Re: [OE-core] [PATCH] libjpeg-turbo: fix build on aarch64 or non-intel build hosts
On Thu, 2017-03-30 at 12:21 +0100, Burton, Ross wrote: > > > Agree with Andre here. I would suggest to update nasm recipe to > > support > > build for other archs as well (tested on arm and it works as > > expected. Will > > require more testing on other archs). > > Yes. Simply deleting the COMPATIBLE_HOST in nasm.bb should be > sufficient, Debian builds nasm for every architecture so that > demonstrates that it builds just fine. > Ok, I had this concern too when originally discussing the approach in #yocto, this sounds better. So just to get things straight, does this mean we no longer need the patch to remove nasm-native from the jpeg turbo dependencies on non- intel targets ? I think "-native" here only means "build an nasm that I can run on the host architecture" then I think this is fine, as jpeg-turbo will just ignore the needlessly built nasm when targetting non-intel platforms. Cheers, -Tristan -- ___ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core
[OE-core] [PATCHv3] libjpeg-turbo: fix build on aarch64 or non-intel build hosts
From f54c719dc0ca3444bc579663b7326d016bf2e1e7 Mon Sep 17 00:00:00 2001
From: Tristan Van Berkom <tristan.vanber...@codethink.co.uk>
Date: Fri, 24 Mar 2017 15:37:36 +
Subject: [PATCH] libjpeg-turbo: Support building on non intel targets.
nasm is intel specific, however libjpeg-turbo is still desirable
on other architectures.
Without this patch, it would seem that nasm is built on an intel
host and then ignored when compiling libjpeg-turbo for an arm target,
however it has libjpeg-turbo requiring nasm be built when building
on an arm/aarch64 host.
This fixes the issue so that nasm-native is not required on the host
when building for an aarch64 target.
Signed-off-by: Tristan Van Berkom <tristan.vanber...@codethink.co.uk>
---
meta/recipes-graphics/jpeg/libjpeg-turbo_1.5.1.bb | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/meta/recipes-graphics/jpeg/libjpeg-turbo_1.5.1.bb
b/meta/recipes-graphics/jpeg/libjpeg-turbo_1.5.1.bb
index 70d38f4a1a..a650d6af2d 100644
--- a/meta/recipes-graphics/jpeg/libjpeg-turbo_1.5.1.bb
+++ b/meta/recipes-graphics/jpeg/libjpeg-turbo_1.5.1.bb
@@ -7,7 +7,8 @@ LIC_FILES_CHKSUM =
"file://cdjpeg.h;endline=13;md5=05bab7c7ad899d85bfba60da1a127
file://jpeglib.h;endline=16;md5=f67d70e547a2662c079781c72f877f72 \
file://djpeg.c;endline=11;md5=b90b6d2b4119f9e5807cd273f525d2af \
"
-DEPENDS = "nasm-native"
+DEPENDS_append_x86-64 = " nasm-native"
+DEPENDS_append_x86= " nasm-native"
SRC_URI = "${SOURCEFORGE_MIRROR}/${BPN}/${BPN}-${PV}.tar.gz \
file://fix-mips.patch"
--
2.11.0
--
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core
[OE-core] [PATCH] libjpeg-turbo: fix build on aarch64 or non-intel build hosts
This is already filed in the appropriate bugzilla:
https://bugzilla.yoctoproject.org/show_bug.cgi?id=11240
Just notifying the list as the custom goes...
libjpeg-turbo in recipes-graphics has an unconditional dependency on
nasm-native, which is intel specific and wont be provided on non intel
arches.
Attaching a patch which fixes this by only requiring nasm-native when
targeting intel architectures.
Cheers,
-Tristan
From d76d57a148dd372ff26b9b902b7b79a4ddb5e7a2 Mon Sep 17 00:00:00 2001
From: Tristan Van Berkom <tristan.vanber...@codethink.co.uk>
Date: Fri, 24 Mar 2017 15:37:36 +
Subject: [PATCH] libjpeg-turbo: Support building on non intel targets.
nasm is intel specific, however libjpeg-turbo is still desirable
on other architectures.
