| Message ID | 1422879584-18485-1-git-send-email-abrodkin@synopsys.com |
|---|---|
| State | Accepted |
| Headers | show |
Dear Alexey Brodkin, On Mon, 2 Feb 2015 15:19:44 +0300, Alexey Brodkin wrote: > Now when new shiny tools are released by Synopsys we're ready for > version update in Buildroot again. > > More details about arc-2014.12 release are available here: > https://github.com/foss-for-synopsys-dwc-arc-processors/toolchain/releases/tag/arc-2014.12 > > Following patches were removed from GCC since they are a part of release > now: > * 200-size_type_unsigned_int.patch > * 300-ptrdiff_type_int.patch > * 400-call-arc_hazard-before-branch-shortening.patch > * 401-fix-length-attribute-for-casesi_load-pattern.patch > * 402-fix-length-of-instructions-that-are-in-delay-slot-and-needs-to-be-predicated.patch > * 403-update-casesi_compact_jump-instruction-length.patch > > But since arc-2014.12 tools are still based on GCC 4.8 following patches > ar still relevant so moving to the new folder to match ARC gcc bump. > * 100-libstdcxx-uclibc-c99.patch > * 910-gcc-poison-system-directories.patch > > Binutils are still based on 2.23 so following patch still makes sense: > * 600-poison-system-directories.patch > > Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com> Applied, thanks! Thomas
Dear Alexey Brodkin, On Mon, 2 Feb 2015 15:19:44 +0300, Alexey Brodkin wrote: > Now when new shiny tools are released by Synopsys we're ready for > version update in Buildroot again. > > More details about arc-2014.12 release are available here: > https://github.com/foss-for-synopsys-dwc-arc-processors/toolchain/releases/tag/arc-2014.12 > > Following patches were removed from GCC since they are a part of release > now: > * 200-size_type_unsigned_int.patch > * 300-ptrdiff_type_int.patch > * 400-call-arc_hazard-before-branch-shortening.patch > * 401-fix-length-attribute-for-casesi_load-pattern.patch > * 402-fix-length-of-instructions-that-are-in-delay-slot-and-needs-to-be-predicated.patch > * 403-update-casesi_compact_jump-instruction-length.patch > > But since arc-2014.12 tools are still based on GCC 4.8 following patches > ar still relevant so moving to the new folder to match ARC gcc bump. > * 100-libstdcxx-uclibc-c99.patch > * 910-gcc-poison-system-directories.patch > > Binutils are still based on 2.23 so following patch still makes sense: > * 600-poison-system-directories.patch > > Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com> > > Cc: Anton Kolesov <akolesov@synopsys.com> > Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> > Cc: Peter Korsgaard <peter@korsgaard.com> It seems like the update of GDB for ARC is causing some build problems: http://autobuild.buildroot.org/results/d62/d62f472d70f60281dfe94cffc46a3e0566fd4a3d/build-end.log It's trying to build a shared libbfd.so using a statically compiled libiberty.a which wasn't built with -fPIC. I don't know if the problem is specific to ARC GDB or not. Can you have a look? Thanks! Thomas
Hi Thomas, On Wed, 2015-03-04 at 12:03 +0100, Thomas Petazzoni wrote: > Dear Alexey Brodkin, > > On Mon, 2 Feb 2015 15:19:44 +0300, Alexey Brodkin wrote: > > Now when new shiny tools are released by Synopsys we're ready for > > version update in Buildroot again. > > > > More details about arc-2014.12 release are available here: > > https://github.com/foss-for-synopsys-dwc-arc-processors/toolchain/releases/tag/arc-2014.12 > > > > Following patches were removed from GCC since they are a part of release > > now: > > * 200-size_type_unsigned_int.patch > > * 300-ptrdiff_type_int.patch > > * 400-call-arc_hazard-before-branch-shortening.patch > > * 401-fix-length-attribute-for-casesi_load-pattern.patch > > * 402-fix-length-of-instructions-that-are-in-delay-slot-and-needs-to-be-predicated.patch > > * 403-update-casesi_compact_jump-instruction-length.patch > > > > But since arc-2014.12 tools are still based on GCC 4.8 following patches > > ar still relevant so moving to the new folder to match ARC gcc bump. > > * 100-libstdcxx-uclibc-c99.patch > > * 910-gcc-poison-system-directories.patch > > > > Binutils are still based on 2.23 so following patch still makes sense: > > * 600-poison-system-directories.patch > > > > Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com> > > > > Cc: Anton Kolesov <akolesov@synopsys.com> > > Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> > > Cc: Peter Korsgaard <peter@korsgaard.com> > > It seems like the update of GDB for ARC is causing some build problems: > > http://autobuild.buildroot.org/results/d62/d62f472d70f60281dfe94cffc46a3e0566fd4a3d/build-end.log > > It's trying to build a shared libbfd.so using a statically compiled > libiberty.a which wasn't built with -fPIC. I don't know if the problem > is specific to ARC GDB or not. Sorry for late reply. I've just tried to re-build with defconfig from this autobuild and for me it got successfully to the end. I'm on master branch @ commit: --->8--- commit 6a3a9784db170ec05b300f228cf8a4bc48d838ec Author: Michael Trimarchi <michael@amarulasolutions.com> Date: Sun Mar 8 09:28:13 2015 +0100 python-spidev: new package --->8--- I'll downgrade to the same commit as mentioned in autobuild: --->8--- commit b9f39b2b389754bb53294311b938310283fda638 Author: Gustavo Zacarias <gustavo@zacarias.com.ar> Date: Tue Mar 3 15:43:09 2015 -0300 whois: bump to version 5.2.5 --->8--- and will give it another try, but frankly I don't expect any differences because I didn't find any sensible commits in between these 2 snapshots. -Alexey
Hi Alexey, Le 13 mars 2015 08:18:48 UTC+01:00, Alexey Brodkin <Alexey.Brodkin@synopsys.com> a écrit : >Hi Thomas, > >On Wed, 2015-03-04 at 12:03 +0100, Thomas Petazzoni wrote: >> Dear Alexey Brodkin, >> >> On Mon, 2 Feb 2015 15:19:44 +0300, Alexey Brodkin wrote: >> > Now when new shiny tools are released by Synopsys we're ready for >> > version update in Buildroot again. >> > >> > More details about arc-2014.12 release are available here: >> > >https://github.com/foss-for-synopsys-dwc-arc-processors/toolchain/releases/tag/arc-2014.12 >> > >> > Following patches were removed from GCC since they are a part of >release >> > now: >> > * 200-size_type_unsigned_int.patch >> > * 300-ptrdiff_type_int.patch >> > * 400-call-arc_hazard-before-branch-shortening.patch >> > * 401-fix-length-attribute-for-casesi_load-pattern.patch >> > * >402-fix-length-of-instructions-that-are-in-delay-slot-and-needs-to-be-predicated.patch >> > * 403-update-casesi_compact_jump-instruction-length.patch >> > >> > But since arc-2014.12 tools are still based on GCC 4.8 following >patches >> > ar still relevant so moving to the new folder to match ARC gcc >bump. >> > * 100-libstdcxx-uclibc-c99.patch >> > * 910-gcc-poison-system-directories.patch >> > >> > Binutils are still based on 2.23 so following patch still makes >sense: >> > * 600-poison-system-directories.patch >> > >> > Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com> >> > >> > Cc: Anton Kolesov <akolesov@synopsys.com> >> > Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> >> > Cc: Peter Korsgaard <peter@korsgaard.com> >> >> It seems like the update of GDB for ARC is causing some build >problems: >> >> >http://autobuild.buildroot.org/results/d62/d62f472d70f60281dfe94cffc46a3e0566fd4a3d/build-end.log >> >> It's trying to build a shared libbfd.so using a statically compiled >> libiberty.a which wasn't built with -fPIC. I don't know if the >problem >> is specific to ARC GDB or not. > >Sorry for late reply. > >I've just tried to re-build with defconfig from this autobuild and for >me it got successfully to the end. > >I'm on master branch @ commit: > --->8--- > commit 6a3a9784db170ec05b300f228cf8a4bc48d838ec > Author: Michael Trimarchi <michael@amarulasolutions.com> > Date: Sun Mar 8 09:28:13 2015 +0100 > > python-spidev: new package > --->8--- > >I'll downgrade to the same commit as mentioned in autobuild: > --->8--- > commit b9f39b2b389754bb53294311b938310283fda638 > Author: Gustavo Zacarias <gustavo@zacarias.com.ar> > Date: Tue Mar 3 15:43:09 2015 -0300 > > whois: bump to version 5.2.5 > --->8--- >and will give it another try, but frankly I don't expect any >differences >because I didn't find any sensible commits in between these 2 >snapshots. > Can you have a look at this patch: https://patchwork.ozlabs.org/patch/449686/ The issue seem to be related to avr32 removal. Best regards, Romain >-Alexey >_______________________________________________ >buildroot mailing list >buildroot@busybox.net >http://lists.busybox.net/mailman/listinfo/buildroot
On Fri, 2015-03-13 at 09:19 +0100, Romain Naour wrote: > Hi Alexey, > Can you have a look at this patch: > https://patchwork.ozlabs.org/patch/449686/ > > The issue seem to be related to avr32 removal. Strange enough I got everything built even from the same git snapshot as it was in autobuilder. That means I cannot really test proposed patch - it won't fix anything for me because everything works already. The only reason I may think of for that difference in behavior is host gcc version. I'm on 64-bit Fedora 21 with gcc 4.9.2: --->8--- gcc version 4.9.2 20150212 (Red Hat 4.9.2-6) (GCC) --->8--- Romain, Thomas what host gcc you/autobuilder use? -Alexey
