[RFT] binutils 2.27

[RFT] binutils 2.27

[Khem Raj]

[-- Attachment #1: Type: text/plain, Size: 342 bytes --]

Hi All

I have put together upgrade for binutils 2.27 here ( kraj/master branch )

https://github.com/kraj/openembedded-core/commit/b98f5761d2285ffd773ec6ecb901b799302ebd6b

I have started bunch of test builds. but I would appreciate any help in testing it out in your
environments and report any issues you see.

Thanks

-Khem


[-- Attachment #2: Message signed with OpenPGP using GPGMail --]
[-- Type: application/pgp-signature, Size: 211 bytes --]

Re: [RFT] binutils 2.27

[Richard Purdie]

On Sun, 2016-08-07 at 01:39 -0700, Khem Raj wrote:
> I have put together upgrade for binutils 2.27 here ( kraj/master
> branch )
> 
> https://github.com/kraj/openembedded-core/commit/b98f5761d2285ffd773e
> c6ecb901b799302ebd6b
> 
> I have started bunch of test builds. but I would appreciate any help
> in testing it out in your
> environments and report any issues you see.

I ran this through the autobuilder:

bitbake world shows issues in libunwind (from gold?):

https://autobuilder.yoctoproject.org/main/builders/nightly-world/builds
/889/steps/BuildImages/logs/stdio

bitbake world with no x11 also shows it:

https://autobuilder.yoctoproject.org/main/builders/nightly-no-x11/build
s/231/steps/BuildImages/logs/stdio

Other than that, things don't seem too bad though.

Cheers,

Richard

Re: [RFT] binutils 2.27

[Khem Raj]

[-- Attachment #1: Type: text/plain, Size: 1368 bytes --]


> On Aug 7, 2016, at 11:28 PM, Richard Purdie <richard.purdie@linuxfoundation.org> wrote:
> 
> On Sun, 2016-08-07 at 01:39 -0700, Khem Raj wrote:
>> I have put together upgrade for binutils 2.27 here ( kraj/master
>> branch )
>> 
>> https://github.com/kraj/openembedded-core/commit/b98f5761d2285ffd773e
>> c6ecb901b799302ebd6b
>> 
>> I have started bunch of test builds. but I would appreciate any help
>> in testing it out in your
>> environments and report any issues you see.
> 
> I ran this through the autobuilder:
> 
> bitbake world shows issues in libunwind (from gold?):
> 

yes, however, use of gold for libunwind in OE was an escape route
where we were relying upon gold’s inability to detect this issue while BFD linker
was detecting it earlier. Now in 2.27 gold has caught up as well.

see
https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=6eeb0170bbb43ffb73e8f01b8b481adde8194c21

so now we have to get to nub of this issue to resolve it.

> https://autobuilder.yoctoproject.org/main/builders/nightly-world/builds
> /889/steps/BuildImages/logs/stdio
> 
> bitbake world with no x11 also shows it:
> 
> https://autobuilder.yoctoproject.org/main/builders/nightly-no-x11/build
> s/231/steps/BuildImages/logs/stdio
> 
> Other than that, things don't seem too bad though.
> 
> Cheers,
> 
> Richard
> 
> 


[-- Attachment #2: Message signed with OpenPGP using GPGMail --]
[-- Type: application/pgp-signature, Size: 211 bytes --]

Re: [RFT] binutils 2.27

[Richard Purdie]

On Mon, 2016-08-08 at 00:24 -0700, Khem Raj wrote:
> > On Aug 7, 2016, at 11:28 PM, Richard Purdie <
> > richard.purdie@linuxfoundation.org> wrote:
> > 
> > On Sun, 2016-08-07 at 01:39 -0700, Khem Raj wrote:
> > > I have put together upgrade for binutils 2.27 here ( kraj/master
> > > branch )
> > > 
> > > https://github.com/kraj/openembedded-core/commit/b98f5761d2285ffd
> > > 773e
> > > c6ecb901b799302ebd6b
> > > 
> > > I have started bunch of test builds. but I would appreciate any
> > > help
> > > in testing it out in your
> > > environments and report any issues you see.
> > 
> > I ran this through the autobuilder:
> > 
> > bitbake world shows issues in libunwind (from gold?):
> > 
> 
> yes, however, use of gold for libunwind in OE was an escape route
> where we were relying upon gold’s inability to detect this issue
> while BFD linker
> was detecting it earlier. Now in 2.27 gold has caught up as well.
> 
> see
> https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=6eeb0170bb
> b43ffb73e8f01b8b481adde8194c21
> 
> so now we have to get to nub of this issue to resolve it.

