Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1418017 > unrolled thread

linux-next: build warnings after merge of the kbuild tree

Started byStephen Rothwell <sfr@canb.auug.org.au>
First post2016-06-09 04:30 +0200
Last post2016-06-09 19:40 +0200
Articles 8 — 5 participants

Back to article view | Back to linux.kernel

Contents

  linux-next: build warnings after merge of the kbuild tree Stephen Rothwell <sfr@canb.auug.org.au> - 2016-06-09 04:30 +0200
    Re: linux-next: build warnings after merge of the kbuild tree Kees Cook <keescook@chromium.org> - 2016-06-09 05:00 +0200
      Re: linux-next: build warnings after merge of the kbuild tree Stephen Rothwell <sfr@canb.auug.org.au> - 2016-06-09 06:10 +0200
        Re: linux-next: build warnings after merge of the kbuild tree Michal Marek <mmarek@suse.cz> - 2016-06-09 13:00 +0200
          Re: linux-next: build warnings after merge of the kbuild tree Emese Revfy <re.emese@gmail.com> - 2016-06-09 19:40 +0200
            Re: linux-next: build warnings after merge of the kbuild tree Kees Cook <keescook@chromium.org> - 2016-06-09 20:00 +0200
        Re: linux-next: build warnings after merge of the kbuild tree Michael Ellerman <mpe@ellerman.id.au> - 2016-06-09 13:20 +0200
    Re: linux-next: build warnings after merge of the kbuild tree Emese Revfy <re.emese@gmail.com> - 2016-06-09 19:40 +0200

#1418017 — linux-next: build warnings after merge of the kbuild tree

FromStephen Rothwell <sfr@canb.auug.org.au>
Date2016-06-09 04:30 +0200
Subjectlinux-next: build warnings after merge of the kbuild tree
Message-ID<rHXTk-8w3-7@gated-at.bofh.it>
Hi Michal,

After merging the kbuild tree, today's linux-next build (x86_64
allmodconfig) produced these warnings:

Cyclomatic Complexity 1 scripts/mod/devicetable-offsets.c:main
Cyclomatic Complexity 1 kernel/bounds.c:foo
Cyclomatic Complexity 1 arch/x86/kernel/asm-offsets_64.c:main
Cyclomatic Complexity 1 arch/x86/kernel/asm-offsets.c:common
Cyclomatic Complexity 5 arch/x86/ia32/audit.c:ia32_classify_syscall

and so on (many, many of these - about 635,000 :-()

Introduced (presumably) by commits

  6b90bd4ba40b ("GCC plugin infrastructure")
  0dae776c6bf3 ("Add Cyclomatic complexity GCC plugin")

I have disabled CONFIG_GCC_PLUGIN_CYC_COMPLEXITY (by making it depend
on CONFIG_BROKEN) until it is not enabled by default.

-- 
Cheers,
Stephen Rothwell

[toc] | [next] | [standalone]

#1418029

FromKees Cook <keescook@chromium.org>
Date2016-06-09 05:00 +0200
Message-ID<rHYmm-en-7@gated-at.bofh.it>
In reply to#1418017
On Wed, Jun 8, 2016 at 7:22 PM, Stephen Rothwell <sfr@canb.auug.org.au> wrote:
> Hi Michal,
>
> After merging the kbuild tree, today's linux-next build (x86_64
> allmodconfig) produced these warnings:
>
> Cyclomatic Complexity 1 scripts/mod/devicetable-offsets.c:main
> Cyclomatic Complexity 1 kernel/bounds.c:foo
> Cyclomatic Complexity 1 arch/x86/kernel/asm-offsets_64.c:main
> Cyclomatic Complexity 1 arch/x86/kernel/asm-offsets.c:common
> Cyclomatic Complexity 5 arch/x86/ia32/audit.c:ia32_classify_syscall
>
> and so on (many, many of these - about 635,000 :-()

Congratulations on having the gcc plugin development headers
successfully installed! ;)

> Introduced (presumably) by commits
>
>   6b90bd4ba40b ("GCC plugin infrastructure")
>   0dae776c6bf3 ("Add Cyclomatic complexity GCC plugin")
>
> I have disabled CONFIG_GCC_PLUGIN_CYC_COMPLEXITY (by making it depend
> on CONFIG_BROKEN) until it is not enabled by default.

Ah, yes, that should default to off. We'll get a fix landed ASAP.

-Kees

-- 
Kees Cook
Chrome OS & Brillo Security

[toc] | [prev] | [next] | [standalone]

#1418051

FromStephen Rothwell <sfr@canb.auug.org.au>
Date2016-06-09 06:10 +0200
Message-ID<rHZs5-1ac-7@gated-at.bofh.it>
In reply to#1418029
Hi Kees,

On Wed, 8 Jun 2016 19:56:38 -0700 Kees Cook <keescook@chromium.org> wrote:
>
> Congratulations on having the gcc plugin development headers
> successfully installed! ;)

Thanks :-)

> Ah, yes, that should default to off. We'll get a fix landed ASAP.

Note that this was an allmodconfig build.  The default is 'n', but
allmodconfig will turn it on (as will allyesconfig).

