<div dir="ltr">Thanks for the details!</div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, Sep 10, 2019 at 8:43 PM Timothy B. Terriberry <<a href="mailto:tterribe@xiph.org">tterribe@xiph.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">Felicia Lim wrote:<br>
> Hi Yves,<br>
> <br>
> At that time, the issue wasn't reproducible in opus master, but there <br>
> had been some significant changes since the last tagged release that <br>
> made it hard to pinpoint a single patch to cherry pick to Chromium, <br>
> hence this local fix.<br>
<br>
I'm not convinced that using a saturating add here is the correct thing <br>
to do, either (as opposed to, say, multiplying out by the coefficients <br>
individually).<br>
<br>
Either way, there are exact analogs of this code in NSQ.c, as well as <br>
all of the SIMD optimizations (x86/NSQ_sse4_1.c and <br>
x86/NSQ_del_dec_sse4_1.c, arm/NSQ_del_dec_neon_intr.c, and <br>
mips/NSQ_del_dec_mipsr1.h). If it is a possible issue in master, the fix <br>
looks incomplete to me, at least.<br>
</blockquote></div><br clear="all"><div><br></div>-- <br><div dir="ltr" class="gmail_signature"><div dir="ltr">Warm regards,<div>Yves</div></div></div>