diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/vector-call-linear-args.ll b/llvm/test/Transforms/LoopVectorize/AArch64/vector-call-linear-args.ll
index ef6b8e1d83f38..ba9d57e1e4a16 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/vector-call-linear-args.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/vector-call-linear-args.ll
@@ -1,26 +1,24 @@
-; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 2
-; RUN: opt < %s -passes=loop-vectorize,instsimplify -force-vector-interleave=1 -S | FileCheck %s
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --filter "call" --version 2
+; RUN: opt < %s -passes=loop-vectorize -force-vector-interleave=1 -S | FileCheck %s --check-prefixes=NEON
+; RUN: opt < %s -mattr=+sve -passes=loop-vectorize -force-vector-interleave=1 -S | FileCheck %s --check-prefixes=SVE_OR_NEON
+; RUN: opt < %s -mattr=+sve -passes=loop-vectorize -force-vector-interleave=1 -S -prefer-predicate-over-epilogue=predicate-dont-vectorize | FileCheck %s --check-prefixes=SVE_TF
 
 target triple = "aarch64-unknown-linux-gnu"
 
 ; A call whose argument can remain a scalar because it's sequential and only the
 ; starting value is required.
-define void @test_linear(ptr noalias %a, ptr readnone %b, i64 %n) #0 {
-; CHECK-LABEL: define void @test_linear
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] {
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
-; CHECK:       for.body:
-; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[GEPB:%.*]] = getelementptr i64, ptr [[B]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    [[CALL:%.*]] = call i64 @foo(ptr [[GEPB]]) #[[ATTR1:[0-9]+]]
-; CHECK-NEXT:    [[GEPA:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    store i64 [[CALL]], ptr [[GEPA]], align 8
-; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]]
-; CHECK:       for.cond.cleanup:
-; CHECK-NEXT:    ret void
+define void @test_linear(ptr noalias %a, ptr readnone %b, i64 %n) {
+; NEON-LABEL: define void @test_linear
+; NEON-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], i64 [[N:%.*]]) {
+; NEON:    [[DATA:%.*]] = call i64 @foo(ptr [[GEPB:%.*]]) #[[ATTR0:[0-9]+]]
+;
+; SVE_OR_NEON-LABEL: define void @test_linear
+; SVE_OR_NEON-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] {
+; SVE_OR_NEON:    [[DATA:%.*]] = call i64 @foo(ptr [[GEPB:%.*]]) #[[ATTR1:[0-9]+]]
+;
+; SVE_TF-LABEL: define void @test_linear
+; SVE_TF-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] {
+; SVE_TF:    [[DATA:%.*]] = call i64 @foo(ptr [[GEPB:%.*]]) #[[ATTR1:[0-9]+]]
 ;
 entry:
   br label %for.body
@@ -28,9 +26,9 @@ entry:
 for.body:
   %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
   %gepb = getelementptr i64, ptr %b, i64 %indvars.iv
-  %call = call i64 @foo(ptr %gepb) #1
+  %data = call i64 @foo(ptr %gepb) #0
   %gepa = getelementptr inbounds i64, ptr %a, i64 %indvars.iv
-  store i64 %call, ptr %gepa
+  store i64 %data, ptr %gepa
   %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
   %exitcond = icmp eq i64 %indvars.iv.next, %n
   br i1 %exitcond, label %for.cond.cleanup, label %for.body
@@ -39,32 +37,30 @@ for.cond.cleanup:
   ret void
 }
 
