Status for Andrew Reiter's work on FreeBSD

05/06/2005 (1004 PDT)
  - Personal tree modifications:
    - Added sysctl's for reporting number of kobj classes initialized &
      number of kobj classes compiled.  Was trying to determine worth of
      protecting kobj_class's reference counts.
    - Added kobj_class_ref() and kobj_class_rele(), but seems like overkill
      from the perspective of one fbsd developer (forget who).
  - I need to track the ref count paths (again) and then look at the kobj
    API to determine if changes can be made without affecting bin compat.
  - Looked into some bugs in the kernel linker (possible bugs) but got
    sidetracked by kobj.  Will finish kobj work first; propose patch.

01/16/2005 (2145 PST)
  - Review and documenting kernel code paths
  - documentation of tools/ and such
  - kernel file and function documentation (cscope of any use?)

01/02/2005 (2340 PST)

 - Rebuilding Firefox made things nice.  Not sure if it was the rebuild
    itself or rebuild with -ggdb :-/  I hate that.
 - Pretty much done with vop_table.pl.  There are tweaks that need to be
    made, but definitely getting there.  Emailed initial output and
    script to phk@ in hopes of some "what the hell are you thinking" types
    of responses.  Need to get working on the save / close stuff regarding
    GraphView and schedgraph.py from jeff@.


12/30/2004 (0300 GMT)

 - Firefox keeps crashing on me :-(  Rebuilding from scratch.. gonna see
    what happens.
 - > 50% done with vop_table.tcl rewrite.  it's in perl... current todo is:
	 - revamp so as to pick up new vop_* routines that are missed from
	   vnode_if.src
	 - redo output to html file so as to fix up table when other vop
	   routines are found that were not found in earlier fs' analyzed.

12/29/2004 (1133 PST)

- In the middle of a re-write of vop_table.tcl.
- Complete the started modifications to schedgraph.py


12/28/2004 (1159 PST)

- IDA Linux works under FreeBSD 6.x with the linux_base-8 compat layer
  installed, also, need linprocfs mounted.

12/27/2004 (1500 PST)

Add usage notes to schedgraph.py

12/07/2004 (08:26AM PST)

Been moving my D400 to CURRENT from 10/25/2004 and have encountered a 
few bugs here and there.  One was a result of a commit on 11/09/2004 where
we set default do_powerstate value to 1... one of my integrated peripherals
didn't like this and would hang on boot.  Setting hw.pci.do_powerstate=0 in
/boot/loader.conf fixes hang.  Thanks for help from Warner Losh; I've sent
him info so there can be code to handle this scenario with the default set
to 1.  Need to work on a few more bugs that I have in my queue, namely the
recursion on the rtentry lock.  Lots of "real" work stuff to do lately... 


11/15/2004 (09:34AM PST)

Failed to do much on Saturday afternoon as I was really tired.  Got some
beginning debug related things going for Audit as well as the beginning of
some design/todo docs that I would like wsalamon to look at and perhaps
we can associate dates with each items.

Really tired this morning and have lots of "real" work to do.  I will also, 
however be looking at the pecoff code and see if it's something we can 
battle axe out of the repo.


11/13/2004 (12:45PM PST)

Busy working lately so not much of a chance to progress on Audit.  However,
coming into work today in hopes of accomplishing some; mostly getting a
few basic things working + a todo plan of action?  Something along those 
lines. 

Reviewing audit code again... Need to refamiliarize myself with where I was
working last.  Going to (again) create a todo list of items as I see things
come up.


11/02/2004 (09:30AM PST)

Just returned from Vancouver, BC ... Beautiful area.  Back to work!
Working on integrating FreeBSD vendor branch changes into trustedbsd 
base branch and then get the changes up to trustedbsd_audit3 branch.  
Wayne Soloman noticed a breakage after my last integ and we believe it was
due to bad timing coming off of the FreeBSD vendor branch.  Let's hope
this one goes ok.  Also, updating my vaio laptop to RELENG_5_3 again in
hopes of finding more bugs.  Until next time...

Also got feedback about pr kern/73066 which I was able to close due to being
invalid.


10/27/2004 (11:11AM PDT)

Started looking at Problem Reports related to 5.3 that I could perhaps
deal with now that we're standing at the end of a gun with regards to
releasing this version.  I submitted a couple yesterday (kern/73177 and
kern/73171) relating to another geom_* panic and the firewire card issues
I've been seeing.  The geom_ panic seems to be that there is a weird loop
being caused that keeps registering devices (? need to look closer ?) and
ends up exhausting memory... kmem_malloc() panics.  

One problem report I responded to was kern/73066.  The guy states that 
his ps/2 mouse is "not working" and without much information ... leaves it
at that.  Ah well, I hope it's just that he did not config atkbdc and not
some greater ordeal.  Until next time..



10/25/2004 (3:14PM PDT)

Played with trustedbsd_audit3 tree a bit over the weekend; mostly code & 
design review.  Committed some silly changes over some man files to remove
references to Mac OS X.  Began integ of trustedbsd_base branch into the
trustedbsd_audit3 branch.  I found that using world + kernel from audit3
disallowed my bge nic from working; also, we'll be doing more and more work
on this branch, so it's time to get it more up to date so that the diffs from
current will not be as massive.  Started the integ / resolve process, so
we'll see what happens (whether or not I'll get it done before I have to leave
work).


10/21/2004 (1:39PM PDT)

Popped in firewire card  into my new dell laptop and received same errors
as I did last week on VAIO.  Going to instrument (again) and this time
save my diff (@#$@$).


10/21/2004 (12:59PM PDT)

While getting my new laptop ready, I, on the VAIO, kldload'd fdc and it
caused a panic in GEOM.  Generated a patch; but it needs to be reviewed
by Poul-Henning prior to use or commit.  It should be ok, however.
  http://people.freebsd.org/~arr/geom_class_loaded.diff

10/15/2004 (2:06PM PDT)

Began instrumenting cardbus code so as to figure out why it seems it is
mapping our firewire card incorrectly.  Since I don't have a setup for this
laptop where I can use GDB easily (work is bad setup), I'm doing idiotic
DEVPRINTF() additions.  Ah well.  Should give me some sense of what's going
on. ... Hope to update this sooner than later!

10/14/2004 (lunchtime)

Was having issues with cardbus.  I had thought kernel was -CURRENT, but in 
fact I missed a commit from Warner Losh that fixed my problems.  Updating and
recompiling fixed cardbus.  Now my Texas Instruments firewire card is
reporting a "bad OHCI version", according to the fwohci driver in CURRENT.
Going to look at linux's ieee1394 driver to see what they do.  Unfortunately,
my firewire systems architecture book is at home.

Until then...