staticdata: Track "CI for external Method" in specsigflags
#58162
+44
−37
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
…lags The list of "newly-inferred" CI's was awkwardly sharing responsibilities for both enqueuing code to be serialized and labelling "external" vs. "internal" properties for CodeInstances so that they can be correctly installed into the cache upon deserialization. This splits the latter responsibility to a `specsigflags` bit, so that this list is used exclusively for enqueuement of code-instances (esp. code instances for foreign Methods that have been triggered via `precompile(...)` statements)
This code is a bit awkward because we'd really like `jl_precompile(...)` to be in charge of telling us all the CodeInstances it generated (and therefore the ones we'd like to serialize), but we only run that function if generating native code.
vchuravy
reviewed
Apr 20, 2025
| _Atomic(uint8_t) specsigflags; // & 0b0001 == specptr is a specialized function signature for specTypes->rettype | ||
| // & 0b0010 == invokeptr matches specptr | ||
| // & 0b0100 == from image | ||
| // & 0b1000 == targets external method (serialized in another pkgimage), implies from image |
vchuravy
reviewed
Apr 20, 2025
| @@ -2446,7 +2455,8 @@ static void jl_update_all_fptrs(jl_serializer_state *s, jl_image_t *image) | |||
| } | |||
| if (specfunc) { | |||
| jl_atomic_store_relaxed(&codeinst->specptr.fptr, fptr); | |||
| jl_atomic_store_relaxed(&codeinst->specsigflags, 0b111); // TODO: set only if confirmed to be true | |||
| uint8_t specsigflags = jl_atomic_load_relaxed(&codeinst->specsigflags); | |||
| jl_atomic_store_relaxed(&codeinst->specsigflags, (specsigflags & 0b1100) | 0b0011); // TODO: set only if confirmed to be true | |||
There was a problem hiding this comment.
Maybe time to use a enum instead of magic constants?
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.
The list of "newly-inferred" CI's was awkwardly sharing responsibilities for enqueuing code for pre-compilation and labelling CodeInstances for correct de-serialization (any CodeInstance for an external Method needs to be specially installed into the cache by
staticdata.jl)This splits the latter responsibility to a
specsigflagsbit, so that this list is used exclusively for enqueuement + serialization of code-instances (esp. code instances for foreign Methods that have been triggered viaprecompile(...)statements + foreign abs-int CI's)Split from #58133