Re: [PATCH]: add tests for pthread initialization macros

["Kaveh R. Ghazi" <ghazi at caipclassic dot rutgers dot edu>]

 > +/*  glibc-2.3.5 defines PTHREAD_{MUTEX|COND}_INITIALIZER
 > + *  incorrectly, so we replace them with versions that
 > + *  correspond to the definition.
 > + */
 > +fix = {
 > +    hackname = pthread_mutex_init;
 > +    files    = pthread.h;
 > +    mach     = *-*-linux-gnu;
 > +    sed      = "/define[ \t]*PTHREAD_MUTEX_INITIALIZER[ \t]*\\\\/,+1"
 > +               "s/{ { 0, } }/{ { 0, 0, 0, 0, 0, 0 } }/\n"
 > +               "/define[ \t]*PTHREAD_COND_INITIALIZER/"
 > +               "s/{ { 0, } }/{ { 0, 0, 0, 0, 0, (void *) 0, 0, 0 }
 > }/";
 > +
 > +    test_text =
 > +    "#define PTHREAD_MUTEX_INITIALIZER \\\\\n"
 > +    "  { { 0, } }\n"
 > +    "#define PTHREAD_COND_INITIALIZER { { 0, } }";
 > +
 > +};
 > +

Uros,

Thanks for the patch, I looked it over and have a couple of notes.
First you probably want to cc: the fixincludes maintainer so he's more
likely to see your patch which was kind of hidden as it came
downstream in my thread about a testcase.  I've copied Bruce in so he
won't miss your post.

Second, this fix seems to address a problem with glibc, not linux
specific.  I don't know that there's a way to select glibc boxes but I
think we'll want to add maybe the bsd's and maybe others in the "mach"
line.

Third, I can't speak for Bruce, but you'll probably be asked to use a
"format" fix rather than a sed fix since the format style is faster.
If you look at fixincludes/README in section 4.4.v, it explains it a
little.

With regards to the testcase, I agree we should add any glibc extra
bits you want to test.  If you want to apply a patch over mine, please
do so.  I'm going to install my testcase as "obvious" in a day or two
so that people have time to comment.  After that you can add yours.

		Thanks,
		--Kaveh
--
Kaveh R. Ghazi			ghazi@caip.rutgers.edu