Patchwork [1/5] kernel.bbclass: blacklist 'perf-dbg' as well for the modules metapackage

login
register
mail settings
Submitter lumag
Date Sept. 17, 2011, 10:18 p.m.
Message ID <1316297897-698-1-git-send-email-dbaryshkov@gmail.com>
Download mbox | patch
Permalink /patch/11625/
State New, archived
Headers show

Comments

lumag - Sept. 17, 2011, 10:18 p.m.
Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
---
 meta/classes/kernel.bbclass |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
Otavio Salvador - Sept. 17, 2011, 10:23 p.m.
I have sent this in July and it was nacked. See at
https://github.com/OSSystems/oe-core/commit/5c53cbf951a11ed92fb2ad0837991db256c11489

On Sat, Sep 17, 2011 at 19:18, Dmitry Eremin-Solenikov
<dbaryshkov@gmail.com> wrote:
> Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
> Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
> ---
>  meta/classes/kernel.bbclass |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/meta/classes/kernel.bbclass b/meta/classes/kernel.bbclass
> index bc1baa0..c577011 100644
> --- a/meta/classes/kernel.bbclass
> +++ b/meta/classes/kernel.bbclass
> @@ -451,7 +451,7 @@ python populate_packages_prepend () {
>        metapkg = "kernel-modules"
>        bb.data.setVar('ALLOW_EMPTY_' + metapkg, "1", d)
>        bb.data.setVar('FILES_' + metapkg, "", d)
> -       blacklist = [ 'kernel-dev', 'kernel-image', 'kernel-base', 'kernel-vmlinux', 'perf', 'kernel-misc' ]
> +       blacklist = [ 'kernel-dev', 'kernel-image', 'kernel-base', 'kernel-vmlinux', 'perf', 'perf-dbg', 'kernel-misc' ]
>        for l in module_deps.values():
>                for i in l:
>                        pkg = module_pattern % legitimize_package_name(re.match(module_regex, os.path.basename(i)).group(1))
> --
> 1.7.2.5
>
>
> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core
>
lumag - Sept. 22, 2011, 12:25 p.m.
On 09/18/2011 02:23 AM, Otavio Salvador wrote:
> I have sent this in July and it was nacked. See at
> https://github.com/OSSystems/oe-core/commit/5c53cbf951a11ed92fb2ad0837991db256c11489

Answering the original question by Saul:
A possible better question is why perf-dbg is getting generated and 
what's in it?

perf and perf-dbg are generated to contain perf tool if it's not 
directly enabled (AFAIK), but you can compile it by running make in 
tools/perf. It has some additional depends (like binutils-dev, 
elfutils-dev, news-dev, etc., so it's not enabled by default. Probably 
we should enable it though.

>
> On Sat, Sep 17, 2011 at 19:18, Dmitry Eremin-Solenikov
> <dbaryshkov@gmail.com>  wrote:
>> Signed-off-by: Koen Kooi<koen@dominion.thruhere.net>
>> Signed-off-by: Dmitry Eremin-Solenikov<dbaryshkov@gmail.com>
>> ---
>>   meta/classes/kernel.bbclass |    2 +-
>>   1 files changed, 1 insertions(+), 1 deletions(-)
>>
>> diff --git a/meta/classes/kernel.bbclass b/meta/classes/kernel.bbclass
>> index bc1baa0..c577011 100644
>> --- a/meta/classes/kernel.bbclass
>> +++ b/meta/classes/kernel.bbclass
>> @@ -451,7 +451,7 @@ python populate_packages_prepend () {
>>         metapkg = "kernel-modules"
>>         bb.data.setVar('ALLOW_EMPTY_' + metapkg, "1", d)
>>         bb.data.setVar('FILES_' + metapkg, "", d)
>> -       blacklist = [ 'kernel-dev', 'kernel-image', 'kernel-base', 'kernel-vmlinux', 'perf', 'kernel-misc' ]
>> +       blacklist = [ 'kernel-dev', 'kernel-image', 'kernel-base', 'kernel-vmlinux', 'perf', 'perf-dbg', 'kernel-misc' ]
>>         for l in module_deps.values():
>>                 for i in l:
>>                         pkg = module_pattern % legitimize_package_name(re.match(module_regex, os.path.basename(i)).group(1))
>> --
>> 1.7.2.5
>>
>>
>> _______________________________________________
>> Openembedded-core mailing list
>> Openembedded-core@lists.openembedded.org
>> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core
>>
>
>
>
Koen Kooi - Sept. 22, 2011, 12:35 p.m.
Op 22 sep. 2011, om 14:25 heeft Dmitry Eremin-Solenikov het volgende geschreven:

