Re: kern/56644: Kernel assertion in audio.c

[mlelstv%serpens.de@localhost (Michael van Elst)]

The following reply was made to PR kern/56644; it has been noted by GNATS.

From: mlelstv%serpens.de@localhost (Michael van Elst)
To: gnats-bugs%netbsd.org@localhost
Cc: 
Subject: Re: kern/56644: Kernel assertion in audio.c
Date: Sat, 29 Jan 2022 06:29:48 -0000 (UTC)

 riastradh%NetBSD.org@localhost (Taylor R Campbell) writes:
 
 > Shouldn't it discard exactly one block's worth of frames instead?  Any
 > smaller amount will make the assertion fail.  Or am I confused, and is
 > the assertion not expected to hold before the branch?
 
 The assertion comes from the assumption that you always process
 full blocks. The code mostly doesn't make explicit guarantees
 about this but the assertions are there to proof the assumption.
 
 Always dropping a full buffer in case of overflow is just another
 case that depends on that assumption.