Cannot add new efi boot entry

DISCUSSION Jiri Slaby 5 participants 15 messages Apr 14, 2013 – Apr 22, 2013 EFI

Share & tools

Jiri Slaby

author

2013-04-14 21:25 UTC #0 lore raw

Hi,

after update to 3.8, every update of the kernel ends up in an unbootable machine. It is due to the following commit:
commit 68d929862e29a8b52a7f2f2f86a0600423b093cd

Author: Matthew Garrett <[email protected]>
Date:   Sat Mar 2 19:40:17 2013 -0500

efi: be more paranoid about available space when creating variables

efibootmgr which tries to add an entry and silently fails when writing to /sys/firmware/efi/vars/new_var with -ENOSPC.

There are many entries in there:

# efibootmgr
BootCurrent: 000D
Timeout: 0 seconds
BootOrder:
0018,0000,0001,0002,0003,0007,0008,0009,000A,000B,000C,000D,000E,000F,0010,0011,0012
Boot0000  Setup
Boot0001  Boot Menu
Boot0002  Diagnostic Splash Screen
Boot0003  Lenovo Diagnostics
Boot0004  Startup Interrupt Menu
Boot0005  ME Configuration Menu
Boot0006  Rescue and Recovery
Boot0007* USB CD
Boot0008* USB FDD
Boot0009* ATAPI CD0
Boot000A* ATA HDD0
Boot000B* ATA HDD1
Boot000C* ATA HDD2
Boot000D* USB HDD
Boot000E* PCI LAN
Boot000F* ATAPI CD1
Boot0010  Other CD
Boot0011* ATA HDD3
Boot0012  Other HDD
Boot0013* IDER BOOT CDROM
Boot0014* IDER BOOT Floppy
Boot0015* ATA HDD
Boot0016* ATAPI CD:
Boot0017* PCI LAN
Boot0018* Linux

Remaining size is about 20k, added entry size is hundreds bytes, store size is 64k.

Obviously lowering the limitation from 1/2 to 1/4 fixes the problem for me because it always worked on my setup to store a new entry...

Any ideas how to overcome that? It would be better to blacklist bad machines rather than whitelist good ones, right?

thanks,

Signature · 2 lines

js
suse labs

joeyli

2013-04-16 09:56 UTC #1 lore raw

Hi Jiri,

於日，2013-04-14 於 23:25 +0200，Jiri Slaby 提到：

Quoted 49 lines

Hi,
after update to 3.8, every update of the kernel ends up in an unbootable machine. It is due to the following commit:
commit 68d929862e29a8b52a7f2f2f86a0600423b093cd
Author: Matthew Garrett <matthew.garrett-05XSO3Yj/[email protected]>
Date:   Sat Mar 2 19:40:17 2013 -0500
efi: be more paranoid about available space when creating variables
efibootmgr which tries to add an entry and silently fails when writing to /sys/firmware/efi/vars/new_var with -ENOSPC.
There are many entries in there:
# efibootmgr
BootCurrent: 000D
Timeout: 0 seconds
BootOrder:
0018,0000,0001,0002,0003,0007,0008,0009,000A,000B,000C,000D,000E,000F,0010,0011,0012
Boot0000  Setup
Boot0001  Boot Menu
Boot0002  Diagnostic Splash Screen
Boot0003  Lenovo Diagnostics
Boot0004  Startup Interrupt Menu
Boot0005  ME Configuration Menu
Boot0006  Rescue and Recovery
Boot0007* USB CD
Boot0008* USB FDD
Boot0009* ATAPI CD0
Boot000A* ATA HDD0
Boot000B* ATA HDD1
Boot000C* ATA HDD2
Boot000D* USB HDD
Boot000E* PCI LAN
Boot000F* ATAPI CD1
Boot0010  Other CD
Boot0011* ATA HDD3
Boot0012  Other HDD
Boot0013* IDER BOOT CDROM
Boot0014* IDER BOOT Floppy
Boot0015* ATA HDD
Boot0016* ATAPI CD:
Boot0017* PCI LAN
Boot0018* Linux
Remaining size is about 20k, added entry size is hundreds bytes, store size is 64k.

