https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118138
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Last reconfirmed| |2024-12-19
Ever confirmed|0 |1
Version|unknown |15.0
Component|tree-optimization |ipa
Status|UNCONFIRMED |NEW
Summary|wrong code at -O3 with |[15 Regression] wrong code
|"-fno-inline" on |at -O3 with "-fno-inline"
|x86_64-linux-gnu |on x86_64-linux-gnu
Target Milestone|--- |15.0
Keywords| |wrong-code
--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Confirmed. The problem is in IPA-CP:
Propagated bits info for function h.constprop/6:
param 0: value = 0xffffffff, mask = 0x0
Propagated bits info for function i.constprop/7:
param 0: value = 0xfffffffffffffffffffffffffffffffffffffff...., mask = 0x0
Propagated bits info for function i/7:
param 0: value = 0xfffffffffffffffffffffffffffffffffffffff...., mask = 0x0
Propagated bits info for function h/6:
param 0: value = 0xffffffff, mask = 0x0
h (and h.constprop) are correct but i (and i.constprop) are incorrect. There
was a sign extend when it should have been a zero extend.
Jump functions of caller h/6:
callsite h/6 -> f/5 :
param 0: PASS THROUGH: 0, op nop_expr
