FTBFS with clang instead of gcc

FTBFS with clang instead of gcc

[Laurent Bigonville]

Hi,

Apparently audit is FTBFS with clang due to the embedded version of
libev with the following error (see: [0]):

../../../../src/libev/ev.c:970:42: error: '_Noreturn' keyword must
precede function declarator

The bug[0] provides a patch. I don't think this has been fixed
upstream yet. There was however a small thread [1] about this on
upstream ML.

Not sure what's the policy regarding patching embedded copy of libev.

Cheers,

Laurent Bigonville

[0] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=787066
[1] http://lists.schmorp.de/pipermail/libev/2015q1/002480.html

Re: FTBFS with clang instead of gcc

[Zbigniew Jędrzejewski-Szmek]

On Sun, Jul 26, 2015 at 03:58:44PM +0200, Laurent Bigonville wrote:
> Hi,
> 
> Apparently audit is FTBFS with clang due to the embedded version of
> libev with the following error (see: [0]):
> 
> ../../../../src/libev/ev.c:970:42: error: '_Noreturn' keyword must
> precede function declarator
> 
> The bug[0] provides a patch. I don't think this has been fixed
> upstream yet. There was however a small thread [1] about this on
> upstream ML.
> 
> Not sure what's the policy regarding patching embedded copy of libev.
The policy is to kill the embedded copy as soon as possible.
Seems like a good time to do that.

Zbyszek

Re: FTBFS with clang instead of gcc

[Steve Grubb]

On Sunday, July 26, 2015 09:49:50 PM Zbigniew Jędrzejewski-Szmek wrote:
> On Sun, Jul 26, 2015 at 03:58:44PM +0200, Laurent Bigonville wrote:
> > Apparently audit is FTBFS with clang due to the embedded version of
> > libev with the following error (see: [0]):
> > 
> > ../../../../src/libev/ev.c:970:42: error: '_Noreturn' keyword must
> > precede function declarator
> > 
> > The bug[0] provides a patch. I don't think this has been fixed
> > upstream yet. There was however a small thread [1] about this on
> > upstream ML.
> > 
> Not sure what's the policy regarding patching embedded copy of libev.
> The policy is to kill the embedded copy as soon as possible.
> Seems like a good time to do that.

Not sure if you've ever had any dealings with upstream libev or not. But let's 
just say its not easy. I apply a patch on top of libev to suit audit's needs.

https://fedorahosted.org/audit/changeset/782

Without this, coverity analysis is much harder. Audit is one of those 
subsystems that you really need to be sure of what its doing and that even 
libraries it uses are correct.

-Steve

--
Linux-audit mailing list
Linux-audit@redhat.com
https://www.redhat.com/mailman/listinfo/linux-audit

Re: FTBFS with clang instead of gcc

[Zbigniew Jędrzejewski-Szmek]

On Mon, Jul 27, 2015 at 12:34:38PM -0400, Steve Grubb wrote:
> On Sunday, July 26, 2015 09:49:50 PM Zbigniew Jędrzejewski-Szmek wrote:
> > On Sun, Jul 26, 2015 at 03:58:44PM +0200, Laurent Bigonville wrote:
> > > Apparently audit is FTBFS with clang due to the embedded version of
> > > libev with the following error (see: [0]):
> > > 
> > > ../../../../src/libev/ev.c:970:42: error: '_Noreturn' keyword must
> > > precede function declarator
> > > 
> > > The bug[0] provides a patch. I don't think this has been fixed
> > > upstream yet. There was however a small thread [1] about this on
> > > upstream ML.
> > > 
> > Not sure what's the policy regarding patching embedded copy of libev.
> > The policy is to kill the embedded copy as soon as possible.
> > Seems like a good time to do that.
> 
> Not sure if you've ever had any dealings with upstream libev or not. But let's 
> just say its not easy. I apply a patch on top of libev to suit audit's needs.
> 
> https://fedorahosted.org/audit/changeset/782
> 
> Without this, coverity analysis is much harder. Audit is one of those 
> subsystems that you really need to be sure of what its doing and that even 
> libraries it uses are correct.
You're right, I should have remembered that libev wants to be a copylib.

Zbyszek

--
Linux-audit mailing list
Linux-audit@redhat.com
https://www.redhat.com/mailman/listinfo/linux-audit

Re: FTBFS with clang instead of gcc

[Steve Grubb]

On Sunday, July 26, 2015 03:58:44 PM Laurent Bigonville wrote:
> Apparently audit is FTBFS with clang due to the embedded version of
> libev with the following error (see: [0]):
> 
> ../../../../src/libev/ev.c:970:42: error: '_Noreturn' keyword must
> precede function declarator
> 
> The bug[0] provides a patch. I don't think this has been fixed
> upstream yet. There was however a small thread [1] about this on
> upstream ML.

Actually, it is fixed upstream.

http://cvs.schmorp.de/libev/ev.c?revision=1.476&view=markup#l1087


> Not sure what's the policy regarding patching embedded copy of libev.

It also appears that 4.20 was release back in June. I'll update the embedded 
copy. Thanks for pointing this out.

-Steve