bug reports on glibc or gcc
I've got some problems installing (compiling) gcc and glibc. (Cross-compiling from sparc-*-linux-gnu to sparc64-*-linux-gnu).
Please give me some exact pointers where to find these problems and their answers. (or where to submit new ones).
Error: sed -e : options requires an argument.
This was to a fault in the automake? script. The generated code should be " sed -e $t ". I could manage a workaround.
Error: memset.S: sparcv9 not compatible with sparcv9a or sparcv9b.
This as a fault to a script error (again). It should pass the CFLAGS to the compiler which it fails to do this time. Simple workaround was to set the the flags within the CC variable.
And my latest error (yet unsolved... :-((( ):
*** No rule to make target `/mnt/tools/glibc-build/misc/setfsuid.o', needed by `/mnt/tools/glibc-build/misc/stamp.o'. Stop.
Of course I spit the internet looking for some information, but most of the time I fail. I did get the solution to the "sparcv9" on the internet, but it was not fixed in my download of glibc from gnu. Neither do I get any result using databases of the bug-glibc mailing lists. Where to find the solution?
Please tell me, or should I just use an intel with the lack of a netboot in the bios? :-p
PS Why does gcc-3.4.3 not have a proper manual for configure?
Glibc sets itself up according to whatever --host is specified as, but you need to override CC and set CFLAGS so the chosen
toolchain produces binaries correctly for the target type you have chosen.
Ensure you set
CC on the environment to configure with correct flags
CC=sparc*-*-linux-gnu-gcc -m64 -mcpu=ultrasparc" \
and supply --host=sparc64-*-linux-gnu .
(You may need to modify --build slightly to trigger glibc to cross-compile if build matches host)
Note: currently no nptl support for sparc64
Been a while since I have attempted a bi-arch sparc build, may run a build over the next day or two, will report back
|All times are GMT -5. The time now is 09:37 PM.|