Permalink
Commits on May 28, 2012
  1. Re-enable IO/APIC

    I had disabled the IO/APIC initialization, but after a bit of testing
    decided it seems best with the initialization on.
    committed May 28, 2012
  2. Fix numerous baremetal bugs

    All tests seem to be working now. There were several subtle bugs that
    required some modifications.
    
    - The event location is assumed to be linear for a lot of our code but
      the apic id is not linear, so some mapping had to be done.
    
    - There was an issue where we forgot to zero out the IDT (It was
      switched from bss to allocated recently). On hardware this caused a
      general protection fault that was quite hard to track down
    
    The outstanding concern is with the ioapic which isn't being
    initialized. It should initialize to reasonable defaults (no IRQ
    routing). The issue is that this IO APIC has a different version than
    the one I setup with, so I am apprehensive about just initializing it
    the same way as the other version.
    committed May 28, 2012
  3. Remove ACPIA

    The ACPICA code was ripped out and replaced by our own implementation
    where we find the table. I discovered that we indeed need to map more
    memory on kd because of the existence of a table in higher memory.
    committed May 25, 2012
Commits on May 25, 2012
  1. Merge pull request #17 from dschatzberg/forSESA

    x86 baremetal update
    sesauser committed May 25, 2012
  2. Boot using KVM on kd

    There were a couple bugs:
    
    1. An optimization of the loading of the gdt using KD's compiler. This
    has been fixed with the use of volatiles
    
    2. The init IPI was not level asserted and not being responded to.
    committed May 25, 2012
Commits on May 24, 2012
  1. notes for running on fusion

    okrieg committed May 24, 2012
  2. fixed parcing of ACPI MADT table to properly tell us the number of co…

    …res to work on vmware as well
    okrieg committed with May 24, 2012
  3. Get msgtst working on bochs

    There was a bug in the pic implementation
    committed May 24, 2012
Commits on May 23, 2012
  1. Fix implementation of lrt_next_event_loc

    Now the cache invalidation doesn't loop indefinitely
    Dan Schatzberg committed May 23, 2012
  2. Add multicore bringup on x86 baremetal

    All cores get brought up now as we do on ulnx. There is still a bug
    with respect to cache invalidation that has to be fixed
    Dan Schatzberg committed May 23, 2012
Commits on May 22, 2012
  1. Fix distcheck

    There were some missing headers in makefiles
    Dan Schatzberg committed May 22, 2012
  2. Refactor for multicore compliance

    This change refactors a lot of the code so that the bringup on a
    single core is more in line with the way it will be once we actually
    add multiple cores.
    Dan Schatzberg committed May 22, 2012
Commits on May 21, 2012
  1. Change lrt_mem_alloc to receive location

    The lrt_mem_allocator now gets told which location to allocate
    from. This is necessary because early on we do not have the facilities
    to get our event location the standard way yet we know the boot core
    is core 0.
    Dan Schatzberg committed May 21, 2012
  2. Refactor x86 baremetal bringup

    Still single core but starting to refactor boot memory vs higher level
    memory allocation. Mem_preinit and event_preinit have been
    implemented.
    Dan Schatzberg committed May 21, 2012
Commits on May 18, 2012
  1. Add event implementation

    msgtst now works (single core of course), event binding and trigger works
    Dan Schatzberg committed May 18, 2012
  2. Fix x86 baremetal compilation issues

    The binary will be compiled and helloworld works (sort of). None of
    the event system is setup but that isn't required for helloworld and
    some others.
    Dan Schatzberg committed May 18, 2012
  3. Modify EBBMissFunc type

    The type is now composed of the other EBB types. I added a static
    assert that checks that it is compatible with lrt_trans_miss_func.
    Dan Schatzberg committed May 18, 2012
  4. Merge pull request #16 from dschatzberg/forSESA

    Small header removal and change lrt_start location
    sesauser committed May 18, 2012
Commits on May 17, 2012
  1. Move lrt_start into l0/lrt/...

    This makes it consistent with the entry point on all platforms
    Dan Schatzberg committed May 17, 2012
  2. Removed trans-def.h

    This header was unnecessary and the definition is moved into trans.h
    Dan Schatzberg committed May 17, 2012
  3. Merge pull request #15 from okrieg/development

    Lots of cleanup of translation system
    sesauser committed May 17, 2012
  4. Fix global translation table zeroing

    We were only zeroing out the memory for core 0 and not the other
    cores, but this could cause the other cores to clear allocated
    entries.
    Dan Schatzberg committed May 17, 2012
  5. - fixed my gmem function

    okrieg committed May 17, 2012
  6. merged changes and got compiling

    Conflicts:
    	apps/eventtst/eventtst.c
    	l0/EventMgrPrim.h
    okrieg committed May 17, 2012
  7. Halfway to trans cleanup

    This is a broken commit so that Orran and I can continue working on
    cleaning up the translation stuff
    Dan Schatzberg committed May 17, 2012
Commits on May 16, 2012
  1. got rid of EventHandler type

    okrieg committed May 16, 2012
  2. - incremental commit for dan

    okrieg committed May 16, 2012
  3. Fix bug in broadcast event trigger

    There was an issue with aliasing the number of event locations and the
    requested event number that would cause a bug. A test needs to be
    written to make sure it works anyhow.
    Dan Schatzberg committed May 16, 2012
  4. Fix missing rc assert

    There was a missing assert in the eventtst on the rc, causing an
    unused variable warning
    Dan Schatzberg committed May 16, 2012
  5. Cleanup trans calls

    There were some remaining calls to the translation initialization that
    needed to be put more in line with the way they should be used.
    Dan Schatzberg committed May 15, 2012
  6. Make distcheck happy

    Now the files left due to the regression will be cleaned up. I also
    added some additional headers that were missing
    Dan Schatzberg committed May 15, 2012
  7. Cleanup trans headers

    Move the l0/types header in l0/lrt/trans.h, we can separate further
    from there. Moved all of l0/sys into l0/lrt/...
    Dan Schatzberg committed May 15, 2012
Commits on May 15, 2012
  1. added exit object to eventtst

    okrieg committed May 15, 2012