-define void @test_linear_with_mask(ptr noalias %a, ptr readnone %b, i64 %n) #0 {
-; CHECK-LABEL: define void @test_linear_with_mask
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
-; CHECK:       for.body:
-; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[GEPB:%.*]] = getelementptr i64, ptr [[B]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    [[CALL:%.*]] = call i64 @foo(ptr [[GEPB]]) #[[ATTR2:[0-9]+]]
-; CHECK-NEXT:    [[GEPA:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    store i64 [[CALL]], ptr [[GEPA]], align 8
-; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]]
-; CHECK:       for.cond.cleanup:
-; CHECK-NEXT:    ret void
+define void @test_linear_with_vector(ptr noalias %a, ptr readnone %b, ptr readonly %c, i64 %n) {
+; NEON-LABEL: define void @test_linear_with_vector
+; NEON-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], ptr readonly [[C:%.*]], i64 [[N:%.*]]) {
+; NEON:    [[DATA:%.*]] = call i32 @baz(i32 [[INPUT:%.*]], ptr [[GEPB:%.*]]) #[[ATTR1:[0-9]+]]
+;
+; SVE_OR_NEON-LABEL: define void @test_linear_with_vector
+; SVE_OR_NEON-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], ptr readonly [[C:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
+; SVE_OR_NEON:    [[DATA:%.*]] = call i32 @baz(i32 [[INPUT:%.*]], ptr [[GEPB:%.*]]) #[[ATTR2:[0-9]+]]
+;
+; SVE_TF-LABEL: define void @test_linear_with_vector
+; SVE_TF-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], ptr readonly [[C:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
+; SVE_TF:    [[DATA:%.*]] = call i32 @baz(i32 [[INPUT:%.*]], ptr [[GEPB:%.*]]) #[[ATTR2:[0-9]+]]
 ;
 entry:
   br label %for.body
 
 for.body:
   %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
-  %gepb = getelementptr i64, ptr %b, i64 %indvars.iv
-  %call = call i64 @foo(ptr %gepb) #2
-  %gepa = getelementptr inbounds i64, ptr %a, i64 %indvars.iv
-  store i64 %call, ptr %gepa
+  %gepc = getelementptr i32, ptr %c, i64 %indvars.iv
+  %input = load i32, ptr %gepc, align 8
+  %gepb = getelementptr i32, ptr %b, i64 %indvars.iv
+  %data = call i32 @baz(i32 %input, ptr %gepb) #1
+  %gepa = getelementptr inbounds i32, ptr %a, i64 %indvars.iv
+  store i32 %data, ptr %gepa, align 8
   %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
   %exitcond = icmp eq i64 %indvars.iv.next, %n
   br i1 %exitcond, label %for.cond.cleanup, label %for.body
@@ -73,36 +69,28 @@ for.cond.cleanup:
   ret void
 }
 
-define void @test_linear_with_vector(ptr noalias %a, ptr readnone %b, ptr readonly %c, i64 %n) #0 {
-; CHECK-LABEL: define void @test_linear_with_vector
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], ptr readonly [[C:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
-; CHECK:       for.body:
-; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[GEPC:%.*]] = getelementptr i32, ptr [[C]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    [[DATA:%.*]] = load i32, ptr [[GEPC]], align 8
-; CHECK-NEXT:    [[GEPB:%.*]] = getelementptr i32, ptr [[B]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 @baz(i32 [[DATA]], ptr [[GEPB]]) #[[ATTR3:[0-9]+]]
-; CHECK-NEXT:    [[GEPA:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    store i32 [[CALL]], ptr [[GEPA]], align 8
-; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]]
-; CHECK:       for.cond.cleanup:
-; CHECK-NEXT:    ret void
+define void @test_linear_bad_stride(ptr noalias %a, ptr readnone %b, i64 %n) {
+; NEON-LABEL: define void @test_linear_bad_stride
+; NEON-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], i64 [[N:%.*]]) {
+; NEON:    [[DATA:%.*]] = call i64 @foo(ptr [[GEPB:%.*]]) #[[ATTR2:[0-9]+]]
+;
+; SVE_OR_NEON-LABEL: define void @test_linear_bad_stride
+; SVE_OR_NEON-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
+; SVE_OR_NEON:    [[DATA:%.*]] = call i64 @foo(ptr [[GEPB:%.*]]) #[[ATTR3:[0-9]+]]
+;
+; SVE_TF-LABEL: define void @test_linear_bad_stride
+; SVE_TF-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
+; SVE_TF:    [[DATA:%.*]] = call i64 @foo(ptr [[GEPB:%.*]]) #[[ATTR3:[0-9]+]]
 ;
 entry:
   br label %for.body
 
