Announcement

Collapse
No announcement yet.

SUSE Has Been Working On An In-Kernel Boot Splash Screen For Linux

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

  • #31
    Originally posted by Adarion View Post
    OMG. More splash screens that hide all the interesting and important information. I hate them. I hate BIOS/UEFI splashes, I had bootloader, kernel or init splashes. I need to see what is going on - especially if there are warnings or something goes really wrong with an error.
    Same general requirement here (need to see stuff to troubleshoot), but I don't see how dumb end users (that can't understand the text anyway) would need to see that stuff too. This is just a kernel option that can be enabled/disabled in the kernel command line.

    The flickering isn't entirely nice, but for that one would have to have something like "K"MS in the firmware (coreboot maybe?). But this needs info on the GPU and the attached display devices and that probably is already a lot of "higher level" code.
    For the love of Stallman, no. Don't add complex high-level stuff in board firmware. It's guaranteed to be implemented like total shit in dozens of different ways requiring workarounds.

    Comment


    • #32
      Originally posted by dwagner View Post
      I'm astonished to see this topic not only is considered worth kernel developer effort and an article on Phoronix, but also receives numerous comments.
      How often do you boot that a "splash screen" is of concern to you?
      Is your laptop always on or on suspend-to-ram?

      Comment


      • #33
        Originally posted by cjcox View Post
        Not sure about the "effort" in question, but I do know my hosts are happier once Plymouth is removed. Bloated pig that does more harm than good. Maybe the systemd folks could get involved so that boot messages are totally useless? They'd be against the graphical boot screen, instead they'd just present a single '.' to let you know it's booting (how useful!!). They would called it "booted". Of course after boot, you could use bootedctl --x9to5 to see the boot messages (because we all know they are worthless to see until after successful boot occurs).
        The boot-time message should obviously be logged to the screen in a binary format.

        Comment


        • #34
          Originally posted by starshipeleven View Post
          For the love of Stallman, no. Don't add complex high-level stuff in board firmware. It's guaranteed to be implemented like total shit in dozens of different ways requiring workarounds.
          This. BIOS is stupid in so many ways, but at least it's very simple and limited, which bounds the stupidity. UEFI, while in some respects looking good on paper, doesn't suffer from many of these limitations and thus enables a whole new level of stupid.

          I think the only sane position is that the firmware should do the absolute bare minimum necessary to fire up something sane under user control. Such as e.g. a minimal Linux kernel which can then, in turn, handle all the fancy stuff like booting the real distro kernel via the network, or on RAID/LVM/ZFS etc. setups.

          UEFI can go DIAF.

          Comment


          • #35
            Originally posted by jrch2k8 View Post
            I can say for certain every ARM kernel have to be compiled specifically for each device and each device have its own personal custom kernel version
            not really. there is generic arm kernel which supports many devices(and more with each release).

            Comment


            • #36
              Originally posted by dwagner View Post
              How often do you boot that a "splash screen" is of concern to you?
              maybe you are just not this patchset's target audience

              Comment


              • #37
                Originally posted by pal666 View Post
                not really. there is generic arm kernel which supports many devices(and more with each release).
                not always outside very old hardware and in many cases you still need a custom kernel or a truck full of patches to make them boot or even in some cases it boots with mainline but a lot of functionality is missing(like cameras, screen, i2c, sensors, gpio, etc.)

                Sure there are few ARM systems that work out of the box like the rpi1 but they can be counted with 1 hand

                Comment


                • #38
                  Originally posted by starshipeleven View Post
                  Is your laptop always on or on suspend-to-ram?
                  Indeed, during those weeks of the year when I use it, my laptop is either on or in S3 sleep.

                  Comment


                  • #39
                    Originally posted by gQuigs View Post
                    Why is this based on FB and not DRM/KMS? IIRC they keep trying to kill FB and this would make another dependency for it.
                    Maybe the framebuffer deprecation is meant only for the user space and the desktop, but for embedded systems it's pretty useful, Linux use in embedded systems is considerable, almost all desktop users don't need or want to see the boot terminal, but is always nice to have options.

                    Comment

                    Working...
                    X