[Bug c/53662] New: Cannot build static gcc on i686 linux gnu with -m64 support.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53662 Bug #: 53662 Summary: Cannot build static gcc on i686 linux gnu with -m64 support. Classification: Unclassified Product: gcc Version: 4.6.3 Status: UNCONFIRMED Severity: normal Priority: P3 Component: c AssignedTo: unassig...@gcc.gnu.org ReportedBy: kre...@email.com Hello there. I am not sure if something is wrong with my setup, but I am unable to build gcc multilib compiler. I have pure 32bit system built from scratch (Linux From Scratch), but I would like to have multilib compiler so I can compile 64bit kernel in 32bit userland (not cross-compile). Long story short, I am unable to bootstrap gcc to support -m64. ~/src/build/prev-gcc$ ./xgcc -v Using built-in specs. COLLECT_GCC=./xgcc Target: i686-pc-linux-gnu Configured with: ../src/configure --prefix=/usr/gcc --libdir=/usr/gcc/lib --libexecdir=/usr/gcc/lib --without-headers --with-newlib --disable-shared --disable-nls --disable-decimal-float --disable-libgomp --disable-libmudflap --disable-libquadmath --disable-libssp --disable-threads --without-ppl --without-cloog --enable-targets=i686-pc-linux-gnu,x86_64-unknown-linux-gnu --with-mpfr-include=/home/armin/src/src/mpfr/src --with-mpfr-lib=/home/armin/src/build/mpfr/.libs --enable-multilib --with-multilib-list=m32,m64,mx32 --enable-bootstrap --enable-languages=c Thread model: single gcc version 4.6.3 (GCC) ~/src/build/prev-gcc$ ./cc1 --help ... -m32[enabled] -m3dnow [disabled] -m64[disabled] ... These are from stage2 and remain same after stage3, even when installed. It just says "Sorry, 64bit mode not implemented" or whatever when I try to compile something with gcc -m64. I have built binutils with --enable-multilib support already, so I am not really sure what I've done wrong. Shared gcc libraries require 64bit glibc and since I don't have multilib compiler, I can't compile it ... So I am in catch22 situation. It looks like you need multilib compiler to compile multilib compiler. Also, I have tried without --with-multilib-lists, without --enable-targets, it was same. Just when I used --enable-targets=all, it failed saying that 64bit mode is unimplemeted after stage1 xgcc has been built. Also, I have tried with system gmp, mpfr and mpc libraries which are same versions as ones that are in gcc source tree (source dirs renamed into mpfr, gmp and mpc respectively). This is configure output from stage1 libgcc when using --enable-targets=all checking whether ln -s works... yes checking for i686-pc-linux-gnu-gcc... /home/armin/src/build/./gcc/xgcc -B/home/armin/src/build/./gcc/ -B/usr/gcc/i686-pc-linux-gnu/bin/ -B/usr/gcc/i686-pc-linux-gnu/lib/ -isystem /usr/gcc/i686-pc-linux-gnu/include -isystem /usr/gcc/i686-pc-linux-gnu/sys-include -m64 checking for suffix of object files... configure: error: in `/home/armin/src/build/i686-pc-linux-gnu/64/libgcc': configure: error: cannot compute suffix of object files: cannot compile See `config.log' for more details. and config.log says this: Assembler messages: Fatal error: no compiled in support for x86_64 configure:3058: $? = 1 So, is this really a problem in gcc bootstrap or my setup? And how else I am able to build multilib compiler just from pure 32bit system? There must be a way right?
[Bug c/53662] Cannot build static gcc on i686 linux gnu with -m64 support.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53662 --- Comment #2 from Armin K. 2012-06-13 21:17:35 UTC --- (In reply to comment #1) > > That means you need a 64-bit capable binutils. > > Just a guess, but you might need to use --target=x86_64-unknown-linux-gnu to > make a cross compiler for x86_64, which will be multilib-capable by default. > To do that you'll need a 64-bit binutils and 64-bit glibc. > > This should really be dealt with on the gcc-help mailing list, I don't think > it's a bug. Sorry, I meant that this might be bug in gcc bootstrap stage. As I said, I have compiled binutils with --enable-multilib ... I could try --enable-64-bit-bfd tough. Also, what is gcc-help mailing list address?
