4a36e96c3f
This simple heuristic uses the estimated live range length combined with the number of registers in the class to switch which heuristic to use. This was taking the raw number of registers in the class, even though not all of them may be available. AMDGPU heavily relies on dynamically reserved numbers of registers based on user attributes to satisfy occupancy constraints, so the raw number is highly misleading. There are still a few problems here. In the original testcase that made me notice this, the live range size is incorrect after the scheduler rearranges instructions, since the instructions don't have the original InstrDist offsets. Additionally, I think it would be more appropriate to use the number of disjointly allocatable registers in the class. For the AMDGPU register tuples, there are a large number of registers in each tuple class, but only a small fraction can actually be allocated at the same time since they all overlap with each other. It seems we do not have a query that corresponds to the number of independently allocatable registers. Relatedly, I'm still debugging some allocation failures where overlapping tuples seem to not be handled correctly. The test changes are mostly noise. There are a handful of x86 tests that look like regressions with an additional spill, and a handful that now avoid a spill. The worst looking regression is likely test/Thumb2/mve-vld4.ll which introduces a few additional spills. test/CodeGen/AMDGPU/soft-clause-exceeds-register-budget.ll shows a massive improvement by completely eliminating a large number of spills inside a loop.
131 lines
5.3 KiB
LLVM
131 lines
5.3 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -mtriple=i386-apple-darwin -disable-cgp-branch-opts | FileCheck %s
|
|
|
|
%struct.DBC_t = type { i32, i8*, i16, %struct.DBC_t*, i8*, i8*, i8*, i8*, i8*, %struct.DBC_t*, i32, i32, i32, i32, i8*, i8*, i8*, i8*, i8*, i32, i32, i32, i32, i32, i32, i32, i32, i16, i16, i32*, i8, i16, %struct.DRVOPT*, i16 }
|
|
%struct.DRVOPT = type { i16, i32, i8, %struct.DRVOPT* }
|
|
%struct.GENV_t = type { i32, i8*, i16, i8*, i8*, i32, i32, i32, i32, %struct.DBC_t*, i16 }
|
|
%struct.pthread_mutex_t = type { i32, [40 x i8] }
|
|
@iodbcdm_global_lock = external global %struct.pthread_mutex_t ; <%struct.pthread_mutex_t*> [#uses=1]
|
|
|
|
define i16 @SQLDriversW(i8* %henv, i16 zeroext %fDir, i32* %szDrvDesc, i16 signext %cbDrvDescMax, i16* %pcbDrvDesc, i32* %szDrvAttr, i16 signext %cbDrvAttrMax, i16* %pcbDrvAttr) nounwind {
|
|
; CHECK-LABEL: SQLDriversW:
|
|
; CHECK: ## %bb.0: ## %entry
|
|
; CHECK-NEXT: pushl %ebp
|
|
; CHECK-NEXT: pushl %ebx
|
|
; CHECK-NEXT: pushl %edi
|
|
; CHECK-NEXT: pushl %esi
|
|
; CHECK-NEXT: subl $12, %esp
|
|
; CHECK-NEXT: movb $1, %al
|
|
; CHECK-NEXT: testb %al, %al
|
|
; CHECK-NEXT: je LBB0_1
|
|
; CHECK-NEXT: ## %bb.3: ## %bb28
|
|
; CHECK-NEXT: movzwl {{[0-9]+}}(%esp), %ecx
|
|
; CHECK-NEXT: movzwl {{[0-9]+}}(%esp), %ebx
|
|
; CHECK-NEXT: movzwl {{[0-9]+}}(%esp), %ebp
|
|
; CHECK-NEXT: movl {{[0-9]+}}(%esp), %edi
|
|
; CHECK-NEXT: movw $-2, %si
|
|
; CHECK-NEXT: jne LBB0_6
|
|
; CHECK-NEXT: ## %bb.4: ## %bb37
|
|
; CHECK-NEXT: movw $0, 40(%edi)
|
|
; CHECK-NEXT: movb $1, %al
|
|
; CHECK-NEXT: testb %al, %al
|
|
; CHECK-NEXT: leal (,%ecx,4), %ecx
|
|
; CHECK-NEXT: leal (,%ebx,4), %edx
|
|
; CHECK-NEXT: subl $12, %esp
|
|
; CHECK-NEXT: movzwl %bp, %eax
|
|
; CHECK-NEXT: movswl %cx, %ecx
|
|
; CHECK-NEXT: movswl %dx, %edx
|
|
; CHECK-NEXT: pushl $87
|
|
; CHECK-NEXT: pushl {{[0-9]+}}(%esp)
|
|
; CHECK-NEXT: pushl %ecx
|
|
; CHECK-NEXT: pushl $0
|
|
; CHECK-NEXT: pushl {{[0-9]+}}(%esp)
|
|
