| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| |
|
|
| |
and method calls set the aggregate; fix test/example hits; docs
|
| |
|
|
|
|
| |
error (spec 4.1.5, b.1.9)
track linited per local alongside the ownership state; snapshots, exit records and joins carry it, and a join keeps a local set only when every reaching path set it. x = e, a declaration initializer, &x (out-parameter), a full array initializer, or using an aggregate as an address set a local; field and element writes do not set the aggregate. while/for bodies that may run zero times do not count; do/while does. compiler, runtime, std, actor and lsp sources initialize the locals the rule flagged; zc0 accepts = {} brace lists (zeroed). tests: tests/errors/uninit_*.zc, tests/conformance/4-statements/uninit_*.zc
|
| |
|
|
| |
longer runs on past the fault (js vm trace check)
|
| |
|
|
| |
reported where it starts (b.2.11); an unreachable droppable cycle drops whole in one collection (6.15.3)
|
| |
|
|
| |
constant expressions (b.1.7, b.2.16); zc0 matches
|
| |
|
|
|
|
|
|
|
| |
rt.MW is the end of the fixed map (PW) since the growing heap landed; the
heap starts right there, so ld16(addr_b(rt.MW)-1) read the heap's first
byte: 0 on the default map by luck, 0x40 on the small map (16384).
Probe the real byte-op bound instead: core answers syscall(9,10) with the
host reservation MW; micro ports, whose bound is their memory size, fall
back to syscall(9,0).
|
| |\
| |
| |
| | |
lists, .u inference), Io.print any args, block-end borrow messages
|
| | |
| |
| |
| | |
a global assigned one (fixes lsp self-check); xref baseline for the new writers
|
| | |
| |
| |
| | |
parameter or a value derived from one; fixtures
|
| | |
| |
| |
| | |
per type); docs, conformance
|
| | |
| |
| |
| | |
reference, fixing a compile hang on a<->b and a leak; ending an owner kills its owned children too (fixes use after reassigning a container); escape fixtures
|
| | |
| |
| |
| | |
then checked as .u)
|
| | |
| |
| |
| | |
(loop-built lists); block-end frees name the owner (why 9) in messages and LSP related info
|
| | |
| |
| |
| | |
lsp keyword completion yields to a same-named symbol; compiler message avoids the word actor
|
| | |
| |
| |
| | |
marked in hover and completion; tmLanguage keywords; hold_lsp_test.js
|
| | |
| |
| |
| | |
values kept by the host and copied in/out at the outermost call, reentrant per-block locks in declaration order (core, micro)
|
| | |
| |
| |
| | |
only the unread bytes; worker stacks 4 MB
|
| | |
| |
| |
| | |
inherits only unread input
|
| | |
| |
| |
| | |
static tls (13 MB of every worker's stack mapping)
|
| | |
| |
| |
| | |
(memfd/shm), o3 entry tables handed over sparse
|
| |/
|
|
| |
wasmtime or node, proven by the fixpoint; seed/MANIFEST hashes; make seed; legacy c bootstrap kept as make genome-c
|
| |
|
|
| |
map's values, and point at a stale src/runtime/rt.zc or map
|
| |
|
|
| |
owner ended and was declared; vscode client passes relatedInformation through
|
| |
|
|
| |
end; borrows clause for prototypes, definitions, externs and function types, checked against the body
|
| |
|
|
| |
owner ended (freed, moved, .d parameter, push, global, reassignment), with fixtures
|
| | |
|
| |
|
|
| |
std.File. and std.Io. their members)
|
| |
|
|
| |
getline) with test, learn example and docs
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
plain blocks
a call to rt.new_k / rt.recycle_leaf with a constant class, or to rt.own,
does its class-list pop or push (or owner store) inline and goes on after
the call; an empty list, memdebug, a held range, a block outside memory, a
translated word or a type/class mismatch takes the ordinary call. in a
region this replaces the o3_invoke round trip. o3 frees of leaf objects now
use recycle_leaf too (new_k stamps the type inside o3 since b86db60).