 for.body:
   %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
-  %gepc = getelementptr i32, ptr %c, i64 %indvars.iv
-  %data = load i32, ptr %gepc, align 8
-  %gepb = getelementptr i32, ptr %b, i64 %indvars.iv
-  %call = call i32 @baz(i32 %data, ptr %gepb) #3
-  %gepa = getelementptr inbounds i32, ptr %a, i64 %indvars.iv
-  store i32 %call, ptr %gepa, align 8
+  %gepb = getelementptr i64, ptr %b, i64 %indvars.iv
+  %data = call i64 @foo(ptr %gepb) #2
+  %gepa = getelementptr inbounds i64, ptr %a, i64 %indvars.iv
+  store i64 %data, ptr %gepa
   %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
   %exitcond = icmp eq i64 %indvars.iv.next, %n
   br i1 %exitcond, label %for.cond.cleanup, label %for.body
@@ -111,32 +99,29 @@ for.cond.cleanup:
   ret void
 }
 
-define void @test_linear_bad_stride(ptr noalias %a, ptr readnone %b, i64 %n) #0 {
-; CHECK-LABEL: define void @test_linear_bad_stride
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
-; CHECK:       for.body:
-; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[GEPB:%.*]] = getelementptr i64, ptr [[B]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    [[CALL:%.*]] = call i64 @foo(ptr [[GEPB]]) #[[ATTR4:[0-9]+]]
-; CHECK-NEXT:    [[GEPA:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    store i64 [[CALL]], ptr [[GEPA]], align 8
-; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]]
-; CHECK:       for.cond.cleanup:
-; CHECK-NEXT:    ret void
+define void @test_linear_wide_stride(ptr noalias %a, ptr readnone %b, i64 %n) {
+; NEON-LABEL: define void @test_linear_wide_stride
+; NEON-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], i64 [[N:%.*]]) {
+; NEON:    [[DATA:%.*]] = call i64 @foo(ptr [[GEPB:%.*]]) #[[ATTR2]]
+;
+; SVE_OR_NEON-LABEL: define void @test_linear_wide_stride
+; SVE_OR_NEON-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
+; SVE_OR_NEON:    [[DATA:%.*]] = call i64 @foo(ptr [[GEPB:%.*]]) #[[ATTR3]]
+;
+; SVE_TF-LABEL: define void @test_linear_wide_stride
+; SVE_TF-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
+; SVE_TF:    [[DATA:%.*]] = call i64 @foo(ptr [[GEPB:%.*]]) #[[ATTR3]]
 ;
 entry:
   br label %for.body
 
 for.body:
   %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
-  %gepb = getelementptr i64, ptr %b, i64 %indvars.iv
-  %call = call i64 @foo(ptr %gepb) #4
+  %double = mul i64 %indvars.iv, 2
+  %gepb = getelementptr i64, ptr %b, i64 %double
+  %data = call i64 @foo(ptr %gepb) #2
   %gepa = getelementptr inbounds i64, ptr %a, i64 %indvars.iv
-  store i64 %call, ptr %gepa
+  store i64 %data, ptr %gepa
   %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
   %exitcond = icmp eq i64 %indvars.iv.next, %n
   br i1 %exitcond, label %for.cond.cleanup, label %for.body
@@ -145,34 +130,29 @@ for.cond.cleanup:
   ret void
 }
 
