Re: conformtest failures on i386

[Siddhesh Poyarekar <siddhesh at redhat dot com>]

On Tue, Jun 24, 2014 at 11:02:26AM +0000, Joseph S. Myers wrote:
> On Tue, 24 Jun 2014, Siddhesh Poyarekar wrote:
> 
> > On Tue, Jun 24, 2014 at 10:37:58AM +0000, Joseph S. Myers wrote:
> > > On Tue, 24 Jun 2014, Siddhesh Poyarekar wrote:
> > > 
> > > > Hi,
> > > > 
> > > > The conformtest run on i386 builds have 40 FAILs in addition to the
> > > > 120 XFAILS.  Are these known are do those need to be fixed?
> > > 
> > > You don't say what any of the actual errors shown in the logs for the 
> > > individual tests are.  The presumption is that any such failure should be 
> > > fixed, as indicating an architecture-specific header bug (unless it 
> > > indicates a problem with the testcase, or that the internal include/ 
> > > headers are missing an _ISOMAC conditional).
> > 
> > I hadn't looked at the actual causes of failures because the failures
> > themselves were not relevant to the code I was testing.  I'll look
> > through the failures and file bugs wherever applicable.
> 
> The few I just looked at were:
> 
>     Namespace violation: "d"
> 
> which may be coming from bits/pthreadtypes.h.

Yeah, all of the failures are the same as above.  How about this for a
fix:

Siddhesh

	* sysdeps/x86/nptl/bits/pthreadtypes.h (pthread_mutex_t):
	Rename member __data.d to __data.__elision_data.

diff --git a/sysdeps/x86/nptl/bits/pthreadtypes.h b/sysdeps/x86/nptl/bits/pthreadtypes.h
index 7f8076b..aad2e14 100644
--- a/sysdeps/x86/nptl/bits/pthreadtypes.h
+++ b/sysdeps/x86/nptl/bits/pthreadtypes.h
@@ -115,10 +115,10 @@ typedef union
       {
 	short __espins;
 	short __elision;
-# define __spins d.__espins
-# define __elision d.__elision
+# define __spins __elision_data.__espins
+# define __elision __elision_data.__elision
 # define __PTHREAD_SPINS         { 0, 0 }
-      } d;
+      } __elision_data;
       __pthread_slist_t __list;
     };
 #endif

Attachment: pgpqolGp9NvxQ.pgp
Description: PGP signature