Quoted 7 lines

Obviously lowering the limitation from 1/2 to 1/4 fixes the problem for me because it always worked on my setup to store a new entry...
Any ideas how to overcome that? It would be better to blacklist bad machines rather than whitelist good ones, right?
thanks,

I think I just got the same situation on my side with Acer machine. I am trying Matthew's new patchset hope can avoid this situation:

https://lkml.org/lkml/2013/4/15/473

Thanks a lot!
Joey Lee

Matt Fleming

2013-04-16 10:11 UTC #2 lore raw

Quoted 4 lines · joeyli

I think I just got the same situation on my side with Acer machine. I am trying Matthew's new patchset hope can avoid this situation:
https://lkml.org/lkml/2013/4/15/473

Please do let us know whether the patches work for you.

joeyli

2013-04-17 11:33 UTC #3 lore raw

於二，2013-04-16 於 11:11 +0100，Matt Fleming 提到：

Quoted 9 lines · joeyli

Quoted 4 lines
I think I just got the same situation on my side with Acer machine. I am trying Matthew's new patchset hope can avoid this situation:
https://lkml.org/lkml/2013/4/15/473
Please do let us know whether the patches work for you.

I applied Matthew's patches to v3.9-rc7 then tested...

The active_size checking works to me on my Acer machine for avoid the storage_size/2 issue, I can use efibootmgr to create new boot entry, now.

But, I found there have the following warning:

[    1.347581] [Firmware Bug]: efi: Inconsistent initial size

Then, I add printk to print the values of var_size and active_size, found the values always 0 after system reboot. I am doing to add more log for locate if this is a BIOS issue.

Thanks a lot!
Joey Lee

Jiri Slaby

author

2013-04-17 14:29 UTC #4 lore raw

In reply to Matt Fleming

Quoted 7 lines · Matt Fleming

Quoted 4 lines · joeyli
I think I just got the same situation on my side with Acer machine. I am trying Matthew's new patchset hope can avoid this situation:
https://lkml.org/lkml/2013/4/15/473
Please do let us know whether the patches work for you.

Hmm, where did efi_query_variable_store in 3/3 come from?

Signature · 2 lines

js
suse labs

joeyli

2013-04-17 14:49 UTC #5 lore raw

於三，2013-04-17 於 16:29 +0200，Jiri Slaby 提到：

Quoted 11 lines · Matt Fleming

Quoted 7 lines · joeyli
Quoted 4 lines
I think I just got the same situation on my side with Acer machine. I am trying Matthew's new patchset hope can avoid this situation:
https://lkml.org/lkml/2013/4/15/473
Please do let us know whether the patches work for you.
Hmm, where did efi_query_variable_store in 3/3 come from?

here:
https://git.kernel.org/cgit/linux/kernel/git/mfleming/efi.git/commit/?h=urgent&id=a6e4d5a03e9e3587e88aba687d8f225f4f04c792

Matthew's 3 patches all merged to Matt's efi git tree.

Thanks
Joey Leeg

joeyli

2013-04-17 14:51 UTC #6 lore raw

於三，2013-04-17 於 22:50 +0800，joeyli 提到：

Quoted 10 lines

於三，2013-04-17 於 16:29 +0200，Jiri Slaby 提到：
Quoted 9 lines · Matt Fleming
Quoted 7 lines · joeyli
Quoted 4 lines
I think I just got the same situation on my side with Acer machine. I am trying Matthew's new patchset hope can avoid this situation:
https://lkml.org/lkml/2013/4/15/473
Please do let us know whether the patches work for you.