-define void @test_linear_wide_stride(ptr noalias %a, ptr readnone %b, i64 %n) #0 {
-; CHECK-LABEL: define void @test_linear_wide_stride
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
-; CHECK:       for.body:
-; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[DOUBLE:%.*]] = mul i64 [[INDVARS_IV]], 2
-; CHECK-NEXT:    [[GEPB:%.*]] = getelementptr i64, ptr [[B]], i64 [[DOUBLE]]
-; CHECK-NEXT:    [[CALL:%.*]] = call i64 @foo(ptr [[GEPB]]) #[[ATTR4]]
-; CHECK-NEXT:    [[GEPA:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    store i64 [[CALL]], ptr [[GEPA]], align 8
-; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]]
-; CHECK:       for.cond.cleanup:
-; CHECK-NEXT:    ret void
+define void @test_linear_mixed_types(ptr noalias %a, ptr readnone %b, ptr readonly %c, i64 %n) {
+; NEON-LABEL: define void @test_linear_mixed_types
+; NEON-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], ptr readonly [[C:%.*]], i64 [[N:%.*]]) {
+; NEON:    [[DATA:%.*]] = call i32 @quux(ptr [[GEPC:%.*]], ptr [[GEPB:%.*]]) #[[ATTR3:[0-9]+]]
+;
+; SVE_OR_NEON-LABEL: define void @test_linear_mixed_types
+; SVE_OR_NEON-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], ptr readonly [[C:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
+; SVE_OR_NEON:    [[DATA:%.*]] = call i32 @quux(ptr [[GEPC:%.*]], ptr [[GEPB:%.*]]) #[[ATTR4:[0-9]+]]
+;
+; SVE_TF-LABEL: define void @test_linear_mixed_types
+; SVE_TF-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], ptr readonly [[C:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
+; SVE_TF:    [[DATA:%.*]] = call i32 @quux(ptr [[GEPC:%.*]], ptr [[GEPB:%.*]]) #[[ATTR4:[0-9]+]]
 ;
 entry:
   br label %for.body
 
 for.body:
   %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
-  %double = mul i64 %indvars.iv, 2
-  %gepb = getelementptr i64, ptr %b, i64 %double
-  %call = call i64 @foo(ptr %gepb) #4
-  %gepa = getelementptr inbounds i64, ptr %a, i64 %indvars.iv
-  store i64 %call, ptr %gepa
+  %gepc = getelementptr i32, ptr %c, i64 %indvars.iv
+  %gepb = getelementptr i64, ptr %b, i64 %indvars.iv
+  %data = call i32 @quux(ptr %gepc, ptr %gepb) #3
+  %gepa = getelementptr inbounds i32, ptr %a, i64 %indvars.iv
+  store i32 %data, ptr %gepa, align 8
   %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
   %exitcond = icmp eq i64 %indvars.iv.next, %n
   br i1 %exitcond, label %for.cond.cleanup, label %for.body
@@ -181,34 +161,29 @@ for.cond.cleanup:
   ret void
 }
 
-define void @test_linear_mixed_types(ptr noalias %a, ptr readnone %b, ptr readonly %c, i64 %n) #0 {
-; CHECK-LABEL: define void @test_linear_mixed_types
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr readnone [[B:%.*]], ptr readonly [[C:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
-; CHECK:       for.body:
-; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[GEPC:%.*]] = getelementptr i32, ptr [[C]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    [[GEPB:%.*]] = getelementptr i64, ptr [[B]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 @quux(ptr [[GEPC]], ptr [[GEPB]]) #[[ATTR5:[0-9]+]]
-; CHECK-NEXT:    [[GEPA:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    store i32 [[CALL]], ptr [[GEPA]], align 8
-; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]]
-; CHECK:       for.cond.cleanup:
-; CHECK-NEXT:    ret void
+define void @test_linear_non_ptr(ptr noalias %a, i64 %n) {
+; NEON-LABEL: define void @test_linear_non_ptr
+; NEON-SAME: (ptr noalias [[A:%.*]], i64 [[N:%.*]]) {
+; NEON:    [[DATA:%.*]] = call i32 @bar(i32 [[TREBLED:%.*]]) #[[ATTR4:[0-9]+]]
+;
+; SVE_OR_NEON-LABEL: define void @test_linear_non_ptr
+; SVE_OR_NEON-SAME: (ptr noalias [[A:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
+; SVE_OR_NEON:    [[DATA:%.*]] = call i32 @bar(i32 [[TREBLED:%.*]]) #[[ATTR5:[0-9]+]]
+;
+; SVE_TF-LABEL: define void @test_linear_non_ptr
+; SVE_TF-SAME: (ptr noalias [[A:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
+; SVE_TF:    [[DATA:%.*]] = call i32 @bar(i32 [[TREBLED:%.*]]) #[[ATTR5:[0-9]+]]
 ;
 entry:
   br label %for.body
 
