Skip to contentSkip to conversation

OOPS at dmar_table_init (2.6.22-rc6-mm1 kernel)

DISCUSSION Pavel Emelianov 4 participants 4 messages Jul 10, 2007 PCI

lore
Share & tools
lore
#0 lore raw

Hi.

While working with Andrew's kernel I faced an OOPS on x86_64 machine. Unfortunately kernel logs are not appearing in serial console by the time oops happens, but I have some info on the screen:

OOPs is at dmar_table_init() here:
ffffffff80532873 <dmar_table_init>:

ffffffff80532873:       41 55                   push   %r13
ffffffff80532875:       41 54                   push   %r12
ffffffff80532877:       55                      push   %rbp
ffffffff80532878:       53                      push   %rbx
ffffffff80532879:       51                      push   %rcx
ffffffff8053287a:       4c 8b 2d 07 21 01 00    mov    73991(%rip),%r13        # ffffffff80544988 <dmar_tbl>
ffffffff80532881:       41 0f b6 45 24          movzbl 0x24(%r13),%eax  <<<<<<<<<<< OOPS, %r13 = NULL
ffffffff80532886:       84 c0                   test   %al,%al
ffffffff80532888:       75 11                   jne    ffffffff8053289b <dmar_table_init+0x28>
ffffffff8053288a:       48 c7 c7 b0 3d 45 80    mov    $0xffffffff80453db0,%rdi
ffffffff80532891:       e8 c2 4a cf ff          callq  ffffffff80227358 <printk>
ffffffff80532896:       e9 f2 01 00 00          jmpq   ffffffff80532a8d <dmar_table_init+0x21a>

Looks like dmar_tbl is NULL.
BUG is 100% reproducible. Is there any other info that can be useful?

Thanks,
Pavel

#1 lore raw
Quoted 22 lines · On Tue, 10 Jul 2007 17:57:25 +0400 Pavel Emelianov wrote:

Hi.

While working with Andrew's kernel I faced an OOPS on x86_64 machine. Unfortunately kernel logs are not appearing in serial console by the time oops happens, but I have some info on the screen:

OOPs is at dmar_table_init() here:
ffffffff80532873 <dmar_table_init>:

ffffffff80532873:       41 55                   push   %r13
ffffffff80532875:       41 54                   push   %r12
ffffffff80532877:       55                      push   %rbp
ffffffff80532878:       53                      push   %rbx
ffffffff80532879:       51                      push   %rcx
ffffffff8053287a:       4c 8b 2d 07 21 01 00    mov    73991(%rip),%r13        # ffffffff80544988 <dmar_tbl>
ffffffff80532881:       41 0f b6 45 24          movzbl 0x24(%r13),%eax  <<<<<<<<<<< OOPS, %r13 = NULL
ffffffff80532886:       84 c0                   test   %al,%al
ffffffff80532888:       75 11                   jne    ffffffff8053289b <dmar_table_init+0x28>
ffffffff8053288a:       48 c7 c7 b0 3d 45 80    mov    $0xffffffff80453db0,%rdi
ffffffff80532891:       e8 c2 4a cf ff          callq  ffffffff80227358 <printk>
ffffffff80532896:       e9 f2 01 00 00          jmpq   ffffffff80532a8d <dmar_table_init+0x21a>
Quoted 2 lines

Looks like dmar_tbl is NULL.
BUG is 100% reproducible. Is there any other info that can be useful?

You could test the patch from
http://lkml.org/lkml/2007/6/29/174 (see "Get diff 1").

---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***

#2 lore raw

(Sorry for the top post)

This bug is fixed and the fix should appear in the next MM release. See this link for the patch
http://marc.info/?l=linux-kernel&m=118313130109808&w=2

-Anil
-----Original Message-----
From: Randy Dunlap [mailto:[email protected]]
Sent: Tuesday, July 10, 2007 9:50 AM
To: Pavel Emelianov
Cc: Raj, Ashok; Li, Shaohua; Keshavamurthy, Anil S; Linux Kernel Mailing List; Andrew Morton
Subject: Re: OOPS at dmar_table_init (2.6.22-rc6-mm1 kernel)

Quoted 7 lines · On Tue, 10 Jul 2007 17:57:25 +0400 Pavel Emelianov wrote:

Hi.

