Announcement

Collapse
No announcement yet.

Linux Developers Look At Upping The GCC Requirements For Building The Kernel

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

  • #11
    Originally posted by starshipeleven View Post
    Nothing. RH has their own dev teams to do that.

    In whatever RHEL. That's a server, if they were working fine 5 years ago they will still be fine now. They only need security/stability patches.
    Got that (more or less). Just missing why *he* is concerned with making changes upstream, in regard to that. Or does he work for RH's dev teams?

    Comment


    • #12
      Originally posted by Xelix View Post
      Why not? One could want a recent kernel to get recent features/support for recent hardware but also keep a conservative userland. In this case, running a recent kernel on an older distribution makes sense.
      ...
      I am not saying that this is what most people will want, but I can definitely picture some users wanting this.
      Well, on RHEL you find a disproportionate amount of users that don't need that.

      RHEL costs like 300+ per physical machine (more or even much more depending on the company-grade support contract you add to it, I don't know the prices for virtual instances), and the whole point of it is guaranteeing that whatever works when the server was turned on the first time will still work exactly the same over the years.

      Servers usually live with the same OS until they get scrapped, which is (was) around a decade, and apart from very rare cases they don't need to support recent hardware (there are also quite a bit of proprietary drivers targeting RHEL), and even more rarely they need more recent kernel features that RH didn't backport.


      I'm just trying to point out that RHEL isn't an average linux distro, it has a very specific use-case in mind, and for that specific use-case you don't really need a very new kernel.

      If you just want to "get recent features/support for recent hardware but also keep a conservative userland", you can use Debian Stable + backport repository, as they keep the latest stable kernel release in there.

      Comment


      • #13
        Originally posted by indepe View Post
        Got that (more or less). Just missing why *he* is concerned with making changes upstream, in regard to that. Or does he work for RH's dev teams?
        Afaik Arnd Bergman does not work for RH, he works for Linaro https://www.linkedin.com/in/arnd-bergmann-6857b36

        Comment


        • #14
          Originally posted by starshipeleven View Post
          RHEL costs like 300+ per physical machine
          or zero in case of centos

          Comment


          • #15
            Originally posted by indepe View Post

            Got that (more or less). Just missing why *he* is concerned with making changes upstream, in regard to that. Or does he work for RH's dev teams?
            No. It is not just Red Hat either. The point is to introduce changes with the least amount of pain for the major users.

            Comment


            • #16
              Originally posted by Xelix View Post
              Why not? One could want a recent kernel to get recent features/support for recent hardware but also keep a conservative userland.
              Because if you do this, you loose RedHat support.
              So those who want to do this would use Fedora/CentOS/Debian/.. not RHEL where you pay for the support..


              Comment


              • #17
                Originally posted by RahulSundaram View Post

                No. It is not just Red Hat either. The point is to introduce changes with the least amount of pain for the major users.
                Of course that's always the point. But how does that apply here?

                He writes that he also found some "actual" bugs, which sounds like an unquestionably good thing.

                Then, in asking whether the kernel's minimum GCC requirement should be "raised" to 4.3 or 4.6, he seems to say that support for 4.3 actually isn't given anymore and re-instating it requires some work, but apparently without waiting for an answer, he starts producing patches meant to back-support GCC 4.3. I'd think that RHEL6, Debian 6, Ubuntu 10 and SLES11 must have already dealt with those issues, and any back and forth now might simply cause double-pain. Unless there is a better explanation. Debian stable is now at 8, and Debian 6 is marked as "obsolete". Now start supporting it upstream?

                Comment


                • #18
                  Originally posted by rubdos View Post

                  To elaborate on @starshipeleven's answer: Red Hat uses a kernel from the 2.X era's and backports bugfixes to it. Same for RHEL 7; they're on Linux 3.10.xxx, which is 3.10 with some btrfs backports and bugfixes.
                  WHY???!?!?!??

                  Why so much ancient stuff? There should be a policy about having newer kernels. This way, even Red Hat would work harder to make mainline more robust inread backporting everywhere.

                  Why support REALLY ANCIENT GCC versions? Those distributions should update certain critical software: Kernel, GCC, FFmpeg (is libav vs FFmpeg mess still alive?), Xorg MESA, LibreOffice, Firefox, Chromium...

                  Why? Because for security and having edecent features.

                  Comment


                  • #19
                    Originally posted by timofonic View Post

                    WHY???!?!?!??

                    Why so much ancient stuff? There should be a policy about having newer kernels. This way, even Red Hat would work harder to make mainline more robust inread backporting everywhere.

                    Why support REALLY ANCIENT GCC versions? Those distributions should update certain critical software: Kernel, GCC, FFmpeg (is libav vs FFmpeg mess still alive?), Xorg MESA, LibreOffice, Firefox, Chromium...

                    Why? Because for security and having edecent features.
                    Because instead of asking if you can run a newer kernel in a company with 1,000 users depending on the mainframe you should ask "should i?"

                    Because newer isnt always better in every use case.

                    IT and Server Administration is a business and not a hobby. Taking risks is a sure way to loose profits when your foundation is working rock solid.

                    Comment


                    • #20
                      It appears the last time a new iso-file for Ubuntu 10 was produced, was in 2012. SLES11 however is still active. So it seems there are two, RHEL6 and SLES11, not four?

                      Are those two asking for functions like ACCESS_ONCE to be changed in that manner? Do they want upstream to support a 10-year old version of GCC? (Life cycle of RHEL.) Or does RH want Fedora with Wayland and Vulkan to have a kernel using modern compiler technology?

                      [Edit: 2012, not 2010]
                      Last edited by indepe; 16 December 2016, 11:37 PM.

                      Comment

                      Working...
                      X