[google] Allow relative profile paths.

[martint, 2011-03-28 17:11:35]

This patch should go into google/main so that relative paths can be
used with FDO.

Bootstraped without regression and verified that my benchmarks correctly
used the generated profiles.

Original patch to trunk included a regression test, but I noticed last
week that it failed in certain test scenarios. Will investigate and
push the change upstream once I resolve that.

Thanks,
Martin


2011-03-28  Martin Thuresson  <martint@google.com>

	* gcc/doc/invoke.texi: Document support for relative profile paths.
	* gcc/libgcov.c	(gcov_exit): Support relative profile paths.

Index: gcc/doc/invoke.texi
===================================================================
--- gcc/doc/invoke.texi	(revision 171049)
+++ gcc/doc/invoke.texi	(working copy)
@@ -7774,7 +7774,7 @@ Set the directory to search for the prof
 This option affects only the profile data generated by
 @option{-fprofile-generate}, @option{-ftest-coverage}, @option{-fprofile-arcs}
 and used by @option{-fprofile-use} and @option{-fbranch-probabilities}
-and its related options.
+and its related options. Both absolute and relative can be used.
 By default, GCC will use the current directory as @var{path}, thus the
 profile data file will appear in the same directory as the object file.
 
Index: gcc/libgcov.c
===================================================================
--- gcc/libgcov.c	(revision 171049)
+++ gcc/libgcov.c	(working copy)
@@ -283,8 +283,9 @@ gcov_exit (void)
 	      }
         }
       /* Update complete filename with stripped original. */
-      if (!IS_DIR_SEPARATOR (*fname) && !HAS_DRIVE_SPEC(fname))
+      if (prefix_length != 0 && !IS_DIR_SEPARATOR (*fname))
 	{
+          /* If prefix is given, add directory separator.  */
 	  strcpy (gi_filename_up, "/");
 	  strcpy (gi_filename_up + 1, fname);
 	}

--
This patch is available for review at http://codereview.appspot.com/4280074

[Diego Novillo, 2011-03-28 17:59:32]

OK.  A nit and a question below.


Diego.

http://codereview.appspot.com/4280074/diff/1/gcc/doc/invoke.texi
File gcc/doc/invoke.texi (right):

http://codereview.appspot.com/4280074/diff/1/gcc/doc/invoke.texi#newcode7777
gcc/doc/invoke.texi:7777: and its related options. Both absolute and relative
can be used.
s/relative/relative paths/

http://codereview.appspot.com/4280074/diff/1/gcc/libgcov.c
File gcc/libgcov.c (right):

http://codereview.appspot.com/4280074/diff/1/gcc/libgcov.c#newcode286
gcc/libgcov.c:286: if (prefix_length != 0 && !IS_DIR_SEPARATOR (*fname))
I'm confused.  why did you take !HAS_DRIVE_SPEC() out?

[martint, 2011-03-28 18:14:19]

http://codereview.appspot.com/4280074/diff/1/gcc/doc/invoke.texi
File gcc/doc/invoke.texi (right):

http://codereview.appspot.com/4280074/diff/1/gcc/doc/invoke.texi#newcode7777
gcc/doc/invoke.texi:7777: and its related options. Both absolute and relative
can be used.
On 2011/03/28 17:59:32, Diego Novillo wrote:
> s/relative/relative paths/

Done. (Hope the updated patch got sent correctly).

http://codereview.appspot.com/4280074/diff/1/gcc/libgcov.c
File gcc/libgcov.c (right):

http://codereview.appspot.com/4280074/diff/1/gcc/libgcov.c#newcode286
gcc/libgcov.c:286: if (prefix_length != 0 && !IS_DIR_SEPARATOR (*fname))
On 2011/03/28 17:59:32, Diego Novillo wrote:
> I'm confused.  why did you take !HAS_DRIVE_SPEC() out?

Earlier in the function, the following code makes sure we don't start the fname
with a DRIVE_SPEC.

      if (prefix_length != 0 && HAS_DRIVE_SPEC(fname))
        fname += 2;

[Diego Novillo, 2011-03-28 19:47:48]

On Mon, Mar 28, 2011 at 14:14,  <martint@google.com> wrote:

> On 2011/03/28 17:59:32, Diego Novillo wrote:
>>
>> I'm confused.  why did you take !HAS_DRIVE_SPEC() out?
>
> Earlier in the function, the following code makes sure we don't start
> the fname with a DRIVE_SPEC.
>
>      if (prefix_length != 0 && HAS_DRIVE_SPEC(fname))
>        fname += 2;

Ah, quite right.  Thanks.


Diego.

[Diego Novillo, 2011-03-28 22:07:40]

Committed to google/main rev 171635.