Fixes the following boot-time warning:

------------[ cut here ]------------
WARNING: at mm/bootmem.c:535 alloc_arch_preferred_bootmem+0x27/0x45()
Modules linked in:
Pid: 0, comm: swapper Not tainted 2.6.30 #1
Call Trace:
 [<ffffffff8102d6e3>] ? warn_slowpath_common+0x73/0xb0
 [<ffffffff810210d9>] ? pvclock_clocksource_read+0x49/0x90
 [<ffffffff812e522f>] ? alloc_arch_preferred_bootmem+0x27/0x45
 [<ffffffff812e5761>] ? ___alloc_bootmem_nopanic+0x39/0xc9
 [<ffffffff812e57fa>] ? ___alloc_bootmem+0x9/0x2f
 [<ffffffff812e9e21>] ? xen_init_IRQ+0x25/0x61
 [<ffffffff812d69ee>] ? start_kernel+0x1b5/0x29e
---[ end trace 4eaa2a86a8e2da22 ]---

[...]

(I myself forgot to put this warning into my initial message too)

Thanks for the fast response and the fix!

Christian.

Signature · 3 lines
BOFH excuse #65:

system needs to be rebooted
#3 lore raw
In reply to Pekka Enberg
Quoted 18 lines · Pekka Enberg

Hi Christian,

Quoted 14 lines · Christian Kujau

Hi there,

this topic has been discussed [0] already and a bug for powerpc[1] has been opened, I'm still seeing this with the latest -git in a x86_64 Xen DomU. The domain continues to boot however, no oopses or crashes so far, it's just this warning during bootup. The thread about the "early boot SLAB" came to a halt, but the warning is still there - is a fix for this still in the works?

Full dmesg/.config: http://nerdbynature.de/bits/2.6.31-rc1/

Thanks,
Christian.

Quoted 7 lines
Quoted 3 lines

[0] http://bugzilla.kernel.org/show_bug.cgi?id=13650 / Badness on the Warp
[1] http://lkml.org/lkml/2009/6/10/396 / Early boot SLAB for 2.6.31

I think you're the first one to report this. The warning is harmless but the following patch should fix things up. Thanks!

Thanks for this. I'd noticed it locally, but I hadn't got around to checking to see if it affected mainline.

Quoted 25 lines

Pekka

Quoted 1 line

From dc231c59194a973b80e6932bb4c162f2076635cb Mon Sep 17 00:00:00 2001

From: Pekka Enberg <[email protected]>
Date: Tue, 30 Jun 2009 11:35:19 +0300
Subject: [PATCH] xen: Use kcalloc() in xen_init_IRQ()

The init_IRQ() function is now called with slab allocator initialized. Therefore, we must not use the bootmem allocator in xen_init_IRQ(). Fixes the following boot-time warning:

  [    0.000000] Linux version 2.6.30 (dummy@sid) (gcc version 4.4.0 (Debian 4.4.0-8) ) #1 Tue Jun 23 02:13:42 CEST 2009
  [    0.000000] Command line: root=/dev/xvda1 console=hvc0 init=/sbin/init
  [    0.000000] KERNEL supported cpus:
  [    0.000000]   Intel GenuineIntel
  [    0.000000]   AMD AuthenticAMD
  [    0.000000]   Centaur CentaurHauls
  [    0.000000] BIOS-provided physical RAM map:
  [    0.000000]  Xen: 0000000000000000 - 00000000000a0000 (usable)
  [    0.000000]  Xen: 00000000000a0000 - 0000000000100000 (reserved)
  [    0.000000]  Xen: 0000000000100000 - 0000000008000000 (usable)
  [    0.000000] DMI not present or invalid.
  [    0.000000] last_pfn = 0x8000 max_arch_pfn = 0x400000000
  [    0.000000] initial memory mapped : 0 - 20000000
Quoted 3 lines
Quoted 21 lines
---
 drivers/xen/events.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)
drivers/xen/events.c +2 -2
index 891d2e9..abad71b 100644
@@ -927,9 +927,9 @@ static struct irq_chip xen_dynamic_chip __read_mostly = {
 void __init xen_init_IRQ(void)
 {
 	int i;
-	size_t size = nr_cpu_ids * sizeof(struct cpu_evtchn_s);
 