 for.body:
   %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
-  %gepc = getelementptr i32, ptr %c, i64 %indvars.iv
-  %gepb = getelementptr i64, ptr %b, i64 %indvars.iv
-  %call = call i32 @quux(ptr %gepc, ptr %gepb) #5
+  %little.iv = trunc i64 %indvars.iv to i32
+  %trebled = mul i32 %little.iv, 3
+  %data = call i32 @bar(i32 %trebled) #4
   %gepa = getelementptr inbounds i32, ptr %a, i64 %indvars.iv
-  store i32 %call, ptr %gepa, align 8
+  store i32 %data, ptr %gepa
   %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
   %exitcond = icmp eq i64 %indvars.iv.next, %n
   br i1 %exitcond, label %for.cond.cleanup, label %for.body
@@ -217,23 +192,18 @@ for.cond.cleanup:
   ret void
 }
 
-define void @test_linear_non_ptr(ptr noalias %a, i64 %n) #0 {
-; CHECK-LABEL: define void @test_linear_non_ptr
-; CHECK-SAME: (ptr noalias [[A:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
-; CHECK:       for.body:
-; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[LITTLE_IV:%.*]] = trunc i64 [[INDVARS_IV]] to i32
-; CHECK-NEXT:    [[TREBLED:%.*]] = mul i32 [[LITTLE_IV]], 3
-; CHECK-NEXT:    [[CALL:%.*]] = call i32 @bar(i32 [[TREBLED]]) #[[ATTR6:[0-9]+]]
-; CHECK-NEXT:    [[GEPA:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    store i32 [[CALL]], ptr [[GEPA]], align 4
-; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]]
-; CHECK:       for.cond.cleanup:
-; CHECK-NEXT:    ret void
+define void @test_linear_non_ptr_neg_stride(ptr noalias %a, i64 %n) {
+; NEON-LABEL: define void @test_linear_non_ptr_neg_stride
+; NEON-SAME: (ptr noalias [[A:%.*]], i64 [[N:%.*]]) {
+; NEON:    [[DATA:%.*]] = call i32 @bar(i32 [[NEGSTRIDE:%.*]]) #[[ATTR5:[0-9]+]]
+;
+; SVE_OR_NEON-LABEL: define void @test_linear_non_ptr_neg_stride
+; SVE_OR_NEON-SAME: (ptr noalias [[A:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
+; SVE_OR_NEON:    [[DATA:%.*]] = call i32 @bar(i32 [[NEGSTRIDE:%.*]]) #[[ATTR6:[0-9]+]]
+;
+; SVE_TF-LABEL: define void @test_linear_non_ptr_neg_stride
+; SVE_TF-SAME: (ptr noalias [[A:%.*]], i64 [[N:%.*]]) #[[ATTR0]] {
+; SVE_TF:    [[DATA:%.*]] = call i32 @bar(i32 [[NEGSTRIDE:%.*]]) #[[ATTR6:[0-9]+]]
 ;
 entry:
   br label %for.body
@@ -241,10 +211,10 @@ entry:
 for.body:
   %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
   %little.iv = trunc i64 %indvars.iv to i32
-  %trebled = mul i32 %little.iv, 3
-  %call = call i32 @bar(i32 %trebled) #6
+  %negstride = mul i32 %little.iv, -5
+  %data = call i32 @bar(i32 %negstride) #5
   %gepa = getelementptr inbounds i32, ptr %a, i64 %indvars.iv
-  store i32 %call, ptr %gepa
+  store i32 %data, ptr %gepa
   %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
   %exitcond = icmp eq i64 %indvars.iv.next, %n
   br i1 %exitcond, label %for.cond.cleanup, label %for.body
@@ -258,18 +228,26 @@ declare i32 @baz(i32, ptr)
 declare i32 @quux(ptr, ptr)
 declare i32 @bar(i32)
 