ZYNOFASTALLOC=1 turns it off; ZYPERFMAP=1 writes a perf map of translated code.
tests/programs/o3_freelist.zc: reuse, zeroing, an empty list, untyped
neighbours, memdebug; jit vs ZYNOFASTALLOC vs ZYO3INTERP.
|
| |
|
|
|
|
|
|
|
|
| |
perf (ZYPERFMAP) on plain trees new/free showed recycle, new_k and own
taking about 70% of the time, mostly in calls to isobj, paylen, bdrop,
bclass and fpush and in the truth values plain code builds for && and ||.
recycle writes isobj, paylen and the class-list push out, calls scan_ptr only
for 8 or more words left, and own and new_k test with nested ifs.
frame_control grows to 22 words: a tag, &dbg and the entries of new_k,
recycle_leaf and own for the x86-64 jit's inline fast paths (next commit).
|
| |
|
|
|
|
|
|
| |
arm64, llvm and the region-interpreter rdrop path now watch the two
prologue words the way the x86-64 jit already did, so icon.replace_function's
lit+jmp patch retires a compiled callee (and clears fn_nat) instead of leaving
a stale stub. three new icon regressions: repeated o3 replace, failed
publication keeps the old entry, and a direct call after warming the region.
|
| |
|
|
|
|
|
| |
Restoring undirtied lafacts when the body might free was wrong: push can
grow an array without dirtying its name, so a[99] after push warned against
a stale length 0. Keep the old lafact_keep rule; reorder explain_prover so
the constant/affine cases run before the touch() loops.
|
| |
|
|
|
|
|
|
| |
Rebased the prover WIP onto main. lafact_keep restores undirtied allocation
facts even when the loop body might free (so a later touch() no longer drops
every length fact). proven_p uses ptr_still so an outer call keeps the check.
entry_fact still supplies the size for the "might free" explain reason.
bounds_prover.zc and explain_prover.zc land with test.sh coverage.
|
| |
|
|
| |
affine i*n+k); unbuilt and untested
|
| |
|
|
|
|
|
| |
Constant-size typed new inside o3 now calls rt.new_k instead of always
taking new_t/new_raw. new_k and recycle_leaf skip the o3 argument-fatten
path the way free/recycle already do. trees o3 new/free on the x86-64
jit dropped from about 4.1 ms to about 2.7 ms per iter (matches plain).
|
| |
|
|
|
|
|
|
|
| |
Port arm64's rcall/rfinish to the x86-64 region jit: call the callee's entry
stub through fn_nat with a built frame, return via linkstk, and fall back to
o3_invoke when the stub is missing or the frame/depth check fails. Clear
fn_nat on rforget and cover the function prologue so icon.replace_function
drops the region (o3_caller_replace). vm_until saves jit_depth around nested
bytecode, matching arm64.
|
| |
|
|
| |
replies and closes the endpoint; the handler reaches the struct through endpoint.state so the struct isn't shared; spawn hint skips mailbox structs
|
| |
|
|
| |
owned std.File (or null) with write and read, and its drop closes the descriptor; Host.writefile is built on it; file_drop.zc, learn and readme
|
| |
|
|
| |
drop capturing its own value), one per collection, each once; before, each waited for the other forever; spec 6.15.3, drop_gc_cycle.zc, drop_gc_self_capture.zc, design doc lists what landed
|
| |
|
|
| |
enters and leaves the region and fattens its result, like a direct call; spec 4.10.3, o3_receiver_method_enter.zc
|
| |
|
|
| |
block they leave (the loop records the o3 depth it began at), so the depth returns to 0 and pending collections and drops run; spec 4.10.3, o3_break_continue_leave.zc
|
| |
|
|
| |
4.11, readme, memdebug_report_drop.zc
|
| |
|
|
| |
it too, on x.spawn() of a mailbox struct for each live owned local with a drop; spec 6.15.4, lsp hint check
|
| |
|
|
| |
a drop, or on an array field of a value with a drop, is error: a: can't move out of a value with drop (fields were already borrowed projections); spec 6.15.5, four conformance programs
|
| |
|
|
| |
value first, running its drop (right side first, then the free, then the store), with hint: x: reassigning frees the old value; a new value that is part of the old one is an error; u, d and captured owners keep the leak error; the new value's size, leaf type and drop are recorded like a declaration's; spec 6.8.5, 6.15.2
|
| |
|
|
| |
left); the map constants in rt.zc and the README table move by 0x8000; gc.zc's reclaim bound is 900 (937 with the heap higher, 952 before) and the run-only section rewrites whatever its number
|
| |
|
|
| |
of the source before it end there: a program joined to the unbounded runtime in a self-hosted run-only image is bounds-checked again (since d55b5fd its index errors went unreported); runtime_image_test expects the trap's exit status 2; spec 4.15.1, 4.x unchecked, 5.15.3
|
| |
|
|
| |
compiler reads mentions actors (test.sh check)
|
| |
|
|
| |
the clearing out of line
|
zinc
