H2 of docs/plan/hybrid.md -- the milestone the plan itself calls the one
that proves the whole idea, and the one that cannot be skipped.
# cat /proc/schinit
1 4 11 os.schos.vgpud
# kill 4
schinitd: restarting os.schos.vgpud
vgpud: claimed
vgpud: pid 8 drew a test pattern from userspace
The test pattern ends in a twelve-block tally of the drawing process's
pid, because a second run that redraws an identical picture is
indistinguishable from a first run that never stopped. Read off the
virtio-gpu scanout across four generations it says 4, 8, 10, 12, and the
DMA arena hands back the same physical address every time.
What it took:
* A halt the kernel can apply to a device it does not understand.
Clearing PCI_CMD_BUS_MASTER is the only one there is, and it has to
happen before the DMA goes back to the arena -- a dead driver's device
keeps reading a framebuffer the kernel is about to give to somebody
else. Claim sets the bit, release clears it; pcitest shows it in a bm
column, with a third state so a failed read is never printed as off.
* A refusal. vgpu_release() asks vga_fb_restore() whether there is a
framebuffer to put the console back on, and declines to hand over the
device if there is not. The rule that the kernel keeps a minimal
driver for everything on the path to a shell is worth writing down
only if something enforces it.
* The DMA arena not being owned by two allocators at once. This is the
bug that stopped H2: freevm() frees every mapped page below PHYSTOP
and the arena is the top 16 MB of exactly that range, so a DMA block
mapped into a driver went onto the page allocator's free list when the
driver died, while pciuser_release_all() handed the same pages back to
the arena. The restarted vgpud asked for its framebuffer, got pages
kalloc() had already given out -- including its own page tables -- and
dma_alloc()'s zeroing wiped them, so it died inside the ioctl on the
return from a syscall whose stack had stopped existing. do_dma() now
records a VMA with no_free set, the way shared memory does.
* One exit status. KeepAlive must restart a driver that crashed and must
not restart one that has looked at the machine and found its device
absent: on hardware with no virtio-gpu that was fifteen spawns in
forty-five seconds, each printing a line to the console, for ever.
SCHINIT_EXIT_NORESTART is 78, sysexits.h's EX_CONFIG. It is not a new
manifest key -- the wait status has always travelled from PID 1 to the
supervisor, and note_exit() simply stopped discarding it.
The milestone's third clause, "the compositor reattaches", is not done
and is not claimed: there is no compositor running. That is recorded in
the plan as a Schcom milestone rather than counted here.
Closes the register row "A released PCI device" in docs/plan/1.0.md.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VQ7Qx8E1q1M1bbCq8QLHS2
143 lines
6 KiB
C
143 lines
6 KiB
C
// SPDX-License-Identifier: Apache-2.0
|
|
//
|
|
// The protocol between PID 1 and the supervisor.
|
|
//
|
|
// docs/architecture/init.md warns about this file specifically: two
|
|
// processes that must agree is a compatibility surface where there was none,
|
|
// and docs/plan/schapi.md §3 is about how those go wrong. So the rules are
|
|
// written here rather than assumed.
|
|
//
|
|
// **Fixed-size messages.** Every message is exactly one struct, and PID 1
|
|
// reads sizeof(struct) or nothing. No length prefix to get wrong, no
|
|
// partial-parse state to be caught in, and -- the reason that matters --
|
|
// **PID 1 never allocates**. A supervisor that has gone mad cannot make PID 1
|
|
// allocate, block on a malloc, or run a parser.
|
|
//
|
|
// **Versioned in the first message.** SCHINIT_PROTO travels in the hello. A
|
|
// mismatch is a refusal, not a best-effort interpretation.
|
|
//
|
|
// **Entirely internal.** This is not SchAPI. Nothing but schinit and
|
|
// schinitd may speak it, and it may change shape whenever both change
|
|
// together, which is what keeping it in one header is for.
|
|
//
|
|
// The asymmetry is the design, not an accident: the supervisor *asks*, PID 1
|
|
// *acts*. Every job on the machine is therefore a child of PID 1 at all times,
|
|
// so a supervisor that dies orphans nothing and reparents nothing.
|
|
|
|
#ifndef SCHINIT_H
|
|
#define SCHINIT_H
|
|
|
|
#define SCHINIT_PROTO 1
|
|
|
|
#define SCHINIT_MAXARGS 16
|
|
#define SCHINIT_MAXENVV 16
|
|
#define SCHINIT_MAXLABEL 64
|
|
#define SCHINIT_MAXPATH 128
|
|
// One flat buffer for argv and envp strings, copied whole. Pointers cannot
|
|
// cross a process boundary, so the spec carries text and PID 1 rebuilds the
|
|
// arrays on its own stack after exec'ing -- which is the only place they are
|
|
// ever valid.