Ok. I think we also have:

https://autobuilder.yoctoproject.org/main/builders/nightly-ppc/builds/8
96
https://autobuilder.yoctoproject.org/main/builders/nightly-ppc/builds/8
96

in that "connmand --help" is segfaulting on ppc. I've not 100%
confirmed its the binutils patch but its seeming likely at this point.

Cheers,

Richard

Re: [RFT] binutils 2.27

[Richard Purdie]

On Mon, 2016-08-08 at 09:14 +0100, Richard Purdie wrote:
> On Mon, 2016-08-08 at 00:24 -0700, Khem Raj wrote:
> > > On Aug 7, 2016, at 11:28 PM, Richard Purdie <
> > > richard.purdie@linuxfoundation.org> wrote:
> > > 
> > > On Sun, 2016-08-07 at 01:39 -0700, Khem Raj wrote:
> > > > I have put together upgrade for binutils 2.27 here (
> > > > kraj/master
> > > > branch )
> > > > 
> > > > https://github.com/kraj/openembedded-core/commit/b98f5761d2285f
> > > > fd
> > > > 773e
> > > > c6ecb901b799302ebd6b
> > > > 
> > > > I have started bunch of test builds. but I would appreciate any
> > > > help
> > > > in testing it out in your
> > > > environments and report any issues you see.
> > > 
> > > I ran this through the autobuilder:
> > > 
> > > bitbake world shows issues in libunwind (from gold?):
> > > 
> > 
> > yes, however, use of gold for libunwind in OE was an escape route
> > where we were relying upon gold’s inability to detect this issue
> > while BFD linker
> > was detecting it earlier. Now in 2.27 gold has caught up as well.
> > 
> > see
> > https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=6eeb0170
> > bb
> > b43ffb73e8f01b8b481adde8194c21
> > 
> > so now we have to get to nub of this issue to resolve it.
> 
> Ok. I think we also have:
> 
> https://autobuilder.yoctoproject.org/main/builders/nightly-ppc/builds
> /8
> 96
> https://autobuilder.yoctoproject.org/main/builders/nightly-ppc/builds
> /8
> 96
> 
> in that "connmand --help" is segfaulting on ppc. I've not 100%
> confirmed its the binutils patch but its seeming likely at this
> point.

Its also sefgaulting on all mips builds:

https://autobuilder.yoctoproject.org/main/builders/nightly-mips/builds/883
https://autobuilder.yoctoproject.org/main/builders/nightly-mips-lsb/builds/863
https://autobuilder.yoctoproject.org/main/builders/nightly-mips64/builds/528

:(

Cheers,

Richard

Re: [RFT] binutils 2.27

[André Draszik]

