<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<META NAME="Generator" CONTENT="MS Exchange Server version 6.5.7652.24">
<TITLE>JIT Stub Problem</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/plain format -->
<P><FONT SIZE=2>I'm having an issue with the Stubs used by the JIT Compiler. I'm not sure if it's a bug or if I'm doing something incorrectly.<BR>
<BR>
I've got a long complicated function with the following basic blocks at the end of it (The complete .ll file is attached):<BR>
<BR>
falseBlock: ; preds = %__exp.exit340<BR>
ret int 617<BR>
<BR>
codeRepl: ; preds = %__exp.exit340<BR>
<BR>
%tmp2.i.i = add int %x, 1 ; <int> [#uses=1]<BR>
<BR>
store int %tmp2.i.i, int* %tmp2.loc<BR>
<BR>
call void %test3_trueBlock_trueBlock.ret.exitStub_newFuncRoot.ce_trueBlock.ret.exitStub.ret.exitStub.ret.exitStub.ret7( )<BR>
%tmp2.reload = load int* %tmp2.loc ; <int> [#uses=1]<BR>
%tmp15 = tail call int (sbyte*, ...)* %printf( sbyte* getelementptr ([8 x sbyte]* %str, int 0, uint 0), int %tmp2.reload ) ; <int> [#uses=0]<BR>
br label %trueBlock.ret<BR>
<BR>
trueBlock.ret: ; preds = %codeRepl<BR>
ret int %tmp2.reload<BR>
<BR>
<BR>
The variable %tmp2.loc is alloca'd further up in the function but the funny stuff happens here. In the codeRepl block above, we store tmp2.i.i into<BR>
tmp2.loc, call a function, then reload the integer stored in tmp2.loc. Without the call instruction there, it works just fine. If we store the value 10 into tmp2.i.i, 10 gets reloaded after the call. However, the call instruction somehow overwrites the data stored in tmp2.loc which surprises me. Using the debugger, we watched that memory location and saw that X86CompilationCallback_SSE was modifying it for some reason. After lookin at<BR>
X86CompilationCallback_SSE, we don't see how it is touching that memory location.<BR>
<BR>
A few other things. If i move the alloca of tmp2.loc into the codeRepl block, the error does not occur. If i add a print directly after the current alloca of temp2.loc, the error does not occur.<BR>
<BR>
In addition, here's what the MachineFunction code and the generated machine code look like for these basic blocks.<BR>
<BR>
falseBlock (0xa60cb18, LLVM BB @0xa5ce378, ID#81):<BR>
Predecessors according to CFG: 0xa60ca80<BR>
%EAX = MOV32ri 617<BR>
%EBX = MOV32rm %EBP, 1, %NOREG, -12<BR>
%EDI = MOV32rm %EBP, 1, %NOREG, -8<BR>
%ESI = MOV32rm %EBP, 1, %NOREG, -4<BR>
%ESP = MOV32rr %EBP<BR>
%EBP = POP32r<BR>
RET<BR>
<BR>
codeRepl (0xa5f4148, LLVM BB @0xa5ce310, ID#82):<BR>
Predecessors according to CFG: 0xa60ca80<BR>
%EDI = INC32r %EDI<BR>
%EAX = MOV32rm %EBP, 1, %NOREG, -268<BR>
MOV32mr %EAX, 1, %NOREG, 0, %EDI<BR>
CALLpcrel32 <ga:test3_trueBlock_trueBlock.ret.exitStub_newFuncRoot.ce_trueBlock.ret.exitStub.ret.exitStub.ret.exitStub.ret7><BR>
%EAX = MOV32rm %EBP, 1, %NOREG, -268<BR>
%ESI = MOV32rm %EAX, 1, %NOREG, 0<BR>
%ESP = SUB32ri %ESP, 8<BR>
MOV32mr %ESP, 1, %NOREG, 4, %ESI<BR>
MOV32mi %ESP, 1, %NOREG, 0, <ga:str><BR>
CALLpcrel32 <ga:printf><BR>
%ESP = ADD32ri8 %ESP, 8<BR>
Successors according to CFG: 0xa60cb58<BR>
<BR>
trueBlock.ret (0xa60cb58, LLVM BB @0xa5d88c8, ID#83):<BR>
Predecessors according to CFG: 0xa5f4148<BR>
%EAX = MOV32rr %ESI<BR>
%EBX = MOV32rm %EBP, 1, %NOREG, -12<BR>
%EDI = MOV32rm %EBP, 1, %NOREG, -8<BR>
%ESI = MOV32rm %EBP, 1, %NOREG, -4<BR>
%ESP = MOV32rr %EBP<BR>
%EBP = POP32r<BR>
RET<BR>
<BR>
<BR>
falseBlock:<BR>
0xf5f6f9f8: mov $0x269,%eax<BR>
0xf5f6f9fd: mov 0xfffffff4(%ebp),%ebx<BR>
0xf5f6fa00: mov 0xfffffff8(%ebp),%edi<BR>
0xf5f6fa03: mov 0xfffffffc(%ebp),%esi<BR>
0xf5f6fa06: mov %ebp,%esp<BR>
0xf5f6fa08: pop %ebp<BR>
0xf5f6fa09: ret<BR>
<BR>
codeRepl:<BR>
0xf5f6fa0a: inc %edi <BR>
0xf5f6fa0b: mov 0xfffffef4(%ebp),%eax <BR>
0xf5f6fa11: mov %edi,(%eax) <BR>
0xf5f6fa13: call 0x83e4c28 <X86CompilationCallback_SSE> <BR>
0xf5f6fa18: mov 0xfffffef4(%ebp),%eax <BR>
0xf5f6fa1e: mov (%eax),%esi <BR>
0xf5f6fa20: sub $0x8,%esp<BR>
0xf5f6fa26: mov %esi,0x4(%esp)<BR>
0xf5f6fa2a: movl $0x946b908,(%esp)<BR>
0xf5f6fa31: call 0x5ba660 <printf><BR>
0xf5f6fa36: add $0x8,%esp<BR>
<BR>
trueBlock.ret:<BR>
0xf5f6fa39: mov %esi,%eax<BR>
0xf5f6fa3b: mov 0xfffffff4(%ebp),%ebx<BR>
0xf5f6fa3e: mov 0xfffffff8(%ebp),%edi<BR>
0xf5f6fa41: mov 0xfffffffc(%ebp),%esi<BR>
0xf5f6fa44: mov %ebp,%esp<BR>
0xf5f6fa46: pop %ebp<BR>
0xf5f6fa47: ret<BR>
<BR>
<BR>
Any help would be appreciated<BR>
<BR>
I have attached the full .ll file. Upon running it should print out TEST 10TEST 10, but it prints out TEST -XXXXXXXXTEST -XXXXXXXXX where XXXXXXX is some large number.<BR>
<BR>
<BR>
<BR>
Thanks,<BR>
Ben<BR>
<BR>
<BR>
<BR>
</FONT>
</P>
</BODY>
</HTML>