Update aie-public to upstream commit 821935bb55449f708f822d0e381164ebbc33483c #137
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…licitLod (#89252) This PR fixes parsing of builtins return types in general and fixes return type when sampling an image with OpImageSampleExplicitLod in particular.
This uses a period instead of a comma; the latter was a typo.
… (#87400) These mostly are checking for various reserved bits being set. The diagnostics for gpu-dependent reserved bits have a bit more context since they seem like the most likely ones to be observed in practice. This commit also improves the error handling mechanism for MCDisassembler::onSymbolStart(). Previously it had a comment stream parameter that was just being ignored by llvm-objdump, now it returns errors using Expected<T>.
Tablegen classes MnemonicAlias, Requires, and VOPC_Real, all define a field 'Predicates'. The prior formulation resulted in the instantiated record inheriting from all three to only have the Predicate set in Requires, i.e. Gen.AssemblerPredicate. This breaks the design of GCNPredicateControl (which is a parent class of VOPC_Real) that allows multiple predicates such as SubtargetPredicate and OtherPredicates to be set on an Instruction. MnemonicAlias does not need to be defined in the same record as VOPC_Real, so we can separate the definitions and remove Requires to avoid the issue. NFCI, but it enables future changes, such as setting multiple predicates on a VOPC_Real.
…ce, Shuffle VectorOps in VectorLinearize (#88204) This PR adds support for converting `vector.extract_strided_slice` and `vector.extract` operations to equivalent `vector.shuffle` operations that operates on linearized (1-D) vectors. `vector.shuffle` operations operating on n-D (n > 1) are also converted to equivalent shuffle operations working on linearized vectors.
The `LLDB_TEST_USE_VENDOR_PACKAGES` has defaulted to `Off` for a while. Either installing `pexpect` or skipping those tests with `-DLLDB_TEST_USER_ARGS=--skip-category=pexpect` seems to be enough that we can fully remove this option. This patch removes the `LLDB_TEST_USE_VENDOR_PACKAGES` cmake configuration as well as the associated code to add `third_party/Python/module` to the python path. I'll do the actual deletion of `third_party/Python/module` in a followup PR in the (unlikely, I hope) event this commit needs to be reverted.
This means that the pass is unusable without some sort of profile. We can revisit this decision later if we want to support running this pass without a profile.
…(#88357) Currently, when moving symbols from one `InputSection` to another (like in ICF) we directly update the symbol's `isec`, `unwindEntry` and `size`. By doing this we lose the original information. This information will be needed in a future change. Since when moving symbols we always set the symbol's `wasCoalesced` and `isec-> replacement`, we can just use this info to conditionally get the information we need at access time.
…on config (#89270)
This PR adds a check within `PutFile` to exit early when both local and destination files have matching MD5 hashes. If they differ, or there is trouble getting the hashes, the regular code path to put the file is run. As I needed this to talk to an `lldb-server` which runs the gdb-remote protocol, I enabled `CalculateMD5` within `Platform/gdb-server` and also found and fixed a parsing bug within it as well. Before this PR, the client is incorrectly parsing the response packet containing the checksum; after this PR, hopefully this is fixed. There is a test for the parsing behavior included in this PR. --------- Co-authored-by: Anthony Ha <[email protected]>
…rst` (#89136) Fixes #89135.
Closes #88067. Adding @nickdesaulniers for review.
…(#89142) Note this also likely fixes a bunch of other cases. We were double-diagnosting in a template because we were generating the expression anyway, so any attempts to instantiate the function would instantiate the expression, thus re-diagnosing it. This patch replaces it with a RecoveryExpr. Additionally, VerifyIntegerConstantExpression couldn't handle the RecoveryExpr, as it requires a non-dependent expression result (which a RecoveryExpr is dependent). Additionally, callers of it use the return value to decide that VerifyIntegerConstantExpression succeeded, so it fails if it sees a RecoveryExpr.
This clause just takes an 'int expr', which is not optional. This patch implements the clause on compute constructs.