-	cpu_evtchn_mask_p = alloc_bootmem(size);
+	cpu_evtchn_mask_p = kcalloc(nr_cpu_ids, sizeof(struct cpu_evtchn_s),
+				    GFP_KERNEL);
 	BUG_ON(cpu_evtchn_mask_p == NULL);
 
 	init_evtchn_cpu_bindings();
#4 lore raw

Hi Jeremy,

Quoted 23 lines · Jeremy Fitzhardinge
Quoted 23 lines
Quoted 1 line

From dc231c59194a973b80e6932bb4c162f2076635cb Mon Sep 17 00:00:00 2001

From: Pekka Enberg <[email protected]>
Date: Tue, 30 Jun 2009 11:35:19 +0300
Subject: [PATCH] xen: Use kcalloc() in xen_init_IRQ()

The init_IRQ() function is now called with slab allocator initialized. Therefore, we must not use the bootmem allocator in xen_init_IRQ(). Fixes the following boot-time warning:

  [    0.000000] Linux version 2.6.30 (dummy@sid) (gcc version 4.4.0 (Debian 4.4.0-8) ) #1 Tue Jun 23 02:13:42 CEST 2009
  [    0.000000] Command line: root=/dev/xvda1 console=hvc0 init=/sbin/init
  [    0.000000] KERNEL supported cpus:
  [    0.000000]   Intel GenuineIntel
  [    0.000000]   AMD AuthenticAMD
  [    0.000000]   Centaur CentaurHauls
  [    0.000000] BIOS-provided physical RAM map:
  [    0.000000]  Xen: 0000000000000000 - 00000000000a0000 (usable)
  [    0.000000]  Xen: 00000000000a0000 - 0000000000100000 (reserved)
  [    0.000000]  Xen: 0000000000100000 - 0000000008000000 (usable)
  [    0.000000] DMI not present or invalid.
  [    0.000000] last_pfn = 0x8000 max_arch_pfn = 0x400000000
  [    0.000000] initial memory mapped : 0 - 20000000
Quoted 4 lines
Quoted 2 lines

Thanks! Who should pick this patch up, btw? I think it ought to go to the xen subsystem tree (wherever that is).

Pekka

#5 lore raw
Quoted 26 lines · Pekka Enberg

Hi Jeremy,

Quoted 23 lines · Jeremy Fitzhardinge
Quoted 23 lines
Quoted 1 line

From dc231c59194a973b80e6932bb4c162f2076635cb Mon Sep 17 00:00:00 2001

From: Pekka Enberg <[email protected]>
Date: Tue, 30 Jun 2009 11:35:19 +0300
Subject: [PATCH] xen: Use kcalloc() in xen_init_IRQ()

The init_IRQ() function is now called with slab allocator initialized. Therefore, we must not use the bootmem allocator in xen_init_IRQ(). Fixes the following boot-time warning:

  [    0.000000] Linux version 2.6.30 (dummy@sid) (gcc version 4.4.0 (Debian 4.4.0-8) ) #1 Tue Jun 23 02:13:42 CEST 2009
  [    0.000000] Command line: root=/dev/xvda1 console=hvc0 init=/sbin/init
  [    0.000000] KERNEL supported cpus:
  [    0.000000]   Intel GenuineIntel
  [    0.000000]   AMD AuthenticAMD
  [    0.000000]   Centaur CentaurHauls
  [    0.000000] BIOS-provided physical RAM map:
  [    0.000000]  Xen: 0000000000000000 - 00000000000a0000 (usable)
  [    0.000000]  Xen: 00000000000a0000 - 0000000000100000 (reserved)
  [    0.000000]  Xen: 0000000000100000 - 0000000008000000 (usable)
  [    0.000000] DMI not present or invalid.
  [    0.000000] last_pfn = 0x8000 max_arch_pfn = 0x400000000
  [    0.000000] initial memory mapped : 0 - 20000000
Quoted 7 lines
Quoted 4 lines
Quoted 2 lines

Thanks! Who should pick this patch up, btw? I think it ought to go to the xen subsystem tree (wherever that is).

Generally those fixes go via tip:x86/urgent - please send the patch with a full changelog, the signoff and acked-by lines, etc.

Thanks,

Ingo