CODEWITHSUNDEEP
gccx86x86-64
Nuclear
Nuclear

Reputation: 1458

Error building 32-bit (x86) gcc-10.2.0 on a 64-bit (x86-64) machine

I am building a 32-bit LFS system, using this tutorial:

https://www.linuxfromscratch.org/~thomas/multilib/index.html

The build is performed on a 64-bit system. As it will run on a 32-bit system, I also need to build all binaries for 32-bit (i place them in a separate directory, so that they don't overwrite the 64-bit ones)

I have built 32-bit gmp, mpc, and mpfr (because gcc requires them) without any problems. Also, no problems building the 64-bit gcc.

However, I have a problem, building the 32-bit gcc-10.2.0 (I remove the build directory after the 64-bit build).

Using the following configuration: (I need -m32 both in CC and CFLAGS, otherwise make fails because it tries to link 64-bit .o files in a 32-bit library, maybe some internal configure scripts do not take the "-m32" option)

CC="gcc -m32" CXX="g++ -m32" CFLAGS="-m32" CXXLAGS="-m32" LDFLAGS="-m32" AR="ar" AS="as" \
            ../configure --prefix=/usr   \
             --enable-languages=c,c++ \
             --disable-bootstrap      \
             --libdir=/usr/lib32 \
             --build=x86_64-lfs-linux-gnu \
             --enable-multilib       \
             --with-multilib-list=m32,m64 \
             --host=i686-pc-linux-gnu \
             --with-system-zlib

during make, I get the following error:

cc   -g -O2 -m32 -O2  -g -O2 -m32 -DIN_GCC    -W -Wall -Wno-narrowing -Wwrite-strings -Wcast-qual -Wno-error=format-diag -Wstrict-prototypes -Wmissing-prototypes -Wno-error=format-diag -Wold-style-definition  -isystem ./include  -fpic -mlong-double-80 -DUSE_ELF_SYMVER  -g -DIN_LIBGCC2 -fbuilding-libgcc -fno-stack-protector  -fpic -mlong-double-80 -DUSE_ELF_SYMVER  -I. -I. -I../.././gcc -I../../../libgcc -I../../../libgcc/. -I../../../libgcc/../gcc -I../../../libgcc/../include -I../../../libgcc/config/libbid -DENABLE_DECIMAL_BID_FORMAT -DHAVE_CC_TLS  -DUSE_TLS -Wno-missing-prototypes -Wno-type-limits -o fixtfti.o -MT fixtfti.o -MD -MP -MF fixtfti.dep  -c ../../../libgcc/soft-fp/fixtfti.c -fvisibility=hidden -DHIDE_EXPORTS
../../../libgcc/soft-fp/fixtfti.c:33:1: error: unknown type name 'TItype'; did you mean 'TFtype'?
   33 | TItype
      | ^~~~~~
      | TFtype
../../../libgcc/soft-fp/fixtfti.c: In function '__fixtfti':
../../../libgcc/soft-fp/fixtfti.c:38:3: error: unknown type name 'UTItype'; did you mean 'UDItype'?

TItype and UDItype are not defined. As I found, they are defined for 64-bit configurations (config/i386/64, config/ia64, etc..)

but I am building a 32-bit gcc and they are not defined but are used.

Any ideas on how to resolve this?

Upvotes: 2

Views: 605

Answers (1)

Nuclear
Nuclear

Reputation: 1458

The solution was to set also the CC_FOR_TARGET and CXX_FOR_TARGET environment variables when configuring GCC. I thought that CC and CXX were enough.

CC="gcc -m32" CC_FOR_TARGET="gcc -m32" CXX="gcc -m32" CXX_FOR_TARGET="gcc -m32" AR="ar"

Upvotes: 2

Related Questions