From: lkcl <lkcl@web>
Date: Tue, 26 Apr 2022 16:27:28 +0000 (+0100)
Subject: (no commit message)
X-Git-Tag: opf_rfc_ls005_v1~2578
X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=dc79c46ae3b6d6add66d6e00737249b9dfcdf1cf;p=libreriscv.git

---

diff --git a/openpower/sv/biginteger/analysis.mdwn b/openpower/sv/biginteger/analysis.mdwn
index a29aa6a6b..76f0a37f5 100644
--- a/openpower/sv/biginteger/analysis.mdwn
+++ b/openpower/sv/biginteger/analysis.mdwn
@@ -409,3 +409,11 @@ selecting whether `RS=RC` or
 allows the instruction to perform full parallel vector div/mod,
 or act in loop-back mode for big-int division by a scalar,
 or for a single scalar 128/64 div/mod.
+
+Just as with `divdeu` on which this instruction is based an overflow
+detection is required.  When the divisor is too small compared to
+the dividend then the result may not fit into 64 bit.  Knuth's
+original algorithm detects overflow and manually places 0xffffffff
+(all ones) into `qhat`. It makes sense for `divrem2du` to do
+this, and also to return an overflow indicator: this can be dobe
+by always setting Rc=1 as a way to save opcode space