
Similar to 3f46e5453d9310b15d974e876f6132e3cf50c4b1, this patch allows the backend to produce a faster access sequence for the local-exec TLS model, where loading from the TOC can be avoided, for local-exec TLS variables that are annotated with the "aix-small-tls" attribute. The expectation is for local-exec TLS variables to be set with this attribute through PGO. Furthermore, the optimized access sequence is only generated for local-exec TLS variables annotated with "aix-small-tls", only if they are less than ~32KB in size.
223 lines
8.1 KiB
LLVM
223 lines
8.1 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
|
|
; RUN: llc -verify-machineinstrs -mcpu=pwr7 -ppc-asm-full-reg-names \
|
|
; RUN: -mtriple powerpc64-ibm-aix-xcoff -mattr=-aix-small-local-exec-tls \
|
|
; RUN: < %s | FileCheck %s --check-prefixes=COMMONCM,SMALLCM64
|
|
; RUN: llc -verify-machineinstrs -mcpu=pwr7 -ppc-asm-full-reg-names \
|
|
; RUN: -mtriple powerpc64-ibm-aix-xcoff --code-model=large \
|
|
; RUN: -mattr=-aix-small-local-exec-tls < %s | \
|
|
; RUN: FileCheck %s --check-prefixes=COMMONCM,LARGECM64
|
|
|
|
; Test that the 'aix-small-tls' global variable attribute generates the
|
|
; optimized small-local-exec TLS sequence. Global variables without this
|
|
; attribute should still generate a TOC-based local-exec access sequence.
|
|
|
|
declare nonnull ptr @llvm.threadlocal.address.p0(ptr nonnull)
|
|
|
|
@a = thread_local(localexec) global [87 x i8] zeroinitializer, align 1 #0
|
|
@a_noattr = thread_local(localexec) global [87 x i8] zeroinitializer, align 1
|
|
@b = thread_local(localexec) global [87 x i16] zeroinitializer, align 2 #0
|
|
@b_noattr = thread_local(localexec) global [87 x i16] zeroinitializer, align 2
|
|
@c = thread_local(localexec) global [87 x i32] zeroinitializer, align 4 #0
|
|
@c_noattr = thread_local(localexec) global [87 x i32] zeroinitializer, align 4
|
|
@d = thread_local(localexec) global [87 x i64] zeroinitializer, align 8 #0
|
|
@d_noattr = thread_local(localexec) global [87 x i64] zeroinitializer, align 8 #0
|
|
|
|
@e = thread_local(localexec) global [87 x double] zeroinitializer, align 8 #0
|
|
@e_noattr = thread_local(localexec) global [87 x double] zeroinitializer, align 8
|
|
@f = thread_local(localexec) global [87 x float] zeroinitializer, align 4 #0
|
|
@f_noattr = thread_local(localexec) global [87 x float] zeroinitializer, align 4
|
|
|
|
define nonnull ptr @AddrTest1() {
|
|
; COMMONCM-LABEL: AddrTest1:
|
|
; COMMONCM: # %bb.0: # %entry
|
|
; COMMONCM-NEXT: addi r3, r13, a[TL]@le+1
|
|
; COMMONCM-NEXT: blr
|
|
entry:
|
|
%tls0 = tail call align 1 ptr @llvm.threadlocal.address.p0(ptr align 1 @a)
|
|
%arrayidx = getelementptr inbounds [87 x i8], ptr %tls0, i64 0, i64 1
|
|
ret ptr %arrayidx
|
|
}
|
|
|
|
define nonnull ptr @AddrTest1_NoAttr() {
|
|
; SMALLCM64-LABEL: AddrTest1_NoAttr:
|
|
; SMALLCM64: # %bb.0: # %entry
|
|
; SMALLCM64-NEXT: ld r3, L..C0(r2) # target-flags(ppc-tprel) @a_noattr
|
|
; SMALLCM64-NEXT: add r3, r13, r3
|
|
; SMALLCM64-NEXT: addi r3, r3, 1
|
|
; SMALLCM64-NEXT: blr
|
|
;
|
|
; LARGECM64-LABEL: AddrTest1_NoAttr:
