diff --git a/issues/a-120000-ms-boot-deadline-fired-132859-ms-late-on-the-t14.md b/issues/a-120000-ms-boot-deadline-fired-132859-ms-late-on-the-t14.md index f87871a4729..a03a151e2cd 100644 --- a/issues/a-120000-ms-boot-deadline-fired-132859-ms-late-on-the-t14.md +++ b/issues/a-120000-ms-boot-deadline-fired-132859-ms-late-on-the-t14.md @@ -96,7 +96,7 @@ What the tail names as the machine's state when the records stop: Between those, cpu5 and then cpu0 repeat `usb-storage: 00:14.0 slot 5 transport broke on SCSI 0x28` and `SCSI 0x28 broke 3 times running; the transport is not coming back on its own`: the boot stick's READ(10) transport -broke at 3.47 s, the root filesystem stopped answering, `logd` and `netd` died +broke at 3.47 s, the root filesystem stopped answering, `logkeeper` and `netstack` died on page faults over file bytes the device would not read, the test runner could not be loaded, and the disk went offline at 5.556 s. A boot with no runner asks for no reboot; the deadline was the only bound left, and it fired diff --git a/issues/a-boot-with-no-kernel-log-is-not-no-boot-complete.md b/issues/a-boot-with-no-kernel-log-is-not-no-boot-complete.md index 803cab82598..95d321b7f5a 100644 --- a/issues/a-boot-with-no-kernel-log-is-not-no-boot-complete.md +++ b/issues/a-boot-with-no-kernel-log-is-not-no-boot-complete.md @@ -9,14 +9,14 @@ opened: 2026-09-08 Run 22's hung mkdir image left the stick holding `attempts` and `loader.log` and **no kernel log file at all** — where every earlier hung boot (run 19's three) left a 325-line file ending at a spawn. `loader.log` shows an ordinary handoff, -so that kernel started; `logd` either never created its file or its first write +so that kernel started; `logkeeper` either never created its file or its first write never completed. The stick after the power cut is `t14-run22/stick-after-cycle.log`, outside the tree. `tests/metal/src/lib.rs` reads that back as the absence of `Boot: complete`, which is the same verdict it gives a boot that ran and stopped late. They are not the same machine: a boot with no kernel log wedged before its -first durable record, which places the failure before `logd`'s first `SYS_FSYNC` +first durable record, which places the failure before `logkeeper`'s first `SYS_FSYNC` returned rather than anywhere in the job list — the single most useful thing the readback could say about the run-22 class of hang. diff --git a/issues/a-claims-own-refusals-are-read-by-nothing.md b/issues/a-claims-own-refusals-are-read-by-nothing.md index edcdb3d8a5f..c7e968b4e23 100644 --- a/issues/a-claims-own-refusals-are-read-by-nothing.md +++ b/issues/a-claims-own-refusals-are-read-by-nothing.md @@ -29,7 +29,7 @@ A one-field mutation of any of them — an index bound compared `<=` rather than green. Owned by whoever next adds a boot config whose own test binary holds a claimable -function: netd holds this machine's only claim, and a test that makes it fail +function: netstack holds this machine's only claim, and a test that makes it fail this way costs the machine its NIC for that boot. Exit condition: a guest arm in which a test binary's claim is refused each of these ways, each one red against a kernel whose corresponding check answers `Ok`. diff --git a/issues/a-client-cannot-tell-soundserver-it-paused.md b/issues/a-client-cannot-tell-soundserver-it-paused.md index dda1cb7d505..78ab452b185 100644 --- a/issues/a-client-cannot-tell-soundserver-it-paused.md +++ b/issues/a-client-cannot-tell-soundserver-it-paused.md @@ -4,20 +4,20 @@ kind: track opened: 2026-08-01 --- -# A client cannot tell soundd it paused, so soundd cannot tell paused from wedged +# A client cannot tell soundserver it paused, so soundserver cannot tell paused from wedged -soundd infers a dead client from its signal event alone: a ring answering +soundserver infers a dead client from its signal event alone: a ring answering `Hungup` is a gone process, and any other ring coalesces into a signal the client has not taken yet. That detects exactly one failure. A client that stops making progress while holding its handle — a wedged thread, a deadlock, a callback that never returns — takes no signal forever and is indistinguishable from one that deliberately paused. The ambiguity is not a -soundd bug: pause is defined as the client's stream thread simply not reading, +soundserver bug: pause is defined as the client's stream thread simply not reading, with no coordination, and the cpal backend confirms it — `pause()` stores a -state word and parks on a futex, and soundd is never told. +state word and parks on a futex, and soundserver is never told. The cost is that `is_streaming` latches on the first delivered period and never -clears, so one paused client holds soundd out of idle for the life of the +clears, so one paused client holds soundserver out of idle for the life of the process: a wake per period forever, the DMA engine running, the codec voice open, and the client's shm, signal event and slot ring pinned. @@ -33,18 +33,18 @@ deferral window the mix loop tolerates — a number to measure, not to invent. **Blocked on the cpal fork**, which needs a clone beside the monorepo and `.cargo/config.toml` path overrides — a file every agent in a working tree -shares. Landing the soundd and SDK halves alone is *worse* than not landing +shares. Landing the soundserver and SDK halves alone is *worse* than not landing them: every cpal client that pauses would declare `STREAMING`, stop producing, and be killed by the new deadline. That is an audible regression. **One edge unblocks three things.** Suspending on no progress is not landable on -its own either, and the reason is worth keeping: soundd's wait has exactly two +its own either, and the reason is worth keeping: soundserver's wait has exactly two sources plus a timeout, and its command channel carries only add/remove/set-volume -— **there is no resume message**. A suspended soundd with the device stopped and +— **there is no resume message**. A suspended soundserver with the device stopped and `timeout = u64::MAX` waits for completions that never come while the resumed client blocks forever on a signal byte nobody writes: a battery defect traded for permanent silence. Killing wedged clients, suspending on no progress and -resuming from suspend all need the same client→soundd notification, sent from +resuming from suspend all need the same client→soundserver notification, sent from cpal's `play()`. The one variant that needs no client change is filed apart, at diff --git a/issues/a-device-function-is-driven-from-its-arrival-to-its-end.md b/issues/a-device-function-is-driven-from-its-arrival-to-its-end.md index cc1f90f2d4b..d958013cc36 100644 --- a/issues/a-device-function-is-driven-from-its-arrival-to-its-end.md +++ b/issues/a-device-function-is-driven-from-its-arrival-to-its-end.md @@ -16,11 +16,9 @@ started. Nothing reaches userland when a device arrives or leaves. Rulings that bound the stages: the ownership and event model (owner, 2026-10-13); PCIe hot-plug as stage 2, native only, with physical arrival -witnessed through `issues/toyos-supports-thunderbolt-and-usb4.md`, and the -claim-slot limit removed by stage 1 if cheap, otherwise by its own pull -request before stage 1's guest smoke -(`issues/a-cargo-run-boot-claims-every-pcidev-slot.md`) (orchestrator, on -#910); and one shared block service to which disk drivers offer their disks +witnessed through `issues/toyos-supports-thunderbolt-and-usb4.md` +(orchestrator, on #910); and one shared block service to which disk drivers +offer their disks (orchestrator, 2026-10-20, replacing #901's `block@` ports). The supervisor starts the `drives` processes before the storage rows and tells the block service how many disk drivers it started; an open of a partition no diff --git a/issues/a-hung-boots-log-partition-is-wiped-by-the-next-runs-flash.md b/issues/a-hung-boots-log-partition-is-wiped-by-the-next-runs-flash.md index 0a161bc80a3..ede0b60dff2 100644 --- a/issues/a-hung-boots-log-partition-is-wiped-by-the-next-runs-flash.md +++ b/issues/a-hung-boots-log-partition-is-wiped-by-the-next-runs-flash.md @@ -46,7 +46,7 @@ and run 56's transcript opens, at 21:29:39Z, with — a copy of `/dev/sda3` taken from outside the loop at 21:15Z, 35651584 bytes, which is the log partition's 69632 sectors (`log p3 at 151552+69632`, the loop's own line about the image) whole. Run 56's `flash` then overwrote it. -The copy held two files, `loader.log` and `attempts`, and no file `logd` +The copy held two files, `loader.log` and `attempts`, and no file `logkeeper` wrote. Its `loader.log` is the loader's first pass alone, 105 lines ending in `Loader log: the kernel handoff begins, so this file ends here`, and its line 20 reads @@ -56,7 +56,7 @@ wrote. Its `loader.log` is the loader's first pass alone, 105 lines ending in — the boot whose `WEDGED` record run 56's loader pass then printed (`issues/a-120000-ms-boot-deadline-fired-132859-ms-late-on-the-t14.md`). So for the one hung boot whose partition was captured, the flash would have -destroyed nothing `logd` wrote, because `logd` wrote nothing: that boot's ring +destroyed nothing `logkeeper` wrote, because `logkeeper` (then `logd`) wrote nothing: that boot's ring tail carries `exit: logd pid=4 code=-1 cpu=2143ms` at 3.471 s. What the other four partitions held is unknown, and the sequence above is why. diff --git a/issues/a-machine-without-an-iommu-refuses-every-claim.md b/issues/a-machine-without-an-iommu-refuses-every-claim.md index 1131ce2c002..974386dc830 100644 --- a/issues/a-machine-without-an-iommu-refuses-every-claim.md +++ b/issues/a-machine-without-an-iommu-refuses-every-claim.md @@ -64,8 +64,8 @@ Exit, in both arms of `iommu_virtio_platform` (`tests/common/iommu.rs`): `faults::refused_claim` reads, so a refusal for any other reason (no MSI-X, no BAR window) fails the arm. Deleting the identity check so that every claim goes through reds it. The non-image claimant is a - `tests/guest/rust-tests` binary, to be added, that init hands the NIC - function under `netcase`; netd is the in-image claimant. + `tests/guest/rust-tests` binary, to be added, that the supervisor hands the NIC + function under `netcase`; netstack is the in-image claimant. - The kernel says "no isolation on this machine" at boot on `Profile::HeadlessNoIommu` and never on `Profile::Headless`. Printing it on every profile reds the `Headless` arm; printing it on none reds the other. diff --git a/issues/a-metal-session-runs-a-pre-flash-gate-first.md b/issues/a-metal-session-runs-a-pre-flash-gate-first.md index 68132671054..cc0c87ebcf0 100644 --- a/issues/a-metal-session-runs-a-pre-flash-gate-first.md +++ b/issues/a-metal-session-runs-a-pre-flash-gate-first.md @@ -140,7 +140,7 @@ that hold nothing against 7 ms finding everything that is there The record this came from also warned about a NIC retry that looks like boot cost; that warning is spent. `sdk/toyos/src/net.rs:272` says the hundred retries at -ten milliseconds are gone — a `netd` connector is live from a process's first +ten milliseconds are gone — a `netstack` connector is live from a process's first instruction — so nothing after `Boot: complete` inflates a reading any more. The panic path's own metal price belongs to the same session: the T14 measured diff --git a/issues/a-package-is-a-directory-under-apps-and-the-installer-is-a-program.md b/issues/a-package-is-a-directory-under-apps-and-the-installer-is-a-program.md index 8ab290b28b8..b311a03b90e 100644 --- a/issues/a-package-is-a-directory-under-apps-and-the-installer-is-a-program.md +++ b/issues/a-package-is-a-directory-under-apps-and-the-installer-is-a-program.md @@ -56,7 +56,7 @@ the binary in a ToyOS guest is this track's harness's job, not gbae's. `pkg install ` stays for local and offline installs, checked against the `SHA256SUMS` beside it. - **Fetching is HTTPS.** GitHub serves releases only over TLS, so `pkg` - carries a TLS client; the network stack under it is netd's. A crate that + carries a TLS client; the network stack under it is netstack's. A crate that does TLS is not our job to write and is widely used; it takes the fork treatment every third-party crate takes. The redirect GitHub answers with is followed once. diff --git a/issues/a-stream-write-spins-on-a-lock-its-own-panic-can-hold.md b/issues/a-stream-write-spins-on-a-lock-its-own-panic-can-hold.md index 7762c571cda..39c23d84cb7 100644 --- a/issues/a-stream-write-spins-on-a-lock-its-own-panic-can-hold.md +++ b/issues/a-stream-write-spins-on-a-lock-its-own-panic-can-hold.md @@ -14,7 +14,7 @@ userland spinlock around each stream's partial line, taken by every - **A thread that runs at a real-time priority spins on a preempted holder.** At smp 1 the holder cannot run until the spinner yields, and the spinner - never does. soundd's mix thread is clear of it (it writes a lane of its own, + never does. soundserver's mix thread is clear of it (it writes a lane of its own, `claim_lane`); any other real-time thread that prints is not. - **A panic raised inside `Held::with` re-enters it.** The panic's message is a write to stderr, which takes the stream's `Held` again on the same thread diff --git a/issues/a-t14-boot-said-rebooting-and-the-host-never-came-back.md b/issues/a-t14-boot-said-rebooting-and-the-host-never-came-back.md index 0ac97aa1f22..cbd45c1cea3 100644 --- a/issues/a-t14-boot-said-rebooting-and-the-host-never-came-back.md +++ b/issues/a-t14-boot-said-rebooting-and-the-host-never-came-back.md @@ -21,7 +21,7 @@ So the kernel ran the whole quiesce and made its last word durable; what follows `Rebooting.` — `xhci::seal_shut`, the reset register — writes nothing the stick keeps. Whether the reset was written, whether firmware came back, and whether the host's Linux came up -without its network (the host is reached over the same I219 this boot's netd +without its network (the host is reached over the same I219 this boot's netstack left faulted) are not separable from anything on the stick. A 454-line copy of the same file that ends at 35.585 s was taken for the whole log; it is a truncated read, not the file. diff --git a/issues/a-t14-boot-wedges-after-a-jobs-exit-and-nothing-said-why.md b/issues/a-t14-boot-wedges-after-a-jobs-exit-and-nothing-said-why.md index 5b83c030f38..edc7f5885fd 100644 --- a/issues/a-t14-boot-wedges-after-a-jobs-exit-and-nothing-said-why.md +++ b/issues/a-t14-boot-wedges-after-a-jobs-exit-and-nothing-said-why.md @@ -33,9 +33,9 @@ No `PANIC:`, no `SEGFAULT`, no fault record, no `LOCK CONTENTION`, no flushed`. Every one of those tripwires would have fired inside 40 s and the machine was down for six minutes. The spawning CPU is `cpu6` in all three. -## The evidence is bounded by logd, and that is the whole difficulty +## The evidence is bounded by logkeeper, and that is the whole difficulty -**The stick's file is what `logd` flushed, not what the kernel emitted.