> On 09/18/2011 02:23 AM, Otavio Salvador wrote:
>> I have sent this in July and it was nacked. See at
>> https://github.com/OSSystems/oe-core/commit/5c53cbf951a11ed92fb2ad0837991db256c11489
> 
> Answering the original question by Saul:
> A possible better question is why perf-dbg is getting generated and what's in it?
> 
> perf and perf-dbg are generated to contain perf tool if it's not directly enabled (AFAIK), but you can compile it by running make in tools/perf. It has some additional depends (like binutils-dev, elfutils-dev, news-dev, etc., so it's not enabled by default. Probably we should enable it though.

the oe core kernel.bbclass builds perf by default
lumag - Sept. 22, 2011, 1 p.m.
On 09/22/2011 04:35 PM, Koen Kooi wrote:
>
> Op 22 sep. 2011, om 14:25 heeft Dmitry Eremin-Solenikov het volgende geschreven:
>
>> On 09/18/2011 02:23 AM, Otavio Salvador wrote:
>>> I have sent this in July and it was nacked. See at
>>> https://github.com/OSSystems/oe-core/commit/5c53cbf951a11ed92fb2ad0837991db256c11489
>>
>> Answering the original question by Saul:
>> A possible better question is why perf-dbg is getting generated and what's in it?
>>
>> perf and perf-dbg are generated to contain perf tool if it's not directly enabled (AFAIK), but you can compile it by running make in tools/perf. It has some additional depends (like binutils-dev, elfutils-dev, news-dev, etc., so it's not enabled by default. Probably we should enable it though.
>
> the oe core kernel.bbclass builds perf by default

according to kernel.bbclass: "perf must be enabled in individual kernel 
recipes"
Koen Kooi - Sept. 22, 2011, 1:17 p.m.
Op 22 sep. 2011, om 15:00 heeft Dmitry Eremin-Solenikov het volgende geschreven:

> On 09/22/2011 04:35 PM, Koen Kooi wrote:
>> 
>> Op 22 sep. 2011, om 14:25 heeft Dmitry Eremin-Solenikov het volgende geschreven:
>> 
>>> On 09/18/2011 02:23 AM, Otavio Salvador wrote:
>>>> I have sent this in July and it was nacked. See at
>>>> https://github.com/OSSystems/oe-core/commit/5c53cbf951a11ed92fb2ad0837991db256c11489
>>> 
>>> Answering the original question by Saul:
>>> A possible better question is why perf-dbg is getting generated and what's in it?
>>> 
>>> perf and perf-dbg are generated to contain perf tool if it's not directly enabled (AFAIK), but you can compile it by running make in tools/perf. It has some additional depends (like binutils-dev, elfutils-dev, news-dev, etc., so it's not enabled by default. Probably we should enable it though.
>> 
>> the oe core kernel.bbclass builds perf by default
> 
> according to kernel.bbclass: "perf must be enabled in individual kernel recipes"

And pretty much everything uses recipes-kernel/linux/linux-tools.inc, which enables it. I had to add do_compile_perf() { : } to my old kernel recipes that have a broken perf, so it's opt-out, not opt-in.
Bruce Ashfield - Sept. 22, 2011, 1:28 p.m.
On Thu, Sep 22, 2011 at 9:17 AM, Koen Kooi <koen@dominion.thruhere.net> wrote:
>
> Op 22 sep. 2011, om 15:00 heeft Dmitry Eremin-Solenikov het volgende geschreven:
>
>> On 09/22/2011 04:35 PM, Koen Kooi wrote:
>>>
>>> Op 22 sep. 2011, om 14:25 heeft Dmitry Eremin-Solenikov het volgende geschreven:
>>>
>>>> On 09/18/2011 02:23 AM, Otavio Salvador wrote:
>>>>> I have sent this in July and it was nacked. See at
>>>>> https://github.com/OSSystems/oe-core/commit/5c53cbf951a11ed92fb2ad0837991db256c11489
>>>>
>>>> Answering the original question by Saul:
>>>> A possible better question is why perf-dbg is getting generated and what's in it?
>>>>
>>>> perf and perf-dbg are generated to contain perf tool if it's not directly enabled (AFAIK), but you can compile it by running make in tools/perf. It has some additional depends (like binutils-dev, elfutils-dev, news-dev, etc., so it's not enabled by default. Probably we should enable it though.
>>>
>>> the oe core kernel.bbclass builds perf by default
>>
>> according to kernel.bbclass: "perf must be enabled in individual kernel recipes"
>
> And pretty much everything uses recipes-kernel/linux/linux-tools.inc, which enables it. I had to add do_compile_perf() { : } to my old kernel recipes that have a broken perf, so it's opt-out, not opt-in.

