[Bug c++/11773] New: GCOV misreports coverage for in-class C++ member functions

[Bug c++/11773] New: GCOV misreports coverage for in-class C++ member functions

[yuri at tsoft dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11773

           Summary: GCOV misreports coverage for in-class C++ member
                    functions
           Product: gcc
           Version: 3.2
            Status: UNCONFIRMED
          Severity: critical
          Priority: P3
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: yuri at tsoft dot com
                CC: gcc-bugs at gcc dot gnu dot org,yuri at tsoft dot com

Code (see uuencoded file src.tgz at the end of the message)
being compiled with command:
> g++ -fprofile-arcs -ftest-coverage *.cc -o tst
Then program is being run:
> ./tst
Then gcov being run:
> gcov -l a.cc
> gcov -l b.cc

WRONG RESULT OBTAINED:
in b.cc.a.h.gcov I find:
 <snip>
                void
      ######    a() {
      ######      printf("AA\n");
                }
 <snip>
this report means that function A::a() isn't used in b.cc
In fact as seen in sources it's invoked from b.cc.

Thanx,
Yuri.

--------- code ----------------------------------------------
begin 644 src.tgz
M'XL(`)X0+#\``^V644^#,!"`>>ZON.#+9A3;44;BC,GP;_A2"@K)9`N%1;/L
MOWNLFT9T\@1;LOM>#CB@38_OBBGUG=,SG$L>!@%&SL-P:N-4[N(>!Q/"YX)/
M)L+!$(3<@:#OB374IE(E@/-1E_G_]Z6E&6)"PV*P_LK3NL\QL)Y\*N71^D^$
M;-5?^$(XP/N<U($+K_]57NA%G:3@*B]SV?=IW)RR]+U*RP+6RSP!/1K/&,N+
MBKVIO!CA`:CR5=^`7A:F`IWA.E[CE?48-@Q@#FJ&(8(8GP)0GFJ>!XB]V![@
M9[=[(]@W;]FIU^(2:?R/3^R_D/YO_WWR?P@Z_&>-^0S]_$OI58D]X&7DWB+/
MA3MN>TY:GS]V_\]Z':-[_V___PDA0_)_"+Z$?S!5DB^][)$QO5#&H.P;MJKC
M1:[O;1=0^RYPT'X^M])O]UTB;N>CGWG=SC\=\ML9-8D38??_\_,_X.3_$!SW
H/^KR/^KP/^KP/R+_"8(@"((@"((@"((@"((@"*(?/@&S1(U$`"@`````
`
end

[Bug c++/11773] GCOV misreports coverage for in-class C++ member functions

[pinskia at physics dot uc dot edu]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11773


pinskia at physics dot uc dot edu changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|3.4                         |---

[Bug c++/11773] GCOV misreports coverage for in-class C++ member functions

[pinskia at physics dot uc dot edu]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11773


pinskia at physics dot uc dot edu changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|critical                    |normal

[Bug c++/11773] GCOV misreports coverage for in-class C++ member functions

[yuri at tsoft dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11773



------- Additional Comments From yuri at tsoft dot com  2003-08-07 07:21 -------
>From yuri@tsoft.com:
CORRECTION (reformulation):
After some experimentation I figured that
####### actually means "used" butwhycounter didn't appear?
Is this a bug?
In another header actual counters appeared.

[Bug c++/11773] GCOV misreports coverage for in-class C++ member functions

[dhazeghi at yahoo dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11773



------- Additional Comments From dhazeghi at yahoo dot com  2003-08-24 22:08 -------
Created an attachment (id=4647)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=4647&action=view)
testcases

Please use bugzilla's "create new attachment" to add files to bug reports,
rather than including them in the message. Not only does it make my life
easier, it also makes the audit trail more intelligible.

[Bug c++/11773] GCOV misreports coverage for in-class C++ member functions

[dhazeghi at yahoo dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11773


dhazeghi at yahoo dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |nathan at codesourcery dot
                   |                            |com


------- Additional Comments From dhazeghi at yahoo dot com  2003-08-24 22:12 -------
Nathan, would you mind clarify whether the reported behavior is a bug? Thanks.

[Bug c++/11773] GCOV misreports coverage for in-class C++ member functions

[nathan at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11773


nathan at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |nathan at gcc dot gnu dot
                   |dot org                     |org
             Status|UNCONFIRMED                 |ASSIGNED


------- Additional Comments From nathan at gcc dot gnu dot org  2003-08-24 22:16 -------
will investigate

[Bug c++/11773] GCOV misreports coverage for in-class C++ member functions

[nathan at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11773



------- Additional Comments From nathan at gcc dot gnu dot org  2003-09-06 18:53 -------
not a bug, but an outcome of inline functions and weak linkage. I used
the current mainline, which mangles gcov file names differently.

You'll see (well I did) that a.cc##a.h contains 2 counts
for A::a. so overall, the execution counts are correct, just not where
you expected them. What is happening is this,
1) you're compiling without optimization, thus function calls are not inlined.
2) both a.cc and b.cc need to emit a body of A::a, which they each call.
3) those A::a functions are emitted with weak linkage, which means one of them
will 'win' in the linking phase.
4) at the linking phase the A::a emitted in a.o is selected, and the call of
A::a in b.o ends up calling the body of A::a that was emitted in a.o
5) thus the body of A::a emitted in b.o is never called.

need to write something in the documentation about this.

[Bug c++/11773] GCOV misreports coverage for in-class C++ member functions

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |documentation
   Last reconfirmed|0000-00-00 00:00:00         |2003-11-28 08:10:49
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11773

[Bug c++/11773] GCOV misreports coverage for in-class C++ member functions

[nathan at gcc dot gnu dot org]

------- Additional Comments From nathan at gcc dot gnu dot org  2003-12-15 14:26 -------
2003-12-15  Nathan Sidwell  <nathan@codesourcery.com>

	PR c++/11773
	* doc/gcov.texi (Gcov and Optimization): Document inline function
	behaviour. Fix some file suffixes.


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11773

[Bug c++/11773] GCOV misreports coverage for in-class C++ member functions

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2003-12-15 14:29 -------
Subject: Bug 11773

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	nathan@gcc.gnu.org	2003-12-15 14:29:11

Modified files:
	gcc            : ChangeLog 
	gcc/doc        : gcov.texi 

Log message:
	PR c++/11773
	* doc/gcov.texi (Gcov and Optimization): Document inline function
	behaviour.  Fix some file suffixes.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.1995&r2=2.1996
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/doc/gcov.texi.diff?cvsroot=gcc&r1=1.18&r2=1.19



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11773