|
|
; LARGECM64: # %bb.0: # %entry
|
|
; LARGECM64-NEXT: addis r3, L..C0@u(r2)
|
|
; LARGECM64-NEXT: ld r3, L..C0@l(r3)
|
|
; LARGECM64-NEXT: add r3, r13, r3
|
|
; LARGECM64-NEXT: addi r3, r3, 1
|
|
; LARGECM64-NEXT: blr
|
|
entry:
|
|
%tls0 = tail call align 1 ptr @llvm.threadlocal.address.p0(ptr align 1 @a_noattr)
|
|
%arrayidx = getelementptr inbounds [87 x i8], ptr %tls0, i64 0, i64 1
|
|
ret ptr %arrayidx
|
|
}
|
|
|
|
define nonnull ptr @AddrTest2() {
|
|
; COMMONCM-LABEL: AddrTest2:
|
|
; COMMONCM: # %bb.0: # %entry
|
|
; COMMONCM-NEXT: addi r3, r13, b[TL]@le+4
|
|
; COMMONCM-NEXT: blr
|
|
entry:
|
|
%tls0 = tail call align 2 ptr @llvm.threadlocal.address.p0(ptr align 2 @b)
|
|
%arrayidx = getelementptr inbounds [87 x i16], ptr %tls0, i64 0, i64 2
|
|
ret ptr %arrayidx
|
|
}
|
|
|
|
define nonnull ptr @AddrTest2_NoAttr() {
|
|
; SMALLCM64-LABEL: AddrTest2_NoAttr:
|
|
; SMALLCM64: # %bb.0: # %entry
|
|
; SMALLCM64-NEXT: ld r3, L..C1(r2) # target-flags(ppc-tprel) @b_noattr
|
|
; SMALLCM64-NEXT: add r3, r13, r3
|
|
; SMALLCM64-NEXT: addi r3, r3, 4
|
|
; SMALLCM64-NEXT: blr
|
|
;
|
|
; LARGECM64-LABEL: AddrTest2_NoAttr:
|
|
; LARGECM64: # %bb.0: # %entry
|
|
; LARGECM64-NEXT: addis r3, L..C1@u(r2)
|
|
; LARGECM64-NEXT: ld r3, L..C1@l(r3)
|
|
; LARGECM64-NEXT: add r3, r13, r3
|
|
; LARGECM64-NEXT: addi r3, r3, 4
|
|
; LARGECM64-NEXT: blr
|
|
entry:
|
|
%tls0 = tail call align 2 ptr @llvm.threadlocal.address.p0(ptr align 2 @b_noattr)
|
|
%arrayidx = getelementptr inbounds [87 x i16], ptr %tls0, i64 0, i64 2
|
|
ret ptr %arrayidx
|
|
}
|
|
|
|
define nonnull ptr @AddrTest3() {
|
|
; COMMONCM-LABEL: AddrTest3:
|
|
; COMMONCM: # %bb.0: # %entry
|
|
; COMMONCM-NEXT: addi r3, r13, c[TL]@le+12
|
|
; COMMONCM-NEXT: blr
|
|
entry:
|
|
%tls0 = tail call align 4 ptr @llvm.threadlocal.address.p0(ptr align 4 @c)
|
|
%arrayidx = getelementptr inbounds [87 x i32], ptr %tls0, i64 0, i64 3
|
|
ret ptr %arrayidx
|
|
}
|
|
|
|
define nonnull ptr @AddrTest3_NoAttr() {
|
|
; SMALLCM64-LABEL: AddrTest3_NoAttr:
|
|
; SMALLCM64: # %bb.0: # %entry
|
|
; SMALLCM64-NEXT: ld r3, L..C2(r2) # target-flags(ppc-tprel) @c_noattr
|
|
; SMALLCM64-NEXT: add r3, r13, r3
|
|
; SMALLCM64-NEXT: addi r3, r3, 12
|
|
; SMALLCM64-NEXT: blr
|
|
;
|
|
; LARGECM64-LABEL: AddrTest3_NoAttr:
|
|
; LARGECM64: # %bb.0: # %entry
|
|
; LARGECM64-NEXT: addis r3, L..C2@u(r2)
|
|
; LARGECM64-NEXT: ld r3, L..C2@l(r3)
|
|
; LARGECM64-NEXT: add r3, r13, r3
|
|
; LARGECM64-NEXT: addi r3, r3, 12
|
|
; LARGECM64-NEXT: blr
|
|
entry:
|
|
%tls0 = tail call align 4 ptr @llvm.threadlocal.address.p0(ptr align 4 @c_noattr)
|
|
%arrayidx = getelementptr inbounds [87 x i32], ptr %tls0, i64 0, i64 3
|
|
ret ptr %arrayidx
|
|
}
|
|
|
|
define nonnull ptr @AddrTest4() {
|
|
; COMMONCM-LABEL: AddrTest4:
|
|
; COMMONCM: # %bb.0: # %entry
|
|
; COMMONCM-NEXT: addi r3, r13, c[TL]@le+56
|
|
; COMMONCM-NEXT: blr
|
|
entry:
|
|
%tls0 = tail call align 8 ptr @llvm.threadlocal.address.p0(ptr align 8 @c)
|
|
%arrayidx = getelementptr inbounds [87 x i64], ptr %tls0, i64 0, i64 7
|
|
ret ptr %arrayidx
|
|
}
|
|
|
|
