1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-23 03:02:36 +01:00
llvm-mirror/test/CodeGen/Hexagon/vararg_align_check.ll

187 lines
9.0 KiB
LLVM
Raw Normal View History

Add support for Linux/Musl ABI Differential revision: https://reviews.llvm.org/D72701 The patch adds a new option ABI for Hexagon. It primary deals with the way variable arguments are passed and is use in the Hexagon Linux Musl environment. If a callee function has a variable argument list, it must perform the following operations to set up its function prologue: 1. Determine the number of registers which could have been used for passing unnamed arguments. This can be calculated by counting the number of registers used for passing named arguments. For example, if the callee function is as follows: int foo(int a, ...){ ... } ... then register R0 is used to access the argument ' a '. The registers available for passing unnamed arguments are R1, R2, R3, R4, and R5. 2. Determine the number and size of the named arguments on the stack. 3. If the callee has named arguments on the stack, it should copy all of these arguments to a location below the current position on the stack, and the difference should be the size of the register-saved area plus padding (if any is necessary). The register-saved area constitutes all the registers that could have been used to pass unnamed arguments. If the number of registers forming the register-saved area is odd, it requires 4 bytes of padding; if the number is even, no padding is required. This is done to ensure an 8-byte alignment on the stack. For example, if the callee is as follows: int foo(int a, ...){ ... } ... then the named arguments should be copied to the following location: current_position - 5 (for R1-R5) * 4 (bytes) - 4 (bytes of padding) If the callee is as follows: int foo(int a, int b, ...){ ... } ... then the named arguments should be copied to the following location: current_position - 4 (for R2-R5) * 4 (bytes) - 0 (bytes of padding) 4. After any named arguments have been copied, copy all the registers that could have been used to pass unnamed arguments on the stack. If the number of registers is odd, leave 4 bytes of padding and then start copying them on the stack; if the number is even, no padding is required. This constitutes the register-saved area. If padding is required, ensure that the start location of padding is 8-byte aligned. If no padding is required, ensure that the start location of the on-stack copy of the first register which might have a variable argument is 8-byte aligned. 5. Decrement the stack pointer by the size of register saved area plus the padding. For example, if the callee is as follows: int foo(int a, ...){ ... } ; ... then the decrement value should be the following: 5 (for R1-R5) * 4 (bytes) + 4 (bytes of padding) = 24 bytes The decrement should be performed before the allocframe instruction. Increment the stack-pointer back by the same amount before returning from the function.
2019-12-27 20:03:01 +01:00
; RUN: llc -march=hexagon -mcpu=hexagonv62 -mtriple=hexagon-unknown-linux-musl -O0 < %s | FileCheck %s
; CHECK-LABEL: foo:
; Check Function prologue.
; Note. All register numbers and offset are fixed.
; Hence, no need of regular expression.
; CHECK: r29 = add(r29,#-24)
; CHECK: r7:6 = memd(r29+#24)
; CHECK: memd(r29+#0) = r7:6
; CHECK: r7:6 = memd(r29+#32)
; CHECK: memd(r29+#8) = r7:6
; CHECK: r7:6 = memd(r29+#40)
; CHECK: memd(r29+#16) = r7:6
; CHECK: memw(r29+#28) = r1
; CHECK: memw(r29+#32) = r2
; CHECK: memw(r29+#36) = r3
; CHECK: memw(r29+#40) = r4
; CHECK: memw(r29+#44) = r5
; CHECK: r29 = add(r29,#24)
%struct.AAA = type { i32, i32, i32, i32 }
%struct.BBB = type { i8, i64, i32 }
%struct.__va_list_tag = type { i8*, i8*, i8* }
@aaa = global %struct.AAA { i32 100, i32 200, i32 300, i32 400 }, align 4
@ddd = global { i8, i64, i32, [4 x i8] } { i8 1, i64 1000000, i32 5, [4 x i8] undef }, align 8
@.str = private unnamed_addr constant [13 x i8] c"result = %d\0A\00", align 1
; Function Attrs: nounwind
define i32 @foo(i32 %xx, %struct.BBB* byval(%struct.BBB) align 8 %eee, ...) #0 {