** So a hang anywhere from that spawn through the reset register write +**The stick's file is what `logkeeper` flushed, not what the kernel emitted.** So a hang anywhere from that spawn through the reset register write leaves exactly this tail, with everything after it still in the record ring and never on the stick. The window is **[the reboot binary's spawn → the reset register write]**, and the candidates in it are all consistent with what the @@ -50,7 +50,7 @@ file shows: program, through the same block path. A wedge in that path stays a candidate; the 2 MiB read that matched the healthy 46–100 ms gap does not. - `quiesce`'s `flush_disks` and `hand_back`, which landed at `88841f56` a few - minutes before the first hang; `hand_back` takes `XHCI` while `logd`'s own + minutes before the first hang; `hand_back` takes `XHCI` while `logkeeper`'s own write to the same stick may still be in flight. Nothing on the stick can separate them, and the black-box page was lost to the diff --git a/issues/a-wait-was-registered-while-its-task-still-read-waiting.md b/issues/a-wait-was-registered-while-its-task-still-read-waiting.md index 7227a228859..f3c35d3d233 100644 --- a/issues/a-wait-was-registered-while-its-task-still-read-waiting.md +++ b/issues/a-wait-was-registered-while-its-task-still-read-waiting.md @@ -15,7 +15,7 @@ Which wait left it set was never established. ## The sightings, all on paths since removed -- Once in logd's `io_uring_enter` on a KVM shard (nightly dispatch +- Once in logkeeper's `io_uring_enter` on a KVM shard (nightly dispatch `31900050723`, job `95049280131`, 2026-08-15), 4.7 s after the boot stick was pulled and the kernel's usb-storage gave the disk up. The same tree was green on the same name minutes earlier; one guest per machine, so host diff --git a/issues/an-unlinked-open-file-answers-stale.md b/issues/an-unlinked-open-file-answers-stale.md index aaa2b8da16b..3579a8089c2 100644 --- a/issues/an-unlinked-open-file-answers-stale.md +++ b/issues/an-unlinked-open-file-answers-stale.md @@ -6,14 +6,6 @@ opened: 2026-10-22 # An unlinked open file answers Stale -A file the file server holds open answers `Stale` from the moment its name is -unlinked or renamed over (`system/fileserver/src/volume.rs`, the module -header), and a test asserts it. POSIX's file lives on until its last close or -unmap, and SQLite relies on that: it opens every temporary file and unlinks it -at once, and its last connection unlinks the `-shm` file while it is still -mapped. So any SQLite statement that spills to a temporary file fails today, -cargo's bundled one included. - **Ruled** (owner, 2026-10-21): removing the last directory name prevents fresh opens through that name, and existing handles and mappings keep the same file until their final release. A @@ -21,10 +13,21 @@ new file made at the old path is a different file and never retargets them. The file server owns the lifetime and the storage's reclaim; the kernel keeps no filesystem identity or naming. -**Exit**: a file unlinked while open, and one renamed over, is read and written -through its open descriptor until the last goes, and its blocks are freed -then; a crash leaves the orphans for the next mount to reclaim; each asserted -by the file server's host volume tests, replacing today's `Stale` test -(`system/fileserver/src/data.rs`). A guest test reads a mapping of an -unlinked file once file mappings exist -(`issues/process-memory-in-4-kib-pages-still-copies-each-programs-text.md`). +Built for DATA and MEMORY: a file unlinked or renamed over while open is read +and written through its descriptor until the last close, and its blocks are +freed then (`unlink_and_replacement_keep_live_inode_identity` in +`system/fileserver/src/data.rs`, `unlink_and_replacement_keep_open_inode_identity` +in `memory.rs`, `persistent_unlink_preserves_dirty_bytes_until_the_last_holder` +in `pages/tests.rs`); a crash leaves the orphans for the next mount to release +(`namespace_operations_survive_a_crash_at_every_request` in +`lib/bcachefs/src/engine/tests/crash.rs`). + +What is left: + +- A FAT volume still answers `Stale`: its `unlink` marks the open file gone + (`system/fileserver/src/fat.rs`, `orphan`). +- No guest test reads a mapping of an unlinked file, which waits for file + mappings + (`issues/process-memory-in-4-kib-pages-still-copies-each-programs-text.md`). + +**Exit**: the guest test of a mapping, and FAT keeping an unlinked open file. diff --git a/issues/both-boots-of-the-manageability-read-image-found-the-t14-powered-off.md b/issues/both-boots-of-the-manageability-read-image-found-the-t14-powered-off.md index e98716bafa4..88b37ebe523 100644 --- a/issues/both-boots-of-the-manageability-read-image-found-the-t14-powered-off.md +++ b/issues/both-boots-of-the-manageability-read-image-found-the-t14-powered-off.md @@ -6,13 +6,13 @@ opened: 2026-09-21 # Both boots of the manageability-read image found the T14 powered off -A read-only netd job — one read of the I219's `EXTCNF_CTRL` (§10.2.2.15, +A read-only netstack job — one read of the I219's `EXTCNF_CTRL` (§10.2.2.15, `0x00F00`), no write, no PHY register touched, then `exit` with the reading encoded in the code — was flashed to the T14 twice, as metal runs 58 and 62. Both times the owner came back to a machine that was **off**: not frozen, not wedged at a panic, powered down. Neither boot left a black box. -The persisted kernel log is the same on both runs: it ends at the first `logd` +The persisted kernel log is the same on both runs: it ends at the first `logkeeper` batch, 192 lines, at around 0.6 s. So the stick carries nothing from the moment the machine went away, and nothing recorded says what took the power. diff --git a/issues/client-ring-depth-is-the-devices-pipeline-depth.md b/issues/client-ring-depth-is-the-devices-pipeline-depth.md index 4d746b85f67..eb8807032af 100644 --- a/issues/client-ring-depth-is-the-devices-pipeline-depth.md +++ b/issues/client-ring-depth-is-the-devices-pipeline-depth.md @@ -6,7 +6,7 @@ opened: 2026-08-01 # Every client's ring geometry is the device's pipeline depth -soundd sets `slot_count = num_buffers`, the kernel driver's `TX_INFLIGHT_MAX`. +soundserver sets `slot_count = num_buffers`, the kernel driver's `TX_INFLIGHT_MAX`. So a device with a different pipeline depth silently changes the shm size, the number of slots a client fills per wake, and the latency floor of *every* client, through a constant no client sees and none can influence. @@ -18,7 +18,7 @@ was assumed. **What to build.