u-boot.git/arch/x86, branch v2011.12

x86: Wrap small helper functions from libgcc to avoid an ABI mismatch

2011-11-29T10:31:24Z

When gcc compiles some 64 bit operations on a 32 bit machine, it generates calls to small functions instead of instructions which do the job directly. Those functions are defined in libgcc and transparently provide whatever functionality was necessary. Unfortunately, u-boot can be built with a non-standard ABI when libgcc isn't. More specifically, u-boot uses -mregparm. When the u-boot and libgcc are linked together, very confusing bugs can crop up, for instance seemingly normal integer division or modulus getting the wrong answer or even raising a spurious divide by zero exception. This change borrows (steals) a technique and some code from coreboot which solves this problem by creating wrappers which translate the calling convention when calling the functions in libgcc. Unfortunately that means that these instructions which had already been turned into functions have even more overhead, but more importantly it makes them work properly. To find all of the functions that needed wrapping, u-boot was compiled without linking in libgcc. All the symbols the linker complained were undefined were presumed to be the symbols that are needed from libgcc. These were a subset of the symbols covered by the coreboot code, so it was used unmodified. To prevent symbols which are provided by libgcc but not currently wrapped (or even known about) from being silently linked against by code generated by libgcc, a new copy of libgcc is created where all the symbols are prefixed with __normal_. Without being purposefully wrapped, these symbols will cause linker errors instead of silently introducing very subtle, confusing bugs. Another approach would be to whitelist symbols from libgcc and strip out all the others. The problem with this approach is that it requires the white listed symbols to be specified three times, once for objcopy, once so the linker inserts the wrapped, and once to generate the wrapper itself, while this implementation needs it to be listed only twice. There isn't much tangible difference in what each approach produces, so this one was preferred. Signed-off-by: Gabe Black

x86: Import the glibc memset implementation

2011-11-29T10:31:17Z

The new implementation is about twice as fast as the old. This is from glibc-2.14, sysdeps/i386/memset.c. Signed-off-by: Gabe Black

x86: Fix a few recently added bugs

2011-11-29T10:30:51Z

Signed-off-by: Gabe Black Signed-off-by: Graeme Russ

x86: Don't relocate symbols which point to things that aren't relocated

2011-11-29T10:30:44Z

This change adds an upper bound for symbols which are fixed up after u-boot is relocated into RAM. This way portions that are left at their original location can be referred to without having to manually fix up any pointers. Signed-off-by: Gabe Black

x86: Fix how the location of the realmode and bios blobs are calculated

2011-11-29T10:30:38Z

There are two blobs embedded into the u-boot image which are linked to run at an address which is different from where they actually end up in the ROM, one called "realmode" and one called "bios". There are realmode_setup and bios_setup functions which prepare those blobs by copying them into the location they're supposed to run from, among other things. During u-boot relocation from ROM to RAM, the text and a few data segments are copied over. The realmode and bios sections are not copied, and so the only place they can be read from is their original location in the ROM. Looking specifically at the bios blob, there are symbols defined in the linker script called __bios_start and __bios_size which are defined to be the start and size of the blob in the ROM. In the bios_setup function, there seem to be two mistakes happening. First, the offset from ROM to RAM is being added to __bios_start which implies that this code expects to use the copy moved to RAM. No such copy is made, so that's wrong. More subtly, when u-boot relocates itself, it goes through all of the relocations stored in .rel.dyn and fixes them up. This has the effect of transforming the __bios_start reference in bios_setup so that it refers to the version in RAM (if one existed) instead of the one in ROM. To correct for that, the offset actually needs to be subtracted out again to translate the address back into the ROM. The net effect is that for both blobs, a + needs to be changed to a -. Signed-off-by: Gabe Black

x86: Misc cleanups

2011-11-29T10:10:12Z

Signed-off-by: Graeme Russ

x86: Misc PCI touchups

2011-11-29T10:10:08Z

Signed-off-by: Graeme Russ

x86: Ensure IDT and GDT remain 16-byte aligned post relocation

2011-11-29T10:10:01Z

Some CPUs have strict alignment requirements for these tables Signed-off-by: Graeme Russ

x86: Provide more configuration granularity

2011-11-29T10:09:52Z

Planned future ports requires more granularity for some options Signed-off-by: Graeme Russ

x86: Add multiboot header

2011-11-29T10:08:02Z

By adding a multiboot header, U-Boot can be loaded by GRUB2. Using GRUB2 to bootstrap U-Boot is useful for using an existing BIOS to get an initial U-Boot port up and running before implementing the low-level reset vector code, SDRAM init, etc. and overwriting the BIOS Signed-off-by: Graeme Russ