Without this patch, it would seem that nasm is built on an intel
host and then ignored when compiling libjpeg-turbo for an arm target,
however it has libjpeg-turbo requiring nasm be built when building
on an arm/aarch64 host.
This fixes the issue so that nasm-native is not required on the host
when building for an aarch64 target.
---
meta/recipes-graphics/jpeg/libjpeg-turbo_1.5.1.bb | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/meta/recipes-graphics/jpeg/libjpeg-turbo_1.5.1.bb b/meta/recipes-graphics/jpeg/libjpeg-turbo_1.5.1.bb
index 70d38f4a1a..a650d6af2d 100644
--- a/meta/recipes-graphics/jpeg/libjpeg-turbo_1.5.1.bb
+++ b/meta/recipes-graphics/jpeg/libjpeg-turbo_1.5.1.bb
@@ -7,7 +7,8 @@ LIC_FILES_CHKSUM = "file://cdjpeg.h;endline=13;md5=05bab7c7ad899d85bfba60da1a127
file://jpeglib.h;endline=16;md5=f67d70e547a2662c079781c72f877f72 \
file://djpeg.c;endline=11;md5=b90b6d2b4119f9e5807cd273f525d2af \
"
-DEPENDS = "nasm-native"
+DEPENDS_append_x86-64 = " nasm-native"
+DEPENDS_append_x86= " nasm-native"
SRC_URI = "${SOURCEFORGE_MIRROR}/${BPN}/${BPN}-${PV}.tar.gz \
file://fix-mips.patch"
--
2.11.0
--
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core
[OE-core] [OE-Core][PATCH] libjpeg-turbo fails to build for aarch64/arm targets on aarch64 host
This is already filed in the appropriate bugzilla:
https://bugzilla.yoctoproject.org/show_bug.cgi?id=11240
Just notifying the list as the custom goes...
libjpeg-turbo in recipes-graphics has an unconditional dependency on
nasm-native, which is intel specific and wont be provided on non intel
arches.
Attaching a patch which fixes this by only requiring nasm-native when
targeting intel architectures.
Cheers,
-Tristan
From 29a82ba25a9c395f5214699ca0afb5bd465ebdf1 Mon Sep 17 00:00:00 2001
From: Tristan Van Berkom <tristan.vanber...@codethink.co.uk>
Date: Fri, 24 Mar 2017 15:37:36 +
Subject: [PATCH] libjpeg-turbo: Support building on non intel targets.
nasm is intel specific, however libjpeg-turbo is still desirable
on other architectures.
Without this patch, it would seem that nasm is built on an intel
host and then ignored when compiling libjpeg-turbo for an arm target,
however it has libjpeg-turbo requiring nasm be built when building
on an arm/aarch64 host.
This fixes the issue so that nasm-native is not required on the host
when building for an aarch64 target.
---
meta/recipes-graphics/jpeg/libjpeg-turbo_1.5.0.bb | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/meta/recipes-graphics/jpeg/libjpeg-turbo_1.5.0.bb b/meta/recipes-graphics/jpeg/libjpeg-turbo_1.5.0.bb
index c9e70d73d2..a02406eaa6 100644
--- a/meta/recipes-graphics/jpeg/libjpeg-turbo_1.5.0.bb
+++ b/meta/recipes-graphics/jpeg/libjpeg-turbo_1.5.0.bb
@@ -7,7 +7,8 @@ LIC_FILES_CHKSUM = "file://cdjpeg.h;endline=13;md5=05bab7c7ad899d85bfba60da1a127
file://jpeglib.h;endline=16;md5=f67d70e547a2662c079781c72f877f72 \
file://djpeg.c;endline=11;md5=b90b6d2b4119f9e5807cd273f525d2af \
"
-DEPENDS = "nasm-native"
+DEPENDS_append_x86-64 = " nasm-native"
+DEPENDS_append_x86= " nasm-native"
SRC_URI = "${SOURCEFORGE_MIRROR}/${BPN}/${BPN}-${PV}.tar.gz"
SRC_URI[md5sum] = "3fc5d9b6a8bce96161659ae7a9939257"
--
2.11.0
--
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core
Re: [OE-core] [PATCH 1/1] webkitgtk: turn off JIT on armv4 and armv7a