On Mon, 2016-08-08 at 09:50 +0100, Richard Purdie wrote:
> On Mon, 2016-08-08 at 09:14 +0100, Richard Purdie wrote:
> > 
> > On Mon, 2016-08-08 at 00:24 -0700, Khem Raj wrote:
> > > 
> > > > 
> > > > On Aug 7, 2016, at 11:28 PM, Richard Purdie <
> > > > richard.purdie@linuxfoundation.org> wrote:
> > > > 
> > > > On Sun, 2016-08-07 at 01:39 -0700, Khem Raj wrote:
> > > > > 
> > > > > I have put together upgrade for binutils 2.27 here (
> > > > > kraj/master
> > > > > branch )
> > > > > 
> > > > > https://github.com/kraj/openembedded-core/commit/b98f5761d2285f
> > > > > fd
> > > > > 773e
> > > > > c6ecb901b799302ebd6b
> > > > > 
> > > > > I have started bunch of test builds. but I would appreciate any
> > > > > help
> > > > > in testing it out in your
> > > > > environments and report any issues you see.
> > > > 
> > > > I ran this through the autobuilder:
> > > > 
> > > > bitbake world shows issues in libunwind (from gold?):
> > > > 
> > > 
> > > yes, however, use of gold for libunwind in OE was an escape route
> > > where we were relying upon gold’s inability to detect this issue
> > > while BFD linker
> > > was detecting it earlier. Now in 2.27 gold has caught up as well.
> > > 
> > > see
> > > https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=6eeb0170
> > > bb
> > > b43ffb73e8f01b8b481adde8194c21
> > > 
> > > so now we have to get to nub of this issue to resolve it.
> > 
> > Ok. I think we also have:
> > 
> > https://autobuilder.yoctoproject.org/main/builders/nightly-ppc/builds
> > /8
> > 96
> > https://autobuilder.yoctoproject.org/main/builders/nightly-ppc/builds
> > /8
> > 96
> > 
> > in that "connmand --help" is segfaulting on ppc. I've not 100%
> > confirmed its the binutils patch but its seeming likely at this
> > point.
> 
> Its also sefgaulting on all mips builds:
> 
> https://autobuilder.yoctoproject.org/main/builders/nightly-mips/builds/883
> https://autobuilder.yoctoproject.org/main/builders/nightly-mips-lsb/builds
> /863
> https://autobuilder.yoctoproject.org/main/builders/nightly-mips64/builds/5
> 28
> 
> :(

binutils 2.27 brings GOLD to mips, but debian has disabled GOLD on all MIPS
for now due to segfaulting... Maybe the same issue...


A.

Re: [RFT] binutils 2.27

[Khem Raj]

[-- Attachment #1: Type: text/plain, Size: 2393 bytes --]


> On Aug 8, 2016, at 2:16 AM, André Draszik <git@andred.net> wrote:
> 
> On Mon, 2016-08-08 at 09:50 +0100, Richard Purdie wrote:
>> On Mon, 2016-08-08 at 09:14 +0100, Richard Purdie wrote:
>>> 
>>> On Mon, 2016-08-08 at 00:24 -0700, Khem Raj wrote:
>>>> 
>>>>> 
>>>>> On Aug 7, 2016, at 11:28 PM, Richard Purdie <
>>>>> richard.purdie@linuxfoundation.org> wrote:
>>>>> 
>>>>> On Sun, 2016-08-07 at 01:39 -0700, Khem Raj wrote:
>>>>>> 
>>>>>> I have put together upgrade for binutils 2.27 here (
>>>>>> kraj/master
>>>>>> branch )
>>>>>> 
>>>>>> https://github.com/kraj/openembedded-core/commit/b98f5761d2285f
>>>>>> fd
>>>>>> 773e
>>>>>> c6ecb901b799302ebd6b
>>>>>> 
>>>>>> I have started bunch of test builds. but I would appreciate any
>>>>>> help
>>>>>> in testing it out in your
>>>>>> environments and report any issues you see.
>>>>> 
>>>>> I ran this through the autobuilder:
>>>>> 
>>>>> bitbake world shows issues in libunwind (from gold?):
>>>>> 
>>>> 
>>>> yes, however, use of gold for libunwind in OE was an escape route
>>>> where we were relying upon gold’s inability to detect this issue
>>>> while BFD linker
>>>> was detecting it earlier. Now in 2.27 gold has caught up as well.
>>>> 
>>>> see
>>>> https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=6eeb0170
>>>> bb
>>>> b43ffb73e8f01b8b481adde8194c21
>>>> 
>>>> so now we have to get to nub of this issue to resolve it.
>>> 
>>> Ok. I think we also have:
>>> 
>>> https://autobuilder.yoctoproject.org/main/builders/nightly-ppc/builds
>>> /8
>>> 96
>>> https://autobuilder.yoctoproject.org/main/builders/nightly-ppc/builds
>>> /8
>>> 96
>>> 
>>> in that "connmand --help" is segfaulting on ppc. I've not 100%
>>> confirmed its the binutils patch but its seeming likely at this
>>> point.
>> 
>> Its also sefgaulting on all mips builds:
>> 
>> https://autobuilder.yoctoproject.org/main/builders/nightly-mips/builds/883
>> https://autobuilder.yoctoproject.org/main/builders/nightly-mips-lsb/builds
>> /863
>> https://autobuilder.yoctoproject.org/main/builders/nightly-mips64/builds/5
>> 28
>> 
>> :(
> 
> binutils 2.27 brings GOLD to mips, but debian has disabled GOLD on all MIPS
> for now due to segfaulting... Maybe the same issue…

