Bug#462753: libao2: need to clear the final period in alsa driver before closing it.

[Morita Sho]

Thank you so much for your detailed explanation!


I have built libao revision 16861 and installed it to /usr/local.
Then I recompiled reproduce_libao_bug.c with the following command:

% gcc -Wall -I/usr/local/include -L/usr/local/lib -lao -ldl -lm -o reproduce_libao_bug
reproduce_libao_bug.c
% ldd reproduce_libao_bug | grep libao
	libao.so.4 => /usr/local/lib/libao.so.4 (0xb7895000)

This executable works properly. No noise at end of the playback.


I also tested with mmap by the following modification.

% diff -u reproduce_libao_bug.c reproduce_libao_bug_mmap.c
--- reproduce_libao_bug.c	2010-02-02 00:32:20.000000000 +0900
+++ reproduce_libao_bug_mmap.c	2010-02-04 09:23:49.000000000 +0900
@@ -64,6 +64,8 @@
 int main(void)
 {
   ao_initialize();
+  ao_option *o=NULL;
+  ao_append_option(&o,"use_mmap","yes");

   printf("reproduce_libao_bug: Using " AO_DRIVER " driver.\n");

@@ -78,7 +80,7 @@
   format.rate = 22254;
   format.byte_format = AO_FMT_LITTLE;

-  device = ao_open_live(driver_id, &format, NULL);
+  device = ao_open_live(driver_id, &format, o);
   if (device == NULL) {
     fprintf(stderr, "Error opening device.\n");
     return 1;
% gcc -Wall -I/usr/local/include -L/usr/local/lib -lao -ldl -lm -o reproduce_libao_bug_mmap
reproduce_libao_bug_mmap.c

This executable works properly, too.

It seems that both tests work fine. Let me know if you need additional tests.


> A question-- if the kernel is built with genuine OSS (no OSS emulation
> in ALSA, actual OSS) does the sound card still have this problem?

Hmm... I'm not sure because I'm not familiar with OSS...


-- 
Morita Sho <morita-pub-en-debian at inz.sakura.ne.jp>