9e6d8dd4e3
sendmsg/recvmsg with SCM_RIGHTS. Item 1 of docs/plan/foundations.md and
mechanism 1 of docs/plan/hybrid.md, which are the same primitive seen from two
directions: Wayland's wl_shm_create_pool passes a file descriptor with no
integer-shaped alternative, and handing a driver process the device it is
allowed to touch is the same act. Both plans stopped here.
Linux's structures byte for byte -- msghdr, cmsghdr, the CMSG_ macros -- because
the programs that need this are not ours. libwayland and toybox both call
recvmsg with Linux's headers and would not notice a friendlier door.
The delivery rule is the whole design. A batch of descriptors is pinned to the
stream position of the first byte of the message that carried it, and that has
two halves: a reader arriving at that position is handed them, and a reader that
has not reached it is stopped SHORT of it. Leave out the second half and a
program asking for 4096 bytes swallows three messages and is handed one
message's descriptors, with nothing in the result saying which. socktest checks
exactly that: two one-byte messages, an eight-byte read, one byte back.
The kernel holds a reference to every descriptor in flight, so a sender may
close its own copy the moment sendmsg returns. Giving that reference back is the
part that leaks if it is not written down, and there are three ways a message
ends:
- received with room installed in the receiver's table
- received without room, or by
plain read(2) closed; MSG_CTRUNC for the first case
- never received, socket closed closed by sockclose
The middle row looks like a choice and is not: holding kernel references for a
receiver that has proved it is not looking is a leak with extra steps.
No garbage collector. A socket descriptor sent over that same socket and never
received holds a reference to the socket that holds the reference to it, and
neither close reaches zero. Linux carries several hundred lines for this.
Nothing on either plan passes a socket over a socket -- Wayland passes memory,
the driver work passes devices -- so it is a bounded leak that is documented
rather than a silent one.
Fourteen checks in cmd/socktest, and every one moves data through the descriptor
rather than inspecting it. The descriptor passed is always a pipe end, because a
pipe announces its own lifetime: a read of a pipe whose writers are all gone
returns 0, which turns "was the kernel's reference released" -- otherwise
invisible from userspace -- into a value a test can compare.
Two bugs fell out, neither in the new code:
- sockpair() never set the socket's domain, so every socketpair() socket
claimed domain 0. Invisible while the data paths asked only whether a socket
was AF_INET or AF_NETLINK, since zero answers "neither" correctly.
SCM_RIGHTS asks "is this AF_UNIX", and zero is wrong for that one.
- pipe_poll() reports POLLHUP but not POLLIN at end of file, so a non-blocking
read of a pipe whose writers had all gone returned EAGAIN forever instead of
0 -- a program that polls and reads spins waiting for an end that already
happened. sockreadable() always had the equivalent condition; the pipe path
did not. Found because the test helper for "is this descriptor closed" is a
non-blocking read of a pipe.
projects/schcom/README.md said SCM_RIGHTS was done for a long time while nothing
in the kernel implemented it, and an entire GUI plan was drawn on that line.
It is true now, which is not the same as it having been right, and the README
says so.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VQ7Qx8E1q1M1bbCq8QLHS2
102 lines
4.8 KiB
Markdown
102 lines
4.8 KiB
Markdown
# Schcom — the SchOS Compositor
|
||
|
||
Owns the display. Every pixel that reaches the screen goes through it.
|
||
|
||
Clients do not draw to the framebuffer; they draw into their own buffers and
|
||
hand them over. Schcom decides what is on top of what, composites the result,
|
||
and presents it. It also owns input: it reads the raw event devices and sends
|
||
each event to whichever client should have it.
|
||
|
||
That is the whole job. Window *placement* is not part of it — that is SchWM's,
|
||
and Schcom exposes a protocol for it rather than implementing policy itself.
|
||
|
||
## Status
|
||
|
||
**v0 runs.** A client connects, is given a shared buffer, draws into it and
|
||
commits; the result is composited and presented. Verified end to end by
|
||
`make schcom-test`, which boots SchOS, runs the compositor and a client, then
|
||
screendumps the framebuffer and asserts the pixels **from the host** — the
|
||
compositor's own report that it composited something proves nothing. It runs
|
||
two overlapping clients and checks that the newer one wins in the overlap: a
|
||
compositor that cannot order two surfaces cannot host a window manager.