They are different issues. for OE ppc also shows same errors with libunwind.

> 
> 
> A.
> 
> 


[-- Attachment #2: Message signed with OpenPGP using GPGMail --]
[-- Type: application/pgp-signature, Size: 211 bytes --]

Re: [RFT] binutils 2.27

[Richard Purdie]

On Mon, 2016-08-08 at 08:21 -0700, Khem Raj wrote:
> They are different issues. for OE ppc also shows same errors with lib
> unwind.

FWIW, I think the connmand segfault on mips and powerpc is hinted at
with this bug:

https://bugs.launchpad.net/ubuntu/+source/lua5.3/+bug/1570055

connmand is a binary linked with a linker version-script. If I build
connmand without that option, it doesn't segfault, at least on mips.

The binutils manuals say you shouldn't build binaries with version
scripts:

http://ftp.gnu.org/old-gnu/Manuals/ld-2.9.1/html_node/ld_25.html 

Its suspected connman might do this to limit the function exposure to
its plugins.

Now sure exactly what needs to happen here without more research but it
certainly hits at where the problem is. I suspect its the same issue on
ppc.

Cheers,

Richard

Re: [RFT] binutils 2.27

[Burton, Ross]

[-- Attachment #1: Type: text/plain, Size: 1074 bytes --]

On 8 August 2016 at 16:59, Richard Purdie <
richard.purdie@linuxfoundation.org> wrote:

> FWIW, I think the connmand segfault on mips and powerpc is hinted at
> with this bug:
>
> https://bugs.launchpad.net/ubuntu/+source/lua5.3/+bug/1570055
>
> connmand is a binary linked with a linker version-script. If I build
> connmand without that option, it doesn't segfault, at least on mips.
>
> The binutils manuals say you shouldn't build binaries with version
> scripts:
>
> http://ftp.gnu.org/old-gnu/Manuals/ld-2.9.1/html_node/ld_25.html
>
> Its suspected connman might do this to limit the function exposure to
> its plugins.
>
> Now sure exactly what needs to happen here without more research but it
> certainly hits at where the problem is. I suspect its the same issue on
> ppc.
>

I've verified that with binutils 2.26 on x86-64, a minimal test case for an
executable using a version script to limit exported symbols to dlopen()d
modules does in fact work.  A toolchain is building now to see what happens
in different environments.

Ross

[-- Attachment #2: Type: text/html, Size: 1782 bytes --]

Re: [RFT] binutils 2.27

[Khem Raj]

[-- Attachment #1.1: Type: text/plain, Size: 1544 bytes --]


