Announcement

Collapse
No announcement yet.

Features You Won't Find With The Mainline Linux 4.15 Kernel

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

  • Features You Won't Find With The Mainline Linux 4.15 Kernel

    Phoronix: Features You Won't Find With The Mainline Linux 4.15 Kernel

    While Linux 4.15 is going to be a mega kernel update with its major new features and changes as we have been covering for the past two weeks, there is some functionality that has yet to see the light of day in mainline...

    http://www.phoronix.com/scan.php?pag...erged-Features

  • #2
    A lot of fixes is still missing thanks to wintel developers (mostly) if you have a Ryzen&RX PC and an older intel laptop.

    Patching kernel for a new Ryzen computer

    To prevent random kernel lock ups, enable RCU_NOCB_CPU and boot the kernel with the rcu_nocbs=0-X command line parameter. X is the cpu thread count -1.

    Apply enable_additional_cpu_optimizations_for_gcc_v4.9+_ kernel_v4.13+.patch.

    So for a non debug kernel, enable ACPI_NO_ERROR_MESSAGES. Exporting ACPI_HELP_APP as terminal environment variable does not work.

    ./include/acpi/platform/acenv.h
    /* acpi_help configuration. Error messages disabled. */

    #ifdef ACPI_HELP_APP
    #define ACPI_NO_ERROR_MESSAGES
    #endif
    +#define ACPI_NO_ERROR_MESSAGES

    Change Invalid PCI ROM header signature message to use the dev_info function in drivers/pci/rom.c


    Intel gpu flip timeout error messages:
    Add to the kernel command line: video=SVIDEO-1:d
    This has solved the bug for me on my Latitude D630 with a 4.9 kernel. It disables the SVIDEO connector but my laptop doesen't even have one so it's fine.

    Comment


    • #3
      You also won't see any work from Con Kolivas.

      Comment


      • #4
        Btrfs Raid 5/6 working without issues.

        Comment


        • #5
          Here's a wild suggestion: Report when (if against all odds!) reiser4 gets merged, instead of keeping mentioning it year after year for every kernel release as if it's some kind of surprise that, no, reiser4 wasn't merged this time either?

          Comment


          • #6
            Originally posted by debianxfce View Post
            A lot of fixes is still missing thanks to wintel developers (mostly) if you have a Ryzen&RX PC and an older intel laptop.
            Is there a webpage somewhere with these patches? It would be nice to have them all gathered together somewhere.

            Comment


            • #7
              reiserfs5, reiserfs6, reiserfs7 and reiserfs8

              Comment


              • #8
                Anyone have suggestions of scenarios where LLVM (clang) for Kernel compilation is useful or, perhaps even, necessary?

                Comment


                • #9
                  Just to keep unnecessary GCC'isms from creeping into kernel?

                  Comment


                  • #10
                    Originally posted by azdaha View Post
                    Anyone have suggestions of scenarios where LLVM (clang) for Kernel compilation is useful or, perhaps even, necessary?
                    Embed.
                    Currently, there seems to be a lot of development going around LLVM (mostly due to its high modularity, but also due to having a license that is more corporate-friendly rather than the end-user-friendly GPL found in GCC).
                    Means that when new CPU architecture arises (as recently with AArch64), there might be a good chance that the corporation spin their own implementation as an LLVM back-end, while GCC has some delays until supporting it.
                    (Even more likely if the company introducing the new architecture is like Apple a little-bit allergic to licenses which are too end-user-friendly / not-enough corporate-friendly).
                    The kernel being compilable with LLVM compilation would make a possible opportunity to start port the Linux kernel to the new arch even before GCC is done supporting the new arch.

                    Also: bug fixes.
                    There might be a few bugs lost somewhere, due to not formally correct C, that happens to work due to how GCC compile the C code, but would lead to compile error or apaprent bugs, due to how LLVM could differently compile the code.
                    (Mostly around behaviours which are "undefined" in C standards. Though recent versions of GCC can throw warnings or errors in these cases)

                    Comment

                    Working...
                    X