Changes to llvm tests to fix Bug 17327

Nick Lewycky nicholas at mxc.ca
Tue Sep 24 22:32:31 PDT 2013


Stephen Checkoway wrote:
>
> On Sep 24, 2013, at 10:09 PM, Thomas O'Dell<thomas_odell at trsolutions.biz>  wrote:
>
>> This is my first commit to llvm. I'm in the process of porting LLVM to
>> X86 QNX Neutrino. I found the attached tests failed because of
>> non-standard expectations from Unix utilities. So attached are the
>> changes (created svn diff) that allow the tests to run.
>
>
> Most of these are trivial>&  /dev/null to>/dev/null 2>&1 that seem obviously correct.
>
> --- test/Transforms/GCOVProfiling/linkagename.ll	(revision 191282)
> +++ test/Transforms/GCOVProfiling/linkagename.ll	(working copy)
> @@ -1,7 +1,8 @@
>   ; RUN: echo '!9 = metadata !{metadata !"%T/linkagename.ll", metadata !0}'>  %t1
>   ; RUN: cat %s %t1>  %t2
>   ; RUN: opt -insert-gcov-profiling -disable-output<  %t2
> -; RUN: grep _Z3foov %T/linkagename.gcno
> +; RUN: echo `cat %T/linkagename.gno`>%t3
> +; RUN: grep _Z3foov %t3
>   ; RUN: rm %T/linkagename.gcno
>
> I don't really understand this one though. What's the nonstandard behavior here? In both cases, it seems to be grepping a file for _Z3foov. You've just added a new file?

Hmm, it introduces a typo.

+; RUN: echo `cat %T/linkagename.gno`>%t3

"gno" vs. "gcno". (Unless there's a matching change elsewhere to the 
gcov support?)

Nick




More information about the llvm-commits mailing list