Revisiting EFI boot stub config file support

Revisiting EFI boot stub config file support

[Shea Levy]

Hi all,

In March last year, I submitted a patch to enable the EFI boot stub to 
read its config arguments from a file on the ESP, both to help with 
booting from livecds and to avoid having to use efibootmgr every time 
you change your boottime arguments. I ended up dropping the submission 
because I ended up coming to the conclusion that tools like rEFInd and 
gummiboot were better suited to the task of boot managing than the 
kernel itself.

Since then, however, I've received enough emails asking about the status 
of the patch that I think there is room for a simpler entrant into the 
field of linux EFI boot management. This could be in-kernel, as per my 
original patch, or a very simple out-of-kernel (but maybe distributed 
in-tree?) efi image that simply reads linux.conf and boots linux with 
those parameters. My personal bias is still toward handling this 
functionality outside of the kernel itself, but before I started the 
work I thought I'd ask here if anyone thought I should instead bring my 
old patch up-to-date.

Thoughts?

Thanks,
Shea Levy

P.S. I am subscribed to linux-efi but not lkml, please include me in any 
replies to lkml.

Re: Revisiting EFI boot stub config file support

[H. Peter Anvin]

On 01/01/2013 07:55 AM, Shea Levy wrote:
> Hi all,
> 
> In March last year, I submitted a patch to enable the EFI boot stub to
> read its config arguments from a file on the ESP, both to help with
> booting from livecds and to avoid having to use efibootmgr every time
> you change your boottime arguments. I ended up dropping the submission
> because I ended up coming to the conclusion that tools like rEFInd and
> gummiboot were better suited to the task of boot managing than the
> kernel itself.
> 
> Since then, however, I've received enough emails asking about the status
> of the patch that I think there is room for a simpler entrant into the
> field of linux EFI boot management. This could be in-kernel, as per my
> original patch, or a very simple out-of-kernel (but maybe distributed
> in-tree?) efi image that simply reads linux.conf and boots linux with
> those parameters. My personal bias is still toward handling this
> functionality outside of the kernel itself, but before I started the
> work I thought I'd ask here if anyone thought I should instead bring my
> old patch up-to-date.
> 

I am willing to accept this (but I am not the first-line maintainer,
Matt is) in the kernel EFI stub as long as there are no hard-coded
filenames involved, no automatic behavior, and no "scanning".  Those are
showstoppers, as is adding to the boot time for nonusers.

	-hpa

Re: Revisiting EFI boot stub config file support

[H. Peter Anvin]

On 01/01/2013 07:55 AM, Shea Levy wrote:
> Hi all,
> 
> In March last year, I submitted a patch to enable the EFI boot stub to
> read its config arguments from a file on the ESP, both to help with
> booting from livecds and to avoid having to use efibootmgr every time
> you change your boottime arguments. I ended up dropping the submission
> because I ended up coming to the conclusion that tools like rEFInd and
> gummiboot were better suited to the task of boot managing than the
> kernel itself.
> 
> Since then, however, I've received enough emails asking about the status
> of the patch that I think there is room for a simpler entrant into the
> field of linux EFI boot management. This could be in-kernel, as per my
> original patch, or a very simple out-of-kernel (but maybe distributed
> in-tree?) efi image that simply reads linux.conf and boots linux with
> those parameters. My personal bias is still toward handling this
> functionality outside of the kernel itself, but before I started the
> work I thought I'd ask here if anyone thought I should instead bring my
> old patch up-to-date.
> 

I am willing to accept this (but I am not the first-line maintainer,
Matt is) in the kernel EFI stub as long as there are no hard-coded
filenames involved, no automatic behavior, and no "scanning".  Those are
showstoppers, as is adding to the boot time for nonusers.

