`base_function["Name"]` crashes with `KeyError` when the model has
a function without an explicit `Name` (e.g. the binary entry stub).
Skip those entries silently; only named functions can be matched
by name anyway.
Replace the hoist-struct-phis pass with a split-struct-phis pass that
breaks struct PHI nodes into per-field PHIs instead of moving them
out, and update both pipelines to use the new name.
Move around the code pertaining to the notification broker, making the
various roles a bit clearer. This also de-couple the broker logic from
the local revng daemon.
Switch out from using uvicorn to hypercorn for the pypeline daemon, make
the facilities create hypercorn commands in stand-alone functions, as
they will be re-used for other servers later on.
Fix the `revng2 project init` command to work with all
`StorageProvider`s, not just the local one. It still specializes for the
local one to allow creating the model file.
Expand the `ProjectMetadata` dataclass to include additional timestamps
and the hash of the pipeline description. Change all the
`StorageProvider`s so that all the new data is saved and returned.
Before this commit the pipeline description suffered from some
non-determinism that led to some values changing between runs of
`revng`. Add extra sorting so to eliminate the non-determinism.
Consolidate the saving of object dependencies and saving the actual
objects' data into a single function called `add_objects`. The previous
methods `add_dependencies`, `add_custom_invalidation_data` and `put`
have been removed.
Due to the drifting differences, split `ScheduledTask` into
`PipeScheduledTask` and `SavepointScheduledTask`, which allows each to
specialize on the task it has to accomplish.
Make the issuing of notification completely out of band with respect to
the `Pipeline`. Now the `StorageProvider` is responsible for providing
notifications to clients. Since some providers are local-only, there is
a `LOCAL_QUEUE` which allows sensing notifications through a local
queue, re-using the local revng daemon.
Split the `invalidate` method in two phases: in `invalidateCheck` it
just checks if the model diff warrants the execution of the actual
`invalidate` method, which requires fetching the custom invalidation
data from storage. The actual `invalidation` method remains the same.
This should make invalidation faster for storage providers that store
the invalidation remotely.
The `PypeGroup` click group did not provide any real benefit:
* It added extra help about pipebox arguments, but text that was not
correct from where the help was coming from (the group does not parse
the arguments). The subcommands still reports it though.
* It forbid using non-PypeGroup and non-PypeCommand commands in the
command tree, but in the long run this is desirable so it should be
dropped
For these reasons it has been dropped.
When one of the tasks completes, it might be due to an exception, call
the `exception()` method so that the exception can actually be caught by
the try block.
The `--format` option relied on `click.Context.get_parameter_source` to
determine if the default value for the `--format` parameter was passed.
A bug in click caused this to behave "correctly", click 8.4.0 fixed the
bug. Change the logic by using `None` as the default value so that it
can be checked without using the parameter source.
A library's exported functions live in model::Function. Walk the
Functions list and insert one Symbol row per ExportedName so the
database covers both shapes.
Also, batch all symbol inserts.
Introduce a new canonicalisation pass that forces a small set of
exponential-dataflow-prone instructions (SelectInst plus the FShl
and FShr intrinsic patterns) to be stored into a dedicated local
variable. Without this, expanding their value into C expressions
during decompilation can blow up into exponential path enumeration
on real binaries.
This is a temporary workaround. A more principled split of
pathological dataflows is planned and will replace this pass.
Run -strip-dead-prototypes as part of the LLVM-pass list invoked
just before the Clifter, so that all unused helpers are stripped
away. In particular, this prevents some LLVM intrinsics that
return structs containing i1 (e.g. the overflow intrinsics) from
ending up in the C output and tripping later passes.
Add a canonicalisation pass that rewrites extractvalue from a
pointer-returning struct as an i8 GEP plus a load. Together with
ArithmeticToGEP this lets later passes consume the result through
the same GEP-based interface they use for ordinary memory accesses.
Introduce the ArithmeticToGEP canonicalisation pass. It detects
pointers (LLVM-typed pointer values, model pointers flagged by the
"revng.pointers" metadata, and pointer-tagged extract values) and
rewrites integer arithmetic computed on top of them as i8 GEPs,
followed by a ptrtoint when the original use needs an integer.
For each obvious pointer the pass also walks backwards through Add
chains and disambiguates the pointer operand of each Add when the
other operand cannot be a pointer, so the rewrite extends to integer
arithmetic that only later flows into a pointer use.
This commit introduces an analysis to import prototypes from a SQLite
database of well-known prototypes, typically built from debug info of
operating systems.
This analysis supersedes import-well-known-models.
Add two situations where the cache is pruned on startup, removing all
the cached objects from storage:
1. the version field does not match the current one
2. the model has been changed out of band
The second one is very important as it allows to use the revng2 CLI
while editing the model manually on disk.
Remove a few `Artifact` classes from `revng.support.artifact`. These
were needed to handle the idiosyncrasies of the legacy pipeline. Some of
the logic has been reworked into the ptml code.
Alessandro Di Federico