Patchwork [RFC] tcmode-default: Set 4.8 as default

login
register
mail settings
Submitter Saul Wold
Date May 12, 2013, 11:05 a.m.
Message ID <1368356739-17625-1-git-send-email-sgw@linux.intel.com>
Download mbox | patch
Permalink /patch/49775/
State Accepted
Commit 7950a307bc7d4104e6cfb09bb2ea267c5da83f2a
Headers show

Comments

Saul Wold - May 12, 2013, 11:05 a.m.
This changes the default GCC to 4.8, it has been tested with OE-Core
and a number of layers have been testing with it for the last few weeks.

There are some minor issues outside of OE-Core. These will be addressed
with time.

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Cc: Khem Raj <raj.khem@gmail.com>

---
 meta/conf/distro/include/tcmode-default.inc | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Khem Raj - May 12, 2013, 4:29 p.m.
Hi Saul

There is one serious problem that I ran into was that qemuarm 3.8 kernel segfaulted
is it working for you ?

-Khem

On May 12, 2013, at 4:05 AM, Saul Wold <sgw@linux.intel.com> wrote:

> This changes the default GCC to 4.8, it has been tested with OE-Core
> and a number of layers have been testing with it for the last few weeks.
> 
> There are some minor issues outside of OE-Core. These will be addressed
> with time.
> 
> Signed-off-by: Saul Wold <sgw@linux.intel.com>
> Cc: Khem Raj <raj.khem@gmail.com>
> 
> ---
> meta/conf/distro/include/tcmode-default.inc | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/meta/conf/distro/include/tcmode-default.inc b/meta/conf/distro/include/tcmode-default.inc
> index 88d109b..f3942b5 100644
> --- a/meta/conf/distro/include/tcmode-default.inc
> +++ b/meta/conf/distro/include/tcmode-default.inc
> @@ -17,7 +17,7 @@ PREFERRED_PROVIDER_virtual/${TARGET_PREFIX}libc-initial = "${TCLIBC}-initial"
> PREFERRED_PROVIDER_virtual/nativesdk-${SDK_PREFIX}libc-initial = "nativesdk-${TCLIBC}-initial"
> PREFERRED_PROVIDER_virtual/gettext ??= "gettext"
> 
> -GCCVERSION ?= "4.7%"
> +GCCVERSION ?= "4.8%"
> SDKGCCVERSION ?= "${GCCVERSION}"
> BINUVERSION ?= "2.23.2"
> EGLIBCVERSION ?= "2.17"
> -- 
> 1.8.0.2
>
Saul Wold - May 12, 2013, 5:35 p.m.
On 05/12/2013 09:29 AM, Khem Raj wrote:
> Hi Saul
>
> There is one serious problem that I ran into was that qemuarm 3.8 kernel segfaulted
> is it working for you ?
>
Nope your right there does appear to be a problem with ARM, I thought we 
where have sanity test issues again, I just completed a build and 
verified that I get a Kernel paninc with 3.8 Kernel and 4.8 GCC.

Not ready yet!

Khem, are you looking into it?  I have CC'ed Bruce for the kernel also.

Sau!


> -Khem
>
> On May 12, 2013, at 4:05 AM, Saul Wold <sgw@linux.intel.com> wrote:
>
>> This changes the default GCC to 4.8, it has been tested with OE-Core
>> and a number of layers have been testing with it for the last few weeks.
>>
>> There are some minor issues outside of OE-Core. These will be addressed
>> with time.
>>
>> Signed-off-by: Saul Wold <sgw@linux.intel.com>
>> Cc: Khem Raj <raj.khem@gmail.com>
>>
>> ---
>> meta/conf/distro/include/tcmode-default.inc | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/meta/conf/distro/include/tcmode-default.inc b/meta/conf/distro/include/tcmode-default.inc
>> index 88d109b..f3942b5 100644
>> --- a/meta/conf/distro/include/tcmode-default.inc
>> +++ b/meta/conf/distro/include/tcmode-default.inc
>> @@ -17,7 +17,7 @@ PREFERRED_PROVIDER_virtual/${TARGET_PREFIX}libc-initial = "${TCLIBC}-initial"
>> PREFERRED_PROVIDER_virtual/nativesdk-${SDK_PREFIX}libc-initial = "nativesdk-${TCLIBC}-initial"
>> PREFERRED_PROVIDER_virtual/gettext ??= "gettext"
>>
>> -GCCVERSION ?= "4.7%"
>> +GCCVERSION ?= "4.8%"
>> SDKGCCVERSION ?= "${GCCVERSION}"
>> BINUVERSION ?= "2.23.2"
>> EGLIBCVERSION ?= "2.17"
>> --
>> 1.8.0.2
>>
>
Khem Raj - May 12, 2013, 9:49 p.m.
On May 12, 2013, at 10:35 AM, Saul Wold <sgw@linux.intel.com> wrote:

