Java requires a class initialization check when any public static method is
called. But when we are calling from another method in the _same_ class, we
already know that the class has been initialized, so we can optimize away the
call to _Jv_InitClass, even if we're not inlining. Here's how:
Reorder the call, so that this (on x86):
.globl _ZN4Test10assertHackEb
.type _ZN4Test10assertHackEb, @function
_ZN4Test10assertHackEb:
.LFB5:
.file 1 "Test.java"
.loc 1 43 0
.LVL0:
pushl %ebp
.LCFI0:
movl %esp, %ebp
.LCFI1:
pushl %ebx
.LCFI2:
subl $16, %esp
.LCFI3:
movb 8(%ebp), %bl
.loc 1 43 0
pushl $_ZN4Test6class$E
.LCFI4:
call _Jv_InitClass
becomes something like this:
.globl _ZN4Test10assertHackEb
.type _ZN4Test10assertHackEb, @function
_ZN4Test10assertHackEb:
.LFB5:
.file 1 "Test.java"
.loc 1 43 0
pushl $_ZN4Test6class$E
.LCFI4:
call _Jv_InitClass
.LVL0:
pushl %ebp
.LCFI0:
movl %esp, %ebp
.LCFI1:
pushl %ebx
.LCFI2:
subl $16, %esp
.LCFI3:
movb 8(%ebp), %bl
and intra-class method calls to that method, call to .LVL0 instead of
_ZN4Test10assertHackEb.
--
Summary: Optimization: Skip _Jv_InitClass for intra-class static
method calls
Product: gcc
Version: 4.0.2
Status: UNCONFIRMED
Severity: enhancement
Priority: P2
Component: java
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: greenrd at greenrd dot org
CC: gcc-bugs at gcc dot gnu dot org,java-prs at gcc dot gnu
dot org
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23500