We can always flip this around if required. When I initially did the kernel
tree based builds for perf it was right in a specific kernel recipe, and then
pulled it out into linux-tools.inc for easier reuse it was the intention that
just including/requiring that .inc file would trigger a build of perf.

I'm open to inhibiting this via the blacklist or via some other suitable
mechanism.

But did we get some include creep (or maybe I'm just missing it) ?
When I just checked my tree, I only see a handful of places that linux-tools.inc
will be pulled in and trigger a build.

Cheers,

Bruce

> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core
>
Koen Kooi - Sept. 22, 2011, 1:56 p.m.
Op 22 sep. 2011, om 15:28 heeft Bruce Ashfield het volgende geschreven:

> On Thu, Sep 22, 2011 at 9:17 AM, Koen Kooi <koen@dominion.thruhere.net> wrote:
>> 
>> Op 22 sep. 2011, om 15:00 heeft Dmitry Eremin-Solenikov het volgende geschreven:
>> 
>>> On 09/22/2011 04:35 PM, Koen Kooi wrote:
>>>> 
>>>> Op 22 sep. 2011, om 14:25 heeft Dmitry Eremin-Solenikov het volgende geschreven:
>>>> 
>>>>> On 09/18/2011 02:23 AM, Otavio Salvador wrote:
>>>>>> I have sent this in July and it was nacked. See at
>>>>>> https://github.com/OSSystems/oe-core/commit/5c53cbf951a11ed92fb2ad0837991db256c11489
>>>>> 
>>>>> Answering the original question by Saul:
>>>>> A possible better question is why perf-dbg is getting generated and what's in it?
>>>>> 
>>>>> perf and perf-dbg are generated to contain perf tool if it's not directly enabled (AFAIK), but you can compile it by running make in tools/perf. It has some additional depends (like binutils-dev, elfutils-dev, news-dev, etc., so it's not enabled by default. Probably we should enable it though.
>>>> 
>>>> the oe core kernel.bbclass builds perf by default
>>> 
>>> according to kernel.bbclass: "perf must be enabled in individual kernel recipes"
>> 
>> And pretty much everything uses recipes-kernel/linux/linux-tools.inc, which enables it. I had to add do_compile_perf() { : } to my old kernel recipes that have a broken perf, so it's opt-out, not opt-in.
> 
> We can always flip this around if required. When I initially did the kernel
> tree based builds for perf it was right in a specific kernel recipe, and then
> pulled it out into linux-tools.inc for easier reuse it was the intention that
> just including/requiring that .inc file would trigger a build of perf.


I'm not complaining about perf being built, I'm complaining that the patch is held up on a question why perf-dbg is getting built. If 'perf' is in the blacklist, 'perf-dbg' should be in as well. If 'perf-dbg' isn't allowed in, take out 'perf' as well, since they are a matched pair.
Bruce Ashfield - Sept. 22, 2011, 2:04 p.m.
On Thu, Sep 22, 2011 at 9:56 AM, Koen Kooi <koen@dominion.thruhere.net> wrote:
>
> Op 22 sep. 2011, om 15:28 heeft Bruce Ashfield het volgende geschreven:
>
>> On Thu, Sep 22, 2011 at 9:17 AM, Koen Kooi <koen@dominion.thruhere.net> wrote:
>>>
>>> Op 22 sep. 2011, om 15:00 heeft Dmitry Eremin-Solenikov het volgende geschreven:
>>>
>>>> On 09/22/2011 04:35 PM, Koen Kooi wrote:
>>>>>
>>>>> Op 22 sep. 2011, om 14:25 heeft Dmitry Eremin-Solenikov het volgende geschreven:
>>>>>
>>>>>> On 09/18/2011 02:23 AM, Otavio Salvador wrote:
>>>>>>> I have sent this in July and it was nacked. See at
>>>>>>> https://github.com/OSSystems/oe-core/commit/5c53cbf951a11ed92fb2ad0837991db256c11489
>>>>>>
>>>>>> Answering the original question by Saul:
>>>>>> A possible better question is why perf-dbg is getting generated and what's in it?
>>>>>>
>>>>>> perf and perf-dbg are generated to contain perf tool if it's not directly enabled (AFAIK), but you can compile it by running make in tools/perf. It has some additional depends (like binutils-dev, elfutils-dev, news-dev, etc., so it's not enabled by default. Probably we should enable it though.
>>>>>
>>>>> the oe core kernel.bbclass builds perf by default
>>>>
>>>> according to kernel.bbclass: "perf must be enabled in individual kernel recipes"
>>>
>>> And pretty much everything uses recipes-kernel/linux/linux-tools.inc, which enables it. I had to add do_compile_perf() { : } to my old kernel recipes that have a broken perf, so it's opt-out, not opt-in.
>>
>> We can always flip this around if required. When I initially did the kernel
>> tree based builds for perf it was right in a specific kernel recipe, and then
>> pulled it out into linux-tools.inc for easier reuse it was the intention that
>> just including/requiring that .inc file would trigger a build of perf.
>
>
> I'm not complaining about perf being built, I'm complaining that the patch is held up on a question why perf-dbg is getting built. If 'perf' is in the blacklist, 'perf-dbg' should be in as well. If 'perf-dbg' isn't allowed in, take out 'perf' as well, since they are a matched pair.