** `slot_count` becomes a client property negotiated in `MSG_STREAM_OPEN`/`MSG_STREAM_OPENED`, which already carries the format -negotiation, with soundd clamping to `[num_buffers.next_power_of_two(), +negotiation, with soundserver clamping to `[num_buffers.next_power_of_two(), MAX_SLOTS]`. A client that asks for nothing gets today's value, so the default is unchanged. `num_buffers` stays a device property and stops leaking. diff --git a/issues/deferred-release-outlives-its-syscall.md b/issues/deferred-release-outlives-its-syscall.md index 1bf8cb53150..a884bb553ad 100644 --- a/issues/deferred-release-outlives-its-syscall.md +++ b/issues/deferred-release-outlives-its-syscall.md @@ -151,7 +151,7 @@ So the sentence below is no longer the whole of it: this is a *semantic* event riding a release the caller cannot wait for, which is what `kernel/src/object/mod.rs`'s own header says must never happen — *"every userland-visible lifecycle event rides `handle_count`"* — and the same shape -reaches soundd, whose cpal clients spend their lives parked in a signal-pipe +reaches soundserver, whose cpal clients spend their lives parked in a signal-pipe read. ## Why it matters beyond a test @@ -278,9 +278,9 @@ live process's `close` of its last handle, and `SYS_PROCESS_KILL` returning before its victim's teardown (`kill_while_blocked`'s writes), both still released on `ZERO_QUEUE`. -## A fourth witness: a device claim, and init waiting it out +## A fourth witness: a device claim, and the supervisor waiting it out -A service swap (`toyos-swap`, `/system/bin/init`) kills a service, `wait`s for +A service swap (`toyos-swap`, `/system/bin/supervisor`) kills a service, `wait`s for it, and mints the service's device claims again for the binary that replaces it. `Device` is a `deferred` row, so the `wait` returning is not the claim being back. Measured on `wt/toyos-swap`, `tests/swapcase` (virtio-net, two @@ -294,8 +294,8 @@ init: netd: pci:1af4:1041 is already claimed ``` The claim was asked for, and refused as held, before the release record — and -the release ran on the other CPU. The replacement netd started with no NIC and -exited, which init's probation caught and answered by restarting the old +the release ran on the other CPU. The replacement netstack started with no NIC and +exited, which the supervisor's probation caught and answered by restarting the old binary five seconds later, when the claim was long back. The supervisor's `CLAIM_RETURN`, `isa_row.rs`'s `claim_after` and diff --git a/issues/desktop-session-put-26ms-of-silence.md b/issues/desktop-session-put-26ms-of-silence.md index 050bb5a3e0a..30f4cba1b37 100644 --- a/issues/desktop-session-put-26ms-of-silence.md +++ b/issues/desktop-session-put-26ms-of-silence.md @@ -8,7 +8,7 @@ opened: 2026-08-07 The owner ran `cargo run` on 2026-08-07 (guest RTC 13:14:51 UTC, tree at or after `43ce73e`), started doom from the terminal, let its demo loop run to -t=69 s, then ran `tone` 44 times. 391 s of serial, 119 soundd stats windows. +t=69 s, then ran `tone` 44 times. 391 s of serial, 119 soundserver stats windows. Every number below is from that capture. **Harm, by the fast tier's own definition.** Two windows during doom: @@ -41,15 +41,15 @@ whose *worst of 30* is 10090. The two distributions are disjoint. `deferred` column separates them.** `deferred` counts a mix cycle declining to submit because a streaming client's ring was empty and there was still playout margin (the `mid_refill` arm of `mix_thread`, now `system/soundserver/src/mix.rs`) — -soundd's restraint, waiting for a producer: +soundserver's restraint, waiting for a producer: ``` doom phase 173 deferrals across 14 of 31 windows tone phase 1 deferral across 1 of 88 windows ``` -Same soundd, same device, same kernel. So the doom-phase underruns are **doom -failing to fill its ring**, held off by soundd until the floor and then paid in +Same soundserver, same device, same kernel. So the doom-phase underruns are **doom +failing to fill its ring**, held off by soundserver until the floor and then paid in silence — not the audio path being late. The 92175 us window sits beside a compositor window of `frames=23` where the steady state is 65-70, so doom stopped presenting at the same moment it stopped producing; both recovered @@ -78,7 +78,7 @@ windows, one per stream start). One candidate mechanism, offered as a hypothesis and not a measurement: the mix loop arms its wait on `target` — the *next future* grid point when the DLL estimate is past due — but records `armed_on = Some(t_est)`, the stale estimate, so the sample is taken -from an instant soundd deliberately did not ask to be woken at and includes +from an instant soundserver deliberately did not ask to be woken at and includes every whole period it skipped. **Where that code is now, and what has moved under it.** soundd is @@ -89,7 +89,7 @@ taken against `armed_on`. Two things have changed around it and neither settles the question: - The site now argues the choice rather than only asserting it: reading the DLL - estimate at wake would score a window's first wake — armed while soundd was + estimate at wake would score a window's first wake — armed while soundserver was idle and asking for no wake time at all — as a missed deadline, so the sample is deliberately the distance from the prediction, "however large". That is an answer about `t_est` versus the DLL's later estimate. It is not an answer diff --git a/issues/desktop-window-child-freeze.md b/issues/desktop-window-child-freeze.md index 3010538f4b2..6531b524eff 100644 --- a/issues/desktop-window-child-freeze.md +++ b/issues/desktop-window-child-freeze.md @@ -149,7 +149,7 @@ above is the measured one and not the whole of it. The test stopped at its **first** probe: the windowed child asked for a window, was answered `NotEndowed`, and printed `WINDOW-CHILD-REFUSED this program was given no compositor`. The client is a harness -binary, no `[programs]` row can name one, and `/system/bin/init` endows a name the +binary, no `[programs]` row can name one, and `/system/bin/supervisor` endows a name the manifest does not carry with nothing. **The endowment travels with the spawn, and that closed it.** diff --git a/issues/dns-resolver-refuses-a-truncated-answer.md b/issues/dns-resolver-refuses-a-truncated-answer.md index 2f3590f5c6f..0cf569e2ad2 100644 --- a/issues/dns-resolver-refuses-a-truncated-answer.md +++ b/issues/dns-resolver-refuses-a-truncated-answer.md @@ -8,11 +8,11 @@ opened: 2026-09-26 `toyos-dns` asks over UDP only and without EDNS (RFC 6891), so a server answers in at most 512 bytes (RFC 1035 §4.2.1); a reply with TC set ends the -lookup with `Failure::Truncated`, which netd names in its log and answers +lookup with `Failure::Truncated`, which netstack names in its log and answers `ERR_OTHER`. An `A` question for an ordinary name does not reach 512 bytes: of the names measured while the resolver was written, the largest reply was 161 bytes (`www.apple.com`, three aliases and one address). A name with a long enough alias chain or enough addresses does. -To build: TCP fallback (RFC 7766) for a truncated reply, from netd's own TCP +To build: TCP fallback (RFC 7766) for a truncated reply, from netstack's own TCP stack, bounded by the lookup's existing waits. diff --git a/issues/every-program-sees-only-the-files-it-was-given.md b/issues/every-program-sees-only-the-files-it-was-given.md index ab53cffe92b..47a3eb0bbb4 100644 --- a/issues/every-program-sees-only-the-files-it-was-given.md +++ b/issues/every-program-sees-only-the-files-it-was-given.md @@ -108,7 +108,7 @@ other, with the same per-session isolation (owner, 2026-11-05). single key list sits in the login authority's view. **Exit**: a program in one user's session cannot name a file in another's, and appending to the key list is inexpressible from any session. -4. **What is not a file.** netd names a socket by an id any +4. **What is not a file.** netstack names a socket by an id any client can name, and memory and process count have no per-session bound, so one session can starve another. Both close before the stage exits. **Exit**: a hostile session can neither reach another diff --git a/issues/hda-ring-fix-unverified-on-metal.md b/issues/hda-ring-fix-unverified-on-metal.md index b94d6e27e19..1a9b39358e5 100644 --- a/issues/hda-ring-fix-unverified-on-metal.md +++ b/issues/hda-ring-fix-unverified-on-metal.md @@ -9,17 +9,17 @@ opened: 2026-08-08 Filed out of the `repeated completion for free buffer` entry when that closed in QEMU. That QEMU close has no positive control either. The only thing in the tree that -can write the line is soundd's own `assert_eq!