https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109878
Bug ID: 109878
Summary: missed simplifications of MAX<a&CST0,a&CST1> and
MIN<a&CST0,a&CST1>
Product: gcc
Version: 14.0
Status: UNCONFIRMED
Keywords: missed-optimization
Severity: enhancement
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: pinskia at gcc dot gnu.org
Target Milestone: ---
Take:
```
int max_and(int a, int b)
{
b = a & 3;
a = a & 1;
if (b > a)
return b;
else
return a;
}
int min_and(int a, int b)
{
b = a & 3;
a = a & 1;
if (b < a)
return b;
else
return a;
}
```
max_and should just be optimized to `a&3` while min_and should be just
optimized to `a&1` The general rule is:
MAX<a&CST0, a & CST1> -> a & CST0 IFF CST0 &CST1 == CST1, that is CST1 is a
true subset of CST0.
I found this on accident while thinking about some generated code in
insn-automata.cc on x86_64.
MIN<a&CST0, a & CST1> -> a & CST0 IFF CST0&CST1 == CST0, that is CST0 is a true
subset of CST1.
