2012-05-31 15:41:20

by Matt Fleming

[permalink] [raw]
Subject: Re: [RFC][PATCH 1/1] x86, efi: EFI boot stub config file support

On Tue, 2012-03-20 at 18:22 -0400, Shea Levy wrote:
> On 03/18/2012 11:49 AM, Shea Levy wrote:
> > On 3/18/12 11:45 AM, Shea Levy wrote:
> >> Hello,
> >>
> >> Inlined is a patch to have the kernel read its parameters from a
> >> config file when booted via the EFI boot stub without any parameters
> >> passed. This patch roughly follows the plan I discussed when
> >> proposing this idea, except now the file is found relative to the
> >> location of the bzImage and there is no additional check to see if
> >> image->parent_handle is NULL (despite the spec saying it will be NULL
> >> when the image was loaded by the firmware itself, it was never NULL
> >> on any of the systems I tested).
> >>
> >> Cheers,
> >> Shea Levy
>
> I haven't yet gotten a reply to this, should I wait until after the
> merge window and resubmit it then?

I'm wondering what use case you're catering for here that can't be
handled with efibootmgr? What problem does this patch fix?


2012-06-02 18:32:07

by Shea Levy

[permalink] [raw]
Subject: Re: [RFC][PATCH 1/1] x86, efi: EFI boot stub config file support


On May 31, 2012, at 11:41, Matt Fleming <[email protected]> wrote:

> On Tue, 2012-03-20 at 18:22 -0400, Shea Levy wrote:
>> On 03/18/2012 11:49 AM, Shea Levy wrote:
>>> On 3/18/12 11:45 AM, Shea Levy wrote:
>>>> Hello,
>>>>
>>>> Inlined is a patch to have the kernel read its parameters from a
>>>> config file when booted via the EFI boot stub without any parameters
>>>> passed. This patch roughly follows the plan I discussed when
>>>> proposing this idea, except now the file is found relative to the
>>>> location of the bzImage and there is no additional check to see if
>>>> image->parent_handle is NULL (despite the spec saying it will be NULL
>>>> when the image was loaded by the firmware itself, it was never NULL
>>>> on any of the systems I tested).
>>>>
>>>> Cheers,
>>>> Shea Levy
>>
>> I haven't yet gotten a reply to this, should I wait until after the
>> merge window and resubmit it then?
>
> I'm wondering what use case you're catering for here that can't be
> handled with efibootmgr? What problem does this patch fix?
>

This fixes cases where efibootmgr doesn't work, either by design (in the case of removable boot media like a livecd) or due to firmware non-conformity (like VirtualBox's EFI emulation).

I had been waiting for the efi_printk stuff to get merged (for some reason I was under the impression it would be in 3.4) before submitting an updated version of this, and in the mean-time have been using the EDK shell + a startup.nsh file for the livecd case.-