It seems to me that the following 'return 0' statements are unnecessary.
This is just an observation, I didn't actually test-run the patch.


Index: mio_rmidi.c
===================================================================
RCS file: /cvs/src/lib/libsndio/mio_rmidi.c,v
retrieving revision 1.6
diff -u -r1.6 mio_rmidi.c
--- mio_rmidi.c	24 Apr 2010 06:15:54 -0000	1.6
+++ mio_rmidi.c	21 Jul 2010 16:13:57 -0000
@@ -136,7 +136,6 @@
 		if (errno != EAGAIN) {
 			DPERROR("rmidi_write: write");
 			hdl->mio.eof = 1;
-			return 0;
 		}
  		return 0;
 	}
Index: mio_thru.c
===================================================================
RCS file: /cvs/src/lib/libsndio/mio_thru.c,v
retrieving revision 1.9
diff -u -r1.9 mio_thru.c
--- mio_thru.c	6 Jul 2010 01:12:45 -0000	1.9
+++ mio_thru.c	21 Jul 2010 16:13:57 -0000
@@ -226,7 +226,6 @@
 		if (errno != EAGAIN) {
 			DPERROR("thru_write: write");
 			hdl->mio.eof = 1;
-			return 0;
 		}
  		return 0;
 	}
Index: sun.c
===================================================================
RCS file: /cvs/src/lib/libsndio/sun.c,v
retrieving revision 1.38
diff -u -r1.38 sun.c
--- sun.c	15 Jul 2010 03:43:11 -0000	1.38
+++ sun.c	21 Jul 2010 16:13:57 -0000
@@ -866,7 +866,6 @@
 		if (errno != EAGAIN) {
 			DPERROR("sun_write: write");
 			hdl->sio.eof = 1;
-			return 0;
 		}
  		return 0;
 	}