Fix fatal error when compiling objects with Flambda#965
Fix fatal error when compiling objects with Flambda#965mshinwell merged 6 commits intoocaml:trunkfrom
Conversation
|
I was a bit afraid that adding 2 instructions to lambda would be heavy, but actually it's really light weight. |
|
@chambart Please review if you could, thanks. |
|
Ok, this looks good is cleaner than using Popaque and might be useful elsewhere. |
|
@chambart Please re-review the case in cmmgen.ml for |
| | (Assignment | Heap_initialization), Pointer -> | ||
| return_unit ( | ||
| addr_array_set (transl env arg1) (transl env arg2) (transl env arg3) | ||
| dbg) |
There was a problem hiding this comment.
Is there a reason not to use caml_initialize instead of caml_modify in case of initialization with a pointer value ?
If this is an error, to avoid potential future problems, I also suggest sharing the logic by something like:
type assignment_kind = Caml_modify | Caml_initialize | Simple
let assignment_kind ptr init =
match init, ptr with
| Assignment, Pointer -> Caml_modify
| Heap_initialization, Pointer -> Caml_initialize
| Assignment, Immediate
| Heap_initialization, Immediate
| Root_initialization, (Immediate | Pointer) -> Simple
|
Rereviewed. This is good to go. |
|
I don't think this issue is fully fixed, as I am getting this error when trying to build core_kernel.v0.9.115.18+07 on the 4.05.0+flambda switch (Linux x64). This should repro: |
|
@drvink could you please enter a report at https://caml.inria.fr/mantis, lest we forget about this problem? |
|
The above is not the same bug. It is the issue discussed at #1169. In the end we decided that this was a bug in |
|
@lpw25 That contradicts what was written in https://caml.inria.fr/mantis/view.php?id=7499
...which referred to this issue on GitHub--my knowledge of OCaml compiler internals is scant, so apologies if this is the ultimately the wrong issue. That said, I'm happy to write a note on Mantis if you guys simply want a note that references receiving this error message in the context of #965, #1169, and the aforementioned Mantis issue. |
|
I don't really agree with that note: the way it is currently written this fatal error is intended to happen when users are using unsafe features. Really this fatal error should be treated the same as the case of mutating an immutable block -- which is currently handled by a warning. I'm going to talk with Mark and Pierre about these cases because personally I don't even think they should warn since the warnings are fragile and can usually be made to appear in safe code using GADTs. It might be worth adding a note on Mantis about this case, although the issue was already discussed in the GPR I linked to. It is worth noting that in this case the compiler has a point, it isn't 100% safe to use that version of Base with 4.05 because it is possible for code to segfault due to the new float array optimisation. |
b11eea1 flambda-backend: Introduce Import_info (ocaml#1036) bc5b135 flambda-backend: Fix `ocamlobjinfo` on flambda2 .cmx files (ocaml#1029) c8babbd flambda-backend: Compilation_unit optimisations (ocaml#1035) e8d3e22 flambda-backend: Use 4.14.0 opam switch for building (includes upgrading ocamlformat to 0.24.1) (ocaml#1030) eb14a86 flambda-backend: Port PR81 from ocaml-jst (ocaml#1024) 131bc12 flambda-backend: Merge ocaml-jst 2022-12-13 (ocaml#1022) 06c189a flambda-backend: Make stack allocation the default (ocaml#1013) 98debd5 flambda-backend: Initial support for value slots not of value kind (ocaml#946) deb1714 flambda-backend: Add is_last flag to closinfo words (ocaml#938) d07fce1 flambda-backend: Disable poll insertion in Configure (ocaml#967) 0f1ce0e flambda-backend: Regenerate ocaml/configure autoconf 2.69 (instead of 2.71) (ocaml#1012) 27132d8 flambda-backend: Fix for spurious typing error related to expanding through functor arguments (ocaml#997) 724fb68 flambda-backend: Use `Compilation_unit.t` instead of `Ident.t` for globals (ocaml#871) 396d5b8 flambda-backend: Add a test for frametable setup in