Fix issue llvm/llvm-project#54624 Add front end flags -gtemplate-alias (also a cc1 flag) and -gno-template-alias to enable/disable usage of the feature. It's enabled by default in the front end for SCE debugger tuning only. GCC emits DW_TAG_typedef for template aliases (as does Clang with this feature disabled), and GDB and LLDB appear not to support DW_TAG_template_alias. The -Xclang option -gsimple-template-names=mangled is treated the same as =full, which is not a regression from current behaviour for template aliases. The current implementation omits defaulted arguments and as a consequence also omits empty parameter packs that come after defaulted arguments. Again, this isn't a regression as the DW_TAG_typedef name doesn't contain defaulted arguments. LLVM support added in llvm/llvm-project#88943 Added template-alias.cpp - Check the metadata construction & interaction with -gsimple-template-names. Added variadic-template-alias.cpp - Check template parameter packs work. Added defaulted-template-alias.cpp - Check defaulted args (don't) work. Modified debug-options.c - Check Clang generates correct cc1 flags.
…n json files (#89281) Moving towards displaying impl status of standard header macros. The macros aren't handled by docgen yet, so I haven't included updated rst files.
The 'vector_length' clause is semantically identical to the 'num_workers' clause, in that it takes a mandatory single int-expr. This is implemented identically to it.
The memory sanitizer doesn't recognize the results of the pipe syscall as being initialized. This patch manually unpoisons that memory.
Tagging bugs as "good first issue" and taking the time to work with new contributors pays dividends.
This patch adds Version2 of the indexed MemProf format. The new format comes with a hash table from CallStackId to actual call stacks llvm::SmallVector<FrameId>. The rest of the format refers to call stacks with CallStackId. This "values + references" model effectively deduplicates call stacks. Without this patch, a large indexed memprof file of mine shrinks from 4.4GB to 1.6GB, a 64% reduction. This patch does not make Version2 generally available yet as I am planning to make a few more changes to the format.
`clang -g -gpubnames` (with optional -gsplit-dwarf) creates the
`.debug_names` section ("per-CU" index). By default lld concatenates
input `.debug_names` sections into an output `.debug_names` section.
LLDB can consume the concatenated section but the lookup performance is
not good.
This patch adds --debug-names to create a per-module index by combining
the per-CU indexes into a single index that covers the entire load
module. The produced `.debug_names` is a replacement for `.gdb_index`.
Type units (-fdebug-types-section) are not handled yet.
Co-authored-by: Fangrui Song <[email protected]>
---------
Co-authored-by: Fangrui Song <[email protected]>
This is partly for readability and partly for consistency with other id-to-frame callbacks.
This change is in line with similar notation in SPIRV. Decoupling version numbering of DXIL and of Shader Model is intended to enable accurate specification of DXIL features that can be targeted by different Shader Model versions.
konstantinschwarz
requested review from
abnikant,
andcarminati,
gbossu,
khallouh,
martien-de-jong,
SagarMaheshwari99 and
stephenneuendorffer
as code owners
|
@konstantinschwarz Is there a reason why the version bump is so old? It looks like that commit landed upstream in April. |
…has been added in fd3eaf7
Fixup needed after 601e102
Fixup needed after b24af43
Fixup needed after 94c988b
Fixup needed after 5e5b656
Fixup needed after 3a106e5
…T x) combine We have a generic combine in 42fe530 that triggers in these cases.
New upstream tests were added that fail due to revert of bfloat16 fallback in GlobalISel and the AIE1 i48 ValueType.
Dead copies are removed after 23d45e5
I'm doing an incremental upgrade to avoid having too many conflicts and fixup commits at once. |
I thought it might be something like this, I just wanted to make sure there wasn't any massively breaking change upstream that prevented us from going beyond this particular commit 😄 Thanks for the clarification! |
konstantinschwarz
force-pushed
the
kschwarz.update.llvm.base
branch
from
fbc00d4
to
f83ee65
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.