[Bug tree-optimization/91758] Clang fails to pass validation after r261089
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91758 --- Comment #9 from Martin Liška --- (In reply to Eric Botcazou from comment #8) > > I've got it. It's a usage of an uninitialize member variable in a class: > > https://bugs.llvm.org/show_bug.cgi?id=40547#c38 > > Thanks for catching this! You're welcome. It took me almost the whole day yesterday, but I learn more about store merging (and read quite some RTL dumps).
[Bug tree-optimization/91758] Clang fails to pass validation after r261089
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91758 --- Comment #8 from Eric Botcazou --- > I've got it. It's a usage of an uninitialize member variable in a class: > https://bugs.llvm.org/show_bug.cgi?id=40547#c38 Thanks for catching this!
[Bug tree-optimization/91758] Clang fails to pass validation after r261089
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91758 --- Comment #6 from Martin Liška --- > @Eric, @Jakub, @Richard: Aren't we missing something similar with the store > merging of bool:1 bit fields? > I can see the cast to 'unsigned char' from 'bool' in GIMPLE. Both should be > 1B and > so that we maybe encorporate random bits? I made a wrong assumption, one can't have value of boolean type different from 0/1. Otherwise, it will be an UBSAN: snippet2.c:34:18: runtime error: load of value 255, which is not a valid value for type 'bool'
[Bug tree-optimization/91758] Clang fails to pass validation after r261089
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91758
Martin Liška changed:
What|Removed |Added
CC||jakub at gcc dot gnu.org
--- Comment #5 from Martin Liška ---
Ok, so without changed source file I see optimized dump changed from:
D.1259546.PaddingInReg = SR.6329_147;
D.1259546.IndirectByVal = SR.6330_121;
D.1259546.IndirectRealign = SR.6331_117;
D.1259546.SRetAfterThis = SR.6332_29;
D.1259546.InReg = SR.6333_124;
D.1259546.CanBeFlattened = SR.6334_129;
MEM[base: __for_begin_145, offset: 8B] = MEM[(struct ABIArgInfo
*)];
into:
_139 = (unsigned char) SR.6329_147;
_49 = (unsigned char) SR.6330_121;
_137 = _49 << 2;
_34 = _137 | _139;
_382 = (unsigned char) SR.6331_117;
_381 = _382 << 3;
_357 = _381 | _34;
_380 = (unsigned char) SR.6332_29;
_377 = _380 << 4;
_376 = _377 | _357;
_375 = (unsigned char) SR.6333_124;
_397 = _375 << 5;
_401 = _397 | _376;
_181 = (unsigned char) SR.6334_129;
_364 = _181 << 6;
_118 = _364 | _401;
_131 = MEM[(struct ABIArgInfo *) + 21B];
_363 = _131 & 130;
_366 = _118 & 125;
_69 = _363 | _366;
MEM[(struct ABIArgInfo *) + 21B] = _69;
MEM[base: __for_begin_145, offset: 8B] = MEM[(struct ABIArgInfo
*)];
which eventually ends up with something like:
addq$32, %rbx #, ivtmp.6357
movzbl 43(%rsp), %ecx # %sfp, SR.6331
movq24(%rsp), %xmm0 # %sfp, tmp336
movl%r15d, -8(%rbx) # SR.6327, MEM[base: __for_begin_145, offset:
8B]
leal0(,%rdi,4), %edx#, tmp337
# /tmp/llvm-project/clang/lib/CodeGen/TargetInfo.cpp:7572:
__builtin_printf ("after: I.info: %d\n", I.info.getInReg ());
leaq.LC156(%rip), %rdi #,
# /tmp/llvm-project/clang/lib/CodeGen/TargetInfo.cpp:7571: I.info =
classifyArgumentType(I.type, State);
movl%r15d, 80(%rsp) # SR.6327, MEM[(struct ABIArgInfo *)
+ 16B]
sall$3, %ecx#, tmp339
orl %r14d, %edx # SR.6329, tmp338
movhps 8(%rsp), %xmm0 # %sfp, tmp336
orl %ecx, %edx # tmp339, tmp340
movzbl 42(%rsp), %ecx # %sfp, tmp341
movups %xmm0, -24(%rbx)# tmp336, MEM[base: __for_begin_145,
offset: 8B]
movaps %xmm0, 64(%rsp) # tmp336, MEM[(struct ABIArgInfo *)]
sall$4, %ecx#, tmp341
orl %ecx, %edx # tmp341, tmp342
orl %eax, %edx # tmp343, tmp344
movl%ebp, %eax # SR.6334, tmp345
sall$6, %eax#, tmp345
orl %eax, %edx # tmp345, tmp346
movzbl 85(%rsp), %eax # MEM[(struct ABIArgInfo *) + 21B],
tmp348
andl$125, %edx #, tmp347
andl$-126, %eax #, tmp348
orl %eax, %edx # tmp348, tmp350
# /tmp/llvm-project/clang/lib/CodeGen/TargetInfo.cpp:7572:
__builtin_printf ("after: I.info: %d\n", I.info.getInReg ());
xorl%eax, %eax #
# /tmp/llvm-project/clang/lib/CodeGen/TargetInfo.cpp:7571: I.info =
classifyArgumentType(I.type, State);
movb%dl, 85(%rsp) # tmp350, MEM[(struct ABIArgInfo *) +
21B]
and the problem is that we load:
0x2776490 <(anonymous
namespace)::LanaiABIInfo::computeInfo(clang::CodeGen::CGFunctionInfo&)
const+400> movzbl 0x2b(%rsp),%ecx
with the following value:
(gdb) p /t $ecx
$1 = 0111
and then we or (0111 << 3) into %edx and so that we end up with the 6th bit
set.