Hi Alexey, ----- Mail original ----- | De: "Alexey Brodkin" <Alexey.Brodkin@synopsys.com> | À: "romain naour" <romain.naour@openwide.fr>, "thomas petazzoni" <thomas.petazzoni@free-electrons.com> | Cc: buildroot@uclibc.org | Envoyé: Vendredi 13 Mars 2015 10:00:34 | Objet: Re: [Buildroot] [PATCH] ARC: bump tools to 2014.12 release | | On Fri, 2015-03-13 at 09:19 +0100, Romain Naour wrote: | > Hi Alexey, | > Can you have a look at this patch: | > https://patchwork.ozlabs.org/patch/449686/ | > | > The issue seem to be related to avr32 removal. | | Strange enough I got everything built even from the same git snapshot | as | it was in autobuilder. | | That means I cannot really test proposed patch - it won't fix | anything | for me because everything works already. | | The only reason I may think of for that difference in behavior is | host | gcc version. I'm on 64-bit Fedora 21 with gcc 4.9.2: | --->8--- | gcc version 4.9.2 20150212 (Red Hat 4.9.2-6) (GCC) | --->8--- | | Romain, Thomas what host gcc you/autobuilder use? Same at home 64-bit Fedora 21 with gcc 4.9.2 :) As I said in the patch comment, I can't reproduce the issue... Best regards, Romain | | -Alexey |
Hi Thomas, On Fri, 2015-03-13 at 10:10 +0100, Romain Naour wrote: > Hi Alexey, > > ----- Mail original ----- > | De: "Alexey Brodkin" <Alexey.Brodkin@synopsys.com> > | À: "romain naour" <romain.naour@openwide.fr>, "thomas petazzoni" <thomas.petazzoni@free-electrons.com> > | Cc: buildroot@uclibc.org > | Envoyé: Vendredi 13 Mars 2015 10:00:34 > | Objet: Re: [Buildroot] [PATCH] ARC: bump tools to 2014.12 release > | > | On Fri, 2015-03-13 at 09:19 +0100, Romain Naour wrote: > | > Hi Alexey, > | > Can you have a look at this patch: > | > https://patchwork.ozlabs.org/patch/449686/ Do you think the patch Romain mentioned may fix http://autobuild.buildroot.org/results/d62/d62f472d70f60281dfe94cffc46a3e0566fd4a3d ? -Alexey
Dear Alexey Brodkin, On Tue, 24 Mar 2015 13:51:08 +0000, Alexey Brodkin wrote: > On Fri, 2015-03-13 at 10:10 +0100, Romain Naour wrote: > > Hi Alexey, > > > > ----- Mail original ----- > > | De: "Alexey Brodkin" <Alexey.Brodkin@synopsys.com> > > | À: "romain naour" <romain.naour@openwide.fr>, "thomas petazzoni" <thomas.petazzoni@free-electrons.com> > > | Cc: buildroot@uclibc.org > > | Envoyé: Vendredi 13 Mars 2015 10:00:34 > > | Objet: Re: [Buildroot] [PATCH] ARC: bump tools to 2014.12 release > > | > > | On Fri, 2015-03-13 at 09:19 +0100, Romain Naour wrote: > > | > Hi Alexey, > > | > Can you have a look at this patch: > > | > https://patchwork.ozlabs.org/patch/449686/ > > Do you think the patch Romain mentioned may fix > http://autobuild.buildroot.org/results/d62/d62f472d70f60281dfe94cffc46a3e0566fd4a3d ? Yes, most likely it will fix this problem. Thomas
Hi Thomas, On Tue, 2015-03-24 at 16:31 +0100, Thomas Petazzoni wrote: > Dear Alexey Brodkin, > > On Tue, 24 Mar 2015 13:51:08 +0000, Alexey Brodkin wrote: > > > On Fri, 2015-03-13 at 10:10 +0100, Romain Naour wrote: > > > Hi Alexey, > > > > > > ----- Mail original ----- > > > | De: "Alexey Brodkin" <Alexey.Brodkin@synopsys.com> > > > | À: "romain naour" <romain.naour@openwide.fr>, "thomas petazzoni" <thomas.petazzoni@free-electrons.com> > > > | Cc: buildroot@uclibc.org > > > | Envoyé: Vendredi 13 Mars 2015 10:00:34 > > > | Objet: Re: [Buildroot] [PATCH] ARC: bump tools to 2014.12 release > > > | > > > | On Fri, 2015-03-13 at 09:19 +0100, Romain Naour wrote: > > > | > Hi Alexey, > > > | > Can you have a look at this patch: > > > | > https://patchwork.ozlabs.org/patch/449686/ > > > > Do you think the patch Romain mentioned may fix > > http://autobuild.buildroot.org/results/d62/d62f472d70f60281dfe94cffc46a3e0566fd4a3d ? > > Yes, most likely it will fix this problem. Are you going to apply it or we'll need to find somebody to try it first? -Alexey
Hi Alexey, Le 24/03/2015 17:03, Alexey Brodkin a écrit : > Hi Thomas, > > On Tue, 2015-03-24 at 16:31 +0100, Thomas Petazzoni wrote: >> Dear Alexey Brodkin, >> >> On Tue, 24 Mar 2015 13:51:08 +0000, Alexey Brodkin wrote: >> >>> On Fri, 2015-03-13 at 10:10 +0100, Romain Naour wrote: >>>> Hi Alexey, >>>> >>>> ----- Mail original ----- >>>> | De: "Alexey Brodkin" <Alexey.Brodkin@synopsys.com> >>>> | À: "romain naour" <romain.naour@openwide.fr>, "thomas petazzoni" <thomas.petazzoni@free-electrons.com> >>>> | Cc: buildroot@uclibc.org >>>> | Envoyé: Vendredi 13 Mars 2015 10:00:34 >>>> | Objet: Re: [Buildroot] [PATCH] ARC: bump tools to 2014.12 release >>>> | >>>> | On Fri, 2015-03-13 at 09:19 +0100, Romain Naour wrote: >>>> | > Hi Alexey, >>>> | > Can you have a look at this patch: >>>> | > https://patchwork.ozlabs.org/patch/449686/ >>> >>> Do you think the patch Romain mentioned may fix >>> http://autobuild.buildroot.org/results/d62/d62f472d70f60281dfe94cffc46a3e0566fd4a3d ? >> >> Yes, most likely it will fix this problem. > > Are you going to apply it or we'll need to find somebody to try it > first? > I'm back on this issue and I found a way to reproduce it. As you can