Magic-Mirror/llvm-capstone
1
0
Fork 0
mirror of https://github.com/capstone-engine/llvm-capstone.git synced 2024-12-14 19:49:36 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

integer/add_sat: Use clang builtin instead of llvm asm

Browse Source 
reviewer: Tom Stellard

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 314702
This commit is contained in:
Jan Vesely 2017-10-02 18:39:00 +00:00
parent 943057a288
commit 1964df8fad
7 changed files with 29 additions and 206 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
libclc/generic/lib/SOURCES
View File

@ -64,8 +64,6 @@ geometric/normalize.cl
integer/abs.cl
integer/abs_diff.cl
integer/add_sat.cl
integer/add_sat_if.ll
integer/add_sat_impl.ll
integer/clz.cl
integer/hadd.cl
integer/mad24.cl

36
libclc/generic/lib/integer/add_sat.cl
View File

@ -12,35 +12,55 @@ _CLC_DECL long __clc_add_sat_s64(long, long);
_CLC_DECL ulong __clc_add_sat_u64(ulong, ulong);
_CLC_OVERLOAD _CLC_DEF char add_sat(char x, char y) {
return __clc_add_sat_s8(x, y);
short r = x + y;
return convert_char_sat(r);
}
_CLC_OVERLOAD _CLC_DEF uchar add_sat(uchar x, uchar y) {
return __clc_add_sat_u8(x, y);
ushort r = x + y;
return convert_uchar_sat(r);
}
_CLC_OVERLOAD _CLC_DEF short add_sat(short x, short y) {
return __clc_add_sat_s16(x, y);
int r = x + y;
return convert_short_sat(r);
}
_CLC_OVERLOAD _CLC_DEF ushort add_sat(ushort x, ushort y) {
return __clc_add_sat_u16(x, y);
uint r = x + y;
return convert_ushort_sat(r);
}
_CLC_OVERLOAD _CLC_DEF int add_sat(int x, int y) {
return __clc_add_sat_s32(x, y);
int r;
if (__builtin_sadd_overflow(x, y, &r))
// The oveflow can only occur if both are pos or both are neg,
// thus we only need to check one operand
return x > 0 ? INT_MAX : INT_MIN;
return r;
}
_CLC_OVERLOAD _CLC_DEF uint add_sat(uint x, uint y) {
return __clc_add_sat_u32(x, y);
uint r;
if (__builtin_uadd_overflow(x, y, &r))
return UINT_MAX;
return r;
}
_CLC_OVERLOAD _CLC_DEF long add_sat(long x, long y) {
return __clc_add_sat_s64(x, y);
long r;
if (__builtin_saddl_overflow(x, y, &r))
// The oveflow can only occur if both are pos or both are neg,
// thus we only need to check one operand
return x > 0 ? LONG_MAX : LONG_MIN;
return r;
}
_CLC_OVERLOAD _CLC_DEF ulong add_sat(ulong x, ulong y) {
return __clc_add_sat_u64(x, y);
ulong r;
if (__builtin_uaddl_overflow(x, y, &r))
return ULONG_MAX;
return r;
}
_CLC_BINARY_VECTORIZE(_CLC_OVERLOAD _CLC_DEF, char, add_sat, char, char)