> On Aug 8, 2016, at 4:16 PM, Burton, Ross <ross.burton@intel.com> wrote:
> 
> 
> On 8 August 2016 at 16:59, Richard Purdie <richard.purdie@linuxfoundation.org <mailto:richard.purdie@linuxfoundation.org>> wrote:
> FWIW, I think the connmand segfault on mips and powerpc is hinted at
> with this bug:
> 
> https://bugs.launchpad.net/ubuntu/+source/lua5.3/+bug/1570055 <https://bugs.launchpad.net/ubuntu/+source/lua5.3/+bug/1570055>
> 
> connmand is a binary linked with a linker version-script. If I build
> connmand without that option, it doesn't segfault, at least on mips.
> 
> The binutils manuals say you shouldn't build binaries with version
> scripts:
> 
> http://ftp.gnu.org/old-gnu/Manuals/ld-2.9.1/html_node/ld_25.html <http://ftp.gnu.org/old-gnu/Manuals/ld-2.9.1/html_node/ld_25.html>
> 
> Its suspected connman might do this to limit the function exposure to
> its plugins.
> 
> Now sure exactly what needs to happen here without more research but it
> certainly hits at where the problem is. I suspect its the same issue on
> ppc.
> 
> I've verified that with binutils 2.26 on x86-64, a minimal test case for an executable using a version script to limit exported symbols to dlopen()d modules does in fact work.  A toolchain is building now to see what happens in different environments.

I could also see it on ppc. backtrace, shows the segfault is in exit path and happens in libc
at this point, I think the problem is how libc is compiled with binutils 2.27, connman itself
is ok.

> 
> Ross


[-- Attachment #1.2: Type: text/html, Size: 2920 bytes --]

[-- Attachment #2: Message signed with OpenPGP using GPGMail --]
[-- Type: application/pgp-signature, Size: 211 bytes --]

Re: [RFT] binutils 2.27

[Richard Purdie]

On Tue, 2016-08-09 at 07:42 -0700, Khem Raj wrote:
> I could also see it on ppc. backtrace, shows the segfault is in exit
> path and happens in libc
> at this point, I think the problem is how libc is compiled with
> binutils 2.27, connman itself
> is ok. 

Its the issue here:

https://sourceware.org/ml/glibc-bugs/2015-01/msg00274.html
https://sourceware.org/bugzilla/show_bug.cgi?id=17908

Basically, if you remove the global _IO_stdin_used symbol, it triggers
compatibility code which crashes.

I've confirmed that if I add that symbol to the version-script in
connman, things work again.

Any idea how we raise the priority of this issue. There are no comments
on the bug despite it having been posted a while ago :(.

Cheers,

Richard

Re: [oe] [RFT] binutils 2.27

[Martin Jansa]

[-- Attachment #1: Type: text/plain, Size: 3648 bytes --]

I'm not sure if it's caused by this binutils upgrade (because openssl in
oe-core is also using version script, imported from debian) or some change
in cmake/openssl, but today all recipes using cmake are failing with:

| DEBUG: Executing shell function do_configure
| cmake: sysroots/x86_64-linux/usr/bin/../lib/libcrypto.so.1.0.0: version
`OPENSSL_1.0.0' not found (required by cmake)
| cmake: sysroots/x86_64-linux/usr/bin/../lib/libssl.so.1.0.0: version
`OPENSSL_1.0.1' not found (required by cmake)
| cmake: sysroots/x86_64-linux/usr/bin/../lib/libssl.so.1.0.0: version
`OPENSSL_1.0.0' not found (required by cmake)
| WARNING: exit code 1 from a shell command.

Maybe it's another symptom of missing openssl-native dependency in
cmake-native (just like libidn-native issue reported here
https://bugzilla.yoctoproject.org/show_bug.cgi?id=9639) - and cmake-native
was originally linked against host's openssl which was built with version
script and binutils-2.24 (Ubuntu-14.04), but then it's executed with libssl
and libcrypto from native sysroot, where the version script didn't work:

