Rene, I understand how the wrapper work. It is a really good tool to inject custom cflags and cxxflags to gcc compiler.
My question was more like how gcc code generation work. Please see the link below because I decided to ask the question on gcc-help mailing list instead. https://www.spinics.net/lists/gcchelp/msg47824.html >> Btw. do you plan to chafe the chromium package you are working on? ;-) I just got qt-5.10 to build w/ webengine support, specifically for pentium4. The work detail was discussed here: https://www.mail-archive.com/[email protected]/msg30524.html There were a few challenges to be able to build it natively on x86_32bit, ie 8GIG memory + alot of disk space. Whoever is interested in getting webengine to run, he/she can following instructions provided in the above URL. TP On Mon, Jan 8, 2018 at 5:18 PM, René Rebe <[email protected]> wrote: > Hi, > > because all the Makefiles and such historically mess with CFLAGS and such, > T2 is using those gcc wrapper to automatically filter the options. > The wrappers therefore by default remove other -m* opts and inject what is > configured in T2 (as you ask for pentium4). > > If you want chromium to mess with the compiler options you could set > GCC_WRAPPER_BYPASS=1 or remove the -m* options from the GCC_WRAPPER_REMOVE > list. > > Btw. do you plan to chafe the chromium package you are working on? ;-) > > René > > On 08 Jan 2018, at 21:55, Toan Pham <[email protected]> wrote: > > > > > Rene, > > I am working on building chromium webengine w/ a t2 compiler that was > configured and optimized for pentium4. The detail of the compile is shown > at the end of this email, version 4.8.5 w/o support for SSE3>= and AVX and > AVX2. The problem I am having is that Chromium (libvpx) requires that the > compiler to support AVX+AVX2, so that those features can detected at > build-time and runtime. I've tried to reconfigured gcc wrapper (see > below) to support a newer architecture in hope that the compiler would be > able to build AVX+AVX2 instructions; however, compile still failed to > recognize/compile AVX assembly code. > > > *GCC_WRAPPER_DEBUG=1 GCC_WRAPPER_INSERT="-march=corei7 -mtune=corei7 -Os > -mssse3 -mavx -mavx2 -mpreferred-stack-boundary=7" > GCC_WRAPPER_REMOVE="-I/usr/local/include -I/usr/include -O -O[0-9s] -mtune* > -march* -mcpu* -Werror -g*" GCC_WRAPPER_APPEND="-s > */X11R6/include?-I/usr/X11R7/include */X11R6/lib?-L/usr/X11R7/lib > */X11R6/lib64?-L/usr/X11R7/lib -mavx -mavx2" make* > > > My question to you is that: is it true that once GCC is built w/ option > "-with-arch=pentium4", it is not possible to tell gcc to build machine code > for higher architectures with updated cflags like "-march=corei7 > -mtune=corei7 -mavx -mavx2"? If the answer is yes, then pretty much a new > 'generic i686' compiler is needed to build chromium. > > Thanks, > > TP > > > > debug-linux26:[webengine]# gcc -v > Using built-in specs. > COLLECT_GCC=gcc > COLLECT_LTO_WRAPPER=/usr/libexec/gcc/i686-t2-linux-gnu/4.8.5/lto-wrapper > Target: i686-t2-linux-gnu > Configured with: ../configure --prefix=/usr --bindir=/usr/bin > --sbindir=/usr/sbin --libdir=/usr/lib --datadir=/usr/share > --includedir=/usr/include --infodir=/usr/info --mandir=/usr/man > --sysconfdir=/etc --localstatedir=/var --disable-debug --with-libpam > --with-pam --enable-libpam --enable-pam --with-mysql-lib=/usr/lib/mysql > --with-mysql-include=/usr/include/mysql --with-mysql-lib=/usr/lib/mysql > --with-mysql-include=/usr/include/mysql --build=i686-t2-linux-gnu > --host=i686-t2-linux-gnu --target=i686-t2-linux-gnu > --enable-install-libiberty=no --disable-install-libiberty > --with-arch=pentium4 --with-arch=pentium4 --enable-__cxa_atexit > --disable-checking --disable-bootstrap --disable-libstdcxx-pch > --disable-multilib --disable-target-libiberty --with-gnu-as --with-gnu-ld > --enable-threads=posix --enable-languages=c,c++,objc,go --enable-shared > Thread model: posix > > ----------------------------------------------------------- > If you wish to unsubscribe from this mailing, send mail to > [email protected] with a subject of: unsubscribe t2 > > > -- > ExactCODE GmbH, Lietzenburger Str. 42, DE-10789 Berlin > <https://maps.google.com/?q=Lietzenburger+Str.+42,+DE-10789+Berlin+%C2%A0DE&entry=gmail&source=g> > DE > <https://maps.google.com/?q=Lietzenburger+Str.+42,+DE-10789+Berlin+%C2%A0DE&entry=gmail&source=g> > Legal: Amtsgericht Berlin (Charlottenburg) HRB 105123B, Tax-ID#: DE251602478 > Managing Director: René Rebe > http://exactcode.com | http://exactscan.com | http://ocrkit.com | > http://t2-project.org | http://rene.rebe.de > >
----------------------------------------------------------- If you wish to unsubscribe from this mailing, send mail to [email protected] with a subject of: unsubscribe t2