Add support for Linux/Musl ABI Differential revision: https://reviews.llvm.org/D72701 The patch adds a new option ABI for Hexagon. It primary deals with the way variable arguments are passed and is use in the Hexagon Linux Musl environment. If a callee function has a variable argument list, it must perform the following operations to set up its function prologue: 1. Determine the number of registers which could have been used for passing unnamed arguments. This can be calculated by counting the number of registers used for passing named arguments. For example, if the callee function is as follows: int foo(int a, ...){ ... } ... then register R0 is used to access the argument ' a '. The registers available for passing unnamed arguments are R1, R2, R3, R4, and R5. 2. Determine the number and size of the named arguments on the stack. 3. If the callee has named arguments on the stack, it should copy all of these arguments to a location below the current position on the stack, and the difference should be the size of the register-saved area plus padding (if any is necessary). The register-saved area constitutes all the registers that could have been used to pass unnamed arguments. If the number of registers forming the register-saved area is odd, it requires 4 bytes of padding; if the number is even, no padding is required. This is done to ensure an 8-byte alignment on the stack. For example, if the callee is as follows: int foo(int a, ...){ ... } ... then the named arguments should be copied to the following location: current_position - 5 (for R1-R5) * 4 (bytes) - 4 (bytes of padding) If the callee is as follows: int foo(int a, int b, ...){ ... } ... then the named arguments should be copied to the following location: current_position - 4 (for R2-R5) * 4 (bytes) - 0 (bytes of padding) 4. After any named arguments have been copied, copy all the registers that could have been used to pass unnamed arguments on the stack. If the number of registers is odd, leave 4 bytes of padding and then start copying them on the stack; if the number is even, no padding is required. This constitutes the register-saved area. If padding is required, ensure that the start location of padding is 8-byte aligned. If no padding is required, ensure that the start location of the on-stack copy of the first register which might have a variable argument is 8-byte aligned. 5. Decrement the stack pointer by the size of register saved area plus the padding. For example, if the callee is as follows: int foo(int a, ...){ ... } ; ... then the decrement value should be the following: 5 (for R1-R5) * 4 (bytes) + 4 (bytes of padding) = 24 bytes The decrement should be performed before the allocframe instruction. Increment the stack-pointer back by the same amount before returning from the function.
2019-12-27 20:03:01 +01:00
entry:
%xx.addr = alloca i32, align 4
%ap = alloca [1 x %struct.__va_list_tag], align 8
%d = alloca i32, align 4
%k = alloca i64, align 8
%ret = alloca i32, align 4
%bbb = alloca %struct.AAA, align 4
store i32 %xx, i32* %xx.addr, align 4
store i32 0, i32* %ret, align 4
%x = getelementptr inbounds %struct.BBB, %struct.BBB* %eee, i32 0, i32 0
%0 = load i8, i8* %x, align 1
%tobool = trunc i8 %0 to i1
br i1 %tobool, label %if.then, label %if.end
if.then: ; preds = %entry
store i32 1, i32* %ret, align 4
br label %if.end
if.end: ; preds = %if.then, %entry
%arraydecay = getelementptr inbounds [1 x %struct.__va_list_tag], [1 x %struct.__va_list_tag]* %ap, i32 0, i32 0
%arraydecay1 = bitcast %struct.__va_list_tag* %arraydecay to i8*
call void @llvm.va_start(i8* %arraydecay1)
%arraydecay2 = getelementptr inbounds [1 x %struct.__va_list_tag], [1 x %struct.__va_list_tag]* %ap, i32 0, i32 0
br label %vaarg.maybe_reg
vaarg.maybe_reg: ; preds = %if.end
%__current_saved_reg_area_pointer_p = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %arraydecay2, i32 0, i32 0
%__current_saved_reg_area_pointer = load i8*, i8** %__current_saved_reg_area_pointer_p
%__saved_reg_area_end_pointer_p = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %arraydecay2, i32 0, i32 1
%__saved_reg_area_end_pointer = load i8*, i8** %__saved_reg_area_end_pointer_p
%1 = ptrtoint i8* %__current_saved_reg_area_pointer to i32
%align_current_saved_reg_area_pointer = add i32 %1, 7
%align_current_saved_reg_area_pointer3 = and i32 %align_current_saved_reg_area_pointer, -8
%align_current_saved_reg_area_pointer4 = inttoptr i32 %align_current_saved_reg_area_pointer3 to i8*