Quoted 7 lines

Quoted 2 lines
Hmm, where did efi_query_variable_store in 3/3 come from?
here:
https://git.kernel.org/cgit/linux/kernel/git/mfleming/efi.git/commit/?h=urgent&id=a6e4d5a03e9e3587e88aba687d8f225f4f04c792
Matthew's 3 patches all merged to Matt's efi git tree.

Forgot mention, in urgent branch

Quoted 5 lines

Thanks
Joey Leeg

Joey Lee

Matt Fleming

2013-04-17 14:51 UTC #7 lore raw

In reply to Jiri Slaby

Quoted 10 lines · Jiri Slaby

Quoted 7 lines · Matt Fleming
Quoted 4 lines · joeyli
I think I just got the same situation on my side with Acer machine. I am trying Matthew's new patchset hope can avoid this situation:
https://lkml.org/lkml/2013/4/15/473
Please do let us know whether the patches work for you.
Hmm, where did efi_query_variable_store in 3/3 come from?

It's in a patch on the 'urgent' branch at,

git://git.kernel.org/pub/scm/linux/kernel/git/mfleming/efi.git

Signature · 1 line

Matt Fleming, Intel Open Source Technology Center

Jiri Slaby

author

2013-04-18 18:07 UTC #8 lore raw

Quoted 8 lines · Matt Fleming

Quoted 7 lines · Jiri Slaby
Quoted 6 lines · Matt Fleming
Quoted 4 lines · joeyli
I think I just got the same situation on my side with Acer machine. I am trying Matthew's new patchset hope can avoid this situation:
https://lkml.org/lkml/2013/4/15/473

Quoted 7 lines

Quoted 3 lines
Quoted 1 line
Please do let us know whether the patches work for you.
Hmm, where did efi_query_variable_store in 3/3 come from?
It's in a patch on the 'urgent' branch at,
git://git.kernel.org/pub/scm/linux/kernel/git/mfleming/efi.git

The 3 patches plus that one work for me.

thanks,

Signature · 2 lines

js
suse labs

Jiri Slaby

author

2013-04-20 07:40 UTC #9 lore raw

Quoted 11 lines · Jiri Slaby

Quoted 10 lines · Matt Fleming
Quoted 9 lines · Jiri Slaby
Quoted 7 lines · Matt Fleming
Quoted 4 lines · joeyli
I think I just got the same situation on my side with Acer machine. I am trying Matthew's new patchset hope can avoid this situation:
https://lkml.org/lkml/2013/4/15/473
Please do let us know whether the patches work for you.

Quoted 7 lines

Quoted 5 lines
Quoted 1 line
Hmm, where did efi_query_variable_store in 3/3 come from?
It's in a patch on the 'urgent' branch at,
git://git.kernel.org/pub/scm/linux/kernel/git/mfleming/efi.git
The 3 patches plus that one work for me.

Actually they don't :(:
open("/sys/firmware/efi/vars/new_var", O_WRONLY) = 3 write(3, "B\0o\0o\0t\0000\0000\0001\0003\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 2084) = -1 ENOSPC (No space left on device)
close(3) = 0

thanks,

Signature · 2 lines

js
suse labs

Matt Fleming

2013-04-22 11:07 UTC #10 lore raw

Quoted 18 lines · Jiri Slaby

Quoted 17 lines · Jiri Slaby
Quoted 15 lines · Matt Fleming
Quoted 10 lines · Jiri Slaby
Quoted 7 lines · Matt Fleming
Quoted 4 lines · joeyli
I think I just got the same situation on my side with Acer machine. I am trying Matthew's new patchset hope can avoid this situation:
https://lkml.org/lkml/2013/4/15/473
Please do let us know whether the patches work for you.
Hmm, where did efi_query_variable_store in 3/3 come from?
It's in a patch on the 'urgent' branch at,
git://git.kernel.org/pub/scm/linux/kernel/git/mfleming/efi.git

Quoted 8 lines

Quoted 1 line
The 3 patches plus that one work for me.
Actually they don't :(:
open("/sys/firmware/efi/vars/new_var", O_WRONLY) = 3 write(3, "B\0o\0o\0t\0000\0000\0001\0003\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 2084) = -1 ENOSPC (No space left on device)
close(3) = 0

Thanks for testing Jiri.