> On 05/12/2013 09:29 AM, Khem Raj wrote:
>> Hi Saul
>> 
>> There is one serious problem that I ran into was that qemuarm 3.8 kernel segfaulted
>> is it working for you ?
>> 
> Nope your right there does appear to be a problem with ARM, I thought we where have sanity test issues again, I just completed a build and verified that I get a Kernel paninc with 3.8 Kernel and 4.8 GCC.
> 
> Not ready yet!
> 
> Khem, are you looking into it?  I have CC'ed Bruce for the kernel also.

haven't looked into the problem in detail. Its one of those either a problem in gcc or gcc uncovering a latent bug in kernel.
Bruce Ashfield - May 12, 2013, 10:23 p.m.
On 13-05-12 5:49 PM, Khem Raj wrote:
>
> On May 12, 2013, at 10:35 AM, Saul Wold <sgw@linux.intel.com> wrote:
>
>> On 05/12/2013 09:29 AM, Khem Raj wrote:
>>> Hi Saul
>>>
>>> There is one serious problem that I ran into was that qemuarm 3.8 kernel segfaulted
>>> is it working for you ?
>>>
>> Nope your right there does appear to be a problem with ARM, I thought we where have sanity test issues again, I just completed a build and verified that I get a Kernel paninc with 3.8 Kernel and 4.8 GCC.
>>
>> Not ready yet!
>>
>> Khem, are you looking into it?  I have CC'ed Bruce for the kernel also.
>
> haven't looked into the problem in detail. Its one of those either a problem in gcc or gcc uncovering a latent bug in kernel.

I'll get myself set up and do a build on Monday, see if anything obvious
pops up (i.e. start with the latest and see if it works, then bisect
back). I'm already poking at mips64, so I can do this in parallel.

Bruce

>
>
ml@communistcode.co.uk - May 13, 2013, 8:41 a.m.
On 12/05/13 23:23, Bruce Ashfield wrote:
> On 13-05-12 5:49 PM, Khem Raj wrote:
>>
>> On May 12, 2013, at 10:35 AM, Saul Wold <sgw@linux.intel.com> wrote:
>>
>>> On 05/12/2013 09:29 AM, Khem Raj wrote:
>>>> Hi Saul
>>>>
>>>> There is one serious problem that I ran into was that qemuarm 3.8 
>>>> kernel segfaulted
>>>> is it working for you ?
>>>>
>>> Nope your right there does appear to be a problem with ARM, I 
>>> thought we where have sanity test issues again, I just completed a 
>>> build and verified that I get a Kernel paninc with 3.8 Kernel and 
>>> 4.8 GCC.
>>>
>>> Not ready yet!
>>>
>>> Khem, are you looking into it?  I have CC'ed Bruce for the kernel also.
>>
>> haven't looked into the problem in detail. Its one of those either a 
>> problem in gcc or gcc uncovering a latent bug in kernel.
>
> I'll get myself set up and do a build on Monday, see if anything obvious
> pops up (i.e. start with the latest and see if it works, then bisect
> back). I'm already poking at mips64, so I can do this in parallel.
>
> Bruce
>
>>
>>
>

I know the beaglebone kernel was segfaulting on me until Koen updated to 
3.8.11, I think it included some important backports. Would definitely 
be worth a try, Koen can point to you to commit which fixed it, I seem 
to have lost the link into the IRC nether.

Cheers,
Koen Kooi - May 13, 2013, 9:29 a.m.
Op 13 mei 2013, om 10:41 heeft Jack Mitchell <ml@communistcode.co.uk> het volgende geschreven:

> On 12/05/13 23:23, Bruce Ashfield wrote:
>> On 13-05-12 5:49 PM, Khem Raj wrote:
>>> 
>>> On May 12, 2013, at 10:35 AM, Saul Wold <sgw@linux.intel.com> wrote:
>>> 
>>>> On 05/12/2013 09:29 AM, Khem Raj wrote:
>>>>> Hi Saul
>>>>> 
>>>>> There is one serious problem that I ran into was that qemuarm 3.8 kernel segfaulted
>>>>> is it working for you ?
>>>>> 
>>>> Nope your right there does appear to be a problem with ARM, I thought we where have sanity test issues again, I just completed a build and verified that I get a Kernel paninc with 3.8 Kernel and 4.8 GCC.
>>>> 
>>>> Not ready yet!
>>>> 
>>>> Khem, are you looking into it?  I have CC'ed Bruce for the kernel also.
>>> 
>>> haven't looked into the problem in detail. Its one of those either a problem in gcc or gcc uncovering a latent bug in kernel.
>> 
>> I'll get myself set up and do a build on Monday, see if anything obvious
>> pops up (i.e. start with the latest and see if it works, then bisect
>> back). I'm already poking at mips64, so I can do this in parallel.
>> 
>> Bruce
>> 
>>> 
>>> 
>> 
> 
> I know the beaglebone kernel was segfaulting on me until Koen updated to 3.8.11, I think it included some important backports. Would definitely be worth a try, Koen can point to you to commit which fixed it, I seem to have lost the link into the IRC nether.