define nonnull ptr @AddrTest4_NoAttr() {
|
|
; SMALLCM64-LABEL: AddrTest4_NoAttr:
|
|
; SMALLCM64: # %bb.0: # %entry
|
|
; SMALLCM64-NEXT: ld r3, L..C2(r2) # target-flags(ppc-tprel) @c_noattr
|
|
; SMALLCM64-NEXT: add r3, r13, r3
|
|
; SMALLCM64-NEXT: addi r3, r3, 56
|
|
; SMALLCM64-NEXT: blr
|
|
;
|
|
; LARGECM64-LABEL: AddrTest4_NoAttr:
|
|
; LARGECM64: # %bb.0: # %entry
|
|
; LARGECM64-NEXT: addis r3, L..C2@u(r2)
|
|
; LARGECM64-NEXT: ld r3, L..C2@l(r3)
|
|
; LARGECM64-NEXT: add r3, r13, r3
|
|
; LARGECM64-NEXT: addi r3, r3, 56
|
|
; LARGECM64-NEXT: blr
|
|
entry:
|
|
%tls0 = tail call align 8 ptr @llvm.threadlocal.address.p0(ptr align 8 @c_noattr)
|
|
%arrayidx = getelementptr inbounds [87 x i64], ptr %tls0, i64 0, i64 7
|
|
ret ptr %arrayidx
|
|
}
|
|
|
|
define nonnull ptr @AddrTest5() {
|
|
; COMMONCM-LABEL: AddrTest5:
|
|
; COMMONCM: # %bb.0: # %entry
|
|
; COMMONCM-NEXT: addi r3, r13, e[TL]@le+48
|
|
; COMMONCM-NEXT: blr
|
|
entry:
|
|
%tls0 = tail call align 8 ptr @llvm.threadlocal.address.p0(ptr align 8 @e)
|
|
%arrayidx = getelementptr inbounds [87 x double], ptr %tls0, i64 0, i64 6
|
|
ret ptr %arrayidx
|
|
}
|
|
|
|
define nonnull ptr @AddrTest5_NoAttr() {
|
|
; SMALLCM64-LABEL: AddrTest5_NoAttr:
|
|
; SMALLCM64: # %bb.0: # %entry
|
|
; SMALLCM64-NEXT: ld r3, L..C3(r2) # target-flags(ppc-tprel) @e_noattr
|
|
; SMALLCM64-NEXT: add r3, r13, r3
|
|
; SMALLCM64-NEXT: addi r3, r3, 48
|
|
; SMALLCM64-NEXT: blr
|
|
;
|
|
; LARGECM64-LABEL: AddrTest5_NoAttr:
|
|
; LARGECM64: # %bb.0: # %entry
|
|
; LARGECM64-NEXT: addis r3, L..C3@u(r2)
|
|
; LARGECM64-NEXT: ld r3, L..C3@l(r3)
|
|
; LARGECM64-NEXT: add r3, r13, r3
|
|
; LARGECM64-NEXT: addi r3, r3, 48
|
|
; LARGECM64-NEXT: blr
|
|
entry:
|
|
%tls0 = tail call align 8 ptr @llvm.threadlocal.address.p0(ptr align 8 @e_noattr)
|
|
%arrayidx = getelementptr inbounds [87 x double], ptr %tls0, i64 0, i64 6
|
|
ret ptr %arrayidx
|
|
}
|
|
|
|
define nonnull ptr @AddrTest6() {
|
|
; COMMONCM-LABEL: AddrTest6:
|
|
; COMMONCM: # %bb.0: # %entry
|
|
; COMMONCM-NEXT: addi r3, r13, f[TL]@le+16
|
|
; COMMONCM-NEXT: blr
|
|
entry:
|
|
%tls0 = tail call align 4 ptr @llvm.threadlocal.address.p0(ptr align 4 @f)
|
|
%arrayidx = getelementptr inbounds [87 x float], ptr %tls0, i64 0, i64 4
|
|
ret ptr %arrayidx
|
|
}
|
|
|
|
define nonnull ptr @AddrTest6_NoAttr() {
|
|
; SMALLCM64-LABEL: AddrTest6_NoAttr:
|
|
; SMALLCM64: # %bb.0: # %entry
|
|
; SMALLCM64-NEXT: ld r3, L..C4(r2) # target-flags(ppc-tprel) @f_noattr
|
|
; SMALLCM64-NEXT: add r3, r13, r3
|
|
; SMALLCM64-NEXT: addi r3, r3, 16
|
|
; SMALLCM64-NEXT: blr
|
|
;
|
|
; LARGECM64-LABEL: AddrTest6_NoAttr:
|
|
; LARGECM64: # %bb.0: # %entry
|
|
; LARGECM64-NEXT: addis r3, L..C4@u(r2)
|
|
; LARGECM64-NEXT: ld r3, L..C4@l(r3)
|
|
; LARGECM64-NEXT: add r3, r13, r3
|
|
; LARGECM64-NEXT: addi r3, r3, 16
|
|
; LARGECM64-NEXT: blr
|
|
entry:
|
|
%tls0 = tail call align 4 ptr @llvm.threadlocal.address.p0(ptr align 4 @f_noattr)
|
|
%arrayidx = getelementptr inbounds [87 x float], ptr %tls0, i64 0, i64 4
|
|
ret ptr %arrayidx
|
|
}
|
|
|
|
attributes #0 = { "aix-small-tls" }
|