Syslinux - Aug 2019 - Getting classic BIOS and UEFI boot on the same disk?

On Fri, Aug 30, 2019 at 12:15 PM Erik Rull <erik.rull at rdsoftware.de>
wrote:
> Hello Greg,
>
> thanks a lot - I will try out your script next week.
> I jut went through the script - there is a line containing:
> bls1 include
>
> I would carefully assume that this line is only valid when patching
> syslinux,
> correct? What will happen if this is read by a non-patched syslinux? Will
> it
> cause errors or will it simply be ignored?
>
>
That is correct. "bls1 include" would cause errors without the patch.
The
main parts of the script that would be relevant to you would be the parts
that create the partition, set its type codes, write the first-stage boot
loader to the first 440 bytes of the disk, and lay out the file system
(e.g. it demonstrates that a single /syslinux.cfg file can be shared
between both the UEFI and BIOS binaries). Much of the rest of it would be
irrelevant for your case.

Gregory Bartholomew wrote:
> That is correct. "bls1 include" would cause errors without the
patch. The main
> parts of the script that would be relevant to you would be the parts that
create
> the partition, set its type codes, write the first-stage boot loader to the
> first 440 bytes of the disk, and lay out the file system (e.g. it
demonstrates
> that a single /syslinux.cfg file can be shared between both the UEFI and
BIOS
> binaries). Much of the rest of it would be irrelevant for your case.
Sounds good. What I did not find - maybe I'm not too deep in this topic -
will
the classic BIOS boot find a proper classic partition table? Or will it find
only the GPT? The oldest system we have in operation has a AMI BIOS dated 2004 -
will it be able to handle the GPT or do I need further modifications?
(I'll have the answer at last next week after trying it out :-))

Best regards,

Erik

I don't think the classic BIOS systems read the partition table at all.
Rather, they just load the first 440 bytes from the drive into memory and
copy the memory address that it was loaded to a CPU register. From there,
it is up to that 440 byte program to decipher the partition table. If you
use syslinux's gptmbr.bin, you should be fine.

On Sat, Aug 31, 2019 at 4:47 AM Erik Rull <erik.rull at rdsoftware.de>
wrote:
> Gregory Bartholomew wrote:
> > That is correct. "bls1 include" would cause errors without
the patch.
> The main
> > parts of the script that would be relevant to you would be the parts
> that create
> > the partition, set its type codes, write the first-stage boot loader
to
> the
> > first 440 bytes of the disk, and lay out the file system (e.g. it
> demonstrates
> > that a single /syslinux.cfg file can be shared between both the UEFI
and
> BIOS
> > binaries). Much of the rest of it would be irrelevant for your case.
>
> Sounds good. What I did not find - maybe I'm not too deep in this topic
-
> will
> the classic BIOS boot find a proper classic partition table? Or will it
> find
> only the GPT? The oldest system we have in operation has a AMI BIOS dated
> 2004 -
> will it be able to handle the GPT or do I need further modifications?
> (I'll have the answer at last next week after trying it out :-))
>
> Best regards,
>
> Erik
>