I'm fairly sure this is the one: 
	
	http://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/commit/?id=84237e8d1b43b896a86e14928993901993ede254

I only tested the 3.8.10 -> 3.811 update in one go, so it could a different patch.

regars,

Koen
Bruce Ashfield - May 13, 2013, 12:31 p.m.
On Mon, May 13, 2013 at 4:41 AM, Jack Mitchell <ml@communistcode.co.uk> wrote:
> On 12/05/13 23:23, Bruce Ashfield wrote:
>>
>> On 13-05-12 5:49 PM, Khem Raj wrote:
>>>
>>>
>>> On May 12, 2013, at 10:35 AM, Saul Wold <sgw@linux.intel.com> wrote:
>>>
>>>> On 05/12/2013 09:29 AM, Khem Raj wrote:
>>>>>
>>>>> Hi Saul
>>>>>
>>>>> There is one serious problem that I ran into was that qemuarm 3.8
>>>>> kernel segfaulted
>>>>> is it working for you ?
>>>>>
>>>> Nope your right there does appear to be a problem with ARM, I thought we
>>>> where have sanity test issues again, I just completed a build and verified
>>>> that I get a Kernel paninc with 3.8 Kernel and 4.8 GCC.
>>>>
>>>> Not ready yet!
>>>>
>>>> Khem, are you looking into it?  I have CC'ed Bruce for the kernel also.
>>>
>>>
>>> haven't looked into the problem in detail. Its one of those either a
>>> problem in gcc or gcc uncovering a latent bug in kernel.
>>
>>
>> I'll get myself set up and do a build on Monday, see if anything obvious
>> pops up (i.e. start with the latest and see if it works, then bisect
>> back). I'm already poking at mips64, so I can do this in parallel.
>>
>> Bruce
>>
>>>
>>>
>>
>
> I know the beaglebone kernel was segfaulting on me until Koen updated to
> 3.8.11, I think it included some important backports. Would definitely be
> worth a try, Koen can point to you to commit which fixed it, I seem to have
> lost the link into the IRC nether.

Luckily I have the latest 3.8.x ready to go, and we are already on 3.8.11, so
as Koen indicated in his follow up, it may be a patch in 3.8.11 + something
else. I'll start with my latest 3.8 tree, and then use 3.9 and 3.10 respectively
until I can find the patch.

That being said, if anyone else bisects and finds the fix, I'll be
happy to merge
a change, or take the confirmation that it's in the pending -stable update.

Bruce

>
> Cheers,
>
> --
>
>   Jack Mitchell (jack@embed.me.uk)
>   Embedded Systems Engineer
>   http://www.embed.me.uk
>
> --
>
>
>
> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core



--
"Thou shalt not follow the NULL pointer, for chaos and madness await
thee at its end"
Martin Jansa - May 19, 2013, 8:24 a.m.
On Sun, May 12, 2013 at 04:05:39AM -0700, Saul Wold wrote:
> This changes the default GCC to 4.8, it has been tested with OE-Core
> and a number of layers have been testing with it for the last few weeks.
> 
> There are some minor issues outside of OE-Core. These will be addressed
> with time.
> 
> Signed-off-by: Saul Wold <sgw@linux.intel.com>
> Cc: Khem Raj <raj.khem@gmail.com>
> 
> ---
>  meta/conf/distro/include/tcmode-default.inc | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/meta/conf/distro/include/tcmode-default.inc b/meta/conf/distro/include/tcmode-default.inc
> index 88d109b..f3942b5 100644
> --- a/meta/conf/distro/include/tcmode-default.inc
> +++ b/meta/conf/distro/include/tcmode-default.inc
> @@ -17,7 +17,7 @@ PREFERRED_PROVIDER_virtual/${TARGET_PREFIX}libc-initial = "${TCLIBC}-initial"
>  PREFERRED_PROVIDER_virtual/nativesdk-${SDK_PREFIX}libc-initial = "nativesdk-${TCLIBC}-initial"
>  PREFERRED_PROVIDER_virtual/gettext ??= "gettext"
>  
> -GCCVERSION ?= "4.7%"
> +GCCVERSION ?= "4.8%"
>  SDKGCCVERSION ?= "${GCCVERSION}"
>  BINUVERSION ?= "2.23.2"
>  EGLIBCVERSION ?= "2.17"

There is also ICE when pixman is built with thumb enabled (e.g. qemuarm
+ ARM_INSTRUCTION_SET = "thumb" set, with thumb disabled it builds fine.