-; vector variants of foo
-declare <vscale x 2 x i64> @foo_linear(ptr, <vscale x 2 x i1>)
-declare <vscale x 2 x i64> @foo_linear_nomask(ptr)
-declare <vscale x 2 x i64> @foo_linear_nomask_2x(ptr)
-declare <vscale x 4 x i32> @baz_vector_and_linear(<vscale x 4 x i32>, ptr)
-declare <vscale x 4 x i32> @quux_linear_mask(ptr, ptr, <vscale x 4 x i1>)
-declare <vscale x 4 x i32> @bar_linear(i32)
-
-attributes #0 = { "target-features"="+sve" }
-attributes #1 = { nounwind "vector-function-abi-variant"="_ZGVsNxl8_foo(foo_linear_nomask)" }
-attributes #2 = { nounwind "vector-function-abi-variant"="_ZGVsMxl8_foo(foo_linear)" }
-attributes #3 = { nounwind "vector-function-abi-variant"="_ZGVsNxvl4_baz(baz_vector_and_linear)" }
-attributes #4 = { nounwind "vector-function-abi-variant"="_ZGVsNxl16_foo(foo_linear_nomask_2x)" }
-attributes #5 = { nounwind "vector-function-abi-variant"="_ZGVsMxl4l8_quux(quux_linear_mask)" }
-attributes #6 = { nounwind "vector-function-abi-variant"="_ZGVsNxl3_bar(bar_linear)" }
+; neon vector variants of foo
+declare <2 x i64> @neon_foo_linear(ptr)
+declare <2 x i64> @neon_foo_linear_2x(ptr)
+declare <4 x i32> @neon_baz_vector_and_linear(<4 x i32>, ptr)
+declare <4 x i32> @neon_quux_linear(ptr, ptr)
+declare <4 x i32> @neon_bar_linear(i32)
+declare <4 x i32> @neon_bar_neg_linear(i32)
+
+; scalable vector variants of foo
+declare <vscale x 2 x i64> @sve_foo_linear(ptr, <vscale x 2 x i1>)
+declare <vscale x 2 x i64> @sve_foo_linear_nomask(ptr)
+declare <vscale x 2 x i64> @sve_foo_linear_nomask_2x(ptr)
+declare <vscale x 4 x i32> @sve_baz_vector_and_linear(<vscale x 4 x i32>, ptr)
+declare <vscale x 4 x i32> @sve_quux_linear_mask(ptr, ptr, <vscale x 4 x i1>)
+declare <vscale x 4 x i32> @sve_bar_linear(i32)
+declare <vscale x 4 x i32> @sve_bar_neg_linear(i32)
+
+attributes #0 = { nounwind "vector-function-abi-variant"="_ZGVsNxl8_foo(sve_foo_linear_nomask),_ZGVsMxl8_foo(sve_foo_linear),_ZGVnN2l8_foo(neon_foo_linear)" }
+attributes #1 = { nounwind "vector-function-abi-variant"="_ZGVsNxvl4_baz(sve_baz_vector_and_linear),_ZGVnN4vl4_baz(neon_baz_vector_and_linear)" }
+attributes #2 = { nounwind "vector-function-abi-variant"="_ZGVsNxl16_foo(sve_foo_linear_nomask_2x),_ZGVnN2l16_foo(neon_foo_linear_2x)" }
+attributes #3 = { nounwind "vector-function-abi-variant"="_ZGVsMxl4l8_quux(sve_quux_linear_mask),_ZGVnN4l4l8_quux(neon_quux_linear)" }
+attributes #4 = { nounwind "vector-function-abi-variant"="_ZGVsNxl3_bar(sve_bar_linear),_ZGVnN4l3_bar(neon_bar_linear)" }
+attributes #5 = { nounwind "vector-function-abi-variant"="_ZGVsNxln5_bar(sve_bar_neg_linear),_ZGVnN4ln5_bar(neon_bar_neg_linear)" }