Aha. Yes, I see that now. I'm generally quite bad at packaging so I sometimes
need to be pointed right at something :)

As for getting it in or out of the tree, I'm only one voice, but I
don't see a downside
to this being on the blacklist as well.

Cheers,

Bruce

> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core
>
lumag - Sept. 28, 2011, 8:47 a.m.
Sorry to bother all of you, what about this patchset?
Richard Purdie - Sept. 28, 2011, 2:50 p.m.
On Sun, 2011-09-18 at 02:18 +0400, Dmitry Eremin-Solenikov wrote:
> Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
> Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
> ---
>  meta/classes/kernel.bbclass |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/meta/classes/kernel.bbclass b/meta/classes/kernel.bbclass
> index bc1baa0..c577011 100644
> --- a/meta/classes/kernel.bbclass
> +++ b/meta/classes/kernel.bbclass
> @@ -451,7 +451,7 @@ python populate_packages_prepend () {
>  	metapkg = "kernel-modules"
>  	bb.data.setVar('ALLOW_EMPTY_' + metapkg, "1", d)
>  	bb.data.setVar('FILES_' + metapkg, "", d)
> -	blacklist = [ 'kernel-dev', 'kernel-image', 'kernel-base', 'kernel-vmlinux', 'perf', 'kernel-misc' ]
> +	blacklist = [ 'kernel-dev', 'kernel-image', 'kernel-base', 'kernel-vmlinux', 'perf', 'perf-dbg', 'kernel-misc' ]
>  	for l in module_deps.values():
>  		for i in l:
>  			pkg = module_pattern % legitimize_package_name(re.match(module_regex, os.path.basename(i)).group(1))

Merged to master, thanks.

Richard
Richard Purdie - Sept. 28, 2011, 2:52 p.m.
On Wed, 2011-09-28 at 12:47 +0400, Dmitry Eremin-Solenikov wrote:
> Sorry to bother all of you, what about this patchset?

I was waiting for Bruce to review them.

I've taken three of them, the forth needs a better patch description as
per Bruce's request.

The MACHINE_PR one I have reservations about and I'll reply to that
thread.

Cheers,

Richard

Patch

diff --git a/meta/classes/kernel.bbclass b/meta/classes/kernel.bbclass
index bc1baa0..c577011 100644
--- a/meta/classes/kernel.bbclass
+++ b/meta/classes/kernel.bbclass
@@ -451,7 +451,7 @@  python populate_packages_prepend () {
 	metapkg = "kernel-modules"
 	bb.data.setVar('ALLOW_EMPTY_' + metapkg, "1", d)
 	bb.data.setVar('FILES_' + metapkg, "", d)
-	blacklist = [ 'kernel-dev', 'kernel-image', 'kernel-base', 'kernel-vmlinux', 'perf', 'kernel-misc' ]
+	blacklist = [ 'kernel-dev', 'kernel-image', 'kernel-base', 'kernel-vmlinux', 'perf', 'perf-dbg', 'kernel-misc' ]
 	for l in module_deps.values():
 		for i in l:
 			pkg = module_pattern % legitimize_package_name(re.match(module_regex, os.path.basename(i)).group(1))