OE @ ~/oe-core/tmp-eglibc/work/armv5te-oe-linux-gnueabi/pixman/1_0.29.4-r0/pixman-0.29.4/test $ make
../arm-oe-linux-gnueabi-libtool  --tag=CC   --mode=compile arm-oe-linux-gnueabi-gcc  -march=armv5te -mthumb -mthumb-interwork --sysroot=/OE/oe-core/tmp-eglibc/sysroots/qemuarm -DHAVE_CONFIG_H -I. -I..  -I../pixman -I../pixman -I/OE/oe-core/tmp-eglibc/sysroots/qemuarm/usr/include/libpng16    -O2 -pipe -g -feliminate-unused-debug-types -Wall -fno-strict-aliasing -fvisibility=hidden -c -o utils-prng.lo utils-prng.c
arm-oe-linux-gnueabi-libtool: compile:  arm-oe-linux-gnueabi-gcc -march=armv5te -mthumb -mthumb-interwork --sysroot=/OE/oe-core/tmp-eglibc/sysroots/qemuarm -DHAVE_CONFIG_H -I. -I.. -I../pixman -I../pixman -I/OE/oe-core/tmp-eglibc/sysroots/qemuarm/usr/include/libpng16 -O2 -pipe -g -feliminate-unused-debug-types -Wall -fno-strict-aliasing -fvisibility=hidden -c utils-prng.c  -fPIC -DPIC -o .libs/utils-prng.o
utils-prng.c: In function 'prng_srand_r':
utils-prng.c:73:1: internal compiler error: Segmentation fault
 }
 ^
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.
make: *** [utils-prng.lo] Error 1
Phil Blundell - May 19, 2013, 8:55 a.m.
On Sun, 2013-05-19 at 10:24 +0200, Martin Jansa wrote:
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <http://gcc.gnu.org/bugs.html> for instructions.

Did you do that?  Please add the PR number to the check message so we
can tell when it's been fixed.

p.
Martin Jansa - May 19, 2013, 8:58 a.m.
On Sun, May 19, 2013 at 09:55:01AM +0100, Phil Blundell wrote:
> On Sun, 2013-05-19 at 10:24 +0200, Martin Jansa wrote:
> > Please submit a full bug report,
> > with preprocessed source if appropriate.
> > See <http://gcc.gnu.org/bugs.html> for instructions.
> 
> Did you do that?  Please add the PR number to the check message so we
> can tell when it's been fixed.

Not yet, I'm trying to strip test case from pixman, but sofar I can
reproduce it only with significant chunk of pixman code.
Phil Blundell - May 19, 2013, 9:16 a.m.
On Sun, 2013-05-19 at 10:58 +0200, Martin Jansa wrote:
> On Sun, May 19, 2013 at 09:55:01AM +0100, Phil Blundell wrote:
> > On Sun, 2013-05-19 at 10:24 +0200, Martin Jansa wrote:
> > > Please submit a full bug report,
> > > with preprocessed source if appropriate.
> > > See <http://gcc.gnu.org/bugs.html> for instructions.
> > 
> > Did you do that?  Please add the PR number to the check message so we
> > can tell when it's been fixed.
> 
> Not yet, I'm trying to strip test case from pixman, but sofar I can
> reproduce it only with significant chunk of pixman code.

That doesn't matter.  Unless the testcase is hundreds of megabytes, you
don't need to make heroic efforts to reduce it.  Just go ahead and file
a bug using whatever file you found the problem in.

p.

Patch

diff --git a/meta/conf/distro/include/tcmode-default.inc b/meta/conf/distro/include/tcmode-default.inc
index 88d109b..f3942b5 100644
--- a/meta/conf/distro/include/tcmode-default.inc
+++ b/meta/conf/distro/include/tcmode-default.inc
@@ -17,7 +17,7 @@  PREFERRED_PROVIDER_virtual/${TARGET_PREFIX}libc-initial = "${TCLIBC}-initial"
 PREFERRED_PROVIDER_virtual/nativesdk-${SDK_PREFIX}libc-initial = "nativesdk-${TCLIBC}-initial"
 PREFERRED_PROVIDER_virtual/gettext ??= "gettext"
 
-GCCVERSION ?= "4.7%"
+GCCVERSION ?= "4.8%"
 SDKGCCVERSION ?= "${GCCVERSION}"
 BINUVERSION ?= "2.23.2"
 EGLIBCVERSION ?= "2.17"