On Thu, 2016-05-19 at 05:16 +0200, Gary Thomas wrote: [...] > Why do you think that changing the instruction set (to thumb2) would > make the JIT work any better? > > If you tell me how, I'll test it on my i.MX6 (Cortex-A9) which is > the platform that inspired the original bug report. FWIW, I ran into this issue as well the first time I tried building webkitgtk with a compiler tuned for armv7a. I did not try many variations but was able to build with a machine configuration with DEFAULTTUNE set to armv7ahf-neon (hard float with neon simd). This link seems to also provide some more information on what configurations can be used to build webkitgtk on arm: http://patchwork.openembedded.org/patch/105081/ Cheers, -Tristan -- ___ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core
Re: [OE-core] [PATCH 3/4] Backporting binutils bug fix to the 2.25 branch
On Tue, 2016-04-26 at 09:34 -0700, Khem Raj wrote: > > On Apr 26, 2016, at 9:31 AM, Tristan Van Berkom <tristan.vanberkom@ > > codethink.co.uk> wrote: > > > > On Mon, 2016-04-25 at 13:55 -0700, Khem Raj wrote: > > > > On Apr 25, 2016, at 1:44 PM, Tristan Van Berkom > > > > <tristan.vanberkom@ > > > > codethink.co.uk> wrote: > > > > > > > > We fail to build webkit on aarch64 due to this binutils bug: > > > > > > > > https://sourceware.org/bugzilla/show_bug.cgi?id=19353 > > > > > > > > Applying patch which fixes this, stripped out changelog entry > > > > from patch to make it apply without error. > > > > > > > > > which branch are you proposing this for? > > > we use 2.26 on master > > > > We are using the jethro branch, sorry for leaving out that detail. > > > > OK, add [jethro] to patch subject line then, so it gets attention of > release maintainers. Thanks for your guidance. I've re-submitted this one to the list with [jethro] in the subject line (this one is obviously jethro specific), I'm not sure if the other two patches I sent apply to master at all - I will try to at least verify if either of those issues are fixed in master. Cheers, -Tristan -- ___ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core
[OE-core] [PATCH][jethro] binutils: backport bug fix to the 2.25 branch for jethro
We fail to build webkit on aarch64 due to this binutils bug:
https://sourceware.org/bugzilla/show_bug.cgi?id=19353
Applying patch which fixes this, stripped out changelog entry
from patch to make it apply without error.
---
meta/recipes-devtools/binutils/binutils-2.25.1.inc | 1 +
...plying-TLSDESC-relocs-without-TLS-segment.patch | 166 +
2 files changed, 167 insertions(+)
create mode 100644
meta/recipes-devtools/binutils/binutils/fix-internal-error-when-applying-TLSDESC-relocs-without-TLS-segment.patch
diff --git a/meta/recipes-devtools/binutils/binutils-2.25.1.inc
b/meta/recipes-devtools/binutils/binutils-2.25.1.inc
index f3817fa..c7db9e0 100644
--- a/meta/recipes-devtools/binutils/binutils-2.25.1.inc
+++ b/meta/recipes-devtools/binutils/binutils-2.25.1.inc
@@ -35,6 +35,7 @@ SRC_URI = "\
file://0015-Fix-dynamic-list-so-that-symbols-not-in-the-list-are.patch \
file://binutils-octeon3.patch \
file://add-thunderx-support-for-gas.patch \
+
file://fix-internal-error-when-applying-TLSDESC-relocs-without-TLS-segment.patch
\
"
S = "${WORKDIR}/git"
diff --git
a/meta/recipes-devtools/binutils/binutils/fix-internal-error-when-applying-TLSDESC-relocs-without-TLS-segment.patch
b/meta/recipes-devtools/binutils/binutils/fix-internal-error-when-applying-TLSDESC-relocs-without-TLS-segment.patch
new file mode 100644
index 000..df9d54c
--- /dev/null
+++
b/meta/recipes-devtools/binutils/binutils/fix-internal-error-when-applying-TLSDESC-relocs-without-TLS-segment.patch
@@ -0,0 +1,166 @@
+From d21f123b0ead1806416cf0dafae12bec4cca8920 Mon Sep 17 00:00:00 2001
+From: Cary Coutant
+Date: Mon, 11 Jan 2016 23:57:44 -0800
+Subject: [PATCH] Fix internal error when applying TLSDESC relocations with no
TLS segment.
