[PATCH] x86: Switch to vga console only if framebuffer details are missing

[PATCH] x86: Switch to vga console only if framebuffer details are missing

[Matt Fleming]

From: Matt Fleming <matt.fleming@intel.com>

The efi_enabled variable has come to mean "Do we have EFI runtime
services available?". However, lack of EFI runtime services does not
mean that we should switch to using the VGA console. Provided that the
boot loader passed the dimensions of the EFI framebuffer there is no
reason we can't use efifb.

There's also no reason to check the memory type of 0xa0000 - whether
or not that memory region is EFI_CONVENTIONAL_MEMORY is immaterial -
the EFI framebuffer device will still work, and checking the EFI
memory type of a memory region on a non-EFI machine is illogical.

Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Matthew Garrett <mjg@redhat.com>
Cc: Olof Johansson <olof@lixom.net>
Cc: Peter Jones <pjones@redhat.com>
Cc: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
---
 arch/x86/kernel/setup.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index f4b9b80..29dc30a 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -1021,7 +1021,7 @@ void __init setup_arch(char **cmdline_p)
 
 #ifdef CONFIG_VT
 #if defined(CONFIG_VGA_CONSOLE)
-	if (!efi_enabled || (efi_mem_type(0xa0000) != EFI_CONVENTIONAL_MEMORY))
+	if (!screen_info.lfb_base || !screen_info.lfb_size)
 		conswitchp = &vga_con;
 #elif defined(CONFIG_DUMMY_CONSOLE)
 	conswitchp = &dummy_con;
-- 
1.7.4.4

Re: [PATCH] x86: Switch to vga console only if framebuffer details are missing

[Alan Cox]

On Mon, 17 Sep 2012 11:29:08 +0100
Matt Fleming <matt@console-pimps.org> wrote:

> From: Matt Fleming <matt.fleming@intel.com>
> 
> The efi_enabled variable has come to mean "Do we have EFI runtime
> services available?". However, lack of EFI runtime services does not
> mean that we should switch to using the VGA console. Provided that the
> boot loader passed the dimensions of the EFI framebuffer there is no
> reason we can't use efifb.

Speed is the obvious one. vga con is fast, efifb is not

Re: [PATCH] x86: Switch to vga console only if framebuffer details are missing

[Matt Fleming]

On Mon, 2012-09-17 at 12:00 +0100, Alan Cox wrote:
> On Mon, 17 Sep 2012 11:29:08 +0100
> Matt Fleming <matt@console-pimps.org> wrote:
> 
> > From: Matt Fleming <matt.fleming@intel.com>
> > 
> > The efi_enabled variable has come to mean "Do we have EFI runtime
> > services available?". However, lack of EFI runtime services does not
> > mean that we should switch to using the VGA console. Provided that the
> > boot loader passed the dimensions of the EFI framebuffer there is no
> > reason we can't use efifb.
> 
> Speed is the obvious one. vga con is fast, efifb is not

Right, and that's an entirely valid reason for wanting vga con over
efifb, but in that case users should just disable the efifb driver.

-- 
Matt Fleming, Intel Open Source Technology Center

Re: [PATCH] x86: Switch to vga console only if framebuffer details are missing

[Matt Fleming]

On Mon, 2012-09-17 at 11:29 +0100, Matt Fleming wrote:
> From: Matt Fleming <matt.fleming@intel.com>
> 
> The efi_enabled variable has come to mean "Do we have EFI runtime
> services available?". However, lack of EFI runtime services does not
> mean that we should switch to using the VGA console. Provided that the
> boot loader passed the dimensions of the EFI framebuffer there is no
> reason we can't use efifb.
> 
> There's also no reason to check the memory type of 0xa0000 - whether
> or not that memory region is EFI_CONVENTIONAL_MEMORY is immaterial -
> the EFI framebuffer device will still work, and checking the EFI
> memory type of a memory region on a non-EFI machine is illogical.
> 
> Cc: H. Peter Anvin <hpa@zytor.com>
> Cc: Matthew Garrett <mjg@redhat.com>
> Cc: Olof Johansson <olof@lixom.net>
> Cc: Peter Jones <pjones@redhat.com>
> Cc: Ingo Molnar <mingo@kernel.org>
> Signed-off-by: Matt Fleming <matt.fleming@intel.com>
> ---
>  arch/x86/kernel/setup.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)

Ping? Anybody got an opinion on this one?

Re: [PATCH] x86: Switch to vga console only if framebuffer details are missing

[Peter Jones]

On Wed, Oct 10, 2012 at 02:43:07PM +0100, Matt Fleming wrote:
> On Mon, 2012-09-17 at 11:29 +0100, Matt Fleming wrote:
> > From: Matt Fleming <matt.fleming@intel.com>
> > 
> > The efi_enabled variable has come to mean "Do we have EFI runtime
> > services available?". However, lack of EFI runtime services does not
> > mean that we should switch to using the VGA console. Provided that the
> > boot loader passed the dimensions of the EFI framebuffer there is no
> > reason we can't use efifb.
> > 
> > There's also no reason to check the memory type of 0xa0000 - whether
> > or not that memory region is EFI_CONVENTIONAL_MEMORY is immaterial -
> > the EFI framebuffer device will still work, and checking the EFI
> > memory type of a memory region on a non-EFI machine is illogical.
> > 
> > Cc: H. Peter Anvin <hpa@zytor.com>
> > Cc: Matthew Garrett <mjg@redhat.com>
> > Cc: Olof Johansson <olof@lixom.net>
> > Cc: Peter Jones <pjones@redhat.com>
> > Cc: Ingo Molnar <mingo@kernel.org>
> > Signed-off-by: Matt Fleming <matt.fleming@intel.com>
> > ---
> >  arch/x86/kernel/setup.c |    2 +-
> >  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> Ping? Anybody got an opinion on this one?

Provided "video=vga" still works to circumvent Alan's objection, I'm all
for it.

-- 
        Peter