Gene Cumm
2010-Dec-18 14:15 UTC
[syslinux] diskstart.inc: PartInfo sub-constants seen as local labels
I'm working on a debugging/diagnostic image in order to debug/diagnose an issue with very recent machines using a UFD (USB Flash Drive) which really _should_ be partitioned but someone decides to be a little too quick and uses it raw. (FAT* file system on /dev/sdb rather than /dev/sdb1). Anyways, I've got a successful first iteration that'll say what the BIOS thinks about geometry and fetches key sectors of the image to visually verify this info. Now, I'm thinking that if it doesn't smell/taste the same as SYSLINUX on FAT*, it's not a valid test as BIOSs commonly use the first sector to guess the intended geometry. I'm copying (for now) the boot sector code from diskstart.inc. Now, nasm is acting like the PartInfo sub-constants (for lack of better/proper terminology) are local labels. I'm guessing it's either bad syntax in diskstart.inc or a parser-related bug in nasm. I'm working around this (for now) by having a label just before these constants. Thoughts? -- -Gene
Gene Cumm
2010-Dec-18 17:29 UTC
[syslinux] diskstart.inc: PartInfo sub-constants seen as local labels
On Sat, Dec 18, 2010 at 09:15, Gene Cumm <gene.cumm at gmail.com> wrote:> I'm working on a debugging/diagnostic image in order to debug/diagnose > an issue with very recent machines using a UFD (USB Flash Drive) which > really _should_ be partitioned but someone decides to be a little too > quick and uses it raw. (FAT* file system on /dev/sdb rather than > /dev/sdb1). > > Anyways, I've got a successful first iteration that'll say what the > BIOS thinks about geometry and fetches key sectors of the image to > visually verify this info. ?Now, I'm thinking that if it doesn't > smell/taste the same as SYSLINUX on FAT*, it's not a valid test as > BIOSs commonly use the first sector to guess the intended geometry. > I'm copying (for now) the boot sector code from diskstart.inc. > > Now, nasm is acting like the PartInfo sub-constants (for lack of > better/proper terminology) are local labels. ?I'm guessing it's either > bad syntax in diskstart.inc or a parser-related bug in nasm. ?I'm > working around this (for now) by having a label just before these > constants. > > Thoughts?I forgot to mention that my build system is Ubuntu 10.04 with NASM from Ubuntu (2.07) and that I tried building NASM 2.09.04 and 2.10rc3 with the same results.