+
+gold/
+ PR gold/19353
+ * aarch64.cc (Target_aarch64::relocate_tls): Don't insist that
+ we have a TLS segment for GD-to-IE optimization.
+ * i386.cc (Target_i386::tls_gd_to_ie): Remove tls_segment parameter.
+ Adjust all calls.
+ (Target_i386::tls_desc_gd_to_ie): Likewise.
+ (Target_i386::relocate_tls): Don't insist that we have a TLS segment
+ for TLSDESC GD-to-IE optimizations.
+ * x86_64.cc (Target_x86_64::tls_gd_to_ie): Remove tls_segment parameter.
+ Adjust all calls.
+ (Target_x86_64::tls_desc_gd_to_ie): Likewise.
+ (Target_x86_64::relocate_tls): Don't insist that we have a TLS segment
+ for TLSDESC GD-to-IE optimizations.
+---
+ gold/aarch64.cc |6 --
+ gold/i386.cc| 14 ++
+ gold/x86_64.cc | 14 ++
+ 4 files changed, 20 insertions(+), 30 deletions(-)
+
+diff --git a/gold/aarch64.cc b/gold/aarch64.cc
+index 20f2f4f..5ad061b 100644
+--- a/gold/aarch64.cc
b/gold/aarch64.cc
+@@ -7422,12 +7422,6 @@ Target_aarch64::Relocate::relocate_tls(
+ }
+ if (tlsopt == tls::TLSOPT_TO_IE)
+ {
+- if (tls_segment == NULL)
+-{
+- gold_assert(parameters->errors()->error_count() > 0
+- || issue_undefined_symbol_error(gsym));
+- return aarch64_reloc_funcs::STATUS_BAD_RELOC;
+-}
+ return tls_desc_gd_to_ie(relinfo, target, rela, r_type,
+view, psymval, got_entry_address,
+address);
+diff --git a/gold/i386.cc b/gold/i386.cc
+index 82886d4..a7168a8 100644
+--- a/gold/i386.cc
b/gold/i386.cc
+@@ -668,7 +668,6 @@ class Target_i386 : public Sized_target<32, false>
+ // Do a TLS General-Dynamic to Initial-Exec transition.
+ inline void
+ tls_gd_to_ie(const Relocate_info<32, false>*, size_t relnum,
+- Output_segment* tls_segment,
+const elfcpp::Rel<32, false>&, unsigned int r_type,
+elfcpp::Elf_types<32>::Elf_Addr value,
+unsigned char* view,
+@@ -687,7 +686,6 @@ class Target_i386 : public Sized_target<32, false>
+ // transition.
+ inline void
+ tls_desc_gd_to_ie(const Relocate_info<32, false>*, size_t relnum,
+-Output_segment* tls_segment,
+ const elfcpp::Rel<32, false>&, unsigned int r_type,
+ elfcpp::Elf_types<32>::Elf_Addr value,
+ unsigned char* view,
+@@ -3054,7 +3052,7 @@ Target_i386::Relocate::relocate_tls(const
Relocate_info<32, false>* relinfo,
+ }
+ if (optimized_type == tls::TLSOPT_TO_IE)
+ {
+-this->tls_gd_to_ie(relinfo, relnum, tls_segment, rel, r_type,
++this->tls_gd_to_ie(relinfo, relnum, rel, r_type,
+got_offset, view, view_size);
+ break;
+ }
+@@ -3116,13 +3114,7 @@ Target_i386::Relocate::relocate_tls(const
Relocate_info<32, false>* relinfo,
+ }
+ if (optimized_type ==
Re: [OE-core] [PATCH 3/4] Backporting binutils bug fix to the 2.25 branch
On Mon, 2016-04-25 at 13:55 -0700, Khem Raj wrote:
> > On Apr 25, 2016, at 1:44 PM, Tristan Van Berkom <tristan.vanberkom@
> > codethink.co.uk> wrote:
> >
> > We fail to build webkit on aarch64 due to this binutils bug:
> >
> > https://sourceware.org/bugzilla/show_bug.cgi?id=19353
> >
> > Applying patch which fixes this, stripped out changelog entry
> > from patch to make it apply without error.