That is the InReg field.
Looking at the corresponding ASM code without the store-merging, there are
quite some
andl$1, %esi#, tmp329
...
andl$1, %ecx#, tmp347
and so on.
@Eric, @Jakub, @Richard: Aren't we missing something similar with the store
merging of bool:1 bit fields?
I can see the cast to 'unsigned char' from 'bool' in GIMPLE. Both should be 1B
and
so that we maybe encorporate random bits?
[Bug tree-optimization/91758] Clang fails to pass validation after r261089
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91758 --- Comment #4 from Martin Liška --- So no, the affective change is: D.1259546.IndirectRealign = SR.6335_125; D.1259546.InReg = SR.6336_130; _163 = (unsigned char) SR.6335_125; _50 = (unsigned char) SR.6336_130; _140 = _50 << 1; _34 = _140 | _163; _383 = MEM[(struct ABIArgInfo *) + 27B]; _382 = _383 & 252; _358 = _34 & 3; _381 = _382 | _358; MEM[(struct ABIArgInfo *) + 27B] = _381; which is also fine from store-merging point of view.
[Bug tree-optimization/91758] Clang fails to pass validation after r261089
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91758
--- Comment #3 from Martin Liška ---
So that's what I have:
1) reduced LLVM test-case:
$ cat /tmp/llvm-project/clang/test/CodeGen/lanai-regparm.c
void f2(int a) __attribute((regparm(0)));
void f0() {
f2(1);
}
2) I applied the following local patch to make the change smaller:
diff --git a/clang/include/clang/CodeGen/CGFunctionInfo.h
b/clang/include/clang/CodeGen/CGFunctionInfo.h
index 1f81072e23d..ef7f7410c9b 100644
--- a/clang/include/clang/CodeGen/CGFunctionInfo.h
+++ b/clang/include/clang/CodeGen/CGFunctionInfo.h
@@ -87,14 +87,14 @@ private:
unsigned AllocaFieldIndex; // isInAlloca()
};
Kind TheKind;
- bool PaddingInReg : 1;
- bool InAllocaSRet : 1;// isInAlloca()
- bool IndirectByVal : 1; // isIndirect()
- bool IndirectRealign : 1; // isIndirect()
- bool SRetAfterThis : 1; // isIndirect()
+ bool PaddingInReg;
+ bool InAllocaSRet;// isInAlloca()
+ bool IndirectByVal; // isIndirect()
+ bool CanBeFlattened; // isDirect()
+ bool SignExt ; // isExtend()
+ bool SRetAfterThis ; // isIndirect()
+ bool IndirectRealign: 1; // isIndirect()
bool InReg : 1; // isDirect() || isExtend() || isIndirect()
- bool CanBeFlattened: 1; // isDirect()
- bool SignExt : 1; // isExtend()
bool canHavePaddingType() const {
return isDirect() || isExtend() || isIndirect() || isExpand();
diff --git a/clang/lib/CodeGen/TargetInfo.cpp
b/clang/lib/CodeGen/TargetInfo.cpp
index f2696a33cfb..b115d9da145 100644
--- a/clang/lib/CodeGen/TargetInfo.cpp
+++ b/clang/lib/CodeGen/TargetInfo.cpp
@@ -7567,7 +7567,10 @@ public:
if (!getCXXABI().classifyReturnType(FI))
FI.getReturnInfo() = classifyReturnType(FI.getReturnType());
for (auto : FI.arguments())
+{
I.info = classifyArgumentType(I.type, State);
+ __builtin_printf ("after: I.info: %d\n", I.info.getInReg ());
+}
}
ABIArgInfo getIndirectResult(QualType Ty, bool ByVal, CCState ) const;
@@ -7658,7 +7661,9 @@ ABIArgInfo LanaiABIInfo::classifyArgumentType(QualType
Ty,
}
if (InReg)
return ABIArgInfo::getDirectInReg();
- return ABIArgInfo::getDirect();
+ ABIArgInfo ret = ABIArgInfo::getDirect();
+ __builtin_printf ("before: %d\n", ret.getInReg ());
+ return ret;
}
namespace {
3) I see the problematic file is:
/tmp/llvm-project/clang/lib/CodeGen/TargetInfo.cpp
4) I took the patch from r261089 and applied it to r255894 and it still fails
5) apparently one needs -O3 to expose the issue
6) If I add following dbg_cnt:
diff --git a/gcc/dbgcnt.def b/gcc/dbgcnt.def
index 0421fae7bc0..3830666dc6c 100644
--- a/gcc/dbgcnt.def