|
||
|
||
What exists:
|
||
|
||
- framebuffer output through `/dev/fb0` and the fbdev ioctls, double-buffered
|
||
- a versioned protocol over a Unix socket, with a handshake that must come first
|
||
- shared-memory buffers, one per surface, owned by the compositor
|
||
- surface / attach / **commit** semantics: nothing is shown mid-draw
|
||
- input from `/dev/input/eventN`, routed to the topmost surface
|
||
- clients that die are cleaned up, including their shm segments
|
||
- **multiple clients**, composited bottom-to-top in creation order
|
||
|
||
What does not exist yet: damage tracking beyond "repaint everything", multiple
|
||
outputs, subsurfaces, and any focus model beyond "topmost". Those are additions
|
||
rather than rewrites, which is the reason for committing to the surface/commit
|
||
model now instead of growing into it later.
|
||
|
||
## Building
|
||
|
||
Not part of the OS build. Needs `make ports` first, for the sysroot's libc:
|
||
|
||
```sh
|
||
make projects # builds and stages into sysroot/bin
|
||
make # rebuilds the image so the binaries are on it
|
||
make schcom-test # boots it and checks the pixels
|
||
```
|
||
|
||
## What it needs from the kernel
|
||
|
||
| Need | State |
|
||
|---|---|
|
||
| Shared memory (`shmget`/`shmat`/`shmdt`) | done |
|
||
| `mmap` of `/dev/fb0` | done |
|
||
| Framebuffer geometry via the fbdev ioctls | done |
|
||
| Unix domain sockets | done |
|
||
| Raw input events — key down/up with modifiers | done |
|
||
| `SCM_RIGHTS` | done — `sendmsg`/`recvmsg`, see [`docs/architecture/networking.md`](../../docs/architecture/networking.md#descriptor-passing) |
|
||
| Mode setting and double buffering | partial — virtio-gpu binds, no modeset API yet |
|
||
|
||
## Design notes
|
||
|
||
- **Buffers are shared memory, not pipes.** A pipe means copying every frame; at
|
||
800×600×32 that is 1.9 MB per frame per client.
|
||
- **The compositor owns the segment, not the client.** It has to survive a
|
||
client crashing mid-frame, and it must not be something a client can resize
|
||
underneath the compositor.
|
||
- **One protocol, versioned from the start.** SchWM and SchDE are separate
|
||
programs, so this is a real compatibility surface on day one. Every message
|
||
carries its own length, so an unknown message is skipped rather than
|
||
desynchronising the stream — that is what lets a v0 compositor survive a v1
|
||
client.
|
||
- **Policy belongs to SchWM.** If Schcom starts deciding where windows go, the
|
||
split has already failed. `CREATE_SURFACE` carries x/y today only because
|
||
there is no SchWM yet to ask.
|
||
|
||
## On Wayland
|
||
|
||
> **Decided: Schcom moves onto Wayland**, replacing the protocol below while
|
||
> keeping the architecture. See [`docs/plan/wayland.md`](../../docs/plan/wayland.md).
|
||
|
||
|
||
The protocol is deliberately Wayland-shaped — surfaces rather than windows,
|
||
buffers attached and made visible only on commit, and the compositor telling
|
||
clients when to draw. Wayland is **MIT-licensed**, so it can be linked against
|
||
without any conflict with Apache-2.0 (see `docs/licensing.md`), and
|
||
supporting it later is intended.
|
||
|
||
The one place v0 knowingly diverges: a buffer handle is a System V shared memory
|
||
id passed as an integer, where Wayland passes a file descriptor over
|
||
`SCM_RIGHTS`. `struct schcom_surface` carries a `handle_kind` field precisely so
|
||
that switching would be a new value rather than a protocol break.
|
||
|
||
**That field could not have saved us.** Wayland's `wl_shm_create_pool` passes a
|
||
descriptor with no alternative form, so *our* protocol being the flexible one
|
||
was never the flexibility that was needed — and for a long time this README
|
||
claimed `SCM_RIGHTS` was done when there was no `sendmsg`/`recvmsg` in the
|
||
kernel at all. An entire GUI plan was drawn on top of that line.
|
||
|
||
It is true now. Item 1 of
|
||
[`docs/plan/foundations.md`](../../docs/plan/foundations.md) is built and
|
||
tested; what a buffer still needs is item 2, a shareable memory object, because
|
||
a descriptor for a System V segment is not a thing that exists.
|