ELFLinkGraphBuilder<ELFT> will hold generic parsing and LinkGraph-building code
that can be shared between JITLink ELF backends for different architectures.
For now it's just a stub. The plan is to incrementally move functionality down
from ELFLinkGraphBuilder_x86_64 into the new template.
Clang can be configured with a different default unwindlib, for example
gcc. In that case, -lunwind will not be present in the output.
Fix this by explicitly specifying libunwind as the unwindlib.
Differential Revision: https://reviews.llvm.org/D104899
If type legalization is going to insert a sign_extend for other users
of X and we can fold the sign_extend into ADDW/MULW/SUBW, it is
better to replace the ANY_EXTEND so we don't end up with a separate
ADD/MUL/SUB instruction for the users of the ANY_EXTEND.
I'm only handling setcc uses right now, but there are other
instructions that force sign_extends like ashr.
There are probably other *W instructions we could use in addition
to ADDW/SUBW/MULW.
My motivating case was a loop terminating compare and a phi use
as seen in the new test file.
Reviewed By: asb
Differential Revision: https://reviews.llvm.org/D104581
When set opt-bisect-limit to some value that is less than ISel pass
in command line and CurBisectNum expired, "DAG to DAG" pass lower
its opt level to O0. However "processimpdefs" and "X86 FP Stackifier"
is not stopped due to the CurBisectNum expiration. So undefined fp0
is generated. This cause crash in the "X86 FP Stackifier" pass,
because Stackifier doesn't expect any undefined fp value.
Here is the scenario that cause compiler crash.
successors: %bb.26
liveins: $r14
ST_FPrr $st0, implicit-def $fpsw, implicit $fpcw
renamable $rdi = MOV64ri @.str.3.16422
renamable $rdx = LEA64r %stack.6, 1, $noreg, 0, $noreg
ADJCALLSTACKDOWN64 0, 0, 0, implicit-def $rsp, implicit-def dead
$eflags, implicit-def $ssp, implicit $rsp, implicit $ssp
dead $esi = MOV32r0 implicit-def dead $eflags, implicit-def $rsi
CALL64pcrel32 @foo, implicit $rsp, implicit $ssp, implicit $rdi,
implicit $rsi, implicit $rdx, implicit-def dead $fp0
renamable $xmm0 = MOVSDrm_alt %stack.10, 1, $noreg, 0, $noreg :: (load 8
from %stack.10)
ADJCALLSTACKUP64 0, 0, implicit-def $rsp, implicit-def dead $eflags,
implicit-def $ssp, implicit $rsp, implicit $ssp
renamable $fp2 = CHS_Fp80 killed undef renamable $fp0, implicit-def
$fpsw
JMP_1 %bb.26
The CALL64pcrel32 mark fp0 dead, so llvm free the stack slot for fp0
and the stack become empty. In the late instruction CHS_Fp80, it use
undefined register fp0, the original code assume there must be a stack
slot for the src register (fp0) without respecting it is undefined,
so llvm report error.
We have some discussion in https://reviews.llvm.org/D104440 and we
decide to fix it in fast ISel. The fix is to lower undefined fp value to
zero value, so that it release the burden of "X86 FP Stackifier" pass.
Thank Craig for the suggestion and the initial patch to fix it.
Differential Revision: https://reviews.llvm.org/D104678
Most tests passed with an extra argument to explicitly enable the pass.
One does not, deleted it as part of this change. I can't see why the codegen
would be different between default on and default off but switched on. It
can be retrieved from the project history.
This would be a revert, but git revert was not clean. Disabling the pass
and leaving it in tree is less likely to cause breakage elsewhere than
patching up the git revert conflicts on unfamiliar code. It'll be landed
without review, as @hsmhsm is believed unavailable at present.
Differential Revision: https://reviews.llvm.org/D104962
__builtin_ctzl takes an unsigned long argument which need not be 64-bit
long on all platforms. Using __builtin_ctzll, which takes an unsigned
long long argument, ensures that 64-bit values will be handled on a
wider range of platforms.
Without this change, the test corresponding to M512 fails in Windows.
