Announcement

Collapse
No announcement yet.

Valgrind Finds Thousands Of Potential Issues With Mesa

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Valgrind Finds Thousands Of Potential Issues With Mesa

    Phoronix: Valgrind Finds Thousands Of Potential Issues With Mesa

    An open-source developer has spotted thousands of potential memory problems with Mesa when using Valgrind...

    http://www.phoronix.com/vr.php?view=MTQzNjk

  • #2
    "David Airlie has been the only Mesa contributor to respond to the Mesa Valgrind posting yet and he suggests some of the errors may be due to kernel ioctls not known by Valgrind. "We access a lot of kernel data via ioctls that valgrind doesn't understand. I'm not saying that is definitely the case but it was in a lot of places."

    Wow, just wow.

    Instead of saying thank you for finding these problems and calling all available programmers to help the debug, the only poster basically says "there isn't any bugs, your program isn't working right, we're prefect and incapable of generating bugs for we're gods. 100% your fault".

    And these people make the drivers? That should scare you a bit. Humility is a good feature in a person, not a bad one

    Comment


    • #3
      Originally posted by dh04000 View Post
      "David Airlie has been the only Mesa contributor to respond to the Mesa Valgrind posting yet and he suggests some of the errors may be due to kernel ioctls not known by Valgrind. "We access a lot of kernel data via ioctls that valgrind doesn't understand. I'm not saying that is definitely the case but it was in a lot of places."

      Wow, just wow.

      Instead of saying thank you for finding these problems and calling all available programmers to help the debug, the only poster basically says "there isn't any bugs, your program isn't working right, we're prefect and incapable of generating bugs for we're gods. 100% your fault".

      And these people make the drivers? That should scare you a bit. Humility is a good feature in a person, not a bad one
      You are seriously lacking reading skills, dude.
      Or you're just a f**king troll.

      Comment


      • #4
        Originally posted by dh04000 View Post
        Wow, just wow.

        Instead of saying thank you for finding these problems and calling all available programmers to help the debug, the only poster basically says "there isn't any bugs, your program isn't working right, we're prefect and incapable of generating bugs for we're gods. 100% your fault".

        And these people make the drivers? That should scare you a bit. Humility is a good feature in a person, not a bad one
        No, he didn't. He simply said that valgrind might be detecting code it doesn't reconize properly as bugs. Nothing else.

        He never even mentioned the other guy's code. I don't get how you reached the conclusion that he lacks humility unless you're trolling.

        Comment


        • #5
          Originally posted by dh04000 View Post
          Wow, just wow.
          Big Fonts! Melodrama!

          Comment


          • #6
            Originally posted by DanL View Post
            Big Fonts! Melodrama!
            What are we yelling about! Loud Noises!

            Comment


            • #7
              opensource drivers are trash, more at 11.

              Comment


              • #8
                Originally posted by dh04000 View Post
                "David Airlie has been the only Mesa contributor to respond to the Mesa Valgrind posting yet and he suggests some of the errors may be due to kernel ioctls not known by Valgrind. "We access a lot of kernel data via ioctls that valgrind doesn't understand. I'm not saying that is definitely the case but it was in a lot of places."

                Wow, just wow.

                Instead of saying thank you for finding these problems and calling all available programmers to help the debug, the only poster basically says "there isn't any bugs, your program isn't working right, we're prefect and incapable of generating bugs for we're gods. 100% your fault".

                And these people make the drivers? That should scare you a bit. Humility is a good feature in a person, not a bad one
                well valgrind as far as i know can't track down ioctls properly[not without special infrastructure kernel side](same applies to oprofile and traces last time i tried) and he is pointing that out since drivers normally uses a loooot of ioctls, he is not discouraging anyone either but pointing out that every ioctl will probably generate a false positive in valgrind.

                so for now this valgrind outputs should be separated in ioctl/non-ioctl and start tracking them down with more advanced tooling if needed.

                aka a perfectly rational response from a developer

                Comment


                • #9
                  Originally posted by jrch2k8 View Post
                  well valgrind as far as i know can't track down ioctls properly[not without special infrastructure kernel side](same applies to oprofile and traces last time i tried) and he is pointing that out since drivers normally uses a loooot of ioctls, he is not discouraging anyone either but pointing out that every ioctl will probably generate a false positive in valgrind.

                  so for now this valgrind outputs should be separated in ioctl/non-ioctl and start tracking them down with more advanced tooling if needed.

                  aka a perfectly rational response from a developer
                  Valgrind picks up a lot of false positives for software that does anything complicated with memory. I'd expect Mesa to fall into that category. Python supplies a Valgrind suppression file because it throws so many invalid problems in the Python interpreter see http://svn.python.org/projects/pytho...EADME.valgrind for the details.

                  Comment


                  • #10
                    This article is seriously confusing.

                    Valgrind can (and frequently does) report one issue many times, which is probably the main reason the number is so big. I understand no more from it than "valgrind has found some issues".

                    Secondly, as pointed out, valgrind does report false positives.

                    Yes, they should probably be checked, but why the hell make a big deal of it like something major is up? Especially when you don't understand what it's about?

                    I know there's a bit of trolling on these forums but sometimes it seems Michael acts like a troll too...

                    Comment


                    • #11
                      Originally posted by chrisb View Post
                      Valgrind picks up a lot of false positives for software that does anything complicated with memory. I'd expect Mesa to fall into that category. Python supplies a Valgrind suppression file because it throws so many invalid problems in the Python interpreter see http://svn.python.org/projects/pytho...EADME.valgrind for the details.
                      Indeed. The headline should have been: Mesa is not providing a suppression file for Valgrind. But that doesn't sound as sensational though I guess.

                      Comment


                      • #12
                        Originally posted by Cyborg16 View Post
                        This article is seriously confusing.

                        Valgrind can (and frequently does) report one issue many times, which is probably the main reason the number is so big. I understand no more from it than "valgrind has found some issues".

                        Secondly, as pointed out, valgrind does report false positives.

                        Yes, they should probably be checked, but why the hell make a big deal of it like something major is up? Especially when you don't understand what it's about?

                        I know there's a bit of trolling on these forums but sometimes it seems Michael acts like a troll too...
                        +1

                        I remember valgrind having fits with threads before. It would rack up errors every time a thread was merged or destroyed. It's actually one of the reasons I went to virtual timers.

                        Comment


                        • #13
                          libdrm needs to be compiled with --enable-valgrind to suppress these warnings. Film at 11.

                          Comment


                          • #14
                            Originally posted by ickle View Post
                            libdrm needs to be compiled with --enable-valgrind to suppress these warnings. Film at 11.
                            Such sass! For a moment I had to check to make sure I wasn't on Tumblr

                            Comment


                            • #15
                              Originally posted by Cyborg16 View Post
                              I know there's a bit of trolling on these forums but sometimes it seems Michael acts like a troll too...
                              Sometimes? I've reached the stage of generally assuming that's what is going on with all these Linkbait titles and melodramatic ZOMG THE SKY IS FALLING!!!!1111oneoneone articles.

                              Comment

                              Working...
                              X