` (`userland/soundd/src/mix.rs`), +can write the line is soundserver's own `assert_eq!` (`system/soundserver/src/mix.rs`), so `hda_client_stall`'s `must_not_say` for it is a second spelling of the `must_be_clean()` beside it, and no arm anywhere provokes a double completion to show the assert can see one. Closing that needs an actuator in the completion -loop, which is soundd's audio path. +loop, which is soundserver's audio path. QEMU's `intel-hda` and the T14's controller are different devices, and the fix was measured on the first. What the next metal boot must show: `tone` playing to -completion with soundd alive, `deferred=0`, and `underruns` nonzero only if the +completion with soundserver alive, `deferred=0`, and `underruns` nonzero only if the client stalls. Read `starve_max` on the same stats line -(`system/soundserver/mixer/src/stats.rs`, printed by `userland/soundd/src/mix.rs`) beside +(`system/soundserver/mixer/src/stats.rs`, printed by `system/soundserver/src/mix.rs`) beside `underruns`: it is the longest unbroken run of underrun periods, so near 1 with a large `underruns` is a client missing by a hair and 8–20 is a stall of one whole 21–53 ms ring. diff --git a/issues/logging-records-from-every-producer-and-a-kernel-that-waits-on-nobody.md b/issues/logging-records-from-every-producer-and-a-kernel-that-waits-on-nobody.md index ae7d3e7c8a0..a6e280b3415 100644 --- a/issues/logging-records-from-every-producer-and-a-kernel-that-waits-on-nobody.md +++ b/issues/logging-records-from-every-producer-and-a-kernel-that-waits-on-nobody.md @@ -6,16 +6,16 @@ opened: 2026-09-26 # Logging: records from every producer, and a kernel that waits on nobody -What is built: every program init starts or launches writes records into a -log ring of its own that init makes and names to `logd` (`sdk/toyos/src/log/`); +What is built: every program the supervisor starts or launches writes records into a +log ring of its own that the supervisor makes and names to `logkeeper` (`sdk/toyos/src/log/`); writing never waits, allocates or makes a syscall, a full ring is a count -`logd` says, and soundd's mix thread writes a lane of its own wait-free. The +`logkeeper` says, and soundserver's mix thread writes a lane of its own wait-free. The clock is a page (`sdk/abi/src/clock.rs`). Between the kernel's boot and its -stop the console UART is its userland driver's, which `logd` feeds the log it -writes, so a program's line reaches the console only through `logd`, under -its name. The kernel waits on `logd` neither at a +stop the console UART is its userland driver's, which `logkeeper` feeds the log it +writes, so a program's line reaches the console only through `logkeeper`, under +its name. The kernel waits on `logkeeper` neither at a panic, which halts the other CPUs first and paints its report, -nor at a stop, which init sequences through `logd`'s flush. `logd` syncs on an +nor at a stop, which the supervisor sequences through `logkeeper`'s flush. `logkeeper` syncs on an alert or an interval, keeps every boot's first part, and holds each program to an allowance. Severity is an ordered ladder. @@ -46,11 +46,11 @@ breaks a verdict. Records grow a key the way journald's `MESSAGE_ID` does, and a verdict matches the key. *Exit:* no verdict in `tests/` or `tests/metal/src/*.rs` reads a record by its prose. -**3. A machine whose `logd` ended says so.** Nothing restarts `logd` and -nothing names its end: init holds no `Process` handle to it, so the rings fill, +**3. A machine whose `logkeeper` ended says so.** Nothing restarts `logkeeper` and +nothing names its end: the supervisor holds no `Process` handle to it, so the rings fill, their writers go on unharmed, and the console says nothing. -*Exit:* init names `logd`'s exit on the console, and a guest test that ends -`logd` finds that line and every other daemon still serving. +*Exit:* the supervisor names `logkeeper`'s exit on the console, and a guest test that ends +`logkeeper` finds that line and every other daemon still serving. **4. The control that stages a kernel writing a file.** `log-writes-the-file` puts a kernel context back to appending records to `/log` through the VFS — @@ -67,12 +67,12 @@ that burst starves a tone playing beside it Constraints a reader would otherwise re-derive: -- A ring lives as long as the process init started with it: `logd` sweeps and +- A ring lives as long as the process the supervisor started with it: `logkeeper` sweeps and retires it at that process's end, so a child the process spawned into its own slots and that outlives it writes into a ring nobody reads. - A shared-memory region is a 2 MiB page, so a ring costs 2 MiB per program — what the pipe it replaced cost once written to. -- The UART's driver takes the log at the UART's own rate, and `logd` holds +- The UART's driver takes the log at the UART's own rate, and `logkeeper` holds what it has not taken in its replay, as for every reader. - A raw `SYS_REBOOT` or `SYS_SHUTDOWN` loses whatever a ring holds unread: the - one caller that has `logd` flush first is init's `power` port. + one caller that has `logkeeper` flush first is the supervisor's `power` port. diff --git a/issues/nothing-reopens-a-diskserver-session-over-a-used-page.md b/issues/nothing-reopens-a-diskserver-session-over-a-used-page.md index f5a811cfbb0..7ff575c5eb5 100644 --- a/issues/nothing-reopens-a-diskserver-session-over-a-used-page.md +++ b/issues/nothing-reopens-a-diskserver-session-over-a-used-page.md @@ -7,7 +7,7 @@ opened: 2026-10-02 # Nothing reopens a diskserver session over a used page A nightly went red on `blockd_io: FAIL /AFTER.BIN after the restart: Io` (run -36753172688, guest 7). blockd answered an open and only then made its ends of +36753172688, guest 7). diskserver answered an open and only then made its ends of the session page. A client that reconnects sends the page its last server wrote, and one that looks before the new server's ends are made hears the dead session's completion. The server's half is a type now: `wire::Opened::over` diff --git a/issues/null-sink-applies-one-connect.md b/issues/null-sink-applies-one-connect.md index e08920d5a0a..09cf16fce53 100644 --- a/issues/null-sink-applies-one-connect.md +++ b/issues/null-sink-applies-one-connect.md @@ -16,7 +16,7 @@ and its `IO_URINGS` are gone (`232f69bb6`), and Read off two T14 boots (`boot7-usba-doom.log`, `boot5-doom-wedge.log`). The shape is the same in both and it is narrow: -- soundd finds no device and presents the null sink; `null sink idle` prints. +- soundserver finds no device and presents the null sink; `null sink idle` prints. - The first client connects. The control thread prints `opening stream`, the **mix loop** prints `client 0 connected (id=0)` — so it woke, drained the command pipe (an event since #1108) and applied the command. @@ -40,12 +40,12 @@ with a terminal opened afterwards. **Eliminated, each with a run behind it.** CPU count. The cpal client path (`null_sink_shipped_client`, adopted from `wt/toyos-hdaprobe` `fa47241`, two -`/system/bin/tone` in series at 1.16 s and 1.15 s). soundd blocking on a client +`/system/bin/tone` in series at 1.16 s and 1.15 s). soundserver blocking on a client (`signal_clients` used `write_nonblock`, and an event's signal never blocks; there is no blocking write in the mix loop). The accept path being held by a stuck client (accept and mix are separate threads and the control thread ran for 14 s afterwards). A CQ overflow (`Poller::wait`'s `dropped` assertion -would have killed soundd, and soundd is alive). A panic anywhere in soundd, for +would have killed soundserver, and soundserver is alive). A panic anywhere in soundserver, for the same reason. **Eliminated by reading, and recorded so it is not re-derived.