On Nov 9, [EMAIL PROTECTED] wrote: > > In putting in some extra consistency checks for the GRUB shell > installer (and for Linux adding BMAP functionality for the stage2), > I seem to have come across an issue... > > The "setup" builtin function will try to install a stage 1.5, but > then overwrite it immediately with a stage2.
First setup doesn't access the drive directly but just calls embed and install which do the real work. The setup command checks if a stage1.5 is there and if it can be embedded in a safe place. If that succeeds it calls install with all required parameters. Note that the install command with an embedded stage1_5 takes several parameters: STAGE2 is actually the location where stage1_5 was embedded, CONFIG_FILE is the location of stage2 (the logic is that stage2 is the config file for stage1_5). REAL_CONFIG_FILE is the location of menu.lst. The naming of the parameters is a bit misleading. The "install" command doesn't generate a sector bitmap for stage2, the CONFIG_FILE parameter. But it needs to write the string CONFIG_FILE and REAL_CONFIG_FILE (this aren't block bitmaps, but plain file names) to the embedded stage1_5 resp. the stage2. The stage2 is on the filesystem, therefore there is the --stage2=... command to make that write more safe. Jochen _______________________________________________ Bug-grub mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/bug-grub
