diff options
author | Jacob Meuser <jakemsr@cvs.openbsd.org> | 2009-09-08 23:30:18 +0000 |
---|---|---|
committer | Jacob Meuser <jakemsr@cvs.openbsd.org> | 2009-09-08 23:30:18 +0000 |
commit | 6aa31a6cfc8d5e5d83670a1418b2c1f7834ce938 (patch) | |
tree | d20c44ecad6ba31186caecd8afdd8dca32442a52 /sys/compat/vax1k | |
parent | 2148386d169a34091eceb12ef3219197881e425b (diff) |
cleanup/fix RIRB processing:
* move the code for azalia_comresp() to be nearer the functions it
uses (azalia_set_command() and azalia_get_response()) for easier
comprehension.
* don't turn off RIRB interrupts in azalia_set_command(), then reenable
them in azalia_get_response(). it's ugly, problematic, and the code
is already at SPL_AUDIO.
* in azalia_get_response(), wait for a solicited response (the response
from azalia_set_command()) instead of "waiting" for just any response
as there may already be an unsolicited response in the RIRB.
* no need to run the unsolicited response handler in azalia_get_response().
since RIRB interrupts aren't being disables, the unsolicited response
handler will be run by the RIRB interrupt handler.
* in azalia_rirb_intr(), when a solicited response is found in the RIRB,
instead of stopping and losing the response, print a debug message
and keep going. there should never be a solicited response in the
RIRB at this point.
* remove some now useless #if 0'd debug cruft.
* minor KNF while here.
Diffstat (limited to 'sys/compat/vax1k')
0 files changed, 0 insertions, 0 deletions