** The mix loop @@ -58,8 +58,8 @@ closed `io_uring::cancel_by_source` lost wake is *not* this: the mix loop's only registration is on the command pipe, now an event, which soundd owns both sides of and nobody closes. -**What settles it is where soundd's mix thread is parked and on what -deadline, and nothing in the tree reads the deadline off a running T14.** soundd's mix thread parked with a sane +**What settles it is where soundserver's mix thread is parked and on what +deadline, and nothing in the tree reads the deadline off a running T14.** soundserver's mix thread parked with a sane deadline says the timer did not fire; parked with an absurd one says the timeout was computed wrong; held by no CPU says nothing ever held it, which would move this into #142's family rather than audio's. The kernel's trace diff --git a/issues/one-manifest-row-grants-every-applet-and-one-is-compared.md b/issues/one-manifest-row-grants-every-applet-and-one-is-compared.md index 11ebf375586..6359e2d0f64 100644 --- a/issues/one-manifest-row-grants-every-applet-and-one-is-compared.md +++ b/issues/one-manifest-row-grants-every-applet-and-one-is-compared.md @@ -40,12 +40,12 @@ applets that have no use for it is not the list this test declares right: [] ``` -Every applet but `tone` in the test image holds a connector to soundd +Every applet but `tone` in the test image holds a connector to soundserver because `tone` needs one. That list is now `DECLARED_OVER_GRANTS`, so the size of the defect cannot grow without a red naming what grew. **What is still missing is the shipped image.** A guest test boots -`tests/images/testcases.toml`, whose `toybox` row is `receives = ["soundd"]` and +`tests/images/testcases.toml`, whose `toybox` row is `receives = ["soundserver"]` and carries no `syscap` at all. The row this record opened on — `system.toml:112`, `syscap = ["power", "roster"]` across every link — is measured by nothing: the differential that would see `/system/bin/echo` holding `Rights::POWER` has to read diff --git a/issues/parallel-tests-red-under-other-suites.md b/issues/parallel-tests-red-under-other-suites.md index 784ef0f8f74..1c056b60af0 100644 --- a/issues/parallel-tests-red-under-other-suites.md +++ b/issues/parallel-tests-red-under-other-suites.md @@ -333,7 +333,7 @@ changes. bytes)`), and green alone three times on each tree, alternating; `swap_crash_rolls_back` in one of the six, and `swap_netd` in a run at `0f27b73d`, both on `the stream's redial was turned away 64 time(s)` with - init's words on netd ending at `accepted`; and in the further run + the supervisor's words on netstack ending at `accepted`; and in the further run `metal_job_reboot` reddened, `quiesce_wakes_on_the_last_park`, on a kernel panic at `arch/tlb.rs:171` (`issues/a-shootdown-panicked-on-a-cpu-the-host-starved.md`). @@ -429,7 +429,7 @@ more refused launches left more live objects behind: [("PipeWrite", 0, 1), both times, in 4 s and 5 s — nowhere near the 192 s ceiling the paragraph above describes. `ALONE: GREEN` both times, and green again when re-run alone by hand (3 s, 2 s). So the name flakes for two different reasons and only one of them -is the ceiling this section corrects; what a contended host does to `/system/bin/init`'s +is the ceiling this section corrects; what a contended host does to `/system/bin/supervisor`'s handle accounting on a *refused* launch is not explained here, and nobody has a mechanism for it. diff --git a/issues/placement-has-no-migration-cost-or-wake-affinity.md b/issues/placement-has-no-migration-cost-or-wake-affinity.md index 6777f3ea7b3..9295f8dc17d 100644 --- a/issues/placement-has-no-migration-cost-or-wake-affinity.md +++ b/issues/placement-has-no-migration-cost-or-wake-affinity.md @@ -33,8 +33,7 @@ lacks: - A run shorter than 8 makes a server with a client in each LLC bounce (simulator, 4 in one run). - A partner moved onto its waker's SMT sibling reads as slow as split. - Leaving threads must name no waker. - - **Measuring it:** where a split pipe pair lands decides `pipe.throughput` - (`issues/a-spawned-thread-starts-in-any-llc-not-its-spawners.md`). Several rows are bimodal per boot on this bench, + - **Measuring it:** where a split pipe pair lands decides `pipe.throughput`. Several rows are bimodal per boot on this bench, so an A/B there needs per-row mode accounting. - **An idle-state input**: see `issues/the-scheduler-and-the-clocks-never-talk.md`. - **Bounded scans.** Each choice walks the rings it needs whole. A steal diff --git a/issues/process-start-skew-on-a-runner.md b/issues/process-start-skew-on-a-runner.md index 76ea0f406ae..9631982b88b 100644 --- a/issues/process-start-skew-on-a-runner.md +++ b/issues/process-start-skew-on-a-runner.md @@ -7,7 +7,7 @@ opened: 2026-08-08 # A userland process reaches its first line half a second after its sibling on a runner, and ~30 ms after it here Fell out of the null-sink probe and is not what it closed. On all three reps the two -programs `tests/images/testcases.toml` starts — soundd and test-runner, spawned 1–3 ms apart +programs `tests/images/testcases.toml` starts — soundserver and test-runner, spawned 1–3 ms apart — printed their first lines **0.53–0.56 s apart**, and which of the two was first flipped between reps. On this host the same pair is ~30 ms apart, in spawn order, every time. The kernel's own boot is the same speed on both (`Boot: diff --git a/issues/redesign-the-log-subsystem.md b/issues/redesign-the-log-subsystem.md index e480730e633..5554f554f2f 100644 --- a/issues/redesign-the-log-subsystem.md +++ b/issues/redesign-the-log-subsystem.md @@ -55,7 +55,7 @@ writers and readers never observe a torn record" and filled at `:158`. The three files the table calls the log's own are gone: `c31e9f97^:kernel/src/log.rs`, `ee8369c9^:kernel/src/drivers/log_ring.rs` and `9ca7631a^:kernel/src/log_file.rs` are none of them -in the tree. The file sink is not a kernel module at all — `/system/bin/logd` is an +in the tree. The file sink is not a kernel module at all — `/system/bin/logkeeper` is an ordinary user process, and "the kernel keeps the record ring and the console; every policy about files — where they go, what they are called, how many there are, what happens when the stick stops answering — is here" diff --git a/issues/soundserver-past-due-wake-max-1.md b/issues/soundserver-past-due-wake-max-1.md index 7005895d2d6..ac0e32a5795 100644 --- a/issues/soundserver-past-due-wake-max-1.md +++ b/issues/soundserver-past-due-wake-max-1.md @@ -12,11 +12,11 @@ computes its poll timeout as `next_period_ns.saturating_sub(now).max(1)`, the is exactly what a grid point already in the past wants, so `0` is the right answer and `1` is a park on a deadline that has passed. A device's wait no longer does this: `armed` arms the next future grid point when the predicted -one has passed. soundd had the pattern in both, and it was the trigger on the fifth of the +one has passed. soundserver had the pattern in both, and it was the trigger on the fifth of the eight T14 boots captured 2026-08-08 while chasing the cpu0 stall, where cpu0 and