On Fri, Dec 01, 2006 at 05:29:11AM -0800, tike64 wrote: > Thanks for hint. Why wasn't -lm needed when host machine was i386?
Glibc on i386 has ldexp in libc.so.6 as well as in libm.so.6, so -lm is not needed. Are you trying to compile it on Linux? > tcc.c:9505:2: warning: #warning add arch specific rt_get_caller_pc() > > Hope that is not critical. It's not critical. I didn't implement it because the TCC stackframe is not the same as the one produced by GCC, so backtraces would show only TCC compiled functions. > arm-linux-gcc -O2 -Wall -c -o libtcc1.o libtcc1.c > libtcc1.c:140:2: #error unsupported CPU type There are several reasons why ARM isn't supported by libtcc1 in the official TCC distribution: 1. TCC has no ARM assembler 2. ARMv4 has no instructions for divisions 3. There are more functions needed in libtcc1 for ARM than for i386 4. libgcc1.a or libgcc1_s.so can be used instead If you don't use the EABI patch that I posted some weeks ago, you can use the libtcc1 from my homepage: http://www.stud.uni-hannover.de/~daniel/tcc/ I prefer to just ln -s libgcc1_s.so to libtcc1.a. > Building bcheck fails also. Why does it try to build that when it > doesn't in case of i386 host? The Makefile is stupid. Bounds checking is not supported for ARM. Feel free to add support for it. Daniel _______________________________________________ Tinycc-devel mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/tinycc-devel