%__new_saved_reg_area_pointer = getelementptr i8, i8* %align_current_saved_reg_area_pointer4, i32 8
%2 = icmp sgt i8* %__new_saved_reg_area_pointer, %__saved_reg_area_end_pointer
br i1 %2, label %vaarg.on_stack, label %vaarg.in_reg
vaarg.in_reg: ; preds = %vaarg.maybe_reg
%3 = bitcast i8* %align_current_saved_reg_area_pointer4 to i64*
store i8* %__new_saved_reg_area_pointer, i8** %__current_saved_reg_area_pointer_p
br label %vaarg.end
vaarg.on_stack: ; preds = %vaarg.maybe_reg
%__overflow_area_pointer_p = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %arraydecay2, i32 0, i32 2
%__overflow_area_pointer = load i8*, i8** %__overflow_area_pointer_p
%4 = ptrtoint i8* %__overflow_area_pointer to i32
%align_overflow_area_pointer = add i32 %4, 7
%align_overflow_area_pointer5 = and i32 %align_overflow_area_pointer, -8
%align_overflow_area_pointer6 = inttoptr i32 %align_overflow_area_pointer5 to i8*
%__overflow_area_pointer.next = getelementptr i8, i8* %align_overflow_area_pointer6, i32 8
store i8* %__overflow_area_pointer.next, i8** %__overflow_area_pointer_p
store i8* %__overflow_area_pointer.next, i8** %__current_saved_reg_area_pointer_p
%5 = bitcast i8* %align_overflow_area_pointer6 to i64*
br label %vaarg.end
vaarg.end: ; preds = %vaarg.on_stack, %vaarg.in_reg
%vaarg.addr = phi i64* [ %3, %vaarg.in_reg ], [ %5, %vaarg.on_stack ]
%6 = load i64, i64* %vaarg.addr
store i64 %6, i64* %k, align 8
%7 = load i64, i64* %k, align 8
%conv = trunc i64 %7 to i32
%div = sdiv i32 %conv, 1000
%8 = load i32, i32* %ret, align 4
%add = add nsw i32 %8, %div
store i32 %add, i32* %ret, align 4
%arraydecay7 = getelementptr inbounds [1 x %struct.__va_list_tag], [1 x %struct.__va_list_tag]* %ap, i32 0, i32 0
%__overflow_area_pointer_p8 = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %arraydecay7, i32 0, i32 2
%__overflow_area_pointer9 = load i8*, i8** %__overflow_area_pointer_p8
%9 = bitcast i8* %__overflow_area_pointer9 to %struct.AAA*
%__overflow_area_pointer.next10 = getelementptr i8, i8* %__overflow_area_pointer9, i32 16
store i8* %__overflow_area_pointer.next10, i8** %__overflow_area_pointer_p8
%10 = bitcast %struct.AAA* %bbb to i8*
%11 = bitcast %struct.AAA* %9 to i8*
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %10, i8* %11, i32 16, i32 4, i1 false)
%d11 = getelementptr inbounds %struct.AAA, %struct.AAA* %bbb, i32 0, i32 3
%12 = load i32, i32* %d11, align 4
%13 = load i32, i32* %ret, align 4
%add12 = add nsw i32 %13, %12
store i32 %add12, i32* %ret, align 4
%arraydecay13 = getelementptr inbounds [1 x %struct.__va_list_tag], [1 x %struct.__va_list_tag]* %ap, i32 0, i32 0
br label %vaarg.maybe_reg14
vaarg.maybe_reg14: ; preds = %vaarg.end
%__current_saved_reg_area_pointer_p15 = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %arraydecay13, i32 0, i32 0
%__current_saved_reg_area_pointer16 = load i8*, i8** %__current_saved_reg_area_pointer_p15
%__saved_reg_area_end_pointer_p17 = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %arraydecay13, i32 0, i32 1
%__saved_reg_area_end_pointer18 = load i8*, i8** %__saved_reg_area_end_pointer_p17
%__new_saved_reg_area_pointer19 = getelementptr i8, i8* %__current_saved_reg_area_pointer16, i32 4
%14 = icmp sgt i8* %__new_saved_reg_area_pointer19, %__saved_reg_area_end_pointer18
br i1 %14, label %vaarg.on_stack21, label %vaarg.in_reg20
vaarg.in_reg20: ; preds = %vaarg.maybe_reg14
%15 = bitcast i8* %__current_saved_reg_area_pointer16 to i32*
store i8* %__new_saved_reg_area_pointer19, i8** %__current_saved_reg_area_pointer_p15
br label %vaarg.end25
vaarg.on_stack21: ; preds = %vaarg.maybe_reg14
%__overflow_area_pointer_p22 = getelementptr inbounds %struct.__va_list_tag, %struct.__va_list_tag* %arraydecay13, i32 0, i32 2
%__overflow_area_pointer23 = load i8*, i8** %__overflow_area_pointer_p22
%__overflow_area_pointer.next24 = getelementptr i8, i8* %__overflow_area_pointer23, i32 4
store i8* %__overflow_area_pointer.next24, i8** %__overflow_area_pointer_p22
store i8* %__overflow_area_pointer.next24, i8** %__current_saved_reg_area_pointer_p15
%16 = bitcast i8* %__overflow_area_pointer23 to i32*
br label %vaarg.end25
vaarg.end25: ; preds = %vaarg.on_stack21, %vaarg.in_reg20
%vaarg.addr26 = phi i32* [ %15, %vaarg.in_reg20 ], [ %16, %vaarg.on_stack21 ]
%17 = load i32, i32* %vaarg.addr26
