Ronald G Minnich <[EMAIL PROTECTED]> writes: > Oh boy. I am looking at this: > void jmp_to_elf_entry(void *entry, unsigned long buffer) > > it verges very quickly to assembly code. So for now we've lost the > machine-independent bits of elf booting, I guess. This goes on the to-do > list.
????? The machine dependent bits? On the coding side, if you don't worry about the case where you can load over LinuxBIOS the assembly code is just a jmp instruction. > I want to move this stuff out to an arch-specific place soon. Any > suggestions? jmp_to_elf_entry is in src/arch/i386/boot. Making hardware main less arch specific should be possible. > This is kind of OK for me for now, as I am still using linuxbiosmain. I > also note that my roadrunner2 ELF-based bios no longer builds: > /users/rminnich/src/bios/freebios/src/arch/i386/lib/hardwaremain.c: In > function `hardwaremain': > /users/rminnich/src/bios/freebios/src/arch/i386/lib/hardwaremain.c:401: > warning: implicit declaration of function `elfboot' > /users/rminnich/src/bios/freebios/src/arch/i386/lib/hardwaremain.c:401: > `streams' undeclared (first use in this function) Hmm. The includes at the top of hardware main, should handle that (i.e.): #if USE_ELF_BOOT #include <rom/read_bytes.h> #include <boot/elf.h> #endif > > Also mkelfImage no longer works for our stuff, I'm still tracking that one > down. What do you mean by does not work? The code has not changed in substantial way for a very long time. It doesn't wokr on the alpha but then it never did. > So, there is work to do for the latest set of patches. > > I'm still waiting for some of you to hold up your hands and tell me which > mainboard you're going to do testing on :-) I pretty much have the supermicro p4dc6, p4dpr, p4dpe, and the tyan s2462. Eric