cpu1 both stopped within 100 ms of `soundd: resumed`. Not fixed here, and deliberately: it is one line in the mix loop, `apic::OneShot`'s -floor makes it safe, and what it changes is when soundd wakes on a late period — +floor makes it safe, and what it changes is when soundserver wakes on a late period — audio timing, which is the owner's call. With the floor, a past-due grid point now costs up to 10 µs of extra lateness against a 2.9 ms period. diff --git a/issues/soundservers-mix-pass-has-no-host-test.md b/issues/soundservers-mix-pass-has-no-host-test.md index 05bf0af1c8d..1f7646f05c6 100644 --- a/issues/soundservers-mix-pass-has-no-host-test.md +++ b/issues/soundservers-mix-pass-has-no-host-test.md @@ -4,16 +4,16 @@ kind: track opened: 2026-07-28 --- -# soundd's mix pass has no host test, and neither negative gate exists +# soundserver's mix pass has no host test, and neither negative gate exists -Both soundd defects found on 2026-07-28 were in pure code needing no device, no +Both soundserver defects found on 2026-07-28 were in pure code needing no device, no syscall and no kernel — and both cost a 120-boot QEMU campaign to find. That is the case for host-testing the daemons, and it is empirical. -Part of it landed: `toyos-abi` is host-tested, soundd has a `#[cfg(test)]` module -covering quantisation, the deferral floor and shape refusal, and netd/sshd got -their own boot configs. What did not land is the pass itself. Everything from the -mix loop's `loop {` to its stats flush — the free-mask fold, the drain branch, +Part of it landed: `toyos-abi` is host-tested, soundserver has a `#[cfg(test)]` module +covering quantisation, the deferral floor and shape refusal, and netstack and +sshserver got their own boot configs. What did not land is the pass itself. +Everything from the mix loop's `loop {` to its stats flush — the free-mask fold, the drain branch, the refill loop, the pending-removal retain, the stats window — runs only in a booted guest. @@ -53,13 +53,12 @@ test and the QEMU gate quantise into the same buckets from one implementation. it. An 11-line pure bug removed the integration gate's detection power once already. -Deliberately out of scope, so nobody re-argues it: sshd (~17 pure lines, and its +Deliberately out of scope, so nobody re-argues it: sshserver (~17 pure lines, and its handler surface cannot be faked without widening a fork); the compositor's `fn main` decomposition (its two worst defects were found by reading and the right -fix for one is to make it unrepresentable); netd's protocol layer (stock -smoltcp — conformance tests there test upstream); and the resampler's numerics +fix for one is to make it unrepresentable); and the resampler's numerics (libm `sin`/`exp` need not agree bit-for-bit between an arm64 host and the x86-64 target). Test the code around the resampler, not its output values. **And the honest ceiling:** none of this touches the wake lateness that causes -the dropouts. It makes soundd's *response* to lateness a controlled variable. +the dropouts. It makes soundserver's *response* to lateness a controlled variable. diff --git a/issues/spawned-process-never-starts.md b/issues/spawned-process-never-starts.md index e18741f866c..ef92c6c05d0 100644 --- a/issues/spawned-process-never-starts.md +++ b/issues/spawned-process-never-starts.md @@ -69,7 +69,7 @@ with no CPU is one that blocked before its first user instruction, and any CPU at all moves the fault into userland startup. The owner ran it on the T14 during boot 10 of `boot5-doom-wedge.log` and **`ps` pid 17 printed nothing and never exited**, and neither did `/system/bin/shutdown` pid 25 or three `doom`s. In -that whole boot the only processes that ever exited were `netd` and one +that whole boot the only processes that ever exited were `netstack` and one `locale`. That is an answer rather than a lost measurement, and it is worth more than the diff --git a/issues/stop-the-device-voice-keep-the-wake.md b/issues/stop-the-device-voice-keep-the-wake.md index bae952f45f2..ea1cc154a96 100644 --- a/issues/stop-the-device-voice-keep-the-wake.md +++ b/issues/stop-the-device-voice-keep-the-wake.md @@ -4,10 +4,10 @@ kind: defect opened: 2026-08-01 --- -# An idle soundd keeps the DMA engine and the codec voice open +# An idle soundserver keeps the DMA engine and the codec voice open Stopping the device voice while keeping the periodic timer wake recovers the DMA engine and the codec — the battery-relevant hardware — and gives up only the wake -itself. Resume still works unchanged, because soundd keeps writing signal bytes, -so it does not need the missing client→soundd message that +itself. Resume still works unchanged, because soundserver keeps writing signal bytes, +so it does not need the missing client→soundserver message that `cpal-backend-hardcodes-the-format` is waiting on. diff --git a/issues/tco2-sts-clearing-is-verified-on-qemu-only.md b/issues/tco2-sts-clearing-is-verified-on-qemu-only.md index 631648da40b..193a90c5527 100644 --- a/issues/tco2-sts-clearing-is-verified-on-qemu-only.md +++ b/issues/tco2-sts-clearing-is-verified-on-qemu-only.md @@ -24,4 +24,4 @@ answered this register's clearing cannot be exercised on the machine at all. **Exit condition**: a T14 boot that resets itself on an armed timer, and the kernel's own `watchdog:` line off the log partition on the boot *after* it — so -it needs a metal run whose boot reaches `logd`, which no run's has. +it needs a metal run whose boot reaches `logkeeper`, which no run's has. diff --git a/issues/the-guest-suite-runs-only-what-no-cheaper-tier-reaches.md b/issues/the-guest-suite-runs-only-what-no-cheaper-tier-reaches.md index 87f50a577f5..ae53045b6e0 100644 --- a/issues/the-guest-suite-runs-only-what-no-cheaper-tier-reaches.md +++ b/issues/the-guest-suite-runs-only-what-no-cheaper-tier-reaches.md @@ -107,7 +107,7 @@ and NVMe. Exit: as stage A. - host `fsync_failed_commit`: an fsync keeps refusing while the device refuses its flush. - host `file_mtime_undated`: a file written with no wall clock is stamped as undated. -- host `blockd_serves_partitions`, `blockd_survives_its_death`, `blockd_serves_nothing`: blockd's +- host `blockd_serves_partitions`, `blockd_survives_its_death`, `blockd_serves_nothing`: diskserver's protocol, its restart, and its refusals with no controller. - host `nvme_large_device`, `nvme_wide_sector`: a device past 2 TiB and an 8 KiB namespace. - host `fsd_two_data`: two DATA partitions are refused by name. @@ -176,8 +176,8 @@ Host: `toyos-desktop`, the console's grid. Metal: the T14's desktop. Exit: as st Host: `toyos-net-tcp`, `toyos-dns`, `toyos-mdns`, `toyos-swap`, `toyos-inspect`, `toyos-logstream`. Metal: the cable. Exit: as stage C. -- host `lan_mdns_answer`, `lan_no_lease`: netd answers for its name, and announces with no lease. -- host `netd_connection_caps`, `netd_listener_forgery`, `netd_hostile_peer`: netd refuses forged +- host `lan_mdns_answer`, `lan_no_lease`: netstack answers for its name, and announces with no lease. +- host `netd_connection_caps`, `netd_listener_forgery`, `netd_hostile_peer`: netstack refuses forged flags, hostile peers and excess connections. - host `netd_slow_reader`, `netd_refused_pipes`, `netd_held_open`: a slow, refused or silent client costs only itself. @@ -193,7 +193,7 @@ Host: `toyos-net-tcp`, `toyos-dns`, `toyos-mdns`, `toyos-swap`, `toyos-inspect`, (`issues/the-host-cannot-reach-the-t14-while-it-runs-toyos.md`). Held on a host by the node's `a_link_that_returns_keeps_a_held_lease_and_announces_it` and `a_link_that_returns_with_no_lease_starts_discovery_over_at_once` (`system/netstack/node/tests/lease.rs`). - metal `sshd_exec`, the arms `lan_talk`'s one command did not reach, each a step of that row's - exchange red when its arm in `userland/sshd/src/main.rs` is reverted; the row and its exchange + exchange red when its arm in `system/sshserver/src/main.rs` is reverted; the row and its exchange are deleted, and this waits on `issues/the-host-cannot-reach-the-t14-while-it-runs-toyos.md`: - a program that is not there ends 127, the shell saying `: not found`; - an unquotable line is the shell's to refuse, its `-c` line read as the shell reads any; diff --git a/issues/the-internet-clients-work-unchanged.md b/issues/the-internet-clients-work-unchanged.md index b0dd28eea41..9bc3f22bd79 100644 --- a/issues/the-internet-clients-work-unchanged.md +++ b/issues/the-internet-clients-work-unchanged.md @@ -11,17 +11,17 @@ works" that means a program written against `std::net`, `rustls` and an HTTP crate such as `ureq` builds for ToyOS unchanged and behaves as it does on any other OS. TLS and HTTP are not written here: the crates are used as published. What this track owns is the layer under them that only ToyOS can supply — -netd, `toyos::net` and std's ToyOS networking in the `rust/` fork. +netstack, `toyos::net` and std's ToyOS networking in the `rust/` fork. ## Stages -1. **Names resolve.** netd's own resolver behind `std::net::ToSocketAddrs`. +1. **Names resolve.** netstack's own resolver behind `std::net::ToSocketAddrs`. Landed (#511). Whether its wire decoding should be `hickory-proto` rather than `toyos-dns`'s own reader is `issues/toyos-dns-decodes-what-a-published-crate-decodes.md`. 2. **TCP holds up.** Bounded connect, resets surfaced, half-close, large and lossy transfers byte-exact, timeouts honoured, a departed client freeing - everything in netd. **Exit**: each is a guest test against the host's own + everything in netstack. **Exit**: each is a guest test against the host's own TCP stack, with a hash over every byte moved. 3. **TLS.** `rustls` with the `ring` provider (owner, 2026-10-02; `graviola` is an option for later) and the Mozilla root set as a pinned data file in diff --git a/issues/the-kernel-keeps-nothing-it-enumerates.md b/issues/the-kernel-keeps-nothing-it-enumerates.md index 30105d2768d..ffb6e013a8a 100644 --- a/issues/the-kernel-keeps-nothing-it-enumerates.md +++ b/issues/the-kernel-keeps-nothing-it-enumerates.md @@ -19,21 +19,21 @@ On top of retention, in dependency order: 1. **Retention plus one query syscall and a topic decoder**, with `hw` and `disk` as its first two consumers. -2. **A diagnostic port on `/system/bin/init`**, answering a program name with a +2. **A diagnostic port on `/system/bin/supervisor`**, answering a program name with a `Process` handle narrowed to `Rights::READ`. `SYS_PROCESS_STATS` takes a `Process` handle and a handle is the whole of the right, so nothing hands a - diagnostic tool a way to sample a daemon: `/system/bin/init` holds the only `Process` + diagnostic tool a way to sample a daemon: `/system/bin/supervisor` holds the only `Process` handles for what `[boot] start` names. So "where is - soundd's / the compositor's / netd's time going?" is unanswerable from a + soundserver's / the compositor's / netstack's time going?" is unanswerable from a shell, and `audio_idle_suspend` name-matches `SYS_SYSINFO` entries out of a byte buffer to sample a running daemon twice. Nothing in the kernel has to change for it — the addressing half is done, and this is the policy half. 3. **A status-query message in the SDK**, reserved out of the message-id space, answered by the compositor first. -4. **A cumulative soundd snapshot** published under a seqlock, and an `audio` +4. **A cumulative soundserver snapshot** published under a seqlock, and an `audio` tool that reads it. Today's stats struct is windowed and reset wholesale in four places, so nothing outside the mix thread can read a total. -5. **netd interface, link and counter reporting**, and a `net` tool — the name +5. **netstack interface, link and counter reporting**, and a `net` tool — the name is taken by an HTTP GET demo that should be renamed `http` first, which is the owner's call. 6. **Master gain, and `audio volume`** — see diff --git a/issues/the-lan-is-not-yet-production-grade.md b/issues/the-lan-is-not-yet-production-grade.md index 443f32698d7..698212f419a 100644 --- a/issues/the-lan-is-not-yet-production-grade.md +++ b/issues/the-lan-is-not-yet-production-grade.md @@ -46,7 +46,7 @@ why, and the boot still leases. management firmware that shares it expect. Ubuntu has had link after every ToyOS boot so far; that becomes a tested property. - **Every received frame is hostile.** The receive path and every parser above - it are fuzzed; a malformed frame ends netd at worst, never the kernel. + it are fuzzed; a malformed frame ends netstack at worst, never the kernel. - **Error counters** (CRC, missed, no-buffer) are read and asserted zero under load. - **No register is left to its reset value, but `TIDV` and the receive addresses past the first, deliberately.** Done for these registers (pull diff --git a/issues/the-logs-server-ending-under-logkeepers-append-is-unmeasured.md b/issues/the-logs-server-ending-under-logkeepers-append-is-unmeasured.md index cd5ed0e177a..7ee5fdfe372 100644 --- a/issues/the-logs-server-ending-under-logkeepers-append-is-unmeasured.md +++ b/issues/the-logs-server-ending-under-logkeepers-append-is-unmeasured.md @@ -4,16 +4,16 @@ kind: tooling opened: 2026-09-27 --- -# The log's server ending under logd's append is unmeasured +# The log's server ending under logkeeper's append is unmeasured `fsd_restart` ends only DATA's server. The log's is restarted by the same row, -and logd holds its file open across it: a handle reopened after an end is +and logkeeper holds its file open across it: a handle reopened after an end is checked against the identity it last saw, and an append the server ended -under is answered `StaleNetworkFileHandle` and not retried. What logd does +under is answered `StaleNetworkFileHandle` and not retried. What logkeeper does then — whether the lines of that append are lost, repeated, or written after a reopen — and whether the log file the host reads back after the stop is whole, has not been run. -**Exit**: a boot that ends the log's server under logd's append (`--end-on` on +**Exit**: a boot that ends the log's server under logkeeper's append (`--end-on` on a path of the log's volume), and the log file read back off the image by a FAT -implementation that is not fsd's, with what logd said about the lost append. +implementation that is not fileserver's, with what logkeeper said about the lost append. diff --git a/issues/the-metal-loop-cannot-judge-a-boot-that-panics.md b/issues/the-metal-loop-cannot-judge-a-boot-that-panics.md index 578c2e8e2fc..2b7a8b29385 100644 --- a/issues/the-metal-loop-cannot-judge-a-boot-that-panics.md +++ b/issues/the-metal-loop-cannot-judge-a-boot-that-panics.md @@ -20,7 +20,7 @@ the T14, and the reason is the loop rather than the kernel. `tests/common/metal.rs`'s `Mode::Drive` tolerates the non-zero exit. 2. `test-late-panic` fires in `kernel_main` after `spawn_init` and before - `enter_idle_loop`, so `logd` has not run: that boot writes no `/log` file at + `enter_idle_loop`, so `logkeeper` has not run: that boot writes no `/log` file at all and `boot.