; CHECK-NEXT: pushl %edx
|
|
; CHECK-NEXT: pushl $0
|
|
; CHECK-NEXT: pushl %eax
|
|
; CHECK-NEXT: pushl %edi
|
|
; CHECK-NEXT: calll _SQLDrivers_Internal
|
|
; CHECK-NEXT: addl $48, %esp
|
|
; CHECK-NEXT: movl %eax, %esi
|
|
; CHECK-NEXT: xorl %eax, %eax
|
|
; CHECK-NEXT: testb %al, %al
|
|
; CHECK-NEXT: je LBB0_1
|
|
; CHECK-NEXT: ## %bb.5:
|
|
; CHECK-NEXT: movzwl {{[0-9]+}}(%esp), %ecx
|
|
; CHECK-NEXT: LBB0_6: ## %done
|
|
; CHECK-NEXT: xorl %eax, %eax
|
|
; CHECK-NEXT: testb %al, %al
|
|
; CHECK-NEXT: je LBB0_7
|
|
; CHECK-NEXT: ## %bb.8: ## %bb167
|
|
; CHECK-NEXT: subl $12, %esp
|
|
; CHECK-NEXT: movl L_iodbcdm_global_lock$non_lazy_ptr, %eax
|
|
; CHECK-NEXT: pushl %eax
|
|
; CHECK-NEXT: calll _pthread_mutex_unlock
|
|
; CHECK-NEXT: addl $16, %esp
|
|
; CHECK-NEXT: movl %esi, %eax
|
|
; CHECK-NEXT: jmp LBB0_2
|
|
; CHECK-NEXT: LBB0_7: ## %bb150
|
|
; CHECK-NEXT: movswl %si, %eax
|
|
; CHECK-NEXT: subl $8, %esp
|
|
; CHECK-NEXT: movswl %cx, %ecx
|
|
; CHECK-NEXT: movswl %bx, %edx
|
|
; CHECK-NEXT: movzwl %bp, %esi
|
|
; CHECK-NEXT: pushl {{[0-9]+}}(%esp)
|
|
; CHECK-NEXT: pushl %ecx
|
|
; CHECK-NEXT: pushl {{[0-9]+}}(%esp)
|
|
; CHECK-NEXT: pushl {{[0-9]+}}(%esp)
|
|
; CHECK-NEXT: pushl %edx
|
|
; CHECK-NEXT: pushl {{[0-9]+}}(%esp)
|
|
; CHECK-NEXT: pushl %esi
|
|
; CHECK-NEXT: pushl %edi
|
|
; CHECK-NEXT: pushl %eax
|
|
; CHECK-NEXT: pushl $1
|
|
; CHECK-NEXT: calll _trace_SQLDriversW
|
|
; CHECK-NEXT: addl $48, %esp
|
|
; CHECK-NEXT: LBB0_1: ## %bb
|
|
; CHECK-NEXT: xorl %eax, %eax
|
|
; CHECK-NEXT: LBB0_2: ## %bb
|
|
; CHECK-NEXT: addl $12, %esp
|
|
; CHECK-NEXT: popl %esi
|
|
; CHECK-NEXT: popl %edi
|
|
; CHECK-NEXT: popl %ebx
|
|
; CHECK-NEXT: popl %ebp
|
|
; CHECK-NEXT: retl
|
|
entry:
|
|
%tmp12 = bitcast i8* %henv to %struct.GENV_t* ; <%struct.GENV_t*> [#uses=1]
|
|
br i1 true, label %bb28, label %bb
|
|
bb: ; preds = %entry
|
|
ret i16 0
|
|
bb28: ; preds = %entry
|
|
br i1 false, label %bb37, label %done
|
|
bb37: ; preds = %bb28
|
|
%tmp46 = getelementptr %struct.GENV_t, %struct.GENV_t* %tmp12, i32 0, i32 10 ; <i16*> [#uses=1]
|
|
store i16 0, i16* %tmp46, align 4
|
|
br i1 false, label %bb74, label %bb92
|
|
bb74: ; preds = %bb37
|
|
br label %bb92
|
|
bb92: ; preds = %bb74, %bb37
|
|
%tmp95180 = shl i16 %cbDrvAttrMax, 2 ; <i16> [#uses=1]
|
|
%tmp100178 = shl i16 %cbDrvDescMax, 2 ; <i16> [#uses=1]
|
|
%tmp113 = tail call i16 @SQLDrivers_Internal( i8* %henv, i16 zeroext %fDir, i8* null, i16 signext %tmp100178, i16* %pcbDrvDesc, i8* null, i16 signext %tmp95180, i16* %pcbDrvAttr, i8 zeroext 87 ) nounwind ; <i16> [#uses=1]
|
|
br i1 false, label %done, label %bb137
|
|
bb137: ; preds = %bb92
|
|
ret i16 0
|
|
done: ; preds = %bb92, %bb28
|
|
%retcode.0 = phi i16 [ -2, %bb28 ], [ %tmp113, %bb92 ] ; <i16> [#uses=2]
|
|
br i1 false, label %bb167, label %bb150
|
|
bb150: ; preds = %done
|
|
%tmp157158 = sext i16 %retcode.0 to i32 ; <i32> [#uses=1]
|
|
tail call void @trace_SQLDriversW( i32 1, i32 %tmp157158, i8* %henv, i16 zeroext %fDir, i32* %szDrvDesc, i16 signext %cbDrvDescMax, i16* %pcbDrvDesc, i32* %szDrvAttr, i16 signext %cbDrvAttrMax, i16* %pcbDrvAttr ) nounwind
|
|
ret i16 0
|
|
bb167: ; preds = %done
|
|
%tmp168 = tail call i32 @pthread_mutex_unlock( %struct.pthread_mutex_t* @iodbcdm_global_lock ) nounwind ; <i32> [#uses=0]
|
|
ret i16 %retcode.0
|
|
}
|
|
|
|
declare i32 @pthread_mutex_unlock(%struct.pthread_mutex_t*)
|
|
|
|
declare i16 @SQLDrivers_Internal(i8*, i16 zeroext , i8*, i16 signext , i16*, i8*, i16 signext , i16*, i8 zeroext ) nounwind
|
|
|
|
declare void @trace_SQLDriversW(i32, i32, i8*, i16 zeroext , i32*, i16 signext , i16*, i32*, i16 signext , i16*)
|