RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-30 15:32:52 +01:00
Code Issues Projects Releases Wiki Activity
d714d8ebe8
llvm-mirror/test/CodeGen/AArch64/extern-weak.ll

55 lines
1.7 KiB
LLVM
Raw Normal View History

Implement external weak (ELF) symbols on AArch64 Weakly defined symbols should evaluate to 0 if they're undefined at link-time. This is impossible to do with the usual address generation patterns, so we should use a literal pool entry to materlialise the address. llvm-svn: 174518
2013-02-06 17:43:33 +01:00
; RUN: llc -mtriple=aarch64-none-linux-gnu -o - < %s | FileCheck %s
AArch64: implement large code model access to global variables. The MOVZ/MOVK instruction sequence may not be the most efficient (a literal-pool load could be better) but adding that would require reinstating the ConstantIslands pass. For now the sequence is correct, and that's enough. Beware, as of commit GNU ld does not appear to support the relocations needed for this. Its primary purpose (for now) will be to support JITed code, since in that case there is no guarantee of where your code will end up in memory relative to external symbols it references. llvm-svn: 181117
2013-05-04 18:53:46 +02:00
; RUN: llc -mtriple=aarch64-none-linux-gnu -code-model=large -o - < %s | FileCheck --check-prefix=CHECK-LARGE %s
Implement external weak (ELF) symbols on AArch64 Weakly defined symbols should evaluate to 0 if they're undefined at link-time. This is impossible to do with the usual address generation patterns, so we should use a literal pool entry to materlialise the address. llvm-svn: 174518
2013-02-06 17:43:33 +01:00
declare extern_weak i32 @var()
define i32()* @foo() {
; The usual ADRP/ADD pair can't be used for a weak reference because it must
; evaluate to 0 if the symbol is undefined. We use a litpool entry.
ret i32()* @var
; CHECK: .LCPI0_0:
; CHECK-NEXT: .xword var
AArch64: remove ConstantIsland pass & put literals in separate section. This implements the review suggestion to simplify the AArch64 backend. If we later discover that we *really* need the extra complexity of the ConstantIslands pass for performance reasons it can be resurrected. llvm-svn: 175258
2013-02-15 10:33:43 +01:00
; CHECK: ldr x0, [{{x[0-9]+}}, #:lo12:.LCPI0_0]
AArch64: implement large code model access to global variables. The MOVZ/MOVK instruction sequence may not be the most efficient (a literal-pool load could be better) but adding that would require reinstating the ConstantIslands pass. For now the sequence is correct, and that's enough. Beware, as of commit GNU ld does not appear to support the relocations needed for this. Its primary purpose (for now) will be to support JITed code, since in that case there is no guarantee of where your code will end up in memory relative to external symbols it references. llvm-svn: 181117
2013-05-04 18:53:46 +02:00
; In the large model, the usual relocations are absolute and can
; materialise 0.
; CHECK-LARGE: movz x0, #:abs_g3:var
; CHECK-LARGE: movk x0, #:abs_g2_nc:var
; CHECK-LARGE: movk x0, #:abs_g1_nc:var
; CHECK-LARGE: movk x0, #:abs_g0_nc:var
Implement external weak (ELF) symbols on AArch64 Weakly defined symbols should evaluate to 0 if they're undefined at link-time. This is impossible to do with the usual address generation patterns, so we should use a literal pool entry to materlialise the address. llvm-svn: 174518
2013-02-06 17:43:33 +01:00
}
AArch64: don't drop GlobalAddress offset when handling extern_weak decls. llvm-svn: 176258
2013-02-28 15:36:24 +01:00
@arr_var = extern_weak global [10 x i32]
define i32* @bar() {
%addr = getelementptr [10 x i32]* @arr_var, i32 0, i32 5
; CHECK: .LCPI1_0:
; CHECK-NEXT: .xword arr_var
; CHECK: ldr [[BASE:x[0-9]+]], [{{x[0-9]+}}, #:lo12:.LCPI1_0]
; CHECK: add x0, [[BASE]], #20
ret i32* %addr
AArch64: implement large code model access to global variables. The MOVZ/MOVK instruction sequence may not be the most efficient (a literal-pool load could be better) but adding that would require reinstating the ConstantIslands pass. For now the sequence is correct, and that's enough. Beware, as of commit GNU ld does not appear to support the relocations needed for this. Its primary purpose (for now) will be to support JITed code, since in that case there is no guarantee of where your code will end up in memory relative to external symbols it references. llvm-svn: 181117
2013-05-04 18:53:46 +02:00
; In the large model, the usual relocations are absolute and can
; materialise 0.
; CHECK-LARGE: movz x0, #:abs_g3:arr_var
; CHECK-LARGE: movk x0, #:abs_g2_nc:arr_var
; CHECK-LARGE: movk x0, #:abs_g1_nc:arr_var
; CHECK-LARGE: movk x0, #:abs_g0_nc:arr_var
AArch64: be more careful resorting to inefficient addressing for weak vars. If an otherwise weak var is actually defined in this unit, it can't be undefined at runtime so we can use normal global variable sequences (ADRP/ADD) to access it. llvm-svn: 176259
2013-02-28 15:36:31 +01:00
}
@defined_weak_var = internal unnamed_addr global i32 0
define i32* @wibble() {
ret i32* @defined_weak_var
; CHECK: adrp [[BASE:x[0-9]+]], defined_weak_var
; CHECK: add x0, [[BASE]], #:lo12:defined_weak_var
AArch64: implement large code model access to global variables. The MOVZ/MOVK instruction sequence may not be the most efficient (a literal-pool load could be better) but adding that would require reinstating the ConstantIslands pass. For now the sequence is correct, and that's enough. Beware, as of commit GNU ld does not appear to support the relocations needed for this. Its primary purpose (for now) will be to support JITed code, since in that case there is no guarantee of where your code will end up in memory relative to external symbols it references. llvm-svn: 181117
2013-05-04 18:53:46 +02:00
; CHECK-LARGE: movz x0, #:abs_g3:defined_weak_var
; CHECK-LARGE: movk x0, #:abs_g2_nc:defined_weak_var
; CHECK-LARGE: movk x0, #:abs_g1_nc:defined_weak_var
; CHECK-LARGE: movk x0, #:abs_g0_nc:defined_weak_var
Convert CodeGen/*/*.ll tests to use the new CHECK-LABEL for easier debugging. No functionality change and all tests pass after conversion. This was done with the following sed invocation to catch label lines demarking function boundaries: sed -i '' "s/^;\( *\)\([A-Z0-9_]*\):\( *\)test\([A-Za-z0-9_-]*\):\( *\)$/;\1\2-LABEL:\3test\4:\5/g" test/CodeGen/*/*.ll which was written conservatively to avoid false positives rather than false negatives. I scanned through all the changes and everything looks correct. llvm-svn: 186258
2013-07-13 22:38:47 +02:00
}
Reference in New Issue Copy Permalink