<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<style type="text/css" style="display:none;"><!-- P {margin-top:0;margin-bottom:0;} --></style>
</head>
<body dir="ltr">
<style type="text/css" style="display:none;"><!-- P {margin-top:0;margin-bottom:0;} --></style>
<div id="divtagdefaultwrapper" style="font-size:12pt;color:#000000;font-family:Calibri,Arial,Helvetica,sans-serif;" dir="ltr">
<p>Sorry, I didn't see the email from Ahmed. I think best option is to revert right now and submit for review a restricted version later as suggested.</p>
<p><span style="font-size: 12pt;">Will do revert.</span><br>
</p>
<p>Thanks </p>
<p>Javed</p>
</div>
<hr style="display:inline-block;width:98%" tabindex="-1">
<div id="divRplyFwdMsg" dir="ltr"><font face="Calibri, sans-serif" style="font-size:11pt" color="#000000"><b>From:</b> Javed Absar<br>
<b>Sent:</b> 30 May 2017 17:43:35<br>
<b>To:</b> Vedant Kumar<br>
<b>Cc:</b> cfe-commits; Diana Picus<br>
<b>Subject:</b> Re: r304201 - [ARM] Fix Neon vector type alignment to 64-bit</font>
<div> </div>
</div>
<div>
<div id="divtagdefaultwrapper" style="font-size: 12pt; color: rgb(0, 0, 0); font-family: Calibri, Arial, Helvetica, sans-serif, EmojiFont, "Apple Color Emoji", "Segoe UI Emoji", NotoColorEmoji, "Segoe UI Symbol", "Android Emoji", EmojiSymbols;" dir="ltr">
<p>Hi.</p>
<p><br>
</p>
<p>I fixed the tests</p>
<p>- union-align.c and bool_casts.cl</p>
<p>Did you mean this error now:</p>
<p><br>
</p>
<p></p>
<pre class="console-output" style="box-sizing: border-box; white-space: pre-wrap; word-wrap: break-word; margin-top: 0px; margin-bottom: 0px; color: rgb(51, 51, 51); font-size: 13px;">FAILED: lib/Target/ARM/ARMGenGlobalISel.inc.tmp "</pre>
<pre class="console-output" style="box-sizing: border-box; white-space: pre-wrap; word-wrap: break-word; margin-top: 0px; margin-bottom: 0px; color: rgb(51, 51, 51); font-size: 13px;"><br></pre>
<pre class="console-output" style="box-sizing: border-box; white-space: pre-wrap; word-wrap: break-word; margin-top: 0px; margin-bottom: 0px; color: rgb(51, 51, 51); font-size: 13px;">because that doesn't seem related to alignment fix.</pre>
<pre class="console-output" style="box-sizing: border-box; white-space: pre-wrap; word-wrap: break-word; margin-top: 0px; margin-bottom: 0px; color: rgb(51, 51, 51); font-size: 13px;">Thanks</pre>
<pre class="console-output" style="box-sizing: border-box; white-space: pre-wrap; word-wrap: break-word; margin-top: 0px; margin-bottom: 0px; color: rgb(51, 51, 51); font-size: 13px;">Javed</pre>
<p></p>
<p><br>
</p>
<br>
<br>
<div style="color: rgb(0, 0, 0);">
<hr tabindex="-1" style="display:inline-block; width:98%">
<div id="divRplyFwdMsg" dir="ltr"><font face="Calibri, sans-serif" color="#000000" style="font-size:11pt"><b>From:</b> vsk@apple.com <vsk@apple.com> on behalf of Vedant Kumar <vsk@apple.com><br>
<b>Sent:</b> 30 May 2017 17:06<br>
<b>To:</b> Javed Absar<br>
<b>Cc:</b> cfe-commits; Diana Picus<br>
<b>Subject:</b> Re: r304201 - [ARM] Fix Neon vector type alignment to 64-bit</font>
<div> </div>
</div>
<div>Hi,
<div class=""><br class="">
</div>
<div class="">
<div class="">I think this change is responsible for a tablgen failure in stage2 builds:</div>
<div class=""><br class="">
</div>