>
>
> which branch are you proposing this for?
> we use 2.26 on master
We are using the jethro branch, sorry for leaving out that detail.
Cheers,
-Tristan
>
> > ---
> > meta/recipes-devtools/binutils/binutils-2.25.1.inc | 1 +
> > ...plying-TLSDESC-relocs-without-TLS-segment.patch | 166
> > +
> > 2 files changed, 167 insertions(+)
> > create mode 100644 meta/recipes-devtools/binutils/binutils/fix-
> > internal-error-when-applying-TLSDESC-relocs-without-TLS-
> > segment.patch
> >
> > diff --git a/meta/recipes-devtools/binutils/binutils-2.25.1.inc
> > b/meta/recipes-devtools/binutils/binutils-2.25.1.inc
> > index f3817fa..c7db9e0 100644
> > --- a/meta/recipes-devtools/binutils/binutils-2.25.1.inc
> > +++ b/meta/recipes-devtools/binutils/binutils-2.25.1.inc
> > @@ -35,6 +35,7 @@ SRC_URI = "\
> > file://0015-Fix-dynamic-list-so-that-symbols-not-in-the-list-
> > are.patch \
> > file://binutils-octeon3.patch \
> > file://add-thunderx-support-for-gas.patch \
> > + file://fix-internal-error-when-applying-TLSDESC-relocs-
> > without-TLS-segment.patch \
> > "
> > S = "${WORKDIR}/git"
> >
> > diff --git a/meta/recipes-devtools/binutils/binutils/fix-internal-
> > error-when-applying-TLSDESC-relocs-without-TLS-segment.patch
> > b/meta/recipes-devtools/binutils/binutils/fix-internal-error-when-
> > applying-TLSDESC-relocs-without-TLS-segment.patch
> > new file mode 100644
> > index 000..df9d54c
> > --- /dev/null
> > +++ b/meta/recipes-devtools/binutils/binutils/fix-internal-error-
> > when-applying-TLSDESC-relocs-without-TLS-segment.patch
> > @@ -0,0 +1,166 @@
> > +From d21f123b0ead1806416cf0dafae12bec4cca8920 Mon Sep 17 00:00:00
> > 2001
> > +From: Cary Coutant <ccout...@gmail.com>
> > +Date: Mon, 11 Jan 2016 23:57:44 -0800
> > +Subject: [PATCH] Fix internal error when applying TLSDESC
> > relocations with no TLS segment.
> > +
> > +gold/
> > + PR gold/19353
> > + * aarch64.cc (Target_aarch64::relocate_tls): Don't insist
> > that
> > + we have a TLS segment for GD-to-IE optimization.
> > + * i386.cc (Target_i386::tls_gd_to_ie): Remove tls_segment
> > parameter.
> > + Adjust all calls.
> > + (Target_i386::tls_desc_gd_to_ie): Likewise.
> > + (Target_i386::relocate_tls): Don't insist that we have a
> > TLS segment
> > + for TLSDESC GD-to-IE optimizations.
> > + * x86_64.cc (Target_x86_64::tls_gd_to_ie): Remove
> > tls_segment parameter.
> > + Adjust all calls.
> > + (Target_x86_64::tls_desc_gd_to_ie): Likewise.
> > + (Target_x86_64::relocate_tls): Don't insist that we have a
> > TLS segment
> > + for TLSDESC GD-to-IE optimizations.