store i32 %17, i32* %d, align 4
%18 = load i32, i32* %d, align 4
%19 = load i32, i32* %ret, align 4
%add27 = add nsw i32 %19, %18
store i32 %add27, i32* %ret, align 4
%arraydecay28 = getelementptr inbounds [1 x %struct.__va_list_tag], [1 x %struct.__va_list_tag]* %ap, i32 0, i32 0
%arraydecay2829 = bitcast %struct.__va_list_tag* %arraydecay28 to i8*
call void @llvm.va_end(i8* %arraydecay2829)
%20 = load i32, i32* %ret, align 4
ret i32 %20
}
; Function Attrs: nounwind
declare void @llvm.va_start(i8*) #1
; Function Attrs: nounwind
declare void @llvm.memcpy.p0i8.p0i8.i32(i8* nocapture, i8* nocapture readonly, i32, i32, i1) #1
; Function Attrs: nounwind
declare void @llvm.va_end(i8*) #1
; Function Attrs: nounwind
define i32 @main() #0 {
entry:
%retval = alloca i32, align 4
%x = alloca i32, align 4
%m = alloca i64, align 8
store i32 0, i32* %retval
store i64 1000000, i64* %m, align 8
%0 = load i64, i64* %m, align 8
%call = call i32 (i32, %struct.BBB*, ...) @foo(i32 1, %struct.BBB* byval(%struct.BBB) align 8 bitcast ({ i8, i64, i32, [4 x i8] }* @ddd to %struct.BBB*), i64 %0, %struct.AAA* byval(%struct.AAA) align 4 @aaa, i32 4)
Add support for Linux/Musl ABI Differential revision: https://reviews.llvm.org/D72701 The patch adds a new option ABI for Hexagon. It primary deals with the way variable arguments are passed and is use in the Hexagon Linux Musl environment. If a callee function has a variable argument list, it must perform the following operations to set up its function prologue: 1. Determine the number of registers which could have been used for passing unnamed arguments. This can be calculated by counting the number of registers used for passing named arguments. For example, if the callee function is as follows: int foo(int a, ...){ ... } ... then register R0 is used to access the argument ' a '. The registers available for passing unnamed arguments are R1, R2, R3, R4, and R5. 2. Determine the number and size of the named arguments on the stack. 3. If the callee has named arguments on the stack, it should copy all of these arguments to a location below the current position on the stack, and the difference should be the size of the register-saved area plus padding (if any is necessary). The register-saved area constitutes all the registers that could have been used to pass unnamed arguments. If the number of registers forming the register-saved area is odd, it requires 4 bytes of padding; if the number is even, no padding is required. This is done to ensure an 8-byte alignment on the stack. For example, if the callee is as follows: int foo(int a, ...){ ... } ... then the named arguments should be copied to the following location: current_position - 5 (for R1-R5) * 4 (bytes) - 4 (bytes of padding) If the callee is as follows: int foo(int a, int b, ...){ ... } ... then the named arguments should be copied to the following location: current_position - 4 (for R2-R5) * 4 (bytes) - 0 (bytes of padding) 4. After any named arguments have been copied, copy all the registers that could have been used to pass unnamed arguments on the stack. If the number of registers is odd, leave 4 bytes of padding and then start copying them on the stack; if the number is even, no padding is required. This constitutes the register-saved area. If padding is required, ensure that the start location of padding is 8-byte aligned. If no padding is required, ensure that the start location of the on-stack copy of the first register which might have a variable argument is 8-byte aligned. 5. Decrement the stack pointer by the size of register saved area plus the padding. For example, if the callee is as follows: int foo(int a, ...){ ... } ; ... then the decrement value should be the following: 5 (for R1-R5) * 4 (bytes) + 4 (bytes of padding) = 24 bytes The decrement should be performed before the allocframe instruction. Increment the stack-pointer back by the same amount before returning from the function.
2019-12-27 20:03:01 +01:00
store i32 %call, i32* %x, align 4
%1 = load i32, i32* %x, align 4
%call1 = call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([13 x i8], [13 x i8]* @.str, i32 0, i32 0), i32 %1)
%2 = load i32, i32* %x, align 4
ret i32 %2
}
declare i32 @printf(i8*, ...) #2
attributes #1 = { nounwind }
!llvm.ident = !{!0}
!0 = !{!"Clang 3.1"}