Well, there are some issues about adding code to syslinux. Mainly there
are no space to add code. I propose to move data to 0000:F000 (now in
0000:5000) and font data to 0000:5000 (now at 0000:E000). So we have
additional 4K to code. I don't understand however why we have to retain
font data... Does EGA card read from system memory all the time?? I
don't think so... Perhaps it's better to read font data at another
location (trackbuf??) and remove this part of data so to gain 8K of
code. Another option would be to compress syslinux (using LZ or other
simple algorithm) in order to reduce disk usage. I would like to add
support for:
- disk > 8GB (this MUST fit in first 512 bytes so I have no size
problem...)
- span disks
I already post a patch for having initrd in a separate disk however I
got no feedback at all.. Usually I use CVS but syslinux seem to have no
such repository...
About span disks. As suggested I would like to use n-m:file syntax
however I don't know the right way to detect disk change. In my patch I
try to read first sector than I read again informations and fat however
if file have same name (initrd.img on first disk and initrd.img on
second) how to detect disk change? Use disk labels or add a empty file
(like disk5.id or disk.5 or whatever) ??
freddy77