see in the build.log, libiberty.a is located in host/usr/lib/libiberty.a But on my machine libiberty.a is located in host/usr/lib64/libiberty.a This is really weird because autobuilders are also 64bits machine (x86_64-unknown-linux-gnu) and there is no reason that libiberty.a is not in the same place... So, I moved libiberty.a from host/usr/lib64/libiberty.a to host/usr/lib/libiberty.a and now I have the very same build failure. When gdb is build with shared libraries, it try to link with it own libiberty.a (host-gdb-7.8.2/bfd/../libiberty/pic -liberty) but it take the one from host/usr/lib/libiberty.a because -LPath/to/host/usr/lib/libiberty.a is before -LPath/to/build/host-gdb-7.8.2/bfd/../libiberty/pic. See the complete gcc line (verbose): /bin/sh ./libtool --tag=CC --mode=link /usr/bin/gcc -W -Wall -Wstrict-prototypes -Wmissing-prototypes -Wshadow -O2 -I/home/naourr/git/buildroot/test/gdb/host/usr/include -release `cat libtool-soversion` -L/home/naourr/git/buildroot/test/gdb/host/lib -L/home/naourr/git/buildroot/test/gdb/host/usr/lib -Wl,-rpath,/home/naourr/git/buildroot/test/gdb/host/usr/lib -o libbfd.la -rpath /home/naourr/git/buildroot/test/gdb/host/usr/x86_64-unknown-linux-gnu/arm-buildroot-linux-uclibcgnueabi/lib archive.lo archures.lo bfd.lo bfdio.lo bfdwin.lo cache.lo coffgen.lo corefile.lo format.lo init.lo libbfd.lo opncls.lo reloc.lo section.lo syms.lo targets.lo hash.lo linker.lo srec.lo binary.lo tekhex.lo ihex.lo stabs.lo stab-syms.lo merge.lo dwarf2.lo simple.lo compress.lo verilog.lo `cat ofiles` -L/home/naourr/git/buildroot/test/gdb/build/host-gdb-7.8.2/bfd/../libiberty/pic -liberty -Wl,-lc,--as-needed,-lm,--no-as-needed -lz This line come from gdb/bfd/Makefile.in: l1239: libbfd.la: $(libbfd_la_OBJECTS) $(libbfd_la_DEPENDENCIES) $(libbfd_la_LINK) $(am_libbfd_la_rpath) $(libbfd_la_OBJECTS) $(libbfd_la_LIBADD) $(LIBS) One solution is to disable shared libraries (preferred). The second is to tweak LDFLAGS in gdb.mk like this: HOST_GDB_CONF_OPTS = \ [...] LDFLAGS="-L$(@D)/bfd/../libiberty/pic $(HOST_LDFLAGS)" I don't see a better fix for gdb. Best regards, Romain
Hi Alexey, Le 05/04/2015 23:53, Romain Naour a écrit : > Hi Alexey, > > Le 24/03/2015 17:03, Alexey Brodkin a écrit : >> Hi Thomas, >> >> On Tue, 2015-03-24 at 16:31 +0100, Thomas Petazzoni wrote: >>> Dear Alexey Brodkin, >>> >>> On Tue, 24 Mar 2015 13:51:08 +0000, Alexey Brodkin wrote: >>> >>>> On Fri, 2015-03-13 at 10:10 +0100, Romain Naour wrote: >>>>> Hi Alexey, >>>>> >>>>> ----- Mail original ----- >>>>> | De: "Alexey Brodkin" <Alexey.Brodkin@synopsys.com> >>>>> | À: "romain naour" <romain.naour@openwide.fr>, "thomas petazzoni" <thomas.petazzoni@free-electrons.com> >>>>> | Cc: buildroot@uclibc.org >>>>> | Envoyé: Vendredi 13 Mars 2015 10:00:34 >>>>> | Objet: Re: [Buildroot] [PATCH] ARC: bump tools to 2014.12 release >>>>> | >>>>> | On Fri, 2015-03-13 at 09:19 +0100, Romain Naour wrote: >>>>> | > Hi Alexey, >>>>> | > Can you have a look at this patch: >>>>> | > https://patchwork.ozlabs.org/patch/449686/ >>>> >>>> Do you think the patch Romain mentioned may fix >>>> http://autobuild.buildroot.org/results/d62/d62f472d70f60281dfe94cffc46a3e0566fd4a3d ? >>> >>> Yes, most likely it will fix this problem. >> >> Are you going to apply it or we'll need to find somebody to try it >> first? >> > > I'm back on this issue and I found a way to reproduce it. > > As you can see in the build.log, libiberty.a is located in host/usr/lib/libiberty.a > But on my machine libiberty.a is located in host/usr/lib64/libiberty.a > > This is really weird because autobuilders are also 64bits machine > (x86_64-unknown-linux-gnu) and there is no reason that libiberty.a is not in the > same place... > So, I moved libiberty.a from host/usr/lib64/libiberty.a to > host/usr/lib/libiberty.a and now I have the very same build failure. This issue reminds me the issue related to config.site import while cross-compiling... See: http://git.buildroot.net/buildroot/commit/?id=f8d4fe376cf65cc4e41423804869a79c2570f674 I think the CONFIG_SITE=/dev/null setting is not taken into account somewhere (binutils or gcc) and the /usr/share/config.site file is sourced. So the libdir value is changed lib64 and that is why libiberty.a is installed to host/usr/lib64. I'm able to reproduce easily the issue on a debian 64bits and not on a Fedora 21 64bits. Best regards, Romain > > When gdb is build with shared libraries, it try to link with it own libiberty.a > (host-gdb-7.8.2/bfd/../libiberty/pic -liberty) but it take the one from > host/usr/lib/libiberty.a because -LPath/to/host/usr/lib/libiberty.a is before > -LPath/to/build/host-gdb-7.8.2/bfd/../libiberty/pic. > > See the complete gcc line (verbose): > /bin/sh ./libtool --tag=CC --mode=link /usr/bin/gcc -W -Wall > -Wstrict-prototypes -Wmissing-prototypes -Wshadow -O2 > -I/home/naourr/git/buildroot/test/gdb/host/usr/include -release `cat > libtool-soversion` -L/home/naourr/git/buildroot/test/gdb/host/lib > -L/home/naourr/git/buildroot/test/gdb/host/usr/lib > -Wl,-rpath,/home/naourr/git/buildroot/test/gdb/host/usr/lib -o libbfd.la -rpath > /home/naourr/git/buildroot/test/gdb/host/usr/x86_64-unknown-linux-gnu/arm-buildroot-linux-uclibcgnueabi/lib > archive.lo archures.lo bfd.lo bfdio.lo bfdwin.lo cache.lo coffgen.lo corefile.lo > format.lo init.lo libbfd.lo opncls.lo reloc.lo section.lo syms.lo targets.lo > hash.lo linker.lo srec.lo binary.lo tekhex.lo ihex.lo stabs.lo stab-syms.lo > merge.lo dwarf2.lo simple.lo compress.lo verilog.lo `cat ofiles` > -L/home/naourr/git/buildroot/test/gdb/build/host-gdb-7.8.2/bfd/../libiberty/pic > -liberty -Wl,-lc,--as-needed,-lm,--no-as-needed -lz > > This line come from gdb/bfd/Makefile.in: l1239: > libbfd.la: $(libbfd_la_OBJECTS) $(libbfd_la_DEPENDENCIES) > $(libbfd_la_LINK) $(am_libbfd_la_rpath) $(libbfd_la_OBJECTS) > $(libbfd_la_LIBADD) $(LIBS) > > One solution is to disable shared libraries (preferred). > > The second is to tweak LDFLAGS in gdb.mk like this: > HOST_GDB_CONF_OPTS = \ > [...] > LDFLAGS="-L$(@D)/bfd/../libiberty/pic $(HOST_LDFLAGS)" > > I don't see a better fix for gdb. > > Best regards, > Romain > _______________________________________________ > buildroot mailing list > buildroot@busybox.net > http://lists.busybox.net/mailman/listinfo/buildroot >
diff --git a/package/binutils/Config.in.host b/package/binutils/Config.in.host index 6dac3c0..0739467 100644 --- a/package/binutils/Config.in.host +++ b/package/binutils/Config.in.host @@ -34,7 +34,7 @@ endchoice config BR2_BINUTILS_VERSION string default "2.18-avr32-1.0.1" if BR2_BINUTILS_VERSION_2_18_AVR32_1_0_1 - default "arc-2014.08" if BR2_arc + default "arc-2014.12" if BR2_arc default "2.22" if BR2_BINUTILS_VERSION_2_22 default "2.23.2" if BR2_BINUTILS_VERSION_2_23_2 default "2.24" if BR2_BINUTILS_VERSION_2_24 diff --git a/package/binutils/arc-2014.08/600-poison-system-directories.patch b/package/binutils/arc-2014.12/600-poison-system-directories.patch similarity index 100% rename from package/binutils/arc-2014.08/600-poison-system-directories.patch rename to package/binutils/arc-2014.12/600-poison-system-directories.patch diff --git a/package/gcc/Config.in.host b/package/gcc/Config.in.host index 0750807..24004c4 100644 --- a/package/gcc/Config.in.host +++ b/package/gcc/Config.in.host @@ -97,7 +97,7 @@ config BR2_GCC_VERSION default "4.7.4" if BR2_GCC_VERSION_4_7_X default "4.8.4" if BR2_GCC_VERSION_4_8_X default "4.9.2" if BR2_GCC_VERSION_4_9_X - default "arc-2014.08" if BR2_GCC_VERSION_4_8_ARC + default "arc-2014.12" if BR2_GCC_VERSION_4_8_ARC config BR2_EXTRA_GCC_CONFIG_OPTIONS string "Additional gcc options" diff --git a/package/gcc/arc-2014.08/200-size_type_unsigned_int.patch b/package/gcc/arc-2014.08/200-size_type_unsigned_int.patch deleted file mode 100644 index a6a0419..0000000 --- a/package/gcc/arc-2014.08/200-size_type_unsigned_int.patch +++ /dev/null @@ -1,24 +0,0 @@ -Fixes SIZE_TYPE to be "unsigned int" instead of "long unsigned int". - -This makes size_t to be "unsigned" ssize_t which makes happy compiler on data -type checks. - -Fix is taken from current development branch of GCC for ARC and will be a -part of the next release of ARC tools, so at that point patch should be dropped. - -https://github.com/foss-for-synopsys-dwc-arc-processors/gcc/commit/249f040299402647525c3f15b79d319fa7acddd3 - -Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com> ---- - ---- a/gcc/config/arc/arc.h -+++ b/gcc/config/arc/arc.h -@@ -487,7 +487,7 @@ if (GET_MODE_CLASS (MODE) == MODE_INT \ - /* Define this as 1 if `char' should by default be signed; else as 0. */ - #define DEFAULT_SIGNED_CHAR 0 - --#define SIZE_TYPE "long unsigned int" -+#define SIZE_TYPE "unsigned int" - #define PTRDIFF_TYPE "long int" - #define WCHAR_TYPE "int" - #define WCHAR_TYPE_SIZE 32 diff --git a/package/gcc/arc-2014.08/300-ptrdiff_type_int.patch b/package/gcc/arc-2014.08/300-ptrdiff_type_int.patch deleted file mode 100644 index 6ee765b..0000000 --- a/package/gcc/arc-2014.08/300-ptrdiff_type_int.patch +++ /dev/null @@ -1,26 +0,0 @@ -Redefine PTRDIFF_TYPE - -Change of SIZE_TYPE from "long unsigned int" to "unsigned int" introduced -regression due to existing PTRDIFF_TYPE. - -Now to fix regression convert PTRDIFF_TYPE to simple "int". - -Fix is taken from current development branch of GCC for ARC and will be a -part of the next release of ARC tools, so at that point patch should be dropped. - -https://github.com/foss-for-synopsys-dwc-arc-processors/gcc/commit/846e92167aa4f486259c9ff44bb4e6cce6097fa4 - -Signed-off-by: Claudiu Zissulescu <claziss@synopsys.com> ---- - ---- a/gcc/config/arc/arc.h -+++ b/gcc/config/arc/arc.h -@@ -488,7 +488,7 @@ if (GET_MODE_CLASS (MODE) == MODE_INT \ - #define DEFAULT_SIGNED_CHAR 0 - - #define SIZE_TYPE "unsigned int" --#define PTRDIFF_TYPE "long int" -+#define PTRDIFF_TYPE "int" - #define WCHAR_TYPE "int" - #define WCHAR_TYPE_SIZE 32 - diff --git a/package/gcc/arc-2014.08/400-call-arc_hazard-before-branch-shortening.patch b/package/gcc/arc-2014.08/400-call-arc_hazard-before-branch-shortening.patch deleted file mode 100644 index 7361e6b..0000000 --- a/package/gcc/arc-2014.08/400-call-arc_hazard-before-branch-shortening.patch +++ /dev/null @@ -1,96 +0,0 @@ -Call arc_hazard before branch shortening - -On attempt to compile busybox (1.22.1) from buildroot master following build -error happened: ---->--- -$ arc-linux-gcc -Iinclude -Ilibbb -include include/autoconf.h -D_FILE_OFFSET_BITS=64 -fstack-protector-all -fno-guess-branch-probability -Os -c -o libbb/xfuncs.o libbb/xfuncs.c -xfuncs.s: Assembler messages: -xfuncs.s:444: Error: operand out of range (128 is not between -128 and 127) ---->--- - -Fix is taken from current development branch of GCC for ARC and will be a -part of the next release of ARC tools, so at that point patch should be dropped. - -https://github.com/foss-for-synopsys-dwc-arc-processors/gcc/commit/37ba2006be0b8c629d2f8ba6c5ec2819bd0269e5 - -Signed-off-by: Claudiu Zissulescu <claziss@synopsys.com> ---- - -diff --git a/gcc/config/arc/arc.c b/gcc/config/arc/arc.c -index 7d64f69..30baae1 100644 ---- a/gcc/config/arc/arc.c -+++ b/gcc/config/arc/arc.c -@@ -6520,6 +6520,16 @@ workaround_arc_anomaly (void) - { - rtx insn, succ0, succ1; - -+ /* For any architecture: call arc_hazard here. */ -+ for (insn = get_insns (); insn; insn = NEXT_INSN (insn)) -+ { -+ succ0 = next_real_insn(insn); -+ if (arc_hazard (insn, succ0)) -+ { -+ emit_insn_before (gen_nopv (), succ0); -+ } -+ } -+ - if (!TARGET_ARC700) - return; - -@@ -8710,7 +8720,7 @@ arc_loop_hazard (rtx pred, rtx succ) - succ_bb = BLOCK_FOR_INSN (NEXT_INSN (label)); - } - -- if (succ_bb && REGNO_REG_SET_P (df_get_live_in (succ_bb), LP_COUNT)) -+ if (succ_bb && REGNO_REG_SET_P (df_get_live_out (succ_bb), LP_COUNT)) - return true; - - return false; -diff --git a/gcc/testsuite/gcc.target/arc/pr9000798619.c b/gcc/testsuite/gcc.target/arc/pr9000798619.c -new file mode 100644 -index 0000000..b08b32d ---- /dev/null -+++ b/gcc/testsuite/gcc.target/arc/pr9000798619.c -@@ -0,0 +1,42 @@ -+/* { dg-do assemble } */ -+/* { dg-options "-Os -fstack-protector-all -fno-guess-branch-probability" } */ -+/* The arc_hazard introduces a NOP which is not taken into account by -+ branch shortening step, resulting in an out of range branch -+ offset (ARC700 architecture). */ -+ -+typedef unsigned char uint8_t; -+extern int *const bb_errno; -+ -+char* hex2bin(char *dst, const char *str, int count) -+{ -+ (*bb_errno) = 22; -+ while (*str && count) { -+ uint8_t val; -+ uint8_t c = *str++; -+ if (((unsigned char)((c) - '0') <= 9)) -+ val = c - '0'; -+ else if ((c|0x20) >= 'a' && (c|0x20) <= 'f') -+ val = (c|0x20) - ('a' - 10); -+ else -+ return ((void *)0); -+ val <<= 4; -+ c = *str; -+ if (((unsigned char)((c) - '0') <= 9)) -+ val |= c - '0'; -+ else if ((c|0x20) >= 'a' && (c|0x20) <= 'f') -+ val |= (c|0x20) - ('a' - 10); -+ else if (c == ':' || c == '\0') -+ val >>= 4; -+ else -+ return ((void *)0); -+ -+ *dst++ = val; -+ if (c != '\0') -+ str++; -+ if (*str == ':') -+ str++; -+ count--; -+ } -+ (*bb_errno) = (*str ? 34 : 0); -+ return dst; -+} diff --git a/package/gcc/arc-2014.08/401-fix-length-attribute-for-casesi_load-pattern.patch b/package/gcc/arc-2014.08/401-fix-length-attribute-for-casesi_load-pattern.patch deleted file mode 100644 index 959c1cc..0000000 --- a/package/gcc/arc-2014.08/401-fix-length-attribute-for-casesi_load-pattern.patch +++ /dev/null @@ -1,44 +0,0 @@ -Fix length attribute for casesi_load pattern. - -Perl compilation was broken: ---->--- -$ arc-linux-gcc -DPERL_CORE -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -matomic -pipe -mdiv-rem -Os -c -o perl.o perl.c -perl.s: Assembler messages: -perl.s:1271: Error: operand out of range (512 is not between -512 and 511) ---->--- - -Fix is taken from current development branch of GCC for ARC and will be a -part of the next release of ARC tools, so at that point patch should be dropped. - -https://github.com/foss-for-synopsys-dwc-arc-processors/gcc/commit/664ccd8d5ac8e047faac599309c9c2867af3a736 - -Signed-off-by: Claudiu Zissulescu <claziss@synopsys.com> ---- - ---- a/gcc/config/arc/arc.md -+++ b/gcc/config/arc/arc.md -@@ -3834,22 +3834,8 @@ - } - }" - [(set_attr "type" "load") -- (set_attr_alternative "iscompact" -- [(cond -- [(ne (symbol_ref "GET_MODE (PATTERN (next_real_insn (operands[3])))") -- (symbol_ref "QImode")) -- (const_string "false") -- (match_test "!ADDR_DIFF_VEC_FLAGS (PATTERN (next_real_insn (operands[3]))).offset_unsigned") -- (const_string "false")] -- (const_string "true")) -- (const_string "false") -- (const_string "false")]) -- (set_attr_alternative "length" -- [(cond -- [(eq_attr "iscompact" "false") (const_int 4)] -- (const_int 2)) -- (const_int 4) -- (const_int 8)])]) -+ (set_attr "iscompact" "false") -+ (set_attr "length" "4,4,8")]) - - ; Unlike the canonical tablejump, this pattern always uses a jump address, - ; even for CASE_VECTOR_PC_RELATIVE. diff --git a/package/gcc/arc-2014.08/402-fix-length-of-instructions-that-are-in-delay-slot-and-needs-to-be-predicated.patch b/package/gcc/arc-2014.08/402-fix-length-of-instructions-that-are-in-delay-slot-and-needs-to-be-predicated.patch deleted file mode 100644 index 7ceeada..0000000 --- a/package/gcc/arc-2014.08/402-fix-length-of-instructions-that-are-in-delay-slot-and-needs-to-be-predicated.patch +++ /dev/null @@ -1,69 +0,0 @@ -Fix length of instructions that are in delay slot and needs to be predicated - -Following problem was reported: ---->--- -arceb-linux-gcc -mq-class -g -O2 -c -o main.o main.i -/tmp/ccudUc8y.s: Assembler messages: -/tmp/ccudUc8y.s:18820: Error: operand out of range (256 is not between -256 and 255) ---->--- - -Fix is taken from current development branch of GCC for ARC and will be a -part of the next release of ARC tools, so at that point patch should be dropped. - -https://github.com/foss-for-synopsys-dwc-arc-processors/gcc/commit/5dfca5504d38293d5264f632c3090ac39c12f72b - -Signed-off-by: Claudiu Zissulescu <claziss@synopsys.com> ---- - -diff --git a/gcc/config/arc/arc-protos.h b/gcc/config/arc/arc-protos.h -index 908d002..212423b 100644 ---- a/gcc/config/arc/arc-protos.h -+++ b/gcc/config/arc/arc-protos.h -@@ -126,4 +126,5 @@ extern void arc_expand_compare_and_swap (rtx *); - - #ifdef RTX_CODE - extern void arc_expand_atomic_op (enum rtx_code, rtx, rtx, rtx, rtx, rtx); -+extern bool arc_bdr_iscond (rtx); - #endif -diff --git a/gcc/config/arc/arc.c b/gcc/config/arc/arc.c -index 68e5552..7d3ded3 100644 ---- a/gcc/config/arc/arc.c -+++ b/gcc/config/arc/arc.c -@@ -9336,6 +9336,23 @@ arc_write_ext_corereg (rtx insn) - return for_each_rtx (&PATTERN (insn), write_ext_corereg_1, 0); - } - -+/* Return true if the insn is in a delay slot and needs to be executed -+ conditionally. */ -+ -+bool -+arc_bdr_iscond (rtx insn) -+{ -+ rtx jump = prev_active_insn (insn); -+ -+ if (!jump || !JUMP_P(jump)) -+ return false; -+ -+ if (INSN_ANNULLED_BRANCH_P (jump) && INSN_FROM_TARGET_P (insn)) -+ return true; -+ -+ return false; -+} -+ - /* This is like the hook, but returns NULL when it can't / won't generate - a legitimate address. */ - -diff --git a/gcc/config/arc/arc.md b/gcc/config/arc/arc.md -index e001659..88b2e37 100644 ---- a/gcc/config/arc/arc.md -+++ b/gcc/config/arc/arc.md -@@ -377,7 +377,8 @@ - (cond [(match_test "GET_CODE (PATTERN (insn)) == COND_EXEC") - (const_int 12)] - (const_int 10)) -- (match_test "GET_CODE (PATTERN (insn)) == COND_EXEC") (const_int 4)] -+ (match_test "GET_CODE (PATTERN (insn)) == COND_EXEC || arc_bdr_iscond (insn)") -+ (const_int 4)] - (const_int 2)) - - (eq_attr "iscompact" "true_limm") diff --git a/package/gcc/arc-2014.08/403-update-casesi_compact_jump-instruction-length.patch b/package/gcc/arc-2014.08/403-update-casesi_compact_jump-instruction-length.patch deleted file mode 100644 index dd188a4..0000000 --- a/package/gcc/arc-2014.08/403-update-casesi_compact_jump-instruction-length.patch +++ /dev/null @@ -1,31 +0,0 @@ -Update casesi_compact_jump instruction length - -Fixes autobuilder issue http://autobuild.buildroot.net/results/bc5/bc5100d6462af4e2805f2bc8d39c87a55d843e2b/ - ---->--- -$ arc-linux-gcc -c -o wps_attr_parse.o -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -pipe -Os -I.../host/usr/arc-buildroot-linux-uclibc/sysroot/usr/include/libnl3/ -I.../build/hostapd-2.2/src -I.../build/hostapd-2.2/src/utils wps_attr_parse.c -wps_attr_parse.s: Assembler messages: -wps_attr_parse.s:860: Error: operand out of range (512 is not between -512 and 511) ---->--- - -Fix is taken from current development branch of GCC for ARC and will be a -part of the next release of ARC tools, so at that point patch should be dropped. - -https://github.com/foss-for-synopsys-dwc-arc-processors/gcc/commit/8e704c43cb50407ec79795f6f7459f09800b9f01 - -Signed-off-by: Claudiu Zissulescu <claziss@synopsys.com> ---- - -diff --git a/gcc/config/arc/arc.md b/gcc/config/arc/arc.md -index 88b2e37..8e18703 100644 ---- a/gcc/config/arc/arc.md -+++ b/gcc/config/arc/arc.md -@@ -3948,7 +3948,7 @@ - output_asm_insn (s, xop); - return \"add_s %2,%2,pcl\n\tj_s%* [%2]\"; - }" -- [(set_attr "length" "10") -+ [(set_attr "length" "12") - (set_attr "type" "jump") - (set_attr "iscompact" "true") - (set_attr "cond" "nocond")]) diff --git a/package/gcc/arc-2014.08/100-libstdcxx-uclibc-c99.patch b/package/gcc/arc-2014.12/100-libstdcxx-uclibc-c99.patch similarity index 100% rename from package/gcc/arc-2014.08/100-libstdcxx-uclibc-c99.patch rename to package/gcc/arc-2014.12/100-libstdcxx-uclibc-c99.patch diff --git a/package/gcc/arc-2014.08/910-gcc-poison-system-directories.patch b/package/gcc/arc-2014.12/910-gcc-poison-system-directories.patch similarity index 100% rename from package/gcc/arc-2014.08/910-gcc-poison-system-directories.patch rename to package/gcc/arc-2014.12/910-gcc-poison-system-directories.patch diff --git a/package/gdb/Config.in.host b/package/gdb/Config.in.host index 7f9df3a..c8d6a94 100644 --- a/package/gdb/Config.in.host +++ b/package/gdb/Config.in.host @@ -52,7 +52,7 @@ config BR2_GDB_VERSION depends on BR2_PACKAGE_GDB || BR2_PACKAGE_HOST_GDB default "6.7.1-avr32-2.1.5" if BR2_GDB_VERSION_6_7_1_AVR32_2_1_5 || \ (!BR2_PACKAGE_HOST_GDB && BR2_avr32) - default "arc-2014.08-gdb" if BR2_arc + default "arc-2014.12-gdb" if BR2_arc default "6be65fb56ea6694a9260733a536a023a1e2d4d57" if BR2_microblaze default "7.7.1" if BR2_GDB_VERSION_7_7 || !BR2_PACKAGE_HOST_GDB default "7.8.2" if BR2_GDB_VERSION_7_8 diff --git a/package/uclibc/Config.in b/package/uclibc/Config.in index 290f19b..020c47c 100644 --- a/package/uclibc/Config.in +++ b/package/uclibc/Config.in @@ -42,7 +42,7 @@ config BR2_UCLIBC_VERSION_STRING string default 0.9.31.1 if BR2_UCLIBC_VERSION_0_9_31 default 0.9.33.2 if BR2_UCLIBC_VERSION_0_9_33 - default "arc-2014.08" if BR2_arc + default "arc-2014.12" if BR2_arc default "7bf35c8b7d4a1f97174eb49f47f33946b282114c" if BR2_UCLIBC_VERSION_XTENSA_GIT default BR2_USE_UCLIBC_SNAPSHOT if BR2_UCLIBC_VERSION_SNAPSHOT
Now when new shiny tools are released by Synopsys we're ready for version update in Buildroot again. More details about arc-2014.12 release are available here: https://github.com/foss-for-synopsys-dwc-arc-processors/toolchain/releases/tag/arc-2014.12 Following patches were removed from GCC since they are a part of release now: * 200-size_type_unsigned_int.patch * 300-ptrdiff_type_int.patch * 400-call-arc_hazard-before-branch-shortening.patch * 401-fix-length-attribute-for-casesi_load-pattern.patch * 402-fix-length-of-instructions-that-are-in-delay-slot-and-needs-to-be-predicated.patch * 403-update-casesi_compact_jump-instruction-length.patch But since arc-2014.12 tools are still based on GCC 4.8 following patches ar still relevant so moving to the new folder to match ARC gcc bump. * 100-libstdcxx-uclibc-c99.patch * 910-gcc-poison-system-directories.patch Binutils are still based on 2.23 so following patch still makes sense: * 600-poison-system-directories.patch Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com> Cc: Anton Kolesov <akolesov@synopsys.com> Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Cc: Peter Korsgaard <peter@korsgaard.com> --- package/binutils/Config.in.host | 2 +- .../600-poison-system-directories.patch | 0 package/gcc/Config.in.host | 2 +- .../arc-2014.08/200-size_type_unsigned_int.patch | 24 ------ package/gcc/arc-2014.08/300-ptrdiff_type_int.patch | 26 ------ ...-call-arc_hazard-before-branch-shortening.patch | 96 ---------------------- ...-length-attribute-for-casesi_load-pattern.patch | 44 ---------- ...-in-delay-slot-and-needs-to-be-predicated.patch | 69 ---------------- ...te-casesi_compact_jump-instruction-length.patch | 31 ------- .../100-libstdcxx-uclibc-c99.patch | 0 .../910-gcc-poison-system-directories.patch | 0 package/gdb/Config.in.host | 2 +- package/uclibc/Config.in | 2 +- 13 files changed, 4 insertions(+), 294 deletions(-) rename package/binutils/{arc-2014.08 => arc-2014.12}/600-poison-system-directories.patch (100%) delete mode 100644 package/gcc/arc-2014.08/200-size_type_unsigned_int.patch delete mode 100644 package/gcc/arc-2014.08/300-ptrdiff_type_int.patch delete mode 100644 package/gcc/arc-2014.08/400-call-arc_hazard-before-branch-shortening.patch delete mode 100644 package/gcc/arc-2014.08/401-fix-length-attribute-for-casesi_load-pattern.patch delete mode 100644 package/gcc/arc-2014.08/402-fix-length-of-instructions-that-are-in-delay-slot-and-needs-to-be-predicated.patch delete mode 100644 package/gcc/arc-2014.08/403-update-casesi_compact_jump-instruction-length.patch rename package/gcc/{arc-2014.08 => arc-2014.12}/100-libstdcxx-uclibc-c99.patch (100%) rename package/gcc/{arc-2014.08 => arc-2014.12}/910-gcc-poison-system-directories.patch (100%)