<div class=""> <a href="http://green.lab.llvm.org/green/job/clang-stage2-configure-Rthinlto_build/2171/" class="">http://green.lab.llvm.org/green/job/clang-stage2-configure-Rthinlto_build/2171/</a></div>
<div class=""><br class="">
</div>
<div class=""><br class="">
</div>
<div>
<blockquote type="cite" class="">
<div class="">On May 30, 2017, at 4:59 AM, Diana Picus via cfe-commits <<a href="mailto:cfe-commits@lists.llvm.org" class="">cfe-commits@lists.llvm.org</a>> wrote:</div>
<br class="Apple-interchange-newline">
<div class="">
<div class="">Hi Javed,<br class="">
<br class="">
I think this broke the bots:<br class="">
<a href="http://lab.llvm.org:8011/builders/clang-cmake-armv7-a15/builds/7620" class="">http://lab.llvm.org:8011/builders/clang-cmake-armv7-a15/builds/7620</a><br class="">
<br class="">
Can you have a look, or revert?<br class="">
</div>
</div>
</blockquote>
<div><br class="">
</div>
<div>+ 1</div>
<div><br class="">
</div>
<div>best,</div>
<div>vedant</div>
<br class="">
<blockquote type="cite" class="">
<div class="">
<div class=""><br class="">
Thanks,<br class="">
Diana<br class="">
<br class="">
On 30 May 2017 at 12:12, Javed Absar via cfe-commits<br class="">
<<a href="mailto:cfe-commits@lists.llvm.org" class="">cfe-commits@lists.llvm.org</a>> wrote:<br class="">
<blockquote type="cite" class="">Author: javed.absar<br class="">
Date: Tue May 30 05:12:15 2017<br class="">
New Revision: 304201<br class="">
<br class="">
URL: <a href="http://llvm.org/viewvc/llvm-project?rev=304201&view=rev" class="">http://llvm.org/viewvc/llvm-project?rev=304201&view=rev</a><br class="">
Log:<br class="">
[ARM] Fix Neon vector type alignment to 64-bit<br class="">
<br class="">
The maximum alignment for ARM NEON data types should be 64-bits as specified<br class="">
in ARM procedure call standard document Sec. A.2 Notes.<br class="">
This patch fixes it from its current larger natural default values, except<br class="">
for Android (so as not to break existing ABI).<br class="">
Reviewed by: Stephen Hines, Renato Golin.<br class="">
Differential Revision: <a href="https://reviews.llvm.org/D33205" class="">https://reviews.llvm.org/D33205</a><br class="">
<br class="">
<br class="">
Modified:<br class="">
cfe/trunk/lib/Basic/Targets.cpp<br class="">
cfe/trunk/test/CodeGen/arm-abi-vector.c<br class="">
cfe/trunk/test/CodeGen/arm-neon-misc.c<br class="">
cfe/trunk/test/CodeGen/arm-swiftcall.c<br class="">
cfe/trunk/test/CodeGen/armv7k-abi.c<br class="">
<br class="">
Modified: cfe/trunk/lib/Basic/Targets.cpp<br class="">
URL: <a href="http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Basic/Targets.cpp?rev=304201&r1=304200&r2=304201&view=diff" class="">
http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Basic/Targets.cpp?rev=304201&r1=304200&r2=304201&view=diff</a><br class="">
==============================================================================<br class="">
--- cfe/trunk/lib/Basic/Targets.cpp (original)<br class="">
+++ cfe/trunk/lib/Basic/Targets.cpp Tue May 30 05:12:15 2017<br class="">
@@ -5382,6 +5382,11 @@ public:<br class="">
// ARM has atomics up to 8 bytes<br class="">
setAtomic();<br class="">
<br class="">
+ if (Triple.getEnvironment() == llvm::Triple::Android)<br class="">
+ MaxVectorAlign = 128; // don't break existing Android ABI<br class="">
