[syslinux] isohybrid boot from logical partition

[Duncan Elliot]

Hi All,

I went down the path of testing SYSLINUX with chain.c32 (instead of
GRUB) and have had some good success.

Initial tests with various ISO's written to logical partition 5 and 6
have worked!

I've only tested with a few ISO's (and only in qemu so far), but early
results look promising.

I'll update with any other progress.

Thanks again for all the help
Duncan

On 30 December 2016 at 19:34, Thomas Schmitt <scdbackup at gmx.net> wrote:
> Hi,
>
> Duncan Elliot wrote:
>> Taking the point that "Syslinux is partition-table-format agnostic"
>
> I don't agree with hpa here. SYSLINUX expects MBR partition table entries
> with absolute addresses. Those are exactly the primary partition entries.
> That's not what i would call agnostic.
>
> But i agree with his assessment that it is very cumbersome to find the
> own partition start if the "DS:SI" protocol hands over a raw partition
> table entry that describes a logical partition.
>
>
>> I'm guessing that any "fix" that should be made, should be on the GRUB
>> side, rather than in the isohybrid side?
>
> One could ask at help-grub at gnu.org whether there is already a feature to
> change the forwarded partition table entry from relative LBA to absolute
> LBA.
> The necessary information is supposed to be available when GRUB prepares
> the chainloading.
>
> Developement of such a feature would need some initial effort.
> Besides learning the assembler language one would have to find and
> understand GRUB's chainloading code and the place where "DS:SI" is
> prepared.
>
>
> Didier Spaier wrote:
>> > Sorry if this is a silly question: why not use a GUID partition
>> > table instead of using logical partitions?
>
> Duncan Elliot wrote:
>> Not a silly question, I had the same idea and had tried it out but it
>> didn't work (same error msg).
>
> This would be a SYSLINUX bug. After all the partok isohybrid MBR does have
> code for reading the start LBA from a GPT entry.
>
> I am wondering about the meaning or reason of "+20" in this assembler line:
>
>   movl    (32+20)(%si), %ecx
>
> Why that extra offset 20 on the byte offset 32 which is specified by
> GPT specs (UEFI 2.4, 5.3.3) ? (Do i get the language wrong ?)
>
> It would be interesting to see the complete content of %si as perceived
> by the isohybrid MBR. (At least up to its byte 59.)
>
> (Beginning to read about the language of "gas" i am oscillating between
> nostalgy and shudders. Three cheers for the C language !)
>
>
> Have a nice day :)
>
> Thomas
>