> > +---
> > + gold/aarch64.cc |6 --
> > + gold/i386.cc| 14 ++
> > + gold/x86_64.cc | 14 ++
> > + 4 files changed, 20 insertions(+), 30 deletions(-)
> > +
> > +diff --git a/gold/aarch64.cc b/gold/aarch64.cc
> > +index 20f2f4f..5ad061b 100644
> > +--- a/gold/aarch64.cc
> > b/gold/aarch64.cc
> > +@@ -7422,12 +7422,6 @@ Target_aarch64<size,
> > big_endian>::Relocate::relocate_tls(
> > + }
> > + if (tlsopt == tls::TLSOPT_TO_IE)
> > + {
> > +- if (tls_segment == NULL)
> > +- {
> > +- gold_assert(parameters->errors()-
> > >error_count() > 0
> > +- ||
> > issue_undefined_symbol_error(gsym));
> > +- return aarch64_reloc_funcs::STATUS_BAD_RELOC;
> > +- }
> > + return tls_desc_gd_to_ie(relinfo, target, rela,
> > r_type,
> > + view, psymval,
> > got_entry_address,
> > + address);
> > +diff --git a/gold/i386.cc b/gold/i386.cc
> > +index 82886d4..a7168a8 100644
> > +--- a/gold/i386.cc
> > b/gold/i386.cc
> > +@@ -668,7 +668,6 @@ class Target_i386 : public Sized_target<32,
> > false>
> > + // D
[OE-core] [PATCH 3/4] Backporting binutils bug fix to the 2.25 branch
We fail to build webkit on aarch64 due to this binutils bug:
https://sourceware.org/bugzilla/show_bug.cgi?id=19353
Applying patch which fixes this, stripped out changelog entry
from patch to make it apply without error.
---
meta/recipes-devtools/binutils/binutils-2.25.1.inc | 1 +
...plying-TLSDESC-relocs-without-TLS-segment.patch | 166 +
2 files changed, 167 insertions(+)
create mode 100644
meta/recipes-devtools/binutils/binutils/fix-internal-error-when-applying-TLSDESC-relocs-without-TLS-segment.patch
diff --git a/meta/recipes-devtools/binutils/binutils-2.25.1.inc
b/meta/recipes-devtools/binutils/binutils-2.25.1.inc
index f3817fa..c7db9e0 100644
--- a/meta/recipes-devtools/binutils/binutils-2.25.1.inc
+++ b/meta/recipes-devtools/binutils/binutils-2.25.1.inc
@@ -35,6 +35,7 @@ SRC_URI = "\
file://0015-Fix-dynamic-list-so-that-symbols-not-in-the-list-are.patch \
file://binutils-octeon3.patch \
file://add-thunderx-support-for-gas.patch \
+
file://fix-internal-error-when-applying-TLSDESC-relocs-without-TLS-segment.patch
\
"
S = "${WORKDIR}/git"
diff --git
a/meta/recipes-devtools/binutils/binutils/fix-internal-error-when-applying-TLSDESC-relocs-without-TLS-segment.patch
b/meta/recipes-devtools/binutils/binutils/fix-internal-error-when-applying-TLSDESC-relocs-without-TLS-segment.patch
new file mode 100644
index 000..df9d54c
--- /dev/null
+++
b/meta/recipes-devtools/binutils/binutils/fix-internal-error-when-applying-TLSDESC-relocs-without-TLS-segment.patch
@@ -0,0 +1,166 @@
+From d21f123b0ead1806416cf0dafae12bec4cca8920 Mon Sep 17 00:00:00 2001
+From: Cary Coutant
+Date: Mon, 11 Jan 2016 23:57:44 -0800
+Subject: [PATCH] Fix internal error when applying TLSDESC relocations with no
TLS segment.
+
+gold/
+ PR gold/19353
+ * aarch64.cc (Target_aarch64::relocate_tls): Don't insist that
+ we have a TLS segment for GD-to-IE optimization.
+ * i386.cc (Target_i386::tls_gd_to_ie): Remove tls_segment parameter.
+ Adjust all calls.
+ (Target_i386::tls_desc_gd_to_ie): Likewise.
+ (Target_i386::relocate_tls): Don't insist that we have a TLS segment
+ for TLSDESC GD-to-IE optimizations.
