On 18/11/2026 19:47, Mike Kelly wrote:
This is part of the code required to add the missing graphical console
output when booting with UEFI. The scrolling performance is visibly
very poor but will improve considerably once I have completed support
for write combining (WC). This code shouldn't require much change for
WC and I think it might be beneficial for Hurd users to make use of it
earlier to test a UEFI boot on their machines. Also since this is a
substantial change, although to code not in active use, it might take
some time to agree to integration. Almost all of the patch is either
new code or changes to unused code so there didn't seem much value in
breaking it into a series of small patches.

There are a few related points that I didn't want to be lost amongst the patch description:

1) The frame buffer code does not make use of the colour information specified by multiboot. It assumes equal sized bit fields for red, green and blue, using 0xaa for each. In the event that this is not true, some colour other than white would be used for the text output. I can fix this if necessary.

2) There is no locking made within the immc.c or fb_disp.c code. This is fine for the immediate console since locking is not necessary at this stage of the boot. Locking will perhaps be necessary within fb_disp.c once multiple CPUs are active. I can't see any locking associated with cnputc anywhere, so printfs might be interleaved?

3) I've made the assumption that the immediate console is not useful on XEN. There is no frame buffer available on Xen, presumably?

4) immc.h is in the git sources but not in the gnumach-xxx.orig.tar.xz. I've added immc.h to the Makefrag.am files, is this sufficient?

5) The multiboot2 header is included in a hurd-i386 kernel build. A multiboot2 boot sets up a frame buffer which is then used by gnumach. The patch disables this configuration on i386 in the case where the immediate console is requested since, I thought at the time, that the latter cannot be used with the frame buffer. I realise now that perhaps it could be used since the frame buffer would presumably have a 32 bit physical address that is identity mapped when using a 32 bit kernel.

Anyway, there is no particular need to use a frame buffer on hurd-i386 since the BIOS EGA display will always be available. Actually, do we want multiboot2 in hurd-i386 at all?

Cheers,

Mike.



Reply via email to