-- 
Cheers,
Stephen Rothwell

[toc] | [prev] | [next] | [standalone]

#1418238

FromMichal Marek <mmarek@suse.cz>
Date2016-06-09 13:00 +0200
Message-ID<rI5QR-5eV-1@gated-at.bofh.it>
In reply to#1418051
Dne 9.6.2016 v 06:05 Stephen Rothwell napsal(a):
> On Wed, 8 Jun 2016 19:56:38 -0700 Kees Cook <keescook@chromium.org> wrote:
>> Ah, yes, that should default to off. We'll get a fix landed ASAP.
> 
> Note that this was an allmodconfig build.  The default is 'n', but
> allmodconfig will turn it on (as will allyesconfig).

I guess we should make GCC_PLUGINS depend on !COMPILE_TEST. Actually I
thought this was already the case, but it is not.

Michal

[toc] | [prev] | [next] | [standalone]

#1418526

FromEmese Revfy <re.emese@gmail.com>
Date2016-06-09 19:40 +0200
Message-ID<rIc5X-13V-21@gated-at.bofh.it>
In reply to#1418238
On Thu, 9 Jun 2016 12:57:16 +0200
Michal Marek <mmarek@suse.cz> wrote:

> Dne 9.6.2016 v 06:05 Stephen Rothwell napsal(a):
> > On Wed, 8 Jun 2016 19:56:38 -0700 Kees Cook <keescook@chromium.org> wrote:
> >> Ah, yes, that should default to off. We'll get a fix landed ASAP.
> > 
> > Note that this was an allmodconfig build.  The default is 'n', but
> > allmodconfig will turn it on (as will allyesconfig).
> 
> I guess we should make GCC_PLUGINS depend on !COMPILE_TEST. Actually I
> thought this was already the case, but it is not.

Is it really necessary to disable all gcc plugins or would it be enough
to disable only the cyc_complexity plugin?

-- 
Emese

[toc] | [prev] | [next] | [standalone]

#1418536

FromKees Cook <keescook@chromium.org>
Date2016-06-09 20:00 +0200
Message-ID<rIcpj-1d0-3@gated-at.bofh.it>
In reply to#1418526
On Thu, Jun 9, 2016 at 10:37 AM, Emese Revfy <re.emese@gmail.com> wrote:
> On Thu, 9 Jun 2016 12:57:16 +0200
> Michal Marek <mmarek@suse.cz> wrote:
>
>> Dne 9.6.2016 v 06:05 Stephen Rothwell napsal(a):
>> > On Wed, 8 Jun 2016 19:56:38 -0700 Kees Cook <keescook@chromium.org> wrote:
>> >> Ah, yes, that should default to off. We'll get a fix landed ASAP.
>> >
>> > Note that this was an allmodconfig build.  The default is 'n', but
>> > allmodconfig will turn it on (as will allyesconfig).
>>
>> I guess we should make GCC_PLUGINS depend on !COMPILE_TEST. Actually I
>> thought this was already the case, but it is not.
>
> Is it really necessary to disable all gcc plugins or would it be enough
> to disable only the cyc_complexity plugin?

I think disabling (depend on !COMPILE_TEST) plugins that have
non-actionable output make sense. For example, in the future, things
like constify or initify may produce warnings that are "real" in the
sense that they have detected situations that should be fixed in the
code. In a perfect world, we would include those fixes ahead of the
new plugin to keep Stephen from going crazy. :)

-Kees

-- 
Kees Cook
Chrome OS & Brillo Security

[toc] | [prev] | [next] | [standalone]

#1418240

FromMichael Ellerman <mpe@ellerman.id.au>
Date2016-06-09 13:20 +0200
Message-ID<rI6ad-5BC-9@gated-at.bofh.it>
In reply to#1418051
On Thu, 2016-06-09 at 14:05 +1000, Stephen Rothwell wrote:
> Hi Kees,
> 
> On Wed, 8 Jun 2016 19:56:38 -0700 Kees Cook <keescook@chromium.org> wrote:
> > 
> > Congratulations on having the gcc plugin development headers
> > successfully installed! ;)
> 
> Thanks :-)
 
And on ppc64le too! (I think)

:)

cheers

[toc] | [prev] | [next] | [standalone]

#1418530

FromEmese Revfy <re.emese@gmail.com>
Date2016-06-09 19:40 +0200
Message-ID<rIc5Y-13V-33@gated-at.bofh.it>
In reply to#1418017
On Thu, 9 Jun 2016 12:22:58 +1000
Stephen Rothwell <sfr@canb.auug.org.au> wrote:

> Hi Michal,
> 
> After merging the kbuild tree, today's linux-next build (x86_64
> allmodconfig) produced these warnings:
> 
> Cyclomatic Complexity 1 scripts/mod/devicetable-offsets.c:main
> Cyclomatic Complexity 1 kernel/bounds.c:foo
> Cyclomatic Complexity 1 arch/x86/kernel/asm-offsets_64.c:main
> Cyclomatic Complexity 1 arch/x86/kernel/asm-offsets.c:common
> Cyclomatic Complexity 5 arch/x86/ia32/audit.c:ia32_classify_syscall
> 
> and so on (many, many of these - about 635,000 :-()
> 
> Introduced (presumably) by commits
> 
>   6b90bd4ba40b ("GCC plugin infrastructure")
>   0dae776c6bf3 ("Add Cyclomatic complexity GCC plugin")
> 
> I have disabled CONFIG_GCC_PLUGIN_CYC_COMPLEXITY (by making it depend
> on CONFIG_BROKEN) until it is not enabled by default.

These aren't warnings. This plugin is a static analyzer. It prints out
the cyclomatic complexity of all functions in the kernel.

I think it would be useful to enable it sometimes and report new functions
with a high enough complexity value. 

-- 
Emese

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web