GCC 7.1 is out.
*** Bug 617272 has been marked as a duplicate of this bug. ***
Please don't notify us of 0-day bumps. The volunteers at Gentoo do their best to get to things in a timely manner
Yes. We always had a tracking bug for porting though, so I wasn't thinking about "I need to notify people" when I created my bug. I was thinking along "we'll need a tracking bug somewhere before progress can happen, and we don't have such a tracking bug by now, so why not create one". I'm not sure if such mentality is discouraged. Maybe I need to write a request to become a volunteer on the toolchain team some day...
I have patches from 6.3.0, tweaked for 7.0.1_rcx. They work with 7.1.0 fine, too. Do I attache them here or will this again trigger dev rants ?
I wouldn't mind to see that patches, but I am not a Gentoo developer, of course. Now, on a serious manner, why not create a topic on the forum under Unsupported section ?
Created attachment 471594 [details, diff] repatched patch-pack for gcc-7.1.0
Created attachment 471596 [details] gcc-7.1.0 ebuild It's an ebuild that I copied from 6.3.0 to 7.0.1_rcx and then to 7.1.0 I don't remember changing anything except PATCH_VER I changed it to 1.1 so I don't collide with gentoo patchset, when they release 7.1
I have pushed an ebuild for gcc-7.1.0 to my overlay [1]. It essentially consists of the same patchset modification as attached to this bug report: - reapplied 13_all_default-ssp-fix.patch - temporarily removed 29_all_arm_armv4t-default.patch - temporarily removed 42_all_superh_default-multilib.patch @toolchain, I am currently giving it a quick run at recompiling a stage-3 in a libvirt-lxc container. Would you mind to add 7.1.0 to the tree in the near future (say next two weeks) if no regressions show up? We need a version in tree to work on bug #617524. [1] https://gitweb.gentoo.org/repo/dev/tamiko.git/
I recompiled world with gcc-7.1.0. Encountered ICE's regarding graphite on: media-libs/webrtc-audio-processing no-graphite media-libs/opus no-graphite dev-libs/libbsd no-graphite media-libs/libvpx no-graphite app-arch/p7zip no-graphite media-libs/adplug no-graphite media-libs/libvorbis no-graphite media-libs/flac no-graphite dev-libs/oniguruma no-graphite media-libs/speex no-graphite media-libs/libsndfile no-graphite media-libs/libtheora no-graphite app-cdr/cdrkit no-graphite media-video/ffmpeg no-graphite media-libs/gegl no-graphite media-plugins/audacious-plugins no-graphite media-gfx/gimp no-graphite dev-libs/icu no-graphite Firefox and thunderbird are affected by bug 580726. Interesting, they compile fine with gcc-6. sandbox-2.10-r4 here. Overall, a much painless transition than gcc-5 and 6 at their's time.
commit ee9917f9bef7e0cd837113745cd4d2329e9e9c51 Author: Matthias Maier <tamiko@gentoo.org> Date: Sun May 7 10:18:14 2017 -0500 sys-devel/gcc: add 7.1.0 to the tree, bug #617260 Package-Manager: Portage-2.3.5, Repoman-2.3.2 (In reply to Cănărău Constantin from comment #9) > I recompiled world with gcc-7.1.0. Encountered ICE's regarding graphite on: > [...] Thanks a lot for testing! Would you mind cross-checking with the buglist for the gcc-6 porting (bug #582084) and if not mentioned there, open a separate bug for each package and add it as a blogger to #617524
I do not know if I should open a bug for every failed packages with graphite as every of them are affected by gcc bug https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68823 Perhaps when a solution will be available upstream ? Every recent gcc new major release came with something similar, for example bug 615876 for 6.3 Thank you for your work!