Reviewed By: gchatelet
Differential Revision: https://reviews.llvm.org/D104897
Bring back the testcase dropped in
1e6303e60c and get it passing by checking
explicitly for `ptr*` in LLParser. Uses `Type::isOpaquePointerTy()` from
ad4bb82809.
Differential Revision: https://reviews.llvm.org/D104938
Word on the grapevine was that the committee had some discussion that
ended with unanimous agreement on eliminating relational function pointer comparisons.
We wanted to be bold and just ban all of them cold turkey.
But then we chickened out at the last second and are going for
eliminating just the spaceship overload candidate instead, for now.
See D104680 for reference.
This should be fine and "safe", because the only possible semantic change this
would cause is that overload resolution could possibly be ambiguous if
there was another viable candidate equally as good.
But to save face a little we are going to:
* Issue an "error" for three-way comparisons on function pointers.
But all this is doing really is changing one vague error message,
from an "invalid operands to binary expression" into an
"ordered comparison of function pointers", which sounds more like we mean business.
* Otherwise "warn" that comparing function pointers like that is totally
not cool (unless we are told to keep quiet about this).
Signed-off-by: Matheus Izvekov <mizvekov@gmail.com>
Reviewed By: rsmith
Differential Revision: https://reviews.llvm.org/D104892
Avoid standing the Objective-C runtime lock by calling
objc_copyRealizedClassList_nolock instead of objc_copyRealizedClassList.
We already guarantee that no other threads can run while we're running
this utility expression, similar to when we parse the data ourselves
from the gdb_objc_realized_classes struct.
Worst case this will crash if the list is getting edited, which won't do
any harm and we'll just try again later.
Differential revision: https://reviews.llvm.org/D104951
This was an oversight of the commit: bb93483c11 that
added support for the Frozen variants. Also added a test case for the way that
currently produces one of these variants (a copy).
Match ML.EXE's behavior for ALIGN, EVEN, and ORG directives both at file level and in STRUCTs.
We currently reject negative offsets passed to ORG inside STRUCTs (in ML.EXE and ML64.EXE, they wrap around as for an unsigned 32-bit integer).
Also, if a STRUCT is declared using an ORG directive, no value of that type can be defined.
Reviewed By: thakis
Differential Revision: https://reviews.llvm.org/D92507
This is an NFC modernization refactoring that replaces the combination
of a bool return + reference argument, with an Optional return value.
Differential Revision: https://reviews.llvm.org/D104404
This fixes a bug at LibCallSimplifier::optimizeMemChr which does the following transformation:
```
// memchr("\r\n", C, 2) != nullptr -> (1 << C & ((1 << '\r') | (1 << '\n')))
// != 0
// after bounds check.
```
As written above, a bounds check on C (whether it is less than integer bitwidth) is done before doing `1 << C` otherwise 1 << C will overflow.
If the bounds check is false, the result of (1 << C & ...) must not be used at all, otherwise the result of shift (which is poison) will contaminate the whole results.
A correct way to encode this is `select i1 (bounds check), (1 << C & ...), false` because select does not allow the unused operand to contaminate the result.
However, this optimization was introducing `and (bounds check), (1 << C & ...)` which cannot do that.
The bug was found from compilation of this C++ code: https://reviews.llvm.org/rG2fd3037ac615#1007197
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D104901
The metadata added in D102361 introduces a module flag that we can check
to determine if the module was compiled with `-fopenmp` enables. We can
now check for the precense of this instead of scanning the call graph
for OpenMP runtime functions.
Depends on D102361
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D102423
This patch adds a module level metadata flag indicating that the module
was compiled with the `-fopenmp` flag. This will make it easier for
passes like OpenMPOpt to determine if it should be run.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D102361
When the default target arch isn't one that is supported as a
windows target, we want to set a suitable architecture (so that
Clang tests that run plain 'llvm-rc' succeed checks for e.g.
"#ifdef _WIN32" even for llvm builds that default to e.g. ppc64).
But if the default target architecture is usable, don't rewrite it.
(Rewriting it, by e.g. "T.setArch(T.getArch())", normalizes the
spelling of the architecture, e.g. changing i686 to i386. Such a
change can make clang unable to find the right sysroot.)
This can't, unfortunately, practically be tested very well because
it is entirely dependent on the default triple of the llvm build.
