Summary:
The patch adds instructions number generated by a solution
to LSR cost under "-lsr-insns-cost" option.
Reviewers: qcolombet, hfinkel
Differential Revision: http://reviews.llvm.org/D28307
From: Evgeny Stupachenko <evstupac@gmail.com>
llvm-svn: 294821
There are no vldN/vstN f16 variants, even with +fullfp16.
We could use the i16 variants, but, in practice, even with +fullfp16,
the f16 sequence leading to the i16 shuffle usually gets scalarized.
We'd need to improve our support for f16 codegen before getting there.
Teach the cost model to consider f16 interleaved operations as
expensive. Otherwise, we are all but guaranteed to end up with
a large block of scalarized vector code.
llvm-svn: 294819
There are no vldN/vstN f16 variants, even with +fullfp16.
We could use the i16 variants, but, in practice, even with +fullfp16,
the f16 sequence leading to the i16 shuffle usually gets scalarized.
We'd need to improve our support for f16 codegen before getting there.
Reject f16 interleaved accesses. If we try to emit the f16 intrinsics,
we'll just end up with a selection failure.
llvm-svn: 294818
Unfortunately some consumers of our .o files produced with -r expect
only one section symbol per section. That is true of at least of go's
own linker.
Combining them is a somewhat convoluted process. We have to create a
symbol for every section since we don't know which ones will be
needed. The relocation sections also have to be written first to
handle the Elf_Rel addend.
I did consider a completely different approach:
We could remove the -r special case of relocation sections when
reading. We would instead have a copyRelocs function that is used
instead of scanRelocs. It would create a DynamicReloc for each
relocation and a RelocationSection for each input relocation section.
A complication of such change is that DynamicReloc would have to take
a section index and a input section instead of a symbol since with
-emit-relocs some DynamicReloc would hold relocations referring to the
dynamic symbol table and other to the static symbol table.
That would be a pretty big change, and if we do it it is probably
better to do it as a refactoring.
llvm-svn: 294816
The recommit includes some changes of testcases. No functional change to the patch.
In RateRegister of existing LSR, if a formula contains a Reg which is a SCEVAddRecExpr,
and this SCEVAddRecExpr's loop is an outerloop, the formula will be marked as Loser
and dropped.
Suppose we have an IR that %for.body is outerloop and %for.body2 is innerloop. LSR only
handle inner loop now so only %for.body2 will be handled.
Using the logic above, formula like
reg(%array) + reg({1,+, %size}<%for.body>) + 1*reg({0,+,1}<%for.body2>) will be dropped
no matter what because reg({1,+, %size}<%for.body>) is a SCEVAddRecExpr type reg related
with outerloop. Only formula like
reg(%array) + 1*reg({{1,+, %size}<%for.body>,+,1}<nuw><nsw><%for.body2>) will be kept
because the SCEVAddRecExpr related with outerloop is folded into the initial value of the
SCEVAddRecExpr related with current loop.
But in some cases, we do need to share the basic induction variable
reg{0 ,+, 1}<%for.body2> among LSR Uses to reduce the final total number of induction
variables used by LSR, so we don't want to drop the formula like
reg(%array) + reg({1,+, %size}<%for.body>) + 1*reg({0,+,1}<%for.body2>) unconditionally.
From the existing comment, it tries to avoid considering multiple level loops at the same time.
However, existing LSR only handles innermost loop, so for any SCEVAddRecExpr with a loop other
than current loop, it is an invariant and will be simple to handle, and the formula doesn't have
to be dropped.
Differential Revision: https://reviews.llvm.org/D26429
llvm-svn: 294814
This was previously searched in CMakeLists.txt unconditionally but as of
r294371 it is only searched in some circumstances. Repeating the search
in config-ix.cmake to make this robust and hopefully fix the macOS
Asan+Ubsan jenkins build.
llvm-svn: 294811
This was marking the loop for deletion after the loop was deleted. This
almost works, except that when we do any kind of debug logging it starts
reading the name of the loop from deleted memory or otherwise blowing
up. This can fail in a bunch of ways. I recently added a test that
*always* does this, and it started failing on the sanitizer bots.