+ * x86_64.cc (Target_x86_64::tls_gd_to_ie): Remove tls_segment parameter.
+ Adjust all calls.
+ (Target_x86_64::tls_desc_gd_to_ie): Likewise.
+ (Target_x86_64::relocate_tls): Don't insist that we have a TLS segment
+ for TLSDESC GD-to-IE optimizations.
+---
+ gold/aarch64.cc |6 --
+ gold/i386.cc| 14 ++
+ gold/x86_64.cc | 14 ++
+ 4 files changed, 20 insertions(+), 30 deletions(-)
+
+diff --git a/gold/aarch64.cc b/gold/aarch64.cc
+index 20f2f4f..5ad061b 100644
+--- a/gold/aarch64.cc
b/gold/aarch64.cc
+@@ -7422,12 +7422,6 @@ Target_aarch64::Relocate::relocate_tls(
+ }
+ if (tlsopt == tls::TLSOPT_TO_IE)
+ {
+- if (tls_segment == NULL)
+-{
+- gold_assert(parameters->errors()->error_count() > 0
+- || issue_undefined_symbol_error(gsym));
+- return aarch64_reloc_funcs::STATUS_BAD_RELOC;
+-}
+ return tls_desc_gd_to_ie(relinfo, target, rela, r_type,
+view, psymval, got_entry_address,
+address);
+diff --git a/gold/i386.cc b/gold/i386.cc
+index 82886d4..a7168a8 100644
+--- a/gold/i386.cc
b/gold/i386.cc
+@@ -668,7 +668,6 @@ class Target_i386 : public Sized_target<32, false>
+ // Do a TLS General-Dynamic to Initial-Exec transition.
+ inline void
+ tls_gd_to_ie(const Relocate_info<32, false>*, size_t relnum,
+- Output_segment* tls_segment,
+const elfcpp::Rel<32, false>&, unsigned int r_type,
+elfcpp::Elf_types<32>::Elf_Addr value,
+unsigned char* view,
+@@ -687,7 +686,6 @@ class Target_i386 : public Sized_target<32, false>
+ // transition.
+ inline void
+ tls_desc_gd_to_ie(const Relocate_info<32, false>*, size_t relnum,
+-Output_segment* tls_segment,
+ const elfcpp::Rel<32, false>&, unsigned int r_type,
+ elfcpp::Elf_types<32>::Elf_Addr value,
+ unsigned char* view,
+@@ -3054,7 +3052,7 @@ Target_i386::Relocate::relocate_tls(const
Relocate_info<32, false>* relinfo,
+ }
+ if (optimized_type == tls::TLSOPT_TO_IE)
+ {
+-this->tls_gd_to_ie(relinfo, relnum, tls_segment, rel, r_type,
++this->tls_gd_to_ie(relinfo, relnum, rel, r_type,
+got_offset, view, view_size);
+ break;
+ }
+@@ -3116,13 +3114,7 @@ Target_i386::Relocate::relocate_tls(const
Relocate_info<32, false>* relinfo,
+ }
+ if (optimized_type ==
[OE-core] [PATCH 1/4] gcc-cross-initial: Split all-gcc and configure-target-libgcc
Serializing this fixes build on aarch64
---
meta/recipes-devtools/gcc/gcc-cross-initial.inc | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/meta/recipes-devtools/gcc/gcc-cross-initial.inc
b/meta/recipes-devtools/gcc/gcc-cross-initial.inc
index c0fa139..98aae45 100644
--- a/meta/recipes-devtools/gcc/gcc-cross-initial.inc
+++ b/meta/recipes-devtools/gcc/gcc-cross-initial.inc
@@ -43,7 +43,8 @@ do_configure_prepend () {
}
do_compile () {
-oe_runmake all-gcc configure-target-libgcc
+oe_runmake all-gcc
+oe_runmake configure-target-libgcc
}
do_install () {
( cd ${B}/${TARGET_SYS}/libgcc; oe_runmake 'DESTDIR=${D}'
install-unwind_h )
--
2.8.0.rc3
--
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core
[OE-core] [meta-oe][PATCH] cross-localedef-native_2.22.bb: Use autotools configure
Use the autotools default configure commands and just tell autotools
where to run configure from.
