Without this, clang's static analyzer finds:
./src/field_5x52_impl.h:425:24: warning: The left operand of '&' is a garbage value
r->n[0] = (r->n[0] & mask0) | (a->n[0] & mask1);
initialize variable #233
pull theuni wants to merge 1 commits into bitcoin-core:master from theuni:fix-uninit changing 1 files +1 −0-
theuni commented at 9:19 PM on April 3, 2015: contributor
-
initialize variable 6534ee1259
-
gmaxwell commented at 9:32 PM on April 3, 2015: contributor
I was kind of lamenting if we should do something about that. GCC -O3 complains too, it's just that it's not smart enough to know that the conditional move will blow it away.
I measure a 2% slowdown from this (bench sign), but it could be measurement noise. (seems kinda high!)
ACK for whatever it's worth from my side.
-
sipa commented at 8:18 AM on April 10, 2015: contributor
I can't measure any difference when locking my CPU to a single frequency. And I mean that very literally:
before:
$ ./bench_sign ecdsa_sign: min 69.9us / avg 69.9us / avg 69.9usafter:
$ ./bench_sign ecdsa_sign: min 69.9us / avg 69.9us / avg 69.9us - sipa merged this on Apr 10, 2015
- sipa closed this on Apr 10, 2015
- sipa referenced this in commit a0d3b89dd6 on Apr 10, 2015