While working with Andrew's kernel I faced an OOPS on x86_64 machine. Unfortunately kernel logs are not appearing in serial console by the time oops happens, but I have some info on the screen:

Quoted 8 lines

OOPs is at dmar_table_init() here:
ffffffff80532873 <dmar_table_init>:

ffffffff80532873:       41 55                   push   %r13
ffffffff80532875:       41 54                   push   %r12
ffffffff80532877:       55                      push   %rbp
ffffffff80532878:       53                      push   %rbx
ffffffff80532879:       51                      push   %rcx
ffffffff8053287a:       4c 8b 2d 07 21 01 00    mov

73991(%rip),%r13 # ffffffff80544988 <dmar_tbl>

Quoted 1 line

ffffffff80532881: 41 0f b6 45 24 movzbl 0x24(%r13),%eax

<<<<<<<<<<< OOPS, %r13 = NULL

Quoted 2 lines
ffffffff80532886:       84 c0                   test   %al,%al
ffffffff80532888:       75 11                   jne

ffffffff8053289b <dmar_table_init+0x28>

Quoted 1 line

ffffffff8053288a: 48 c7 c7 b0 3d 45 80 mov

$0xffffffff80453db0,%rdi

Quoted 1 line

ffffffff80532891: e8 c2 4a cf ff callq

ffffffff80227358 <printk>

Quoted 1 line

ffffffff80532896: e9 f2 01 00 00 jmpq

ffffffff80532a8d <dmar_table_init+0x21a>

Quoted 3 lines

Looks like dmar_tbl is NULL.
BUG is 100% reproducible. Is there any other info that can be useful?

You could test the patch from
http://lkml.org/lkml/2007/6/29/174 (see "Get diff 1").

---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code
***

#3 lore raw
In reply to Pavel Emelianov
Quoted 22 lines · On Tue, 10 Jul 2007 17:57:25 +0400 Pavel Emelianov <[email protected]> wrote:

Hi.

While working with Andrew's kernel I faced an OOPS on x86_64 machine. Unfortunately kernel logs are not appearing in serial console by the time oops happens, but I have some info on the screen:

OOPs is at dmar_table_init() here:
ffffffff80532873 <dmar_table_init>:

ffffffff80532873:       41 55                   push   %r13
ffffffff80532875:       41 54                   push   %r12
ffffffff80532877:       55                      push   %rbp
ffffffff80532878:       53                      push   %rbx
ffffffff80532879:       51                      push   %rcx
ffffffff8053287a:       4c 8b 2d 07 21 01 00    mov    73991(%rip),%r13        # ffffffff80544988 <dmar_tbl>
ffffffff80532881:       41 0f b6 45 24          movzbl 0x24(%r13),%eax  <<<<<<<<<<< OOPS, %r13 = NULL
ffffffff80532886:       84 c0                   test   %al,%al
ffffffff80532888:       75 11                   jne    ffffffff8053289b <dmar_table_init+0x28>
ffffffff8053288a:       48 c7 c7 b0 3d 45 80    mov    $0xffffffff80453db0,%rdi
ffffffff80532891:       e8 c2 4a cf ff          callq  ffffffff80227358 <printk>
ffffffff80532896:       e9 f2 01 00 00          jmpq   ffffffff80532a8d <dmar_table_init+0x21a>
Quoted 3 lines

Looks like dmar_tbl is NULL.
BUG is 100% reproducible. Is there any other info that can be useful?

This?

From: "Keshavamurthy, Anil S" <[email protected]>

Check for dmar_tbl pointer as this can be NULL on systems with no Intel VT-d support.

---

 drivers/pci/dmar.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/pci/dmar.c renamed from drivers/pci/dmar.c~intel-iommu-dmar-detection-and-parsing-logic-fix-intel-dmar-crash-on-amd-x86_64 +3 -1
diff -puN drivers/pci/dmar.c~intel-iommu-dmar-detection-and-parsing-logic-fix-intel-dmar-crash-on-amd-x86_64 drivers/pci/dmar.c
@@ -260,6 +260,8 @@ parse_dmar_table(void)
 	int ret = 0;
 
 	dmar = (struct acpi_table_dmar *)dmar_tbl;
+	if (!dmar)
+		return -ENODEV;
 
 	if (!dmar->width) {
 		printk (KERN_WARNING PREFIX "Zero: Invalid DMAR haw\n");
@@ -301,7 +303,7 @@ int __init dmar_table_init(void)
 
 	parse_dmar_table();
 	if (list_empty(&dmar_drhd_units)) {
-		printk(KERN_ERR PREFIX "No DMAR devices found\n");
+		printk(KERN_INFO PREFIX "No DMAR devices found\n");
 		return -ENODEV;
 	}
 	return 0;

_