| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
|
|
|
|
|
|
|
| |
Occam's razor shows that copying the entire ports add-on over the libc
sources is a bad idea compared to simply adding a link to the add-on in
the libc source directory.
See also: <http://www.eglibc.org/archives/patches/msg01137.html>.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Building EGLIBC with the static GCC would result in an error such as
this:
/tmp/[...].s: Assembler messages:
/tmp/[...].s:168: Error: invalid swi expression
/tmp/[...].s:168: Error: value of 983045 too large for field of 2 byt
es at 170
Apparently, it makes no sense for GCC to generate Thumb code with the
ARM Architecture Procedure Call Standard (AAPCS).
|
| |
|
|
EGLIBC fails to build:
../sysdeps/unix/syscall.S: Assembler messages:
../sysdeps/unix/syscall.S:28: Error: cannot represent SWI relocation in this object file format
make[2]: *** [/home/pj/bbl/sys.cortexa8/fsmnt/src/eglibc-build/misc/syscall.o] Error 1
make[2]: Leaving directory `/home/pj/bbl/sys.cortexa8/fsmnt/src/eglibc-2.15/libc/misc'
make[1]: *** [misc/subdir_lib] Error 2
make[1]: Leaving directory `/home/pj/bbl/sys.cortexa8/fsmnt/src/eglibc-2.15/libc'
make: *** [all] Error 2
Apparently SWI (software interrupt) vector table relocation is only
available in certain ARM cores (including Cortex-A8).
|