	-hpa

Re: Revisiting EFI boot stub config file support

[Shea Levy]

On Jan 1, 2013, at 12:24, "H. Peter Anvin" <hpa-YMNOUZJC4hwAvxtiuMwx3w@public.gmane.org> wrote:

> On 01/01/2013 07:55 AM, Shea Levy wrote:
>> Hi all,
>> 
>> In March last year, I submitted a patch to enable the EFI boot stub to
>> read its config arguments from a file on the ESP, both to help with
>> booting from livecds and to avoid having to use efibootmgr every time
>> you change your boottime arguments. I ended up dropping the submission
>> because I ended up coming to the conclusion that tools like rEFInd and
>> gummiboot were better suited to the task of boot managing than the
>> kernel itself.
>> 
>> Since then, however, I've received enough emails asking about the status
>> of the patch that I think there is room for a simpler entrant into the
>> field of linux EFI boot management. This could be in-kernel, as per my
>> original patch, or a very simple out-of-kernel (but maybe distributed
>> in-tree?) efi image that simply reads linux.conf and boots linux with
>> those parameters. My personal bias is still toward handling this
>> functionality outside of the kernel itself, but before I started the
>> work I thought I'd ask here if anyone thought I should instead bring my
>> old patch up-to-date.
> 
> I am willing to accept this (but I am not the first-line maintainer,
> Matt is) in the kernel EFI stub as long as there are no hard-coded
> filenames involved, no automatic behavior, and no "scanning".  Those are
> showstoppers, as is adding to the boot time for nonusers.

Hm, ok. Does the boot stub have access to CONFIG_CMDLINE? Or could another compile-time setting be added so that distros can ship livecds that automatically use a config file found at a hard-coded location?

> 
>    -hpa
> 
> 

Re: Revisiting EFI boot stub config file support

[Shea Levy]

On Jan 1, 2013, at 12:24, "H. Peter Anvin" <hpa@zytor.com> wrote:

> On 01/01/2013 07:55 AM, Shea Levy wrote:
>> Hi all,
>> 
>> In March last year, I submitted a patch to enable the EFI boot stub to
>> read its config arguments from a file on the ESP, both to help with
>> booting from livecds and to avoid having to use efibootmgr every time
>> you change your boottime arguments. I ended up dropping the submission
>> because I ended up coming to the conclusion that tools like rEFInd and
>> gummiboot were better suited to the task of boot managing than the
>> kernel itself.
>> 
>> Since then, however, I've received enough emails asking about the status
>> of the patch that I think there is room for a simpler entrant into the
>> field of linux EFI boot management. This could be in-kernel, as per my
>> original patch, or a very simple out-of-kernel (but maybe distributed
>> in-tree?) efi image that simply reads linux.conf and boots linux with
>> those parameters. My personal bias is still toward handling this
>> functionality outside of the kernel itself, but before I started the
>> work I thought I'd ask here if anyone thought I should instead bring my
>> old patch up-to-date.
> 
> I am willing to accept this (but I am not the first-line maintainer,
> Matt is) in the kernel EFI stub as long as there are no hard-coded
> filenames involved, no automatic behavior, and no "scanning".  Those are
> showstoppers, as is adding to the boot time for nonusers.

Hm, ok. Does the boot stub have access to CONFIG_CMDLINE? Or could another compile-time setting be added so that distros can ship livecds that automatically use a config file found at a hard-coded location?

> 
>    -hpa
> 
> 

Re: Revisiting EFI boot stub config file support

[Matthew Garrett]

On Wed, Jan 02, 2013 at 08:26:29AM -0500, Shea Levy wrote:

> Hm, ok. Does the boot stub have access to CONFIG_CMDLINE? Or could 
> another compile-time setting be added so that distros can ship livecds 
> that automatically use a config file found at a hard-coded location?

Just pass the config file as the first argument?

-- 
Matthew Garrett | mjg59-1xO5oi07KQx4cg9Nei1l7Q@public.gmane.org

Re: Revisiting EFI boot stub config file support

[Matthew Garrett]

On Wed, Jan 02, 2013 at 08:26:29AM -0500, Shea Levy wrote:

> Hm, ok. Does the boot stub have access to CONFIG_CMDLINE? Or could 
> another compile-time setting be added so that distros can ship livecds 
> that automatically use a config file found at a hard-coded location?

