Take the following example:
int f(int i, int j)
{
if (i <= j - 1)
if (i >= 0)
{
return i >= j;
}
return 0;
}
Note this might show up somewhere but I don't know if it actually does (I made
this up while looking into some missed optimizations due the patch which fixes
PR 23666).
The problem if I am looking at this correctly is that:
Visiting statement:
D.1631_3 = j_2 - 1;
Found new range for D.1631_3: VARYING
we say the range for D.1631_3 is varying but shouldn't we say the range is [j_2
- 1, j_2 - 1] ?
That should fix the issue as we then we get the upper bound correctly as shown
by:
int f(int i, int j)
{
if (i < j)
if (i >= 0)
{
return i >= j;
}
return 0;
}
Note as I said before, I don't know how many times this shows but I don't doubt
that people would write i <= j -1 instead of i < j all the time to show what it
really does for integers and loops.
--
Summary: missed VRP opportunity
Product: gcc
Version: 4.2.0
Status: UNCONFIRMED
Keywords: missed-optimization
Severity: enhancement
Priority: P3
Component: tree-optimization
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: pinskia at gcc dot gnu dot org
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25145