55
libclc/generic/lib/integer/add_sat_if.ll
View File

@ -1,55 +0,0 @@
declare i8 @__clc_add_sat_impl_s8(i8 %x, i8 %y)
define i8 @__clc_add_sat_s8(i8 %x, i8 %y) nounwind readnone alwaysinline {
%call = call i8 @__clc_add_sat_impl_s8(i8 %x, i8 %y)
ret i8 %call
}
declare i8 @__clc_add_sat_impl_u8(i8 %x, i8 %y)
define i8 @__clc_add_sat_u8(i8 %x, i8 %y) nounwind readnone alwaysinline {
%call = call i8 @__clc_add_sat_impl_u8(i8 %x, i8 %y)
ret i8 %call
}
declare i16 @__clc_add_sat_impl_s16(i16 %x, i16 %y)
define i16 @__clc_add_sat_s16(i16 %x, i16 %y) nounwind readnone alwaysinline {
%call = call i16 @__clc_add_sat_impl_s16(i16 %x, i16 %y)
ret i16 %call
}
declare i16 @__clc_add_sat_impl_u16(i16 %x, i16 %y)
define i16 @__clc_add_sat_u16(i16 %x, i16 %y) nounwind readnone alwaysinline {
%call = call i16 @__clc_add_sat_impl_u16(i16 %x, i16 %y)
ret i16 %call
}
declare i32 @__clc_add_sat_impl_s32(i32 %x, i32 %y)
define i32 @__clc_add_sat_s32(i32 %x, i32 %y) nounwind readnone alwaysinline {
%call = call i32 @__clc_add_sat_impl_s32(i32 %x, i32 %y)
ret i32 %call
}
declare i32 @__clc_add_sat_impl_u32(i32 %x, i32 %y)
define i32 @__clc_add_sat_u32(i32 %x, i32 %y) nounwind readnone alwaysinline {
%call = call i32 @__clc_add_sat_impl_u32(i32 %x, i32 %y)
ret i32 %call
}
declare i64 @__clc_add_sat_impl_s64(i64 %x, i64 %y)
define i64 @__clc_add_sat_s64(i64 %x, i64 %y) nounwind readnone alwaysinline {
%call = call i64 @__clc_add_sat_impl_s64(i64 %x, i64 %y)
ret i64 %call
}
declare i64 @__clc_add_sat_impl_u64(i64 %x, i64 %y)
define i64 @__clc_add_sat_u64(i64 %x, i64 %y) nounwind readnone alwaysinline {
%call = call i64 @__clc_add_sat_impl_u64(i64 %x, i64 %y)
ret i64 %call
}

83
libclc/generic/lib/integer/add_sat_impl.ll
View File

@ -1,83 +0,0 @@
declare {i8, i1} @llvm.sadd.with.overflow.i8(i8, i8)
declare {i8, i1} @llvm.uadd.with.overflow.i8(i8, i8)
define i8 @__clc_add_sat_impl_s8(i8 %x, i8 %y) nounwind readnone alwaysinline {
%call = call {i8, i1} @llvm.sadd.with.overflow.i8(i8 %x, i8 %y)
%res = extractvalue {i8, i1} %call, 0
%over = extractvalue {i8, i1} %call, 1
%x.msb = ashr i8 %x, 7
%x.limit = xor i8 %x.msb, 127
%sat = select i1 %over, i8 %x.limit, i8 %res
ret i8 %sat
}
define i8 @__clc_add_sat_impl_u8(i8 %x, i8 %y) nounwind readnone alwaysinline {
%call = call {i8, i1} @llvm.uadd.with.overflow.i8(i8 %x, i8 %y)
%res = extractvalue {i8, i1} %call, 0
%over = extractvalue {i8, i1} %call, 1
%sat = select i1 %over, i8 -1, i8 %res
ret i8 %sat
}
declare {i16, i1} @llvm.sadd.with.overflow.i16(i16, i16)
declare {i16, i1} @llvm.uadd.with.overflow.i16(i16, i16)
define i16 @__clc_add_sat_impl_s16(i16 %x, i16 %y) nounwind readnone alwaysinline {
%call = call {i16, i1} @llvm.sadd.with.overflow.i16(i16 %x, i16 %y)
%res = extractvalue {i16, i1} %call, 0
%over = extractvalue {i16, i1} %call, 1
%x.msb = ashr i16 %x, 15
%x.limit = xor i16 %x.msb, 32767
%sat = select i1 %over, i16 %x.limit, i16 %res
ret i16 %sat
}
define i16 @__clc_add_sat_impl_u16(i16 %x, i16 %y) nounwind readnone alwaysinline {
%call = call {i16, i1} @llvm.uadd.with.overflow.i16(i16 %x, i16 %y)
%res = extractvalue {i16, i1} %call, 0
%over = extractvalue {i16, i1} %call, 1
%sat = select i1 %over, i16 -1, i16 %res
ret i16 %sat
}
declare {i32, i1} @llvm.sadd.with.overflow.i32(i32, i32)
declare {i32, i1} @llvm.uadd.with.overflow.i32(i32, i32)
define i32 @__clc_add_sat_impl_s32(i32 %x, i32 %y) nounwind readnone alwaysinline {
%call = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %x, i32 %y)
%res = extractvalue {i32, i1} %call, 0
%over = extractvalue {i32, i1} %call, 1
%x.msb = ashr i32 %x, 31
%x.limit = xor i32 %x.msb, 2147483647
%sat = select i1 %over, i32 %x.limit, i32 %res
ret i32 %sat
}
define i32 @__clc_add_sat_impl_u32(i32 %x, i32 %y) nounwind readnone alwaysinline {
%call = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %x, i32 %y)
%res = extractvalue {i32, i1} %call, 0
%over = extractvalue {i32, i1} %call, 1
%sat = select i1 %over, i32 -1, i32 %res
ret i32 %sat
}
declare {i64, i1} @llvm.sadd.with.overflow.i64(i64, i64)
declare {i64, i1} @llvm.uadd.with.overflow.i64(i64, i64)
define i64 @__clc_add_sat_impl_s64(i64 %x, i64 %y) nounwind readnone alwaysinline {
%call = call {i64, i1} @llvm.sadd.with.overflow.i64(i64 %x, i64 %y)
%res = extractvalue {i64, i1} %call, 0
%over = extractvalue {i64, i1} %call, 1
%x.msb = ashr i64 %x, 63
%x.limit = xor i64 %x.msb, 9223372036854775807
%sat = select i1 %over, i64 %x.limit, i64 %res
ret i64 %sat
}
define i64 @__clc_add_sat_impl_u64(i64 %x, i64 %y) nounwind readnone alwaysinline {
%call = call {i64, i1} @llvm.uadd.with.overflow.i64(i64 %x, i64 %y)
%res = extractvalue {i64, i1} %call, 0
%over = extractvalue {i64, i1} %call, 1
%sat = select i1 %over, i64 -1, i64 %res
ret i64 %sat
}

