Refactor and support multiple affinity register for a task
- Support multiple affinity register for a task
- Move iterator loop generate logic to OMPIRBuilder
- Extract iterator loop body convertion logic
- Refactor buildAffinityData by hoisting the creation of affinity_list
- IteratorsOp -> IteratorOp
- Add mlir to llvmir test
[lldb/test] Use filecheck_log to read log files from the host platform (#183422)
Some tests were using `self.filecheck` with `platform shell cat {log}`
to validate test behavior through log inspection.
This doesn't work when running the testsuite against a remote platform
since the logs are saved on the host's filesystem.
This patch refactors those call sites to use the new `filecheck_log`
helper, which ensures the log file is always read from the host
platform.
Signed-off-by: Med Ismail Bennani <ismail at bennani.ma>
[NFC][IROutliner] Add test showing llvm.typeid.for not Outlined (#183406)
This intrinsic is not valid to outline because it relies on
function-local information about types and landing pads during
lowering/codegen. The code is currently correct because the IROutliner
uses the CodeExtractor, which will not extract this intrinsic.
See llvm/llvm-project#38893 (aka llvm.org/PR39545)
[utils] update how auto-updated tests are displayed when the test is retried (#181097)
This changes how test updater output is displayed to make it less
confusing for tests with ALLOW_RETRIES. Previously it was merged into
the output string for the test result, but that hides it in many less
verbose modes, so now it's displayed separately. The FIXED status is
added, which is the same as FLAKYPASS except it highlights that a test
was auto-updated before it passed.
From PR https://github.com/llvm/llvm-project/pull/181097
Revert "[lldb/test] Fix tests reading log from remote platform instead of host" (#183419)
Reverts llvm/llvm-project#183413 because this is still fragile.
[lldb/test] Fix tests reading log from remote platform instead of host (#183413)
Some tests are using logs to validate that a test behaves correctly
however they used `platform shell cat {log}` to read the logfile.
This doesn't work when running the testsuite against a remote platform
since the logs are saved on the host's filesystem.
This patch addresses those failures by making sure we read the log file
from the host platform.
Signed-off-by: Med Ismail Bennani <ismail at bennani.ma>
[openacc] Change function wrapMultiBlockRegionWithSCFExecuteRegion to non-static (#183409)
This change updates function `wrapMultiBlockRegionWithSCFExecuteRegion`
to be non-static.
[MISched] Advance HazardRec past stalls before calling EmitInstruction (#182977)
There are three calls to bumpCycle in bumpNode. Prior to the first call,
we calculate NextCycle as the next cycle in which all of a given
instruction's required hardware resources (as defined by the SchedModel)
are available. Any gap between this calculated NextCycle and CurrCycle
measures stalls that must occur before we can schedule the given
instruction.
The second and third call handle adjustments that occur during or after
issuing of the instruction (e.g. if the number of microops exceeds the
issue width).
According to the documentation of HazardRec->EmitInstruction, we should
call this method when an instruction is emitted: "This callback is
invoked when an instruction is emitted, to advance the hazard state."
In the context of bumpNode, this implies that it should be called after
we bumpCycle for stalls that must occur before issue of the
[18 lines not shown]