The fix is to mark the loop as deleted in the loop PM infrastructure
before we remove the loop. We can do this by passing the updater into
the routine. That also lets us simplify a bunch of other interface
components here for a net win.
llvm-svn: 294810
Summary:
Since struct rtentry is an internal kernel-only structure on FreeBSD,
and SIOCADDRT and SIOCDELRT are not supported anyway, stop including
socketvar.h and attempting to get at the definition of struct rtentry,
and move the line with struct_rtentry_sz to the SANIZER_LINUX block.
Reviewers: kcc, kutuzov.viktor.84, emaste
Reviewed By: kcc, emaste
Subscribers: emaste, llvm-commits, kubamracek
Differential Revision: https://reviews.llvm.org/D29832
llvm-svn: 294806
What we want to actually control this behavior is something more local
than an EvalutationMode. Please see the linked revision for more
discussion on why/etc.
This fixes PR31843.
Differential Revision: https://reviews.llvm.org/D29469
llvm-svn: 294800
Summary: This patch adds whitelist for RHEL6 and RHEL7 kernels that are known to have the CVE fixed.
Reviewers: koriakin, kcc
Reviewed By: kcc
Subscribers: kubamracek
Differential Revision: https://reviews.llvm.org/D29825
llvm-svn: 294799
The summary information includes all uses of llvm.type.test and
llvm.type.checked.load intrinsics that can be used to devirtualize calls,
including any constant arguments for virtual constant propagation.
Differential Revision: https://reviews.llvm.org/D29734
llvm-svn: 294795
This is necessary to avoid warnings from GCC.
InstCombineLoadStoreAlloca.cpp:238:7: error: 'PointerReplacer' declared
with greater visibility than the type of its field 'PointerReplacer::IC'
llvm-svn: 294794
Summary:
GET_CALLER_PC doesn't work properly on 31-bit s390, as pointers are 31-bit, the MSB bit can be set when the return address is copied into integer.
This causes e.g. errors like:
#0 0xfdadb129 (<unknown module>)
#1 0x7da5e1d1 in __asan::GetStackTraceWithPcBpAndContext(__sanitizer::BufferedStackTrace*, unsigned long, unsigned long, unsigned long, void*, bool) ../../../../../libsanitizer/asan/asan_stack.h:50
#2 0x7da5e1d1 in __asan::ErrorGeneric::Print() ../../../../../libsanitizer/asan/asan_errors.cc:482
#3 0x7db178d5 in __asan::ErrorDescription::Print() ../../../../../libsanitizer/asan/asan_errors.h:360
#4 0x7db178d5 in __asan::ScopedInErrorReport::~ScopedInErrorReport() ../../../../../libsanitizer/asan/asan_report.cc:167
#5 0x7db178d5 in __asan::ReportGenericError(unsigned long, unsigned long, unsigned long, unsigned long, bool, unsigned long, unsigned int, bool) ../../../../../libsanitizer/asan/asan_report.cc:397
#6 0x7dadb14f in __interceptor_memcmp ../../../../../libsanitizer/sanitizer_common/sanitizer_common_interceptors.inc:626
#7 0x400cf5 in main /home/jakub/gcc/gcc/testsuite/c-c++-common/asan/memcmp-1.c:14
#8 0x7d807215 in __libc_start_main (/lib/libc.so.6+0x1a215)
#9 0x4007ed (/home/jakub/gcc/obj/gcc/testsuite/memcmp-1.exe+0x4007ed)
The actual return PC inside __interceptor_memcmp was 0x7dadb129 rather than 0xfdadb129.
Reviewers: koriakin, kcc
Reviewed By: kcc
Subscribers: kubamracek
Differential Revision: https://reviews.llvm.org/D29824
llvm-svn: 294793
This makes this code much more similar to what ThinLTO is
using (also API wise), so now we can probably use a single
code path instead of copying stuff around.
llvm-svn: 294792
Summary:
Symbol __tls_get_addr_internal is a GLIBC_PRIVATE private symbol on s390{,x}, the glibc folks aren't very happy about asan using it.