+ else<br class="">
+ MaxVectorAlign = 64; // AAPCS<br class="">
+<br class="">
// Do force alignment of members that follow zero length bitfields. If<br class="">
// the alignment of the zero-length bitfield is greater than the member<br class="">
// that follows it, `bar', `bar' will be aligned as the type of the<br class="">
<br class="">
Modified: cfe/trunk/test/CodeGen/arm-abi-vector.c<br class="">
URL: <a href="http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm-abi-vector.c?rev=304201&r1=304200&r2=304201&view=diff" class="">
http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm-abi-vector.c?rev=304201&r1=304200&r2=304201&view=diff</a><br class="">
==============================================================================<br class="">
--- cfe/trunk/test/CodeGen/arm-abi-vector.c (original)<br class="">
+++ cfe/trunk/test/CodeGen/arm-abi-vector.c Tue May 30 05:12:15 2017<br class="">
@@ -133,20 +133,20 @@ double test_5c(__char5 *in) {<br class="">
<br class="">
double varargs_vec_9c(int fixed, ...) {<br class="">
// CHECK: varargs_vec_9c<br class="">
-// CHECK: [[VAR:%.*]] = alloca <9 x i8>, align 16<br class="">
+// CHECK: [[VAR:%.*]] = alloca <9 x i8>, align 8<br class="">
// CHECK: [[ALIGN:%.*]] = and i32 {{%.*}}, -8<br class="">
// CHECK: [[AP_ALIGN:%.*]] = inttoptr i32 [[ALIGN]] to i8*<br class="">
// CHECK: [[AP_NEXT:%.*]] = getelementptr inbounds i8, i8* [[AP_ALIGN]], i32 16<br class="">
// CHECK: [[AP_CAST:%.*]] = bitcast i8* [[AP_ALIGN]] to <9 x i8>*<br class="">
// CHECK: [[T0:%.*]] = load <9 x i8>, <9 x i8>* [[AP_CAST]], align 8<br class="">
-// CHECK: store <9 x i8> [[T0]], <9 x i8>* [[VAR]], align 16<br class="">
+// CHECK: store <9 x i8> [[T0]], <9 x i8>* [[VAR]], align 8<br class="">
// APCS-GNU: varargs_vec_9c<br class="">
-// APCS-GNU: [[VAR:%.*]] = alloca <9 x i8>, align 16<br class="">
+// APCS-GNU: [[VAR:%.*]] = alloca <9 x i8>, align 8<br class="">
// APCS-GNU: [[AP:%.*]] = load i8*,<br class="">
// APCS-GNU: [[AP_NEXT:%.*]] = getelementptr inbounds i8, i8* [[AP]], i32 16<br class="">
// APCS-GNU: [[AP_CAST:%.*]] = bitcast i8* [[AP]] to <9 x i8>*<br class="">
// APCS-GNU: [[VEC:%.*]] = load <9 x i8>, <9 x i8>* [[AP_CAST]], align 4<br class="">
-// APCS-GNU: store <9 x i8> [[VEC]], <9 x i8>* [[VAR]], align 16<br class="">
+// APCS-GNU: store <9 x i8> [[VEC]], <9 x i8>* [[VAR]], align 8<br class="">
// ANDROID: varargs_vec_9c<br class="">
// ANDROID: [[VAR:%.*]] = alloca <9 x i8>, align 16<br class="">
// ANDROID: [[ALIGN:%.*]] = and i32 {{%.*}}, -8<br class="">
@@ -246,15 +246,15 @@ double test_3s(__short3 *in) {<br class="">
<br class="">
double varargs_vec_5s(int fixed, ...) {<br class="">
// CHECK: varargs_vec_5s<br class="">
-// CHECK: [[VAR_ALIGN:%.*]] = alloca <5 x i16>, align 16<br class="">
+// CHECK: [[VAR_ALIGN:%.*]] = alloca <5 x i16>, align 8<br class="">
// CHECK: [[ALIGN:%.*]] = and i32 {{%.*}}, -8<br class="">
// CHECK: [[AP_ALIGN:%.*]] = inttoptr i32 [[ALIGN]] to i8*<br class="">
// CHECK: [[AP_NEXT:%.*]] = getelementptr inbounds i8, i8* [[AP_ALIGN]], i32 16<br class="">
// CHECK: [[AP_CAST:%.*]] = bitcast i8* [[AP_ALIGN]] to <5 x i16>*<br class="">
// CHECK: [[VEC:%.*]] = load <5 x i16>, <5 x i16>* [[AP_CAST]], align 8<br class="">
