On 27 May 2011 15:08, Stefano Ceccherini <stefano.ceccherini@xxxxxxxxx> wrote: > Il giorno 27/mag/2011 19.26, "Adrien Destugues" > <pulkomandy@xxxxxxxxxxxxxxxxx> ha scritto: >> Can someone have a look at this new bootloader code in ticket #3441 ? it >> seems it helps with quite some computers where the current one doesn't work. >> Maybe we should put it in alpha3, or at least provide it as an alternative >> in the release notes in some way ? >> Not sure I can read x86 assembly well enough... > > I think it shouldn't go in the alpha since it's completely new code, and in > such a critical component like the boot loader. If it really helps on so > many computers it should go in the trunk. I updated the existing boot code (the one on src/add-ons/kernel/partitioning_systems/intel/PartitionMapWriter.cpp) with the suggestions from Starsseed. According to him, the new code works on a machine which wouldn't boot before, and he'll test it on more machines next week. Actually, src/bin/writembr/mbr.S could be entirely replaced by this, so we don't have redundant MBR codes on the tree. mbr.S is the one used by the anyboot image, taken straight from FreeBSD. What I originally did was take this MBR and add GPT support in. Now I fixed some (of my original) bugs and incorporated those suggestions that enabled more machines to boot. Since it's mostly tried and tested code, it might lower the barrier to adopt on trunk/alpha3. It's also more extensively commented, so more eyes could check for errors, even those not so versed in x86 assembly (like myself: I made most of those comments for myself while studying the original FreeBSD code). Code on http://dev.haiku-os.org/ticket/4028 Cheers, A.