From a150899221a5badff9740703b754f901b4f52762 Mon Sep 17 00:00:00 2001 From: cinap_lenrek Date: Thu, 14 Apr 2011 17:27:24 +0000 Subject: fill /acme --- acme/bin/source/acd/outline | 32 ++++++++++++++++++++++++++++++++ 1 file changed, 32 insertions(+) create mode 100644 acme/bin/source/acd/outline (limited to 'acme/bin/source/acd/outline') diff --git a/acme/bin/source/acd/outline b/acme/bin/source/acd/outline new file mode 100644 index 000000000..5ab337d9c --- /dev/null +++ b/acme/bin/source/acd/outline @@ -0,0 +1,32 @@ +acd is composed of four procs + +wineventproc (win.c:/^wineventproc) + reads acme window events, sends them along w->cevent. + +cdstatusproc (mmc.c:/^cdstatusproc) + reads cd status once per second, sending + status updates to d->cstatus. + detects disk changes, sends new tocs to d->ctocdisp. + +cddbproc (cddb.c:/^cddbproc) + reads tocs from d->cdbreq, if it finds + translations in the cddb, sends new tocs to d->cdbreply. + +eventwatcher (main.c:/^eventwatcher) + the main event loop. + reads status from d->cstatus. + reads events from w->cevent. + reads new tocs to display from d->ctocdisp. + sends new tocs to translate to d->cdbreq. + reads new translated tocs from d->cdbreply. + +an interesting bug in the original design: + both cdstatusproc and the eventwatcher proc + issue scsi commands. (the eventwatcher responds to + things such as Play, Stop, etc., as well as advancing the track.) + + the sd(3) driver did not expect overlapped commands, + and crashed. + + this has been fixed by making the scsi(2) commands threadsafe, + and making the sd(3) driver more robust. -- cgit v1.2.3