I have a SuperMicro C2SBX motherboard with BIOS revision 1.0b. With
vt-d enabled in the BIOS, Linux gets into an endless loop printing
"DMAR:Unknown DMAR structure type" when booting. Here is the DMAR ACPI
table:
DMAR @ 0x7fe86dec
0000: 44 4d 41 52 98 00 00 00 01 6f 49 6e 74 65 6c 20 DMAR.....oIntel
0010: 4f 45 4d 44 4d 41 52 20 00 00 04 06 4c 4f 48 52 OEMDMAR ....LOHR
0020: 01 00 00 00 23 00 00 00 00 00 00 00 00 00 00 00 ....#...........
0030: 01 00 58 00 00 00 00 00 00 a0 e8 7f 00 00 00 00 ..X.............
0040: ff ff ef 7f 00 00 00 00 01 08 00 00 00 00 1d 00 ................
0050: 01 08 00 00 00 00 1d 01 01 08 00 00 00 00 1d 02 ................
0060: 01 08 00 00 00 00 1d 07 01 08 00 00 00 00 1a 00 ................
0070: 01 08 00 00 00 00 1a 01 01 08 00 00 00 00 1a 02 ................
0080: 01 08 00 00 00 00 1a 07 01 08 00 00 00 00 1a 07 ................
0090: c0 00 68 00 04 10 66 60 ..h...f`
Here are the messages printed by the kernel:
DMAR:Host address width 36
DMAR:RMRR base: 0x000000007fe8a000 end: 0x000000007fefffff
DMAR:Unknown DMAR structure type
DMAR:Unknown DMAR structure type
DMAR:Unknown DMAR structure type
...
Although I not very familiar with ACPI, to me it looks like
struct acpi_dmar_header::length == 0x0058 is incorrect, causing
parse_dmar_table() to look at an invalid offset on the next loop. This
offset happens to have struct acpi_dmar_header::length == 0x0000, which
prevents the loop from ever terminating. This patch checks for this
condition and bails out instead of looping forever.
Signed-off-by: Tony Battersby <[email protected]>
---
I previously sent this to linux-acpi on August 18, 2008, but got no
reply. Please apply this patch upstream (2.6.29 if possible).
--- linux-2.6.28-git2/drivers/pci/dmar.c.orig 2008-12-24 18:26:37.000000000 -0500
+++ linux-2.6.28-git2/drivers/pci/dmar.c 2008-12-30 11:01:17.000000000 -0500
@@ -339,6 +339,14 @@ parse_dmar_table(void)
entry_header = (struct acpi_dmar_header *)(dmar + 1);
while (((unsigned long)entry_header) <
(((unsigned long)dmar) + dmar_tbl->length)) {
+ /* Avoid looping forever on bad ACPI tables */
+ if (entry_header->length == 0) {
+ printk(KERN_WARNING PREFIX
+ "Invalid 0-length structure\n");
+ ret = -EINVAL;
+ break;
+ }
+
dmar_table_print_dmar_entry(entry_header);
switch (entry_header->type) {
(copying relevant people)
Tony Battersby wrote:
> I have a SuperMicro C2SBX motherboard with BIOS revision 1.0b. With
> vt-d enabled in the BIOS, Linux gets into an endless loop printing
> "DMAR:Unknown DMAR structure type" when booting. Here is the DMAR ACPI
> table:
>
> DMAR @ 0x7fe86dec
> 0000: 44 4d 41 52 98 00 00 00 01 6f 49 6e 74 65 6c 20 DMAR.....oIntel
> 0010: 4f 45 4d 44 4d 41 52 20 00 00 04 06 4c 4f 48 52 OEMDMAR ....LOHR
> 0020: 01 00 00 00 23 00 00 00 00 00 00 00 00 00 00 00 ....#...........
> 0030: 01 00 58 00 00 00 00 00 00 a0 e8 7f 00 00 00 00 ..X.............
> 0040: ff ff ef 7f 00 00 00 00 01 08 00 00 00 00 1d 00 ................
> 0050: 01 08 00 00 00 00 1d 01 01 08 00 00 00 00 1d 02 ................
> 0060: 01 08 00 00 00 00 1d 07 01 08 00 00 00 00 1a 00 ................
> 0070: 01 08 00 00 00 00 1a 01 01 08 00 00 00 00 1a 02 ................
> 0080: 01 08 00 00 00 00 1a 07 01 08 00 00 00 00 1a 07 ................
> 0090: c0 00 68 00 04 10 66 60 ..h...f`
>
> Here are the messages printed by the kernel:
>
> DMAR:Host address width 36
> DMAR:RMRR base: 0x000000007fe8a000 end: 0x000000007fefffff
> DMAR:Unknown DMAR structure type
> DMAR:Unknown DMAR structure type
> DMAR:Unknown DMAR structure type
> ...
>
> Although I not very familiar with ACPI, to me it looks like
> struct acpi_dmar_header::length == 0x0058 is incorrect, causing
> parse_dmar_table() to look at an invalid offset on the next loop. This
> offset happens to have struct acpi_dmar_header::length == 0x0000, which
> prevents the loop from ever terminating. This patch checks for this
> condition and bails out instead of looping forever.
>
> Signed-off-by: Tony Battersby <[email protected]>
> ---
>
> I previously sent this to linux-acpi on August 18, 2008, but got no
> reply. Please apply this patch upstream (2.6.29 if possible).
>
> --- linux-2.6.28-git2/drivers/pci/dmar.c.orig 2008-12-24 18:26:37.000000000 -0500
> +++ linux-2.6.28-git2/drivers/pci/dmar.c 2008-12-30 11:01:17.000000000 -0500
> @@ -339,6 +339,14 @@ parse_dmar_table(void)
> entry_header = (struct acpi_dmar_header *)(dmar + 1);
> while (((unsigned long)entry_header) <
> (((unsigned long)dmar) + dmar_tbl->length)) {
> + /* Avoid looping forever on bad ACPI tables */
> + if (entry_header->length == 0) {
> + printk(KERN_WARNING PREFIX
> + "Invalid 0-length structure\n");
> + ret = -EINVAL;
> + break;
> + }
> +
> dmar_table_print_dmar_entry(entry_header);
>
> switch (entry_header->type) {
>
>
>
--
error compiling committee.c: too many arguments to function
Avi Kivity <[email protected]> wrote:
> (copying relevant people)
>
> Tony Battersby wrote:
>> + "Invalid 0-length structure\n");
This line is everything the reader of your message will see, (unless it happens
not to be the first ACPI error). I would not be able to tell the cause without
grepping the kernel source.
Bodo Eggert wrote:
> Avi Kivity <[email protected]> wrote:
>
>
>> (copying relevant people)
>>
>> Tony Battersby wrote:
>>
>
>
>>> + "Invalid 0-length structure\n");
>>>
>
> This line is everything the reader of your message will see, (unless it happens
> not to be the first ACPI error). I would not be able to tell the cause without
> grepping the kernel source.
>
>
>
Well, there is a PREFIX of "DMAR:" in the message too. And I do not
think it is any less informative than any of the other error messages in
the same function, e.g. "Invalid DMAR haw", or "Unknown DMAR structure
type".
Tony