You can disable the EFI storage checks in -rc8 by using the efi_no_storage_paranoia kernel parameter, but it's only recommended to do that if you machine ends up in these kinds of situations where garbage collection isn't be initiated, or your machine ships with > 50% of the variable store in use from the factory.

Matthew, what kind of info do you need from Jiri to improve the storage limit algorithm?

Signature · 1 line

Matt Fleming, Intel Open Source Technology Center

joeyli

2013-04-22 11:31 UTC #11 lore raw

於一，2013-04-22 於 12:07 +0100，Matt Fleming 提到：

Quoted 28 lines · Jiri Slaby

Quoted 18 lines · Jiri Slaby
Quoted 17 lines · Matt Fleming
Quoted 15 lines · Jiri Slaby
Quoted 10 lines · Matt Fleming
Quoted 7 lines · joeyli
Quoted 4 lines
I think I just got the same situation on my side with Acer machine. I am trying Matthew's new patchset hope can avoid this situation:
https://lkml.org/lkml/2013/4/15/473
Please do let us know whether the patches work for you.
Hmm, where did efi_query_variable_store in 3/3 come from?
It's in a patch on the 'urgent' branch at,
git://git.kernel.org/pub/scm/linux/kernel/git/mfleming/efi.git
Quoted 8 lines
Quoted 1 line
The 3 patches plus that one work for me.
Actually they don't :(:
open("/sys/firmware/efi/vars/new_var", O_WRONLY) = 3 write(3, "B\0o\0o\0t\0000\0000\0001\0003\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 2084) = -1 ENOSPC (No space left on device)
close(3) = 0

Quoted 6 lines

Thanks for testing Jiri.
You can disable the EFI storage checks in -rc8 by using the efi_no_storage_paranoia kernel parameter, but it's only recommended to do that if you machine ends up in these kinds of situations where garbage collection isn't be initiated, or your machine ships with > 50%

Does garbage collection only trigger by UEFI BIOS when system reboot? or garbage collection also triggered at run time?

There have a problem is if we delete boot entry when the flash space tight(maybe 99%, just remaining_size < max_size), then we can not create new boot entry unless we reboot system for trigger garbage collection. But we already lost available boot entry.

Quoted 5 lines

of the variable store in use from the factory.
Matthew, what kind of info do you need from Jiri to improve the storage limit algorithm?

Thanks a lot!
Joey Lee

Matt Fleming

2013-04-22 12:20 UTC #12 lore raw

Quoted 2 lines · joeyli

Does garbage collection only trigger by UEFI BIOS when system reboot? or garbage collection also triggered at run time?

That is implementation specific. AFAIK, there are implementations for both schemes.

Signature · 1 line

Matt Fleming, Intel Open Source Technology Center

Jiri Slaby

author

2013-04-22 12:25 UTC #13 lore raw

Quoted 6 lines · Matt Fleming

Quoted 2 lines · joeyli
Does garbage collection only trigger by UEFI BIOS when system reboot? or garbage collection also triggered at run time?
That is implementation specific. AFAIK, there are implementations for both schemes.

Just a point from the abstract level. Is there a way to find out? And what do Windows do?

thanks,

Signature · 2 lines

js
suse labs

Matthew Garrett

2013-04-22 16:43 UTC #14 lore raw

Quoted 10 lines · Jiri Slaby

Quoted 6 lines · Matt Fleming
Quoted 2 lines · joeyli
Does garbage collection only trigger by UEFI BIOS when system reboot? or garbage collection also triggered at run time?
That is implementation specific. AFAIK, there are implementations for both schemes.
Just a point from the abstract level. Is there a way to find out? And what do Windows do?

Windows kills Samsungs.

Signature · 1 line

Matthew Garrett | [email protected]