buggy BIOSes (Re: pretty colors in gfxterm)

buggy BIOSes (Re: pretty colors in gfxterm)

[Robert Millan]

> Besides that, it is really necessary to check what workarounds for buggy 
> BIOSes are missing in GRUB 2. The current bootstrap code on PC is based on 
> somewhere around GRUB 0.96, so it is a bit outdated.

I just had a look.  Is there something other than:

2005-03-15  Yoshinori K. Okuji  <okuji@enbug.org>

	* stage2/fsys_fat.c (fat_mount): Ignore the 3rd bit of a media
	descriptor, because some BIOSes overwrite this value, according
	to the storage mode (e.g. USB Floppy or USB HDD).

[...]

2005-02-15  Yoshinori K. Okuji  <okuji@enbug.org>

	* stage2/builtins.c (install_func): If DEST_DRIVE is a hard
	disk, enable the workaround in Stage 1 by replacing the jmp
	with double nop's.

	* stage1/stage1.h (STAGE1_BOOT_DRIVE_CHECK): New macro.
	(STAGE1_BOOT_DRIVE_MASK): Removed.

	* stage1/stage1.S (boot_drive_check): New label. This implements
	a different workaround for buggy BIOSes which don't pass boot
	drive correctly. This is effective for BIOSes which pass a value
	without the seventh bit (0x80).
	(boot_drive_mask): Removed.

-- 
Robert Millan

<GPLv2> I know my rights; I want my phone call!
<DRM> What use is a phone call, if you are unable to speak?
(as seen on /.)

Re: buggy BIOSes (Re: pretty colors in gfxterm)

[Yoshinori K. Okuji]

On Friday 31 August 2007 16:47, Robert Millan wrote:
> > Besides that, it is really necessary to check what workarounds for buggy
> > BIOSes are missing in GRUB 2. The current bootstrap code on PC is based
> > on somewhere around GRUB 0.96, so it is a bit outdated.
>
> I just had a look.  Is there something other than:

I also looked at the ChangeLog, and found nothing else. So can we assume that 
all workarounds have been applied to GRUB 2 as well?

Thanks,
Okuji

Re: buggy BIOSes (Re: pretty colors in gfxterm)

[Robert Millan]

On Sun, Sep 02, 2007 at 12:02:10PM +0200, Yoshinori K. Okuji wrote:
> On Friday 31 August 2007 16:47, Robert Millan wrote:
> > > Besides that, it is really necessary to check what workarounds for buggy
> > > BIOSes are missing in GRUB 2. The current bootstrap code on PC is based
> > > on somewhere around GRUB 0.96, so it is a bit outdated.
> >
> > I just had a look.  Is there something other than:
> 
> I also looked at the ChangeLog, and found nothing else. So can we assume that 
> all workarounds have been applied to GRUB 2 as well?

In GRUB 2 we have:

2005-03-18  Yoshinori K. Okuji  <okuji@enbug.org>

	* fs/fat.c (grub_fat_mount): Ignore the 3rd bit of a media
	descriptor. This is ported from GRUB Legacy.

but I couldn't find this one from GRUB Legacy:

2005-02-15  Yoshinori K. Okuji  <okuji@enbug.org>

	* stage2/builtins.c (install_func): If DEST_DRIVE is a hard
	disk, enable the workaround in Stage 1 by replacing the jmp
	with double nop's.

	* stage1/stage1.h (STAGE1_BOOT_DRIVE_CHECK): New macro.
	(STAGE1_BOOT_DRIVE_MASK): Removed.

	* stage1/stage1.S (boot_drive_check): New label. This implements
	a different workaround for buggy BIOSes which don't pass boot
	drive correctly. This is effective for BIOSes which pass a value
	without the seventh bit (0x80).
	(boot_drive_mask): Removed.

As for other changes, I can't find them.  Did you have anything else in mind
when you mentioned these workarounds?

-- 
Robert Millan

<GPLv2> I know my rights; I want my phone call!
<DRM> What use is a phone call, if you are unable to speak?
(as seen on /.)

Re: buggy BIOSes (Re: pretty colors in gfxterm)

[Yoshinori K. Okuji]

On Sunday 02 September 2007 12:52, Robert Millan wrote:
> but I couldn't find this one from GRUB Legacy:
>
> 2005-02-15  Yoshinori K. Okuji  <okuji@enbug.org>
>
> 	* stage2/builtins.c (install_func): If DEST_DRIVE is a hard
> 	disk, enable the workaround in Stage 1 by replacing the jmp
> 	with double nop's.
>
> 	* stage1/stage1.h (STAGE1_BOOT_DRIVE_CHECK): New macro.
> 	(STAGE1_BOOT_DRIVE_MASK): Removed.
>
> 	* stage1/stage1.S (boot_drive_check): New label. This implements
> 	a different workaround for buggy BIOSes which don't pass boot
> 	drive correctly. This is effective for BIOSes which pass a value
> 	without the seventh bit (0x80).
> 	(boot_drive_mask): Removed.

This has already been included, as far as I see the code.

> As for other changes, I can't find them.  Did you have anything else in
> mind when you mentioned these workarounds?

No. The code in GRUB 2 looks more up-to-date than I thought. :)

Okuji