Additionally, only recent glibc versions have it, older versions just have __tls_get_offset and nothing else.
The patch doesn't drop the __tls_get_addr_internal interception altogether, but changes it so that it calls real __tls_get_offset function instead (and much more importantly,
that __tls_get_offset interception calls the real __tls_get_offset function).
This way it should work also on glibc 2.18 and earlier. See http://gcc.gnu.org/PR79341 for further details.
Reviewers: kcc, koriakin
Reviewed By: kcc, koriakin
Subscribers: kubamracek, mehdi_amini
Tags: #sanitizers
Differential Revision: https://reviews.llvm.org/D29735
llvm-svn: 294790
For function-scope variables with large initialisation list, FE usually
generates a global variable to hold the initializer, then generates
memcpy intrinsic to initialize the alloca. InstCombiner::visitAllocaInst
identifies such allocas which are accessed only by reading and replaces
them with the global variable. This is done by casting the global variable
to the type of the alloca and replacing all references.
However, when the global variable is in a different address space which
is disjoint with addr space 0 (e.g. for IR generated from OpenCL,
global variable cannot be in private addr space i.e. addr space 0), casting
the global variable to addr space 0 results in invalid IR for certain
targets (e.g. amdgpu).
To fix this issue, when the global variable is not in addr space 0,
instead of casting it to addr space 0, this patch chases down the uses
of alloca until reaching the load instructions, then replaces load from
alloca with load from the global variable. If during the chasing
bitcast and GEP are encountered, new bitcast and GEP based on the global
variable are generated and used in the load instructions.
Differential Revision: https://reviews.llvm.org/D27283
llvm-svn: 294786
It's actually meaningful and useful to allow such variables to have no
initializer, but we are strictly following the standard here until the C++
committee reaches consensus on allowing this.
llvm-svn: 294785
Summary:
This patch starts the implementation as discuss in the following RFC: http://lists.llvm.org/pipermail/llvm-dev/2016-October/106532.html
When optimization duplicates code that will scale down the execution count of a basic block, we will record the duplication factor as part of discriminator so that the offline process tool can find the duplication factor and collect the accurate execution frequency of the corresponding source code. Two important optimization that fall into this category is loop vectorization and loop unroll. This patch records the duplication factor for these 2 optimizations.
The recording will be guarded by a flag encode-duplication-in-discriminators, which is off by default.
Reviewers: probinson, aprantl, davidxl, hfinkel, echristo
Reviewed By: hfinkel
Subscribers: mehdi_amini, anemet, mzolotukhin, llvm-commits
Differential Revision: https://reviews.llvm.org/D26420
llvm-svn: 294782
Summary:
powerpc64 big-endian is not supported, but I believe that most logic can
be shared, except for xray_powerpc64.cc.
Also add a function InvalidateInstructionCache to xray_util.h, which is
copied from llvm/Support/Memory.cpp. I'm not sure if I need to add a unittest,
and I don't know how.
Reviewers: dberris, echristo, iteratee, kbarton, hfinkel
Subscribers: mehdi_amini, nemanjai, mgorny, llvm-commits
Differential Revision: https://reviews.llvm.org/D29742
llvm-svn: 294781
Summary:
The implementation, however, is in different arch-specific files, unless it's emulated.
Reviewers: dberris, pelikan, javed.absar
Subscribers: aemerson, llvm-commits
Differential Revision: https://reviews.llvm.org/D29796
llvm-svn: 294777
Since r274013, we've been looking through bitcasts on broadcast inputs.
In the scalar-folding case (from a load, build_vector, or sc2vec),
the input type didn't matter, as we'd simply bitcast the resulting
scalar back.
However, when broadcasting a 128-bit-lane-aligned element, we create an
EXTRACT_SUBVECTOR. Use proper types, by creating an extract_subvector
of the original input type.
llvm-svn: 294774
in this case for CPU_SUBTYPE_ARM64_ALL.
For this cpusubtype it should default to a cyclone CPU
to give proper disassembly without a -mcpu= flag.
rdar://27767188
llvm-svn: 294771
