Hi,
latest linus tree kernel hangs on my Thinkpad T420.
I bisected this issue, culprit commit is below:
commit 1acba98f810a14b1255e34bc620594f83de37e36
Author: Matthew Garrett <[email protected]>
Date: Sun Jun 2 18:12:25 2013 -0400
UEFI: Don't pass boot services regions to SetVirtualAddressMap()
We need to map boot services regions during startup in order to avoid
firmware bugs, but we shouldn't be passing those regions to
SetVirtualAddressMap(). Ensure that we're only passing regions that are
marked as being mapped at runtime.
Signed-off-by: Matthew Garrett <[email protected]>
Signed-off-by: Matt Fleming <[email protected]>
Reverting this patch, the kernel boot just fine.
Another problem is: With this patch applied I tried noefi boot, but
kernel paniced, looks like efivar_init depends on efi runtime. Below
patches works for me about noefi boot though I'm not sure if it's a
right fix, please review.
--- linux-2.6.orig/drivers/firmware/efi/vars.c
+++ linux-2.6/drivers/firmware/efi/vars.c
@@ -370,6 +370,8 @@ int efivar_init(int (*func)(efi_char16_t
efi_guid_t vendor_guid;
int err = 0;
+ if (! efi_enabled(EFI_RUNTIME_SERVICES))
+ return -ENODEV;
variable_name = kzalloc(variable_name_size, GFP_KERNEL);
if (!variable_name) {
printk(KERN_ERR "efivars: Memory allocation failed.\n");
--
Thanks
Dave
On 10/07/13 07:36, Dave Young wrote:
> Another problem is: With this patch applied I tried noefi boot, but
> kernel paniced, looks like efivar_init depends on efi runtime. Below
> patches works for me about noefi boot though I'm not sure if it's a
> right fix, please review.
It makes more sense to check in efivars.c, since none of that file can
function without runtime serivces.
Could you try the following patch?
---
diff --git a/drivers/firmware/efi/efivars.c b/drivers/firmware/efi/efivars.c
index 8bd1bb6..8a7432a 100644
--- a/drivers/firmware/efi/efivars.c
+++ b/drivers/firmware/efi/efivars.c
@@ -583,6 +583,9 @@ int efivars_sysfs_init(void)
struct kobject *parent_kobj = efivars_kobject();
int error = 0;
+ if (!efi_enabled(EFI_RUNTIME_SERVICES))
+ return -ENODEV;
+
/* No efivars has been registered yet */
if (!parent_kobj)
return 0;
On 07/10/2013 08:36 PM, Matt Fleming wrote:
> On 10/07/13 07:36, Dave Young wrote:
>> Another problem is: With this patch applied I tried noefi boot, but
>> kernel paniced, looks like efivar_init depends on efi runtime. Below
>> patches works for me about noefi boot though I'm not sure if it's a
>> right fix, please review.
>
> It makes more sense to check in efivars.c, since none of that file can
> function without runtime serivces.
>
> Could you try the following patch?
It works, thanks for the patch.
Tested-by: Dave Young <[email protected]>
>
> ---
>
> diff --git a/drivers/firmware/efi/efivars.c b/drivers/firmware/efi/efivars.c
> index 8bd1bb6..8a7432a 100644
> --- a/drivers/firmware/efi/efivars.c
> +++ b/drivers/firmware/efi/efivars.c
> @@ -583,6 +583,9 @@ int efivars_sysfs_init(void)
> struct kobject *parent_kobj = efivars_kobject();
> int error = 0;
>
> + if (!efi_enabled(EFI_RUNTIME_SERVICES))
> + return -ENODEV;
> +
> /* No efivars has been registered yet */
> if (!parent_kobj)
> return 0;
>
--
Thanks
Dave