Before this patch, flags such as undef were dropped by TII::insertBranch
(used by BranchFolding pass), resulting in the following error from
machine verifier:
*** Bad machine code: Reading virtual register without a def ***
- function: hoge
- basic block: %bb.0 bb (0x562e9c240e68)
- instruction: CBranch %2:int1regs, %bb.3
- operand 0: %2:int1regs
Differential Revision: https://reviews.llvm.org/D113001
This patch adds the substring information to the fircg.ext_embox and
fircg.ext_rebox operations.
Substring is used for CHARACTER types.
This patch is part of the upstreaming effort from fir-dev branch.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D112807
Co-authored-by: Eric Schweitz <eschweitz@nvidia.com>
This patch attempts to fix a compiler crash that occurs when long
double type is used with -mno-x87 compiler option.
The option disables x87 target feature, which in turn disables x87
registers, so CG cannot select them for x86_fp80 LLVM IR type. Long
double is lowered as x86_fp80 for some targets, so it leads to a
crash.
The option seems to contradict the SystemV ABI, which requires long
double to be represented as a 80-bit floating point, and it also
requires to use x87 registers.
To avoid that, `long double` type is disabled when -mno-x87 option is
set. In addition to that, `float` and `double` also use x87 registers
for return values on 32-bit x86, so they are disabled as well.
Differential Revision: https://reviews.llvm.org/D98895
In D71220 a pattern was added to replace shuffle's insertelement operand
if inserted scalar is not demanded. The pattern was added only for
the case where the shuffle's mask size is equal to element's vector size.
However, that condition is not required because the pattern does not
change the shuffle vector size.
This patch extends the pattern to also include cases where shuffle's mask
size is not equal to element's vector size.
Differential Revision: https://reviews.llvm.org/D112318
This better decouples transfer read/write from vector-only rewrite of conv.
This form is close to ready to plop into a new vector.conv op and the vector.transfer operations to be generalized as part of generic vectorization once the properties ConvolutionOpInterface are inferred from the indexing maps.
This also results in a nice perf boost in the dw == 1 cases.
Differential revision: https://reviews.llvm.org/D112822
This refactoring prepares conv1d vectorization for a future integration into
the generic codegen path.
Once transfer_read / transfer_write vectorization also supports sliding windows,
the special pattern for conv can disappear.
This will also likely need a vector.conv operation.
Differential Revision: https://reviews.llvm.org/D112797
This reverts commit 5cbec88cbf.
Vitaly said that 2faac77f26 actually works.
Sanitizer's armv7-linux-androideabi24 configuration has other issues which haven't been identified yet, but that's unrelated to the empty symbol name issue.
If we assume SPMD-mode during the fixpoint iteration we have to execute
the kernel in SPMD-mode. If we change our mind during manifest there is
the chance of a mismatch between the simplification, e.g., of
`__kmpc_is_spmd_exec_mode` calls, and the execution mode. This problem
was introduced in D109438.
This patch is compromise to resolve the problem purely in OpenMP-opt
while trying to keep the benefits of D109438 around. This might not
always work, see `get_hardware_num_threads_in_block_fold` but it often
does. At the same time we do keep value specialization and execution
mode in sync.
Proper solutions to this problem should be considered. I believe a new
execution mode is the easiest way forward (Singleton-SPMD).
Alternatively, SPMD-mode execution can be used with a way to provide a
new thread_limit (here 1) to the runtime. This is more general and could
be useful if we see `num_threads` clauses or workshared loops with small
trip counts in the kernel. In either proposal we need to disable the
guarding for the kernel (which was the motivation for D109438).
Reviewed By: jhuber6
Differential Revision: https://reviews.llvm.org/D112894
Before we had aligned barriers the `__kmpc_barrier_simple_spmd` was
OK to be used in the custom state machine. Now that SPMD barriers are
assumed to be aligned we need to use a "generic" barrier in places
that are not aligned.
Reviewed By: tianshilei1992
Differential Revision: https://reviews.llvm.org/D112893
A lot of NVVM intrinsics can use the default intrinsic attributes (e.g.,
nosync, nofree, ...) as well as `speculatable`. The latter is important
if we want to recompute intrinsics results instead of communicating them
via memory.
I did use default attributes for almost all `readnone` attributes but
speculatable only where I had reasonable confidence they cannot
experience UB. That said, someone should double check.
TODO: There seem to be various intrinsics marked `Commutative` which
should not, e.g., fma and div.
Reviewed By: tra
Differential Revision: https://reviews.llvm.org/D109987
Global symbols cannot have any name so we need to sanitize the string
first. Also remove an assertion that is not actually necessary nor
true in general.
Reviewed By: ggeorgakoudis
Differential Revision: https://reviews.llvm.org/D112892
When we pick state 0 to initialize state but thread N is going to be the
"main thread", in generic mode, we would require extra synchronization.
Instead, we should pick the main thread to initialize state in generic
mode and any thread in SPMD mode.
Reviewed By: tianshilei1992
Differential Revision: https://reviews.llvm.org/D112874
The function to generate S_MOV_B64_IMM_PSEUDO was recently modified to
optimize AGPR to AGPR copy but it missed checking for the SGPR
clobbering for the S_MOV_B64_IMM_PSEUDO generation.
