ci: add pkg.pr.new preview workflow#2
Merged
Merged
Conversation
…tests to subdirectories
feat(secure-exec): add Python runtime with Pyodide driver
commit: |
NathanFlurry
added a commit
that referenced
this pull request
Jul 1, 2026
SECURE_EXEC_XTRACE gates an [xtrace] hex dump of every unix-socket payload (tagged dir + remote path) at the sidecar read/write boundary; experiments/wasm-gui/scripts/xdecode.py decodes the X11 stream per direction (request opcodes; reply/event/error type + sequence). Zero cost when unset. Used on the 2-party lxpanel<->Xvfb render stall: data delivery is BYTE-COMPLETE both ways (client sends == server receives, and vice versa — nothing lost/unflushed). The server replies up to seq 301 then stops; lxpanels final requests in the stuck first-GTK-draw dispatch are an XRENDER drawing batch plus GetInputFocus (op 0x2b, expects a reply). Both ends then sit idle in net_poll. So the blocker is NOT data loss and NOT a GTK draw spin — it is a cross-VM poll-wakeup edge case where, mid first draw, the server stops processing/replying to a delivered request and both guests park in poll. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
NathanFlurry
added a commit
that referenced
this pull request
Jul 1, 2026
… render diagnosis Tool #2 (X11 wire tap + xdecode.py) and tool #3 (wasm-frame symbolizer for livelocked guests) are committed and cracked the render diagnosis: both guests spin in net_poll, data delivery is complete, the X server sits idle in WaitForSomething without replying to a delivered first-draw RENDER request. The earlier prediction that C names survive the build was wrong (fpcast-emu + -Oz strip them) — which is exactly why tool #3 was needed. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
NathanFlurry
added a commit
that referenced
this pull request
Jul 1, 2026
…er root to an unknown ext opcode - scripts/xreassemble.py: reassembles the FULL client/server X11 byte streams (handling the setup request/reply + BigRequests) and matches requests to replies by sequence, so the exact unanswered request is visible. SECURE_EXEC_XTRACE=<n> now sets the per-payload capture cap (default 256; large for full reassembly). - ActiveUnixSocket::poll: try_recv first, so a guest non-blocking drain (net.poll(fd,0)) never misses a delivered request via the recv_timeout(ZERO)-returns-Timeout-with-data-pending std gotcha. Strictly more correct (did not by itself unblock the render). DIAGNOSIS (tools #2/#3): lxpanel creates the panel toplevel + sets its DOCK/_NET_WM properties but the toplevel is NEVER mapped — GTK queued its MapWindow in the GLib loop, which is stuck in its first dispatch. lxpanel waits for a GetInputFocus reply (op 0x2b) that never arrives; the server received all bytes (delivery is byte-symmetric) yet sits idle in net_poll. Root lead: the first-draw batch sends requests with major opcode 0x89=137, but Xvfb's registered extensions use opcodes 128/130/132/134/138 — NO extension owns 137. An unknown-opcode request mis-handled by the server desyncs the request stream (server waits for bytes to finish a phantom request; client waits for its reply) = the deadlock. Next: confirm which client lib emits op 137 / fix Xvfb's unknown-opcode handling (platform layer). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
NathanFlurry
added a commit
that referenced
this pull request
Jul 1, 2026
…fpcast arity tuning + marshallers A deep feasibility research agent (read the toolchain + V8 runtime) proved the user-named "typed-function-references" does NOT fix the ~13.4s GObject cascade: call_ref type-checks exactly like call_indirect, so a GObject UB cast mismatch still traps -- typed-func-refs/GC cannot run the mismatched calls faster. The real cost is binaryen fpcast-emu uniform max-arity padding + per-call trampoline, not the call opcode. V8-here runs GC-wasm fine (V8 130 default-on), so not blocked, just the wrong layer. Also corrected: no max-func-params@128 in-repo; link-xapp.sh:24 uses BARE --fpcast-emu (binaryen default arity). REAL Root-1 fix (cheapest-first): (1) QUICK WIN hours/one-line -- --pass-arg=max-func-params@<measured-true-max> (padding scales with arity; GObject closures 1-4 ptr args); (2) days -- force per-signature C marshallers to kill g_cclosure_marshal_generic then drop/scope --fpcast-emu; (3) weeks -- per-arity trampoline tables. Spec Root-1 item + PERF-FINDINGS.md + status-log T69 updated. Also captured: #2 in-place pwrite caveat is real (RootFileSystem uses the default VirtualFileSystem::pwrite read_file-clone via self.overlay; fix needs dispatch-chain forwarding). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
NathanFlurry
added a commit
that referenced
this pull request
Jul 1, 2026
…pivot to multi-app levers Measured @64/@28/@26 (re-fpcast from saved prefpcast, no relink): css-bench CSS parse 894->681ms (24%, dense GObject), but gtk-hello REAL-APP construction only 3734->3635ms (2.6% -- a simple app is one-time-font/theme bound, not arity bound). Both render-correct, smaller binary. The arity quick win was already banked @128->@64 (3.7x); @64->@28 is marginal + overrides a deliberate 2.5x safety margin (measured max 25) -> keeping the @64 default. Root-1 arity = done; remaining Root-1 lever = per-signature marshallers (days). MEASUREMENT-DRIVEN PIVOT: single-app construction (~3.7s, tuned) is not the multi-app bottleneck; Root-2 serialization is. Next: #2 in-place pwrite (framebuffer throughput, cuts service-thread load) then Root-2 multiplex. Numbers in M8-STATUS-LOG T70. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
NathanFlurry
added a commit
that referenced
this pull request
Jul 1, 2026
…ration research + ROOT2 trace forward XUBUNTU-SPEC.md: added the living LINUX-SUPPORT-LEDGER.md constraint (catalog every native/platform-layer fix that makes unmodified Linux software run; append on each fix; treat as a required progress artifact). GOBJECT-FPCAST-INTEGRATION.md (research subagent): the GObject wasm fpcast cost is binaryen --fpcast-emu uniform arity padding (taxes the correctly-typed vfunc majority too). Ranked fix: #1 NOW (hours, low-risk) drop max-func-params @64 -> ~@32 toward measured true-max @25 (too-low fails at BUILD time, safe); #2 (weeks, the structural fix) a vendored binaryen selective-fpcast pass using wasm-gc ref.test (V8 130 has it) to keep correctly-typed call_indirect native via call_ref and adapt only mismatched sites (= emscripten#23952 / Pyodide ref.test). Rejected: Pyodide 0.19 JS trampolines (needs finite patchable sites; breaks sync-import V8 model), clang thunk-gen (custom clang), typed marshallers (signal emission not construction). host/src/main.rs: forward SECURE_EXEC_ROOT2_TRACE to the sidecar (built earlier with native CC). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Summary
Validation