natdynlinked libraries (ocaml#983) b73ab12 flambda-backend: Fix invocation of `caml_shared_startup` in native dynlink (ocaml#980) 7c7d75a flambda-backend: Fix split_default_wrapper which did not trigger anymore with flambda2 (ocaml#970) 8fb75bd flambda-backend: Port ocaml#11727 and ocaml#11732 (ocaml#965) fdb7987 flambda-backend: Fix include functor issue after 4.14 merge. (ocaml#948) 9745cdb flambda-backend: Print -dprofile/-dtimings output to stdout like 4.12 (ocaml#943) 5f51f21 flambda-backend: Merge pull request ocaml#932 from mshinwell/4.14-upgrade 841687d flambda-backend: Run make alldepend in ocaml/ (ocaml#936) 72a7658 flambda-backend: Remove reformatting changes only in dynlink/dune (preserving PR889 and adjusting to minimise diff) 6d758cd flambda-backend: Revert whitespace changes in dune files, to match upstream c86bf6e flambda-backend: Remove duplicate tests for polling 971dbeb flambda-backend: Testsuite fixes 32f8356 flambda-backend: Topeval fix for symbols patch befea01 flambda-backend: Compilation fixes / rectify merge faults a84543f flambda-backend: Merge ocaml-jst 8e65056 flambda-backend: Merge ocaml-jst 4d70045 flambda-backend: Remove filename from system frametable (amd64) (ocaml#920) 5e57b7d flambda-backend: Bugfix for runtime frame_descr logic for C frames (ocaml#918) 6423d5e flambda-backend: Merge pull request ocaml#914 from mshinwell/merge-ocaml-jst-2022-10-24 ead605c flambda-backend: Add a missing Extract_exception (ocaml#916) c8f1481 flambda-backend: Resolve conflicts and add specialise/specialised attributes to Builtin_attributes cf4d0d3 flambda-backend: Merge fixes (ocaml#21) c2f742f flambda-backend: Re-enable some tests for Flambda2 (ocaml#881) 3d38d13 flambda-backend: Long frames in frametable (ocaml#797) 85aec7b flambda-backend: Add loop attribute to Builtin_attributes c0f16e3 flambda-backend: Compilation fixes 90dea23 flambda-backend: Merge flambda-backend/main 5acc6ea flambda-backend: Fixes after merge e501946 flambda-backend: Merge ocaml-jst 115083b flambda-backend: Merge ocaml-jst 9943b2e flambda-backend: Revert "Revert "Transform tail-recursive functions into recursive continuations (ocaml#893)"" (ocaml#909) ce339f1 flambda-backend: Fix alloc modes and call kinds for overapplications (ocaml#902) e6a317c flambda-backend: Revert "Transform tail-recursive functions into recursive continuations (ocaml#893)" 853c488 flambda-backend: Transform tail-recursive functions into recursive continuations (ocaml#893) 5a977e4 flambda-backend: Fix missing End_region primitives on switch arms (ocaml#898) 7fa7f9d flambda-backend: Add missing dependencies to Dune files (ocaml#889) 3cd36f0 flambda-backend: Have Lambda `Pgetglobal` and `Psetglobal` take `Compilation_unit.t` (ocaml#896) 7565915 flambda-backend: [@poll error] attribute (ocaml#745) 9eb9448 flambda-backend: Backport the main safepoints PRs (ocaml#740) 689bdda flambda-backend: Add strict mode for ocamldep (ocaml#892) git-subtree-dir: ocaml git-subtree-split: b11eea1
6 participants
(Was PR#7426.)
Translclassuses array operations on object blocks because there is no primitive to set the field of a block using a computed index. At present these "arrays" are said to have kindPaddrarraywhich causes an error in the Flambda simplifier in the event that a float is assigned to them. I think this error is correct. I believe the reason is as follows:Double_array_tag. (Since we're dealing with arrays not blocks, all elements have the same type.)Paddrarrayoperations with float operands.One solution might be to use
Pgenarrayinstead, but this might penalise code using objects; also, it's still a lie to the optimizers (which we should particularly avoid now that Flambda exists). Instead I've done what @garrigue suggested and introduced two new primitives for getting and setting fields of blocks where the offset is not necessarily known at compile time. I have not yet added code to the Flambda simplifier to handle the newPfield_computedlike the existingPfieldwhen the offset turns out to be known; this can be done later.