Just pass the config file as the first argument?

-- 
Matthew Garrett | mjg59@srcf.ucam.org

Re: Revisiting EFI boot stub config file support

[Shea Levy]

On Jan 2, 2013, at 12:43, Matthew Garrett <mjg59-1xO5oi07KQx4cg9Nei1l7Q@public.gmane.org> wrote:

> On Wed, Jan 02, 2013 at 08:26:29AM -0500, Shea Levy wrote:
> 
>> Hm, ok. Does the boot stub have access to CONFIG_CMDLINE? Or could 
>> another compile-time setting be added so that distros can ship livecds 
>> that automatically use a config file found at a hard-coded location?
> 
> Just pass the config file as the first argument?
> 

But you can't set args when booting from external media (using e.g. \boot\efi\bootx64.efi), at least not on any system I've seen. And even if you could, that'd over complicate the install process for new users.

> -- 
> Matthew Garrett | mjg59-1xO5oi07KQx4cg9Nei1l7Q@public.gmane.org
> --
> To unsubscribe from this list: send the line "unsubscribe linux-efi" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Revisiting EFI boot stub config file support

[Shea Levy]

On Jan 2, 2013, at 12:43, Matthew Garrett <mjg59@srcf.ucam.org> wrote:

> On Wed, Jan 02, 2013 at 08:26:29AM -0500, Shea Levy wrote:
> 
>> Hm, ok. Does the boot stub have access to CONFIG_CMDLINE? Or could 
>> another compile-time setting be added so that distros can ship livecds 
>> that automatically use a config file found at a hard-coded location?
> 
> Just pass the config file as the first argument?
> 

But you can't set args when booting from external media (using e.g. \boot\efi\bootx64.efi), at least not on any system I've seen. And even if you could, that'd over complicate the install process for new users.

> -- 
> Matthew Garrett | mjg59@srcf.ucam.org
> --
> To unsubscribe from this list: send the line "unsubscribe linux-efi" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Revisiting EFI boot stub config file support

[Matthew Garrett]

On Wed, Jan 02, 2013 at 12:52:21PM -0500, Shea Levy wrote:

> But you can't set args when booting from external media (using e.g. 
> \boot\efi\bootx64.efi), at least not on any system I've seen. And even 
> if you could, that'd over complicate the install process for new 
> users.

So have an initial wrapper that launches the kernel and passes an 
argument. You're going to need something equivalent if you want to 
support Secure Boot.

-- 
Matthew Garrett | mjg59-1xO5oi07KQx4cg9Nei1l7Q@public.gmane.org

Re: Revisiting EFI boot stub config file support

[Matthew Garrett]

On Wed, Jan 02, 2013 at 12:52:21PM -0500, Shea Levy wrote:

> But you can't set args when booting from external media (using e.g. 
> \boot\efi\bootx64.efi), at least not on any system I've seen. And even 
> if you could, that'd over complicate the install process for new 
> users.

So have an initial wrapper that launches the kernel and passes an 
argument. You're going to need something equivalent if you want to 
support Secure Boot.

-- 
Matthew Garrett | mjg59@srcf.ucam.org

Re: Revisiting EFI boot stub config file support

[H. Peter Anvin]

On 01/02/2013 09:43 AM, Matthew Garrett wrote:
> On Wed, Jan 02, 2013 at 08:26:29AM -0500, Shea Levy wrote:
> 
>> Hm, ok. Does the boot stub have access to CONFIG_CMDLINE? Or could 
>> another compile-time setting be added so that distros can ship livecds 
>> that automatically use a config file found at a hard-coded location?
> 
> Just pass the config file as the first argument?
> 

CONFIG_CMDLINE can certainly become available to the stub... it's a
compile-time constant.

	-hpa