$ ldd sysroots/x86_64-linux/usr/bin/cmake
sysroots/x86_64-linux/usr/bin/cmake:
sysroots/x86_64-linux/usr/bin/../lib/libcrypto.so.1.0.0: version
`OPENSSL_1.0.0' not found (required by sysroots/x86_64-linux/usr/bin/cmake)
sysroots/x86_64-linux/usr/bin/cmake:
sysroots/x86_64-linux/usr/bin/../lib/libssl.so.1.0.0: version
`OPENSSL_1.0.1' not found (required by sysroots/x86_64-linux/usr/bin/cmake)
sysroots/x86_64-linux/usr/bin/cmake:
sysroots/x86_64-linux/usr/bin/../lib/libssl.so.1.0.0: version
`OPENSSL_1.0.0' not found (required by sysroots/x86_64-linux/usr/bin/cmake)
        linux-vdso.so.1 =>  (0x00007ffed2fce000)
        libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x00007fb631276000)
        libssl.so.1.0.0 =>
sysroots/x86_64-linux/usr/bin/../lib/libssl.so.1.0.0 (0x00007fb63100c000)
        libcrypto.so.1.0.0 =>
sysroots/x86_64-linux/usr/bin/../lib/libcrypto.so.1.0.0 (0x00007fb630bc1000)
        libstdc++.so.6 => /usr/lib/x86_64-linux-gnu/libstdc++.so.6
(0x00007fb6308bd000)
        libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x00007fb6305b7000)
        libgcc_s.so.1 => /lib/x86_64-linux-gnu/libgcc_s.so.1
(0x00007fb6303a1000)
        libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007fb62ffdc000)
        /lib64/ld-linux-x86-64.so.2 (0x00007fb63147a000)

Which is sort of confirmed by removing libssl.so and libcrypto.so from
native sysroot, I can execute cmake binary again.

On Tue, Aug 9, 2016 at 5:24 PM, Richard Purdie <
richard.purdie@linuxfoundation.org> wrote:

> On Tue, 2016-08-09 at 07:42 -0700, Khem Raj wrote:
> > I could also see it on ppc. backtrace, shows the segfault is in exit
> > path and happens in libc
> > at this point, I think the problem is how libc is compiled with
> > binutils 2.27, connman itself
> > is ok.
>
> Its the issue here:
>
> https://sourceware.org/ml/glibc-bugs/2015-01/msg00274.html
> https://sourceware.org/bugzilla/show_bug.cgi?id=17908
>
> Basically, if you remove the global _IO_stdin_used symbol, it triggers
> compatibility code which crashes.
>
> I've confirmed that if I add that symbol to the version-script in
> connman, things work again.
>
> Any idea how we raise the priority of this issue. There are no comments
> on the bug despite it having been posted a while ago :(.
>
> Cheers,
>
> Richard
> --
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.openembedded.org/mailman/listinfo/openembedded-devel
>

[-- Attachment #2: Type: text/html, Size: 4941 bytes --]

Re: [RFT] binutils 2.27

[Khem Raj]

[-- Attachment #1: Type: text/plain, Size: 1192 bytes --]


> On Aug 7, 2016, at 11:28 PM, Richard Purdie <richard.purdie@linuxfoundation.org> wrote:
> 
> On Sun, 2016-08-07 at 01:39 -0700, Khem Raj wrote:
>> I have put together upgrade for binutils 2.27 here ( kraj/master
>> branch )
>> 
>> https://github.com/kraj/openembedded-core/commit/b98f5761d2285ffd773e
>> c6ecb901b799302ebd6b
>> 
>> I have started bunch of test builds. but I would appreciate any help
>> in testing it out in your
>> environments and report any issues you see.
> 
> I ran this through the autobuilder:
> 
> bitbake world shows issues in libunwind (from gold?):
> 
> https://autobuilder.yoctoproject.org/main/builders/nightly-world/builds
> /889/steps/BuildImages/logs/stdio
> 
> bitbake world with no x11 also shows it:
> 
> https://autobuilder.yoctoproject.org/main/builders/nightly-no-x11/build
> s/231/steps/BuildImages/logs/stdio
> 
> Other than that, things don't seem too bad though.
> 


I posted couple of patches for libunwind and webkitgtk recipes which should take care of build issues.

https://patchwork.openembedded.org/patch/128733/
https://patchwork.openembedded.org/patch/128731/



> Cheers,
> 
> Richard
> 
> 


[-- Attachment #2: Message signed with OpenPGP using GPGMail --]
[-- Type: application/pgp-signature, Size: 211 bytes --]