[OE-core] [PATCH] linux-yocto-3.0: Fix compile failure in perf

Thu Jan 5 21:20:21 UTC 2012

On Thu, Jan 5, 2012 at 12:59 PM, Bruce Ashfield
<bruce.ashfield at gmail.com> wrote:
> On Thu, Jan 5, 2012 at 3:55 PM, Khem Raj <raj.khem at gmail.com> wrote:
>> On Thu, Jan 5, 2012 at 12:13 PM, Bruce Ashfield
>> <bruce.ashfield at gmail.com> wrote:
>>> On Thu, Jan 5, 2012 at 3:06 PM, Khem Raj <raj.khem at gmail.com> wrote:
>>>> Signed-off-by: Khem Raj <raj.khem at gmail.com>
>>>> ---
>>>>  ...Undef-before-redefining-__attribute_const.patch |   33 ++++++++++++++++++++
>>>>  meta/recipes-kernel/linux/linux-yocto_3.0.bb       |    4 ++-
>>>>  2 files changed, 36 insertions(+), 1 deletions(-)
>>>>  create mode 100644 meta/recipes-kernel/linux/linux-yocto/0001-compiler.h-Undef-before-redefining-__attribute_const.patch
>>>
>>> Khem,
>>>
>>> I just built multiple linux-yocto kernels, for all the qemu targets
>>> and haven't seen this
>>> error at all. Do you have any special configuration that would be
>>> triggering this ?
>>
>> yes I am building with eglibc 2.15 ( recipes I have not posted yet)
>> and this defines is reported to be defined in usr/include/sys/cdefs.h
>> which is a header from libc.
>
> Aha. Perfect. This explains it completely.
>
>>
>>>
>>> The change itself looks sane, and in fact makes sense, since the rest
>>> of the file
>>> has the same protection for the new defines. Nothing has changed that
>>> should have triggered
>>> this to become an issue. I'm just trying to figure out why it has
>>> popped up now.
>>>
>>> I also checked and the 3.2 kernel still has the same issue, so pushing
>>> it upstream makes
>>> sense as well.

btw. I made the change thinking that perf utils want it that way to I undefine
the previous definitions before redefining. But if it was a placeholder where
it was defined because something needed it and there was no definitions from
standard include headers then I think right fix would be to check if its defined
before redefining it. In both cases the warning will go away

However with current patch the behaviour remains same as it was but if we check
for definition before redefining then that may change.

>>>
>>> And I see you asked on IRC, so I'll answer here. Submit a patch to the
>>> kernel like
>>> any upstream kernel change. Simply send the patch right from your
>>> linux working tree,
>>
>> You mean send to oe-core ml here right ?
>
> Yep!
>
>>
>>> and I'll pickup the change, merge it into the tree, push it out and
>>> send a pull request
>>> with updated SRCREVs that collect all the queued changes.
>>>
>>
>> Will you be able to bean this patch out from the one I sent or should
>> I resend it ?
>> the patch itself should be git am'able
>
> I'll grab it, no need to resend. I'm doing 3.0.16 merging right now,
> so I can collect
> it with that set of changes.
>
> Cheers,
>
> Bruce
>
>>
>>> Cheers,
>>>
>>> Bruce
>>>
>>>>
>>>> diff --git a/meta/recipes-kernel/linux/linux-yocto/0001-compiler.h-Undef-before-redefining-__attribute_const.patch b/meta/recipes-kernel/linux/linux-yocto/0001-compiler.h-Undef-before-redefining-__attribute_const.patch
>>>> new file mode 100644
>>>> index 0000000..b85cc97
>>>> --- /dev/null
>>>> +++ b/meta/recipes-kernel/linux/linux-yocto/0001-compiler.h-Undef-before-redefining-__attribute_const.patch
>>>> @@ -0,0 +1,33 @@
>>>> +From 6f92ab6de9d8daeb575949bbbcbc7bcdcebc60af Mon Sep 17 00:00:00 2001
>>>> +From: Khem Raj <raj.khem at gmail.com>
>>>> +Date: Thu, 5 Jan 2012 11:42:35 -0800
>>>> +Subject: [PATCH] compiler.h: Undef before redefining __attribute_const__
>>>> +
>>>> +This is required to avoid warnings like
>>>> +util/include/linux/compiler.h:8:0: error: "__attribute_const__" redefined [-Werror]
>>>> +
>>>> +Signed-off-by: Khem Raj <raj.khem at gmail.com>
>>>> +---
>>>> + tools/perf/util/include/linux/compiler.h |    4 +++-
>>>> + 1 files changed, 3 insertions(+), 1 deletions(-)
>>>> +
>>>> +diff --git a/tools/perf/util/include/linux/compiler.h b/tools/perf/util/include/linux/compiler.h
>>>> +index 791f9dd..f17ecef 100644
>>>> +--- a/tools/perf/util/include/linux/compiler.h
>>>> ++++ b/tools/perf/util/include/linux/compiler.h
>>>> +@@ -4,9 +4,11 @@
>>>> + #ifndef __always_inline
>>>> + #define __always_inline       inline
>>>> + #endif
>>>> ++#undef __user
>>>> + #define __user
>>>> ++#undef __attribute_const__
>>>> + #define __attribute_const__
>>>> +-
>>>> ++#undef __used
>>>> + #define __used                __attribute__((__unused__))
>>>> +
>>>> + #endif
>>>> +--
>>>> +1.7.5.4
>>>> +
>>>> diff --git a/meta/recipes-kernel/linux/linux-yocto_3.0.bb b/meta/recipes-kernel/linux/linux-yocto_3.0.bb
>>>> index b650c84..2e87f7f 100644
>>>> --- a/meta/recipes-kernel/linux/linux-yocto_3.0.bb
>>>> +++ b/meta/recipes-kernel/linux/linux-yocto_3.0.bb
>>>> @@ -23,7 +23,9 @@ SRCREV_meta ?= "c979f1365b1eb74e882b2cbbc8407ec536ab6eb8"
>>>>  PR = "r2"
>>>>  PV = "${LINUX_VERSION}+git${SRCPV}"
>>>>
>>>> -SRC_URI = "git://git.yoctoproject.org/linux-yocto-3.0;protocol=git;nocheckout=1;branch=${KBRANCH},meta;name=machine,meta"
>>>> +SRC_URI = "git://git.yoctoproject.org/linux-yocto-3.0;protocol=git;nocheckout=1;branch=${KBRANCH},meta;name=machine,meta \
>>>> +           file://0001-compiler.h-Undef-before-redefining-__attribute_const.patch \
>>>> +           "
>>>>
>>>>  COMPATIBLE_MACHINE = "(qemuarm|qemux86|qemuppc|qemumips|qemux86-64)"
>>>>
>>>> --
>>>> 1.7.5.4
>>>>
>>>>
>>>> _______________________________________________
>>>> Openembedded-core mailing list
>>>> Openembedded-core at 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"
>>>
>>> _______________________________________________
>>> Openembedded-core mailing list
>>> Openembedded-core at lists.openembedded.org
>>> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core
>>
>> _______________________________________________
>> Openembedded-core mailing list
>> Openembedded-core at 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"
>
> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core at lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core