1
libclc/ptx/lib/OVERRIDES
View File

@ -1,2 +1 @@
integer/add_sat_if.ll
integer/sub_sat_if.ll

3
libclc/ptx/lib/SOURCES
View File

@ -1,2 +1 @@
integer/add_sat.ll
integer/sub_sat.ll
integer/sub_sat.ll

55
libclc/ptx/lib/integer/add_sat.ll
View File

@ -1,55 +0,0 @@
declare i8 @__clc_add_sat_impl_s8(i8 %x, i8 %y)
define ptx_device i8 @__clc_add_sat_s8(i8 %x, i8 %y) nounwind readnone alwaysinline {
%call = call i8 @__clc_add_sat_impl_s8(i8 %x, i8 %y)
ret i8 %call
}
declare i8 @__clc_add_sat_impl_u8(i8 %x, i8 %y)
define ptx_device i8 @__clc_add_sat_u8(i8 %x, i8 %y) nounwind readnone alwaysinline {
%call = call i8 @__clc_add_sat_impl_u8(i8 %x, i8 %y)
ret i8 %call
}
declare i16 @__clc_add_sat_impl_s16(i16 %x, i16 %y)
define ptx_device i16 @__clc_add_sat_s16(i16 %x, i16 %y) nounwind readnone alwaysinline {
%call = call i16 @__clc_add_sat_impl_s16(i16 %x, i16 %y)
ret i16 %call
}
declare i16 @__clc_add_sat_impl_u16(i16 %x, i16 %y)
define ptx_device i16 @__clc_add_sat_u16(i16 %x, i16 %y) nounwind readnone alwaysinline {
%call = call i16 @__clc_add_sat_impl_u16(i16 %x, i16 %y)
ret i16 %call
}
declare i32 @__clc_add_sat_impl_s32(i32 %x, i32 %y)
define ptx_device i32 @__clc_add_sat_s32(i32 %x, i32 %y) nounwind readnone alwaysinline {
%call = call i32 @__clc_add_sat_impl_s32(i32 %x, i32 %y)
ret i32 %call
}
declare i32 @__clc_add_sat_impl_u32(i32 %x, i32 %y)
define ptx_device i32 @__clc_add_sat_u32(i32 %x, i32 %y) nounwind readnone alwaysinline {
%call = call i32 @__clc_add_sat_impl_u32(i32 %x, i32 %y)
ret i32 %call
}
declare i64 @__clc_add_sat_impl_s64(i64 %x, i64 %y)
define ptx_device i64 @__clc_add_sat_s64(i64 %x, i64 %y) nounwind readnone alwaysinline {
%call = call i64 @__clc_add_sat_impl_s64(i64 %x, i64 %y)
ret i64 %call
}
declare i64 @__clc_add_sat_impl_u64(i64 %x, i64 %y)
define ptx_device i64 @__clc_add_sat_u64(i64 %x, i64 %y) nounwind readnone alwaysinline {
%call = call i64 @__clc_add_sat_impl_u64(i64 %x, i64 %y)
ret i64 %call
}