Re: code review 5933055: nix: sync with plan 9 (issue 5933055)

[0in...@gmail.com]

Hello nixi...@gmail.com, ne...@lsub.org, quan...@quanstro.net (cc:
nix...@googlegroups.com),

Please take another look.

http://codereview.appspot.com/5933055/

[erik quanstrom]

i just spent a couple of hrs reviewing these changes,
then reitveld barfed on my shoes and lost all the changes.

Sorry, an error occurred...
Unhandled exception.
Details: too many values to unpack [ValueError]
Please post a message to the discussion group if you have any questions.

unfortunately, there isn't time to recount all of the comments.
i'll try to documente a few of the quick ones first.

i don't understand the stray references to "9boot".
i am not going to make any of these changes locally
because i plan on using cniap's 9boot to replace 9load.
(it isn't enough at this point, but it does use undi which
is awsome)

i also think the changes to usbohci don't look correct.
- disabling interrupts in the interrupt routine looks bogus.
the part shouldn't be issuing interrupts between issuing an
interrupt and rx'ing the eoi. ohci is *level triggered* anyway,
disabling interrupts can't do anything. if this helps, it is masking a problem.

- all the added coherence() statements appear to be masking
problems. they are all inserted after non-posted pci writes,
which are already coherent.

i would nack the ohci change, and ask for clarification.

- erik

[erik quanstrom]

can you resubmit this cl without the fortunes file, and make sure
that nothing's broken about the patch.

thanks.

- erik

[0in...@gmail.com]

*** Abandoned ***

http://codereview.appspot.com/5933055/