Differential Revision: https://reviews.llvm.org/D113005
Two things in this diff:
1) Warn on the invalid range, currently three types of checking, see the detailed message in the code.
2) In some situation, llvm-profgen gives lots of warnings on the truncated stacks which is noisy. This change provides a switch to `--show-detailed-warning` to skip the warnings. Alternatively, we use a summary for those warning and show the percentage of cases with those issues.
Example of warning summary.
```
warning: 0.05%(1120/2428958) cases with issue: Profile context truncated due to missing probe for call instruction.
warning: 0.00%(2/178637) cases with issue: Range does not belong to any functions, likely from external function.
```
Reviewed By: hoy
Differential Revision: https://reviews.llvm.org/D111902
Data references in a loop should not access elements over the
statically allocated size. So we can infer a loop max trip count
from this undefined behavior.
Reviewed By: reames, mkazantsev, nikic
Differential Revision: https://reviews.llvm.org/D109821
The changes in D80163 defered the assignment of MachineMemOperand (MMO)
until the X86ExpandPseudo pass. This will result in crash due to prolog
insert point been sunk across the pseudo instruction VASTART_SAVE_XMM_REGS.
Moving the assignment to the creation of the node can avoid the problem.
Reviewed By: rnk
Differential Revision: https://reviews.llvm.org/D112859
To correctly use Query, one had to first call collectInterferingVRegs to
pre-cache the query result, then call interferingVRegs. Failing the
former, interferingVRegs could be stale. This did cause a bug which was
addressed in D98232, but the underlying usability issue of the Query API
wasn't.
This patch addresses the latter by making collectInterferingVRegs an
implementation detail, and having interferingVRegs play both roles. One
side-effect of this is that interferingVRegs is not const anymore.
Differential Revision: https://reviews.llvm.org/D112882
This is similar to D98669 but I don't know whether and why ARM needs it.
I suspect this may fix the sanitizer-x86_64-linux-android bot which runs
--target=armv7-linux-androideabi24 tests
(https://lab.llvm.org/buildbot/#/builders/77)
Someone needs to investigate the root cause and I am not sure this hack
fixes the bot.
The current setup of LinalgTransformationFilter allows a
transformation to trigger when either
1) The StringAttr is not set and no filter identifier is specified.
2) The StringAttr is set and its value matches (one of) the provided
identifier.
This misses the case where the transformation should trigger either
when the attribute is not set or its value matches (one of) the
provided identifier. Since `Identifier` does not allow empty strings,
add a boolean option to match when the attribute is not set. This
option is by default off.
Differential Revision: https://reviews.llvm.org/D113057
* Properly specify reference type in enumerator_iter
* Fix constness of iterator_adaptor_base::operator*
Differential Revision: https://reviews.llvm.org/D112981
This builder exposed a somehow "unsafe" API: it pretends we can
construct an InsertOnRangeOp from a range of SSA values, even though
this will crash if these aren't the result of `arith.constant` since
the operation actually needs attribute values (a build method can't
fail gracefully).
That means that the caller must check for the producer, at which
point they can just assemble the attribute array directly and call
the existing builder.
The existing call-sites were even in a worse state here: they would
actually create a constant operation that wouldn't be used and only
serve to carry the attribute through the builder API.
Differential Revision: https://reviews.llvm.org/D112946
The 2-D case can be rewritten to generate quite fewer instructions and a single vector.shuffle which seems to provide a nice performance boost.
Add this arrow to our quiver by exposing it with a new vector transform option.
Differential Revision: https://reviews.llvm.org/D113062
Followup to https://reviews.llvm.org/D112988
Sorry, I broke this test. The test was verifying the bad behavior
of --source-quietly that the previous change fixed -- namely that
it still echos the initial list of startup commands while
sourcing them.
Updated the test to verify that --source-quietly is quiet, rather than
loud.
Reviewed By: JDevlieghere
Differential Revision: https://reviews.llvm.org/D113047
Improve error handling for the lang objc tagged-pointer info. Rather
than failing silently, report an error if we couldn't convert an
argument to an address or resolve the class descriptor.
(lldb) lang objc tagged-pointer info 0xbb6404c47a587764
error: could not get class descriptor for 0xbb6404c47a587764
(lldb) lang objc tagged-pointer info n1
error: could not convert 'n1' to a valid address
Differential revision: https://reviews.llvm.org/D112945
As pointed out in Bug 52371, the Solaris version of
`MemoryMappingLayout::Next` completely failed to handle `readlink` errors
or properly NUL-terminate the result.
This patch fixes this. Originally provided in the PR with slight
formatting changes.
Tested on `amd64-pc-solaris2.11`.
Differential Revision: https://reviews.llvm.org/D112998
This reverts commit 2d7fba5f95.
The patch was reverted because it caused regression with rocThrust
due to ambiguity of template specialization.
For details please see https://reviews.llvm.org/D109496
All heuristics for variable accesses require both access sizes to
be known, so check this once at the start, rather than for each
particular heuristic.
Testing the unsupported pattern can trigger the invalid parameter handler,
which depending on CRT configuration can abort the process.
Differential Revision: https://reviews.llvm.org/D112352
rkayaith