Announcement

Collapse
No announcement yet.

Allwinner Continues Work On Linux Patches To Dump Kernel Errors To Block Devices

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

  • Allwinner Continues Work On Linux Patches To Dump Kernel Errors To Block Devices

    Phoronix: Allwinner Continues Work On Linux Patches To Dump Kernel Errors To Block Devices

    While Allwinner Technology isn't known as one of the most gracious contributors to the Linux kernel, their continued work on the "pstore_block" kernel patches will be of interest to many especially in the ARM/embedded space and just not for those using Allwinner SoCs...

    http://www.phoronix.com/scan.php?pag...store-Block-V6

  • #2
    I always wondered by it was so hard to write a dump to disk. I mean, create a small partition, pass it via a kernel cmd line argument, then kernel can gather the required information during normal start, setup a few direct write calls that just need to be triggered. Should be easy.

    Comment


    • #3
      Dump kernel logs or kernel image?
      Why to disc? I'd choose a less fault friendly interface if the kernel has gone to bits.
      Serial console log when debugging? Has always worked for me.

      Is this a customer site debugging possibility they are talking about?

      Comment


      • #4
        Originally posted by milkylainen View Post
        Dump kernel logs or kernel image?
        Why to disc? I'd choose a less fault friendly interface if the kernel has gone to bits.
        Serial console log when debugging? Has always worked for me.

        Is this a customer site debugging possibility they are talking about?
        Doesn't work when you have no serial pin hooked up, and debugging a kernel panic, though. I tried to use pstore (and last_dmesg, which stores it to ram) a few months earlier, when I started works towards upstreaming a device for postmarketos. It didn't have dedicated pstore, it could have been handy to be able to specify a block device (with offsets, please ^^"). But yeah, otherwise, serial is usually the way to go, unless you're debugging serial, or a panic that happens earlier. I don't think it's customer-facing.

        Comment


        • #5
          Originally posted by [email protected] View Post

          Doesn't work when you have no serial pin hooked up, and debugging a kernel panic, though. I tried to use pstore (and last_dmesg, which stores it to ram) a few months earlier, when I started works towards upstreaming a device for postmarketos. It didn't have dedicated pstore, it could have been handy to be able to specify a block device (with offsets, please ^^"). But yeah, otherwise, serial is usually the way to go, unless you're debugging serial, or a panic that happens earlier. I don't think it's customer-facing.
          Didn't pstore ram work for you? Dump it to disk once the machine has rebooted? Maybe the bootloader trashed all RAM on init?

          Comment

          Working...
          X