@canarauc Interestingly enough, I'm using graphite and I have not had any ICE's. @tamiko I'm getting issues building grub with gcc-7.1.0. specifically, configure tests aren't working even though '-nostdlib' is in LDFLAGS. It looks like 13_all_default-ssp-fix.patch is not working. sample from config.log: ``` configure:25487: gcc -o conftest -Os -Wall -W -Wshadow -Wpointer-arith -Wundef -Wchar-subscripts -Wcomment -Wdeprecated-declarations -Wdisabled-optimization -Wdiv-by-zero -Wfloat-equal -Wformat-extra-args -Wformat-security -Wformat-y2k -Wimplicit -Wimplicit-function-declaration -Wimplicit-int -Wmain -Wmissing-braces -Wmissing-format-attribute -Wmultichar -Wparentheses -Wreturn-type -Wsequence-point -Wshadow -Wsign-compare -Wswitch -Wtrigraphs -Wunknown-pragmas -Wunused -Wunused-function -Wunused-label -Wunused-parameter -Wunused-value -Wunused-variable -Wwrite-strings -Wnested-externs -Wstrict-prototypes -g -Wredundant-decls -Wmissing-prototypes -Wmissing-declarations -march=i386 -m32 -mrtd -mregparm=3 -falign-jumps=1 -falign-loops=1 -falign-functions=1 -freg-struct-return -mno-mmx -mno-sse -mno-sse2 -mno-sse3 -mno-3dnow -msoft-float -fno-dwarf2-cfi-asm -mno-stack-arg-probe -fno-asynchronous-unwind-tables -fno-unwind-tables -Wall -W -DGRUB_MACHINE_PCBIOS=1 -DGRUB_MACHINE=I386_PC -m32 -Wl,-arch,i386 -nostdlib -static conftest.c >&5 /tmp/ccBADkIp.o: In function `__main': /home/adb/Downloads/grub-2.02/conftest.c:350: undefined reference to `__stack_chk_fail_local' /tmp/ccBADkIp.o: In function `main': /home/adb/Downloads/grub-2.02/conftest.c:358: undefined reference to `__stack_chk_fail_local' /usr/lib/gcc/x86_64-gentoo-linux-musl/7.1.0/../../../../x86_64-gentoo-linux-musl/bin/ld: conftest: hidden symbol `__stack_chk_fail_local' isn't defined /usr/lib/gcc/x86_64-gentoo-linux-musl/7.1.0/../../../../x86_64-gentoo-linux-musl/bin/ld: final link failed: Bad value collect2: error: ld returned 1 exit status ```
After staring at the gcc source code for about an hour, I think i found it: In 13_all_default_ssp-fix.patch there's a typo: ``` --- a/gcc/gcc.c +++ b/gcc/gcc.c @@ -1145,6 +1145,7 @@ %{-help=*:--help=%*}\ %{!fsyntax-only:%{S:%W{o*}%{!o*:-o %b.s}}}\ %{fsyntax-only:-o %j} %{-param*}\ + %{nostlib:-nostdlib}\ %{coverage:-fprofile-arcs -ftest-coverage}\ %{fprofile-arcs|fprofile-generate*|coverage:\ %{!fprofile-update=single:\ We disable ssp on -nostdlib and -ffreestanding as we did on older gcc and change the buffer size. ``` "nostlib" should be "nostdlib" Please fix this ASAP
Are you using lto ? Also bug 580726 is present when using both lto and graphite.
(In reply to Cănărău Constantin from comment #14) > Are you using lto ? Also bug 580726 is present when using both lto and > graphite. Yes, I'm using lto. My C/XXFLAGS are "-march=native -O2 -pipe -flto=2 -floop-nest-optimize -fuse-linker-plugin"
Those are possible culprits then: -fno-fat-lto-objects -fuse-ld=gold -fgraphite-identity -ftree-vectorize -fgcse-las. I'll have a look tomorrow.
(In reply to Cănărău Constantin from comment #16) > Those are possible culprits then: -fno-fat-lto-objects -fuse-ld=gold > -fgraphite-identity -ftree-vectorize -fgcse-las. > I'll have a look tomorrow. Just FYI, "-fgraphite-identity" is redundant with "-floop-nest-optimize" in GCC 7.1 (if you look at the source code.) "-fno-fat-lto-objects" has been the default behaviour for LTO since.. GCC 5...? I'd say it's most likely "-fuse-ld=gold" that's causing you issues -- for that matter, why didn't you just set ld to bfd using `binutils-config --linked ld.gold`? I'm pretty sure "-ftree-vectorize" "-fgcse-las" are pretty safe, but more likely to give you runtime issues.
> I'd say it's most likely "-fuse-ld=gold" that's causing you issues -- for > that matter, why didn't you just set ld to bfd using `binutils-config > --linked ld.gold`? *ld to gold using `binutils-config --linker ld.gold`
(In reply to Aric Belsito from comment #13) > After staring at the gcc source code for about an hour, I think i found it: > > In 13_all_default_ssp-fix.patch there's a typo: > > [...] > > Please fix this ASAP *ouch* commit b358e3be9843e80cd2a752686cba21a838be33c1 Author: Matthias Maier <tamiko@gentoo.org> Date: Mon May 8 15:42:45 2017 -0500 sys-devel/gcc: Bump patchset for gcc-7.1.0, bug #617260 comment 13 This fixes a typo in patch 13_all_default_ssp-fix.patch Package-Manager: Portage-2.3.5, Repoman-2.3.2
(In reply to Cănărău Constantin from comment #11) > I do not know if I should open a bug for every failed packages with graphite > as every of them are affected by gcc bug > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68823 > Perhaps when a solution will be available upstream ? Can you please open *one* bug for gcc-7.1.0 for the graphite issue?