-// CHECK: store <5 x i16> [[VEC]], <5 x i16>* [[VAR_ALIGN]], align 16<br class="">
+// CHECK: store <5 x i16> [[VEC]], <5 x i16>* [[VAR_ALIGN]], align 8<br class="">
// APCS-GNU: varargs_vec_5s<br class="">
-// APCS-GNU: [[VAR:%.*]] = alloca <5 x i16>, align 16<br class="">
+// APCS-GNU: [[VAR:%.*]] = alloca <5 x i16>, align 8<br class="">
// APCS-GNU: [[AP:%.*]] = load i8*,<br class="">
// APCS-GNU: [[AP_NEXT:%.*]] = getelementptr inbounds i8, i8* [[AP]], i32 16<br class="">
// APCS-GNU: [[AP_CAST:%.*]] = bitcast i8* [[AP]] to <5 x i16>*<br class="">
<br class="">
Modified: cfe/trunk/test/CodeGen/arm-neon-misc.c<br class="">
URL: <a href="http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm-neon-misc.c?rev=304201&r1=304200&r2=304201&view=diff" class="">
http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm-neon-misc.c?rev=304201&r1=304200&r2=304201&view=diff</a><br class="">
==============================================================================<br class="">
--- cfe/trunk/test/CodeGen/arm-neon-misc.c (original)<br class="">
+++ cfe/trunk/test/CodeGen/arm-neon-misc.c Tue May 30 05:12:15 2017<br class="">
@@ -32,3 +32,11 @@ void t2(uint64_t *src1, uint8_t *src2, u<br class="">
*dst = q;<br class="">
// CHECK: store <2 x i64><br class="">
}<br class="">
+<br class="">
+// Neon types have 64-bit alignment<br class="">
+int32x4_t gl_b;<br class="">
+void t3(int32x4_t *src) {<br class="">
+// CHECK: @t3<br class="">
+ gl_b = *src;<br class="">
+// CHECK: store <4 x i32> {{%.*}}, <4 x i32>* @gl_b, align 8<br class="">
+}<br class="">
<br class="">
Modified: cfe/trunk/test/CodeGen/arm-swiftcall.c<br class="">
URL: <a href="http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm-swiftcall.c?rev=304201&r1=304200&r2=304201&view=diff" class="">
http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm-swiftcall.c?rev=304201&r1=304200&r2=304201&view=diff</a><br class="">
==============================================================================<br class="">
--- cfe/trunk/test/CodeGen/arm-swiftcall.c (original)<br class="">
+++ cfe/trunk/test/CodeGen/arm-swiftcall.c Tue May 30 05:12:15 2017<br class="">
@@ -343,7 +343,7 @@ typedef union {<br class="">
} union_hom_fp_partial;<br class="">
TEST(union_hom_fp_partial)<br class="">
// CHECK-LABEL: define void @test_union_hom_fp_partial()<br class="">
-// CHECK: [[TMP:%.*]] = alloca [[REC:%.*]], align 16<br class="">
+// CHECK: [[TMP:%.*]] = alloca [[REC:%.*]], align 8<br class="">
// CHECK: [[CALL:%.*]] = call [[SWIFTCC]] [[UAGG:{ float, float, float, float }]] @return_union_hom_fp_partial()<br class="">
// CHECK: [[CAST_TMP:%.*]] = bitcast [[REC]]* [[TMP]] to [[AGG:{ float, float, float, float }]]*<br class="">
// CHECK: [[T0:%.*]] = getelementptr inbounds [[AGG]], [[AGG]]* [[CAST_TMP]], i32 0, i32 0<br class="">
@@ -376,7 +376,7 @@ typedef union {<br class="">
} union_het_fpv_partial;<br class="">
TEST(union_het_fpv_partial)<br class="">
// CHECK-LABEL: define void @test_union_het_fpv_partial()<br class="">
-// CHECK: [[TMP:%.*]] = alloca [[REC:%.*]], align 16<br class="">
+// CHECK: [[TMP:%.*]] = alloca [[REC:%.*]], align 8<br class="">
// CHECK: [[CALL:%.*]] = call [[SWIFTCC]] [[UAGG:{ i32, i32, float, float }]] @return_union_het_fpv_partial()<br class="">