Differential Revision: https://reviews.llvm.org/D104589
Modify the D13209 logic: for a script inside the sysroot, if an absolute path
does not exist, report an error instead of falling back to the path without the
sysroot prefix.
This matches GNU ld, which makes sense to me: we don't want to find an arbitrary
file in the host.
Reviewed By: ikudrin
Differential Revision: https://reviews.llvm.org/D104894
Add support for the .reloc directive along the lines of
other back-ends.
This fixes a regression after https://reviews.llvm.org/D104080
was merged, since that patch presupposed support for .reloc.
Types should be defined in function scope instead of a local lexical scope. Field types should be defined inside in its parent type scope.
We were seeing a type defined in a local scope causing trouble to the dwarf emitter where a context is required to be a funciton scope, a namespace or a global scope.
Reviewed By: aprantl
Differential Revision: https://reviews.llvm.org/D104937
A recent change that extended semantic analysis for actual arguments
that associate with procedure dummy arguments exposed some bugs in
regression test suites due to points of confusion in symbol table
handling in situations where a generic interface contains a specific
procedure of the same name. When passing that name as an actual
argument, for example, it's necessary to take this possibility into
account because the symbol for the generic interface shadows the
symbol of the same name for the specific procedure, which is
what needs to be checked. So add a small utility that bypasses
the symbol for a generic interface in this case, and use it
where needed.
Differential Revision: https://reviews.llvm.org/D104929
This add as a fold of sub(0, splat(sub(0, x))) -> splat(x). This can
come up in the lowering of right shifts under AArch64, where we generate
a shift left of a negated number.
Differential Revision: https://reviews.llvm.org/D103755
We don't need to have the compare output a value and then copy it
to FPSW for use by FNSTSW. Instead we can just have the compare
output Glue and glue the FNSTSW to it. InstrEmitter effectively
performed this optimization when emitting the Machine IR. Doing
it directly simplifies the codes and reduces the work in
InstrEmitter. There's no change in the machine IR at the end of
isel before and after this change.
[libomptarget][amdgpu] Build openmp for two more targets
The 4800U APU is a gfx902 and the MI100 accelerator is a gfx908.
Both numbers are listed in ROCT topology.c
Reviewed By: jhuber6
Differential Revision: https://reviews.llvm.org/D104922
Depends On D104780
Recursive work splitting instead of sequential async tasks submission gives ~20%-30% speedup in microbenchmarks.
Algorithm outline:
1. Collapse scf.parallel dimensions into a single dimension
2. Compute the block size for the parallel operations from the 1d problem size
3. Launch parallel tasks
4. Each parallel task reconstructs its own bounds in the original multi-dimensional iteration space
5. Each parallel task computes the original parallel operation body using scf.for loop nest
Reviewed By: herhut
Differential Revision: https://reviews.llvm.org/D104850
Specify the `!async.group` size (the number of tokens that will be added to it) at construction time. `async.await_all` operation can potentially race with `async.execute` operations that keep updating the group, for this reason it is required to know upfront how many tokens will be added to the group.
Reviewed By: ftynse, herhut
Differential Revision: https://reviews.llvm.org/D104780
If we have a umul.with.overflow where the multiply result is not used and one of the operands is a constant, we can perform the overflow check cheaper with a comparison then by performing the multiply and extracting the overflow flag.
(Noticed when looking at the conditions SCEV emits for overflow checks.)
Differential Revision: https://reviews.llvm.org/D104665
This option is already supported by update_test_checks.py, but it can
also be useful in update_cc_test_checks.py. For example, I'd like to
use it in OpenMP offload codegen tests to check global variables like
`.offload_maptypes*`.
Reviewed By: jdoerfert, arichardson, ggeorgakoudis
Differential Revision: https://reviews.llvm.org/D104714
For each of the x.with.overflow variants, if only the overflow bit is consumed, we can generate a direct overflow comparison. This precommits tests for each of the variants and tries to cover interesting cornercases.
This change is NFC upstream. We pass in the loop's block to the kernel
rewriter explicitly, instead of assuming it's the loop's top block. This
change is made for downstream targets where this assumption doesn't hold.
Differential Revision: https://reviews.llvm.org/D104811
LLVM GN Syncbot