The positional arguments passed to _PyStack_UnpackDict are already
kept alive by the caller, so we can avoid the extra reference count
operations by using borrowed references instead of creating new ones.
This reduces reference count contention in the free-threaded build
when calling functions with keyword arguments. In particular, this
avoids contention on the type argument to `__new__` when instantiating
namedtuples with keyword arguments.
Add a FRAME_SUSPENDED_YIELD_FROM_LOCKED state that acts as a brief
lock, preventing other threads from transitioning the frame state
while gen_getyieldfrom reads the yield-from object off the stack.
Now that the specializing interpreter works with free threading,
replace ENABLE_SPECIALIZATION_FT with ENABLE_SPECIALIZATION and
replace requires_specialization_ft with requires_specialization.
Also limit the uniquely referenced check to FOR_ITER_RANGE. It's not
necessary for FOR_ITER_GEN and would cause test_for_iter_gen to fail.
This makes generator frame state transitions atomic in the free
threading build, which avoids segfaults when trying to execute
a generator from multiple threads concurrently.
There are still a few operations that aren't thread-safe and may crash
if performed concurrently on the same generator/coroutine:
* Accessing gi_yieldfrom/cr_await/ag_await
* Accessing gi_frame/cr_frame/ag_frame
* Async generator operations
This PR implements frame caching in the RemoteUnwinder class to significantly reduce memory reads when profiling remote processes with deep call stacks.
When cache_frames=True, the unwinder stores the frame chain from each sample and reuses unchanged portions in subsequent samples. Since most profiling samples capture similar call stacks (especially the parent frames), this optimization avoids repeatedly reading the same frame data from the target process.
The implementation adds a last_profiled_frame field to the thread state that tracks where the previous sample stopped. On the next sample, if the current frame chain reaches this marker, the cached frames from that point onward are reused instead of being re-read from remote memory.
The sampling profiler now enables frame caching by default.
* Factor out bodies of the largest uops, to reduce jit code size.
* Factor out common assert, also reducing jit code size.
* Limit size of jitted code for a single executor to 1MB.
Only raises if the stack pointer is both below the limit *and* above the stack base.
This prevents false positives for user-space threads, as the stack pointer will be outside those bounds
if the stack has been swapped.
This PR changes the current JIT model from trace projection to trace recording. Benchmarking: better pyperformance (about 1.7% overall) geomean versus current https://raw.githubusercontent.com/facebookexperimental/free-threading-benchmarking/refs/heads/main/results/bm-20251108-3.15.0a1%2B-7e2bc1d-JIT/bm-20251108-vultr-x86_64-Fidget%252dSpinner-tracing_jit-3.15.0a1%2B-7e2bc1d-vs-base.svg, 100% faster Richards on the most improved benchmark versus the current JIT. Slowdown of about 10-15% on the worst benchmark versus the current JIT. **Note: the fastest version isn't the one merged, as it relies on fixing bugs in the specializing interpreter, which is left to another PR**. The speedup in the merged version is about 1.1%. https://raw.githubusercontent.com/facebookexperimental/free-threading-benchmarking/refs/heads/main/results/bm-20251112-3.15.0a1%2B-f8a764a-JIT/bm-20251112-vultr-x86_64-Fidget%252dSpinner-tracing_jit-3.15.0a1%2B-f8a764a-vs-base.svg
Stats: 50% more uops executed, 30% more traces entered the last time we ran them. It also suggests our trace lengths for a real trace recording JIT are too short, as a lot of trace too long aborts https://github.com/facebookexperimental/free-threading-benchmarking/blob/main/results/bm-20251023-3.15.0a1%2B-eb73378-CLANG%2CJIT/bm-20251023-vultr-x86_64-Fidget%252dSpinner-tracing_jit-3.15.0a1%2B-eb73378-pystats-vs-base.md .
This new JIT frontend is already able to record/execute significantly more instructions than the previous JIT frontend. In this PR, we are now able to record through custom dunders, simple object creation, generators, etc. None of these were done by the old JIT frontend. Some custom dunders uops were discovered to be broken as part of this work gh-140277
The optimizer stack space check is disabled, as it's no longer valid to deal with underflow.
Pros:
* Ignoring the generated tracer code as it's automatically created, this is only additional 1k lines of code. The maintenance burden is handled by the DSL and code generator.
* `optimizer.c` is now significantly simpler, as we don't have to do strange things to recover the bytecode from a trace.
* The new JIT frontend is able to handle a lot more control-flow than the old one.
* Tracing is very low overhead. We use the tail calling interpreter/computed goto interpreter to switch between tracing mode and non-tracing mode. I call this mechanism dual dispatch, as we have two dispatch tables dispatching to each other. Specialization is still enabled while tracing.
* Better handling of polymorphism. We leverage the specializing interpreter for this.
Cons:
* (For now) requires tail calling interpreter or computed gotos. This means no Windows JIT for now :(. Not to fret, tail calling is coming soon to Windows though https://github.com/python/cpython/pull/139962
Design:
* After each instruction, the `record_previous_inst` function/label is executed. This does as the name suggests.
* The tracing interpreter lowers bytecode to uops directly so that it can obtain "fresh" values at the point of lowering.
* The tracing version behaves nearly identical to the normal interpreter, in fact it even has specialization! This allows it to run without much of a slowdown when tracing. The actual cost of tracing is only a function call and writes to memory.
* The tracing interpreter uses the specializing interpreter's deopt to naturally form the side exit chains. This allows it to side exit chain effectively, without repeating much code. We force a re-specializing when tracing a deopt.
* The tracing interpreter can even handle goto errors/exceptions, but I chose to disable them for now as it's not tested.
* Because we do not share interpreter dispatch, there is should be no significant slowdown to the original specializing interpreter on tailcall and computed got with JIT disabled. With JIT enabled, there might be a slowdown in the form of the JIT trying to trace.
* Things that could have dynamic instruction pointer effects are guarded on. The guard deopts to a new instruction --- `_DYNAMIC_EXIT`.
Add PyUnstable_ThreadState_SetStackProtection() and
PyUnstable_ThreadState_ResetStackProtection() functions
to set the stack base address and stack size of a Python
thread state.
Co-authored-by: Petr Viktorin <encukou@gmail.com>
There were a few thread-safety issues when profiling or tracing all
threads via PyEval_SetProfileAllThreads or PyEval_SetTraceAllThreads:
* The loop over thread states could crash if a thread exits concurrently
(in both the free threading and default build)
* The modification of `c_profilefunc` and `c_tracefunc` wasn't
thread-safe on the free threading build.
Move PYOS_LOG2_STACK_MARGIN, PYOS_STACK_MARGIN,
PYOS_STACK_MARGIN_BYTES and PYOS_STACK_MARGIN_SHIFT macros to
pycore_pythonrun.h internal header. Add underscore (_) prefix to the
names to make them private. Rename _PYOS to _PyOS.
For several builtin functions, we now fall back to __main__.__dict__ for the globals
when there is no current frame and _PyInterpreterState_IsRunningMain() returns
true. This allows those functions to be run with Interpreter.call().
The affected builtins:
* exec()
* eval()
* globals()
* locals()
* vars()
* dir()
We take a similar approach with "stateless" functions, which don't use any
global variables.
Ken Jin
