Bug 887

Summary:	valgrind errors in ALL tests on glibc-2.10
Product:	ns-3	Reporter:	Andrey Mazo <ahippo>
Component:	general	Assignee:	ns-bugs <ns-bugs>
Status:	RESOLVED INVALID
Severity:	normal	CC:	ahippo, craigdo, tomh
Priority:	P3
Version:	pre-release
Hardware:	PC
OS:	Linux
Attachments:	./test.py --grind --retain --verbose on

Description Andrey Mazo 2010-04-20 06:52:33 UTC

./test.py --grind summary:
0 of 126 tests passed (0 passed, 6 skipped, 0 failed, 0 crashed, 120 valgrind errors)

1) ns-3-dev revision 3aa30fa758f5
2) ./waf configure -d optimized --disable-python (debug/release are also affected)
3) ./test.py --grind

Tested on following machines:
*) gcc 4.3.4, glibc 2.10.1, valgrind 3.4.1, kernel 2.6.32, 64-bit
*) gcc 4.3.4, glibc 2.10.1, valgrind 3.5.0, kernel 2.6.33, 32-bit
*) gcc 4.4.1, glibc 2.10.1, valgrind 3.5.0, kernel 2.6.31, 32-bit
o) gcc 4.3.4, glibc 2.9_p20081201, valgrind 3.4.1, kernel 2.6.31, 64-bit

Tests pass successfully on o) machine only.

Comment 1 Andrey Mazo 2010-04-20 07:59:24 UTC

Created attachment 838 [details]
./test.py --grind --retain --verbose on 

on machine
gcc 4.3.4, glibc 2.10.1, valgrind 3.4.1, kernel 2.6.32, 64-bit

Comment 2 Craig Dowell 2010-04-20 14:26:33 UTC

What is valgrind complaining about?  (i.e., ./test.py -v -g)

I am running kernel 2.6.32, glibc 2.11.1, valgrind 3.5.0 with no problem.

Comment 3 Craig Dowell 2010-04-20 14:35:48 UTC

Sorry, I didn't notice the attachment.

You can always add the offending library errors to the suppressions file (testpy.supp).

Comment 4 Andrey Mazo 2010-04-21 06:20:59 UTC

(In reply to comment #3)
> Sorry, I didn't notice the attachment.
> 
> You can always add the offending library errors to the suppressions file
> (testpy.supp).

I've reinstalled valgrind and it seems to solve the bunch of invalid read errors.
Looks like valgrind produces false errors in case, when glibc was updated after valgrind.

Suppose to close the bug as WORKSFORME.

Comment 5 Andrey Mazo 2010-10-02 09:45:25 UTC

(In reply to comment #4)
> (In reply to comment #3)
> > Sorry, I didn't notice the attachment.
> > 
> > You can always add the offending library errors to the suppressions file
> > (testpy.supp).
> 
> I've reinstalled valgrind and it seems to solve the bunch of invalid read
> errors.
> Looks like valgrind produces false errors in case, when glibc was updated after
> valgrind.
> 
> Suppose to close the bug as WORKSFORME.

Well, the same problem has appeared several times on various Gentoo installations.
And every time just rebuilding valgrind solved the problem.
Looks like it's a Gentoo/valgrind bug, not ns-3 one.
So I'd propose to close this bug.

Comment 6 Tom Henderson 2011-08-09 14:16:08 UTC

moving this to the wiki and marking resolved/invalid since it is not an ns-3 problem:

https://www.nsnam.org/wiki/index.php/Troubleshooting#Valgrind_errors