// CHECK: [[CAST_TMP:%.*]] = bitcast [[REC]]* [[TMP]] to [[AGG:{ i32, i32, float, float }]]*<br class="">
// CHECK: [[T0:%.*]] = getelementptr inbounds [[AGG]], [[AGG]]* [[CAST_TMP]], i32 0, i32 0<br class="">
@@ -413,7 +413,7 @@ TEST(int4)<br class="">
<br class="">
TEST(int8)<br class="">
// CHECK-LABEL: define {{.*}} @return_int8()<br class="">
-// CHECK: [[RET:%.*]] = alloca [[REC:<8 x i32>]], align 32<br class="">
+// CHECK: [[RET:%.*]] = alloca [[REC:<8 x i32>]], align 8<br class="">
// CHECK: [[VAR:%.*]] = alloca [[REC]], align<br class="">
// CHECK: store<br class="">
// CHECK: load<br class="">
@@ -457,7 +457,7 @@ TEST(int8)<br class="">
<br class="">
TEST(int5)<br class="">
// CHECK-LABEL: define {{.*}} @return_int5()<br class="">
-// CHECK: [[RET:%.*]] = alloca [[REC:<5 x i32>]], align 32<br class="">
+// CHECK: [[RET:%.*]] = alloca [[REC:<5 x i32>]], align 8<br class="">
// CHECK: [[VAR:%.*]] = alloca [[REC]], align<br class="">
// CHECK: store<br class="">
// CHECK: load<br class="">
<br class="">
Modified: cfe/trunk/test/CodeGen/armv7k-abi.c<br class="">
URL: <a href="http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/armv7k-abi.c?rev=304201&r1=304200&r2=304201&view=diff" class="">
http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/armv7k-abi.c?rev=304201&r1=304200&r2=304201&view=diff</a><br class="">
==============================================================================<br class="">
--- cfe/trunk/test/CodeGen/armv7k-abi.c (original)<br class="">
+++ cfe/trunk/test/CodeGen/armv7k-abi.c Tue May 30 05:12:15 2017<br class="">
@@ -83,11 +83,11 @@ typedef struct {<br class="">
OddlySizedStruct return_oddly_sized_struct() {}<br class="">
<br class="">
// CHECK: define <4 x float> @test_va_arg_vec(i8* %l)<br class="">
-// CHECK: [[ALIGN_TMP:%.*]] = add i32 {{%.*}}, 15<br class="">
-// CHECK: [[ALIGNED:%.*]] = and i32 [[ALIGN_TMP]], -16<br class="">
+// CHECK: [[ALIGN_TMP:%.*]] = add i32 {{%.*}}, 7<br class="">
+// CHECK: [[ALIGNED:%.*]] = and i32 [[ALIGN_TMP]], -8<br class="">
// CHECK: [[ALIGNED_I8:%.*]] = inttoptr i32 [[ALIGNED]] to i8*<br class="">
// CHECK: [[ALIGNED_VEC:%.*]] = bitcast i8* [[ALIGNED_I8]] to <4 x float><br class="">
-// CHECK: load <4 x float>, <4 x float>* [[ALIGNED_VEC]], align 16<br class="">
+// CHECK: load <4 x float>, <4 x float>* [[ALIGNED_VEC]], align 8<br class="">
float32x4_t test_va_arg_vec(__builtin_va_list l) {<br class="">
return __builtin_va_arg(l, float32x4_t);<br class="">
}<br class="">
<br class="">
<br class="">
_______________________________________________<br class="">
cfe-commits mailing list<br class="">
<a href="mailto:cfe-commits@lists.llvm.org" class="">cfe-commits@lists.llvm.org</a><br class="">
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits<br class="">
</blockquote>
_______________________________________________<br class="">
cfe-commits mailing list<br class="">
<a href="mailto:cfe-commits@lists.llvm.org" class="">cfe-commits@lists.llvm.org</a><br class="">
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits<br class="">
</div>
</div>
</blockquote>
</div>
<br class="">
</div>
</div>
</div>
</div>
</div>
IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose,
or store or copy the information in any medium. Thank you.
</body>
</html>