[Bug c++/80655] New: -Werror=format-truncation inconsistency between x86_32 and x86_64
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80655
Bug ID: 80655
Summary: -Werror=format-truncation inconsistency between x86_32
and x86_64
Product: gcc
Version: 7.1.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c++
Assignee: unassigned at gcc dot gnu.org
Reporter: krejzi at email dot com
Target Milestone: ---
When building SPIRV-Tools with gcc-7.1.0 (stock, self-built), it builds fine
for 64 bit, but not for 32 bit (multilib setup).
The part of the code in question can be seen at [1]. I am getting the following
error when building SPIRV-Tools with "gcc -m32" and "g++ -m32"
In file included from
/home/armin/src/pacman/pkgbuild/vulkan-loader/src/Vulkan-LoaderAndValidationLayers32-sdk-1.0.46.0/external/spirv-tools/source/opt/ir_loader.cpp:17:0:
/home/armin/src/pacman/pkgbuild/vulkan-loader/src/Vulkan-LoaderAndValidationLayers32-sdk-1.0.46.0/external/spirv-tools/source/opt/log.h:
In function ‘void spvtools::Logf(const MessageConsumer&, spv_message_level_t,
const char*, const spv_position_t&, const char*, Args&& ...) [with Args =
{const SpvOp_&}]’:
/home/armin/src/pacman/pkgbuild/vulkan-loader/src/Vulkan-LoaderAndValidationLayers32-sdk-1.0.46.0/external/spirv-tools/source/opt/log.h:113:13:
error: null destination pointer [-Werror=format-truncation=]
snprintf(longer_message.data(), longer_message.size(), format,
^~
std::forward(args)...);
The very same package built fine with gcc-6.3.0.
[1]
https://github.com/KhronosGroup/SPIRV-Tools/blob/87a3f651e2416c830cb1eab410b3616068395985/source/opt/log.h#L111
[Bug c++/80655] -Werror=format-truncation inconsistency between x86_32 and x86_64
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80655 --- Comment #2 from Armin K. --- Created attachment 41333 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=41333&action=edit Preprocessed source files Here are the requested files. I apologize for having to compress them, but a single file exceeded bugzilla file size limit (1.1 MB).
[Bug c++/80655] -Werror=format-truncation inconsistency between x86_32 and x86_64
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80655 --- Comment #4 from Armin K. --- There is a log file in the tarball with all the options. Package is built with cmake, so there's lot of them.
[Bug preprocessor/65713] C Preprocessor generates invalid output
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65713 --- Comment #1 from Armin K. --- Created attachment 35275 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35275&action=edit gcc 5.0.0 snapshot output
[Bug preprocessor/65713] C Preprocessor generates invalid output
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65713 --- Comment #2 from Armin K. --- Created attachment 35276 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35276&action=edit File used to produce the first two files Two files are generated with gcc -E _mkerrcodes.h
[Bug preprocessor/65713] New: C Preprocessor generates invalid output
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65713 Bug ID: 65713 Summary: C Preprocessor generates invalid output Product: gcc Version: 5.0 Status: UNCONFIRMED Severity: blocker Priority: P3 Component: preprocessor Assignee: unassigned at gcc dot gnu.org Reporter: krejzi at email dot com Created attachment 35274 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35274&action=edit gcc-4.9.2 output Using GCC snapshot 5.0.0 20150405 (experimental) and trying to build libgpg-error-1.18, which generates a header file using the output from gcc -E on a header file. Output from gcc-4.9.2 and the mentioned snapshot is different, which causes libgpg-error to fail to compile. gcc-4.9.2 gives the output from first.txt mentioned snapshot gives the output from second.txt The header file used in both examples is the third attachment.
[Bug preprocessor/65713] C Preprocessor generates invalid output
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65713 --- Comment #4 from Armin K. --- Yes, that works. I'll take it to libgpg-error upstream. Thanks for the quick response and please close this bug.