This fixes the build when running on an aarch64 host, which the prebuilt
configure scripts with glibc 2.22 do not recognize.
---
meta/recipes-core/glibc/cross-localedef-native_2.22.bb | 9 -
1 file changed, 4 insertions(+), 5 deletions(-)
diff --git a/meta/recipes-core/glibc/cross-localedef-native_2.22.bb
b/meta/recipes-core/glibc/cross-localedef-native_2.22.bb
index 3aefe74..7e484fa 100644
--- a/meta/recipes-core/glibc/cross-localedef-native_2.22.bb
+++ b/meta/recipes-core/glibc/cross-localedef-native_2.22.bb
@@ -9,6 +9,10 @@ LIC_FILES_CHKSUM =
"file://LICENSES;md5=e9a558e243b36d3209f380deb394b213 \
file://COPYING.LIB;md5=4fbd65380cdd255951079008b364516c"
+# Tell autotools that we're working in the localedef directory
+#
+AUTOTOOLS_SCRIPT_PATH = "${S}/localedef"
+
inherit native
inherit autotools
@@ -49,11 +53,6 @@ S = "${WORKDIR}/git"
EXTRA_OECONF = "--with-glibc=${S}"
CFLAGS += "-fgnu89-inline -std=gnu99 -DIS_IN\(x\)='0'"
-do_configure () {
- ${S}/localedef/configure ${EXTRA_OECONF}
-}
-
-
do_install() {
install -d ${D}${bindir}
install -m 0755 ${B}/localedef ${D}${bindir}/cross-localedef
--
2.8.0.rc3
--
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core
Re: [OE-core] [PATCH] cross-localedef-native_2.22.bb: Use autotools configure
Hi,
This patch is also discussed in bug 9508:
https://bugzilla.yoctoproject.org/show_bug.cgi?id=9508
Sorry for messing up the subject line, it should have been [OE-core],
this is just my first git send-email and I copy pasted from:
http://www.openembedded.org/wiki/How_to_submit_a_patch_to_OpenEmbedded
Cheers,
-Tristan
On Mon, 2016-04-25 at 16:36 -0400, Tristan Van Berkom wrote:
> Use the autotools default configure commands and just tell autotools
> where to run configure from.
>
> This fixes the build when running on an aarch64 host, which the
> prebuilt
> configure scripts with glibc 2.22 do not recognize.
> ---
> meta/recipes-core/glibc/cross-localedef-native_2.22.bb | 9 -
> 1 file changed, 4 insertions(+), 5 deletions(-)
>
> diff --git a/meta/recipes-core/glibc/cross-localedef-native_2.22.bb
> b/meta/recipes-core/glibc/cross-localedef-native_2.22.bb
> index 3aefe74..7e484fa 100644
> --- a/meta/recipes-core/glibc/cross-localedef-native_2.22.bb
> +++ b/meta/recipes-core/glibc/cross-localedef-native_2.22.bb
> @@ -9,6 +9,10 @@ LIC_FILES_CHKSUM = "file://LICENSES;md5=e9a558e243b3
> 6d3209f380deb394b213 \
> file://COPYING.LIB;md5=4fbd65380cdd255951079008b364516c"
>
>
> +# Tell autotools that we're working in the localedef directory
> +#
> +AUTOTOOLS_SCRIPT_PATH = "${S}/localedef"
> +
> inherit native
> inherit autotools
>
> @@ -49,11 +53,6 @@ S = "${WORKDIR}/git"
> EXTRA_OECONF = "--with-glibc=${S}"
> CFLAGS += "-fgnu89-inline -std=gnu99 -DIS_IN\(x\)='0'"
>
> -do_configure () {
> - ${S}/localedef/configure ${EXTRA_OECONF}
> -}
> -
> -
> do_install() {
> install -d ${D}${bindir}
> install -m 0755 ${B}/localedef ${D}${bindir}/cross-localedef
--
___
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.openembedded.org/mailman/listinfo/openembedded-core