+++ b/gcc/dbgcnt.def
@@ -195,3 +195,4 @@ DEBUG_COUNTER (tree_sra)
DEBUG_COUNTER (vect_loop)
DEBUG_COUNTER (vect_slp)
DEBUG_COUNTER (dom_unreachable_edges)
+DEBUG_COUNTER (store_merging)
diff --git a/gcc/gimple-ssa-store-merging.c b/gcc/gimple-ssa-store-merging.c
index 3c63e75fcf6..2369fd4bf5d 100644
--- a/gcc/gimple-ssa-store-merging.c
+++ b/gcc/gimple-ssa-store-merging.c
@@ -166,6 +166,7 @@
#include "rtl.h"
#include "expr.h" /* For get_bit_range. */
#include "optabs-tree.h"
+#include "dbgcnt.h"
#include "selftest.h"
/* The maximum size (in bits) of the stores this pass should generate. */
@@ -3898,7 +3899,8 @@ imm_store_chain_info::output_merged_stores ()
bool ret = false;
FOR_EACH_VEC_ELT (m_merged_store_groups, i, merged_store)
{
- if (output_merged_store (merged_store))
+ if (dbg_cnt (store_merging)
+ && output_merged_store (merged_store))
{
unsigned int j;
store_immediate_info *store;
I can bisect that to one store merging transformation:
before:
MEM[(struct SmallVectorBase *)].Size = 0;
MEM[(struct SmallVectorBase *)].Capacity = 3;
after:
MEM[(unsigned int *) + 8B] = 12884901888;
if (SizeInRegs_144 > 3)
The transformation looks fine to me and it must be an issue in RTL, because
this is the only difference I see in tree optimized dump file.
The change happens in {anonymous}::ARCABIInfo::computeInfo (const struct
ARCABIInfo * const this, struct CGFunctionInfo & FI)
function. And the valgrind also points to the same function:
valgrind /tmp/llvm-project/build/bin/clang -cc1 -triple lanai-unknown-unknown
-mregparm 4 /tmp/llvm-project/clang/test/CodeGen/lanai-regparm.c -emit-llvm
==766== Memcheck, a memory error detector
==766== Copyright (C) 2002-2017, and GNU GPL'd, by Julian Seward et al.
==766== Using Valgrind-3.15.0 and LibVEX; rerun with -h for copyright info
==766== Command: /tmp/llvm-project/build/bin/clang -cc1 -triple
lanai-unknown-unknown -mregparm 4
/tmp/llvm-project/clang/test/CodeGen/lanai-regparm.c -emit-llvm
==766==
before: 0
==766== Conditional jump or move depends on uninitialised value(s)
==766==at 0x7295287: __vfprintf_internal (vfprintf-internal.c:1644)
==766==by 0x7280C7A: printf (printf.c:33)
==766==by 0x2776A28: (anonymous
[Bug tree-optimization/91758] Clang fails to pass validation after r261089
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91758 Martin Liška changed: What|Removed |Added Status|WAITING |ASSIGNED CC||marxin at gcc dot gnu.org Assignee|unassigned at gcc dot gnu.org |marxin at gcc dot gnu.org --- Comment #2 from Martin Liška --- I've reduced that and I'm working on that right now.
[Bug tree-optimization/91758] Clang fails to pass validation after r261089
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91758 Eric Botcazou changed: What|Removed |Added Status|UNCONFIRMED |WAITING Last reconfirmed||2019-09-12 CC||ebotcazou at gcc dot gnu.org Ever confirmed|0 |1 --- Comment #1 from Eric Botcazou --- > The failures first started after: > https://gcc.gnu.org/viewcvs/gcc?view=revision=261089 > gimple-ssa-store-merging.c: Include gimple-fold.h. > > While subsequent gcc tweaks resulted in different tests failing, this commit > appears to be the genesis of all of the failures. > > Steps to Reproduce: > 1. git clone https://github.com/llvm/llvm-project.git > 2. cd llvm-project > 3. mkdir build && cd build > 4. CC=gcc CXX=g++ cmake -DLLVM_ENABLE_PROJECTS=clang > -DCMAKE_BUILD_TYPE=Release -G "Unix Makefiles" ../llvm > 5. make check-all -j4 Thanks for reporting the problem, but we need more information and a reproducer, see the instructions at https://gcc.gnu.org/bugs/
