What: 94th HelenOS project meeting
When: Dec 17, 2016
Where: http://www.maneska.com/
Who: MD, VH, JJ
Scribe: JJ

Round table:

MD:
  - temporary fix for #669
  - planning to make a HelenOS release after January 15

VH:
  - WiP: rewrite of the CI script into Python

JJ:
  - sun4v status: the sun4v support will be hopefully merged into QEMU
    after 2.8, HelenOS already runs on this new QEMU sun4v machine

  - sun4u status
    - demo of lp:~jakub/helenos/sun4u
    - main idea: be able to have usable sun4u QEMU machine with
      serial console
    - collateral damage: all platforms should be able to have at least
      a serial userspace console merely by providing a serial port
      driver which puts itself into the 'serial' category
    - IRQ window
      - needed for IRQ remapping on sparc64 (PCI irq 0x2b becomes
        system-wide IRQ 0x7eb)
      - similar to the current PIO window provided by the parent driver
      - not currently implemented due to a workaround in the ISA driver
        (use dedicated configuration file for EBUS with the IRQ already
        remapped)
    - EBUS vs. ISA
      - using the ISA driver for EBUS with alternate configuration
    - serial devices used as input and output devices and waiting for
      them
      - input currently considers all serial character devices that
        put themselves into the 'serial' category
      - output currently considers the first character device to put
        itself into the 'serial' category
      - need to have a list (in a per-platform config file or boot
        arguments) that says what serial devices can be used as the
        console and consider/wait only for those

  - Ultra 5+60 update:
    - added external NVRAM battery, both systems appear stable

  - FOSDEM
    - JJ will do a HelenOS yearly update talk at FOSDEM 2017

  - unrelated thoughts and recommendations
    - we should support multiple answerboxes per task and pair them
      with async ports
    - we could replace kernel's B+tree with the RB-tree from uspace for
      more accurate memory reservations (unlike the B+tree, the RB-tree
      does not allocate memory on its own)
    - we should: provide async_ wrappers for purely asynchronous IPC
      with callbacks


_______________________________________________
HelenOS-devel mailing list
HelenOS-devel@lists.modry.cz
http://lists.modry.cz/listinfo/helenos-devel

Reply via email to