|
|
#define SCHINIT_STRBUF 1024
|
|
|
|
// supervisor -> PID 1
|
|
#define SCHINIT_HELLO 1 // struct schinit_hello
|
|
#define SCHINIT_SPAWN 2 // struct schinit_spawn
|
|
#define SCHINIT_PONG 3 // struct schinit_hello -- answer to a PING
|
|
|
|
// PID 1 -> supervisor
|
|
#define SCHINIT_WELCOME 128 // struct schinit_hello, echoed with PID 1's version
|
|
#define SCHINIT_SPAWNED 129 // struct schinit_reply -- pid, or negative on failure
|
|
#define SCHINIT_EXITED 130 // struct schinit_reply -- a child has been reaped
|
|
#define SCHINIT_PING 131 // struct schinit_hello -- answer within the deadline
|
|
|
|
// Exit statuses a job can use to tell the supervisor something.
|
|
//
|
|
// KeepAlive restarts a job that crashed or was killed, which is the whole
|
|
// point of it for a driver. It must not restart one that has looked at the
|
|
// machine and reported there is nothing here for it to do: a driver whose
|
|
// device is absent fails identically every time, and a supervisor that tries
|
|
// anyway spends the rest of the uptime doing so -- 15 spawns in 45 seconds,
|
|
// on hardware with no virtio-gpu, each one printing a line to the console.
|
|
//
|
|
// 78 is sysexits.h's EX_CONFIG, "something was found in an unconfigured or
|
|
// misconfigured state", which is exactly what a driver with no device has
|
|
// found. Borrowed rather than invented, so a program written against any
|
|
// other Unix already means the right thing by it.
|
|
//
|
|
// Note what this is *not*: it is not a new manifest key and not a new policy
|
|
// knob. The wait status has always travelled from PID 1 to the supervisor in
|
|
// struct schinit_reply, and note_exit() has always thrown it away. This is
|
|
// the other half of a thing that was already built.
|
|
#define SCHINIT_EXIT_NORESTART 78
|
|
|
|
// The heartbeat.
|
|
//
|
|
// docs/architecture/init.md, "The heartbeat" -- the failure the two-init scheme
|
|
// could not see:
|
|
// **the common way an init breaks is not dying, it is hanging.** wait() never
|
|
// returns for a process that is merely stuck, so a supervisor watching a
|
|
// wedged PID 1 -- or PID 1 watching a wedged supervisor -- reports everything
|
|
// normal forever.
|
|
//
|
|
// So each must be heard from. PID 1 sends a PING every interval and expects a
|
|
// PONG; silence past the deadline means the other end is stuck rather than
|
|
// busy, and stuck is the case worth acting on.
|
|
//
|
|
// Measured in seconds of uptime(), which counts ticks. Ticks can be *lost*
|
|
// when interrupts are off too long -- see kernel/core/time.c -- which makes a
|
|
// tick-measured deadline longer in real time than it looks, never shorter.
|
|
// That is the safe direction: a machine under load waits longer before
|
|
// declaring a hang, rather than declaring one that has not happened.
|
|
#define SCHINIT_PING_SECS 2
|
|
#define SCHINIT_DEADLINE_SECS 10
|
|
|
|
struct schinit_hello {
|
|
unsigned int op;
|
|
unsigned int proto;
|
|
unsigned int generation; // how many times PID 1 has been re-exec'd
|
|
unsigned int reserved;
|
|
};
|
|
|
|
// A job, flattened. `args` and `envs` index into `strbuf`, which holds
|
|
// NUL-terminated strings back to back.
|
|
struct schinit_spawn {
|
|
unsigned int op;
|
|
unsigned int cookie; // echoed in the reply, so the supervisor
|
|
// can match a spawn to the job it asked for
|
|
char label[SCHINIT_MAXLABEL];
|
|
char program[SCHINIT_MAXPATH];
|
|
|
|
char stdin_path[SCHINIT_MAXPATH];
|
|
char stdout_path[SCHINIT_MAXPATH];
|
|
char stderr_path[SCHINIT_MAXPATH];
|
|
|
|
unsigned int own_tty; // claim the terminal as controlling
|
|
unsigned int nargs;
|
|
unsigned int nenvs;
|
|
unsigned short args[SCHINIT_MAXARGS]; // offsets into strbuf
|
|
unsigned short envs[SCHINIT_MAXENVV];
|
|
char strbuf[SCHINIT_STRBUF];
|
|
};
|
|
|
|
struct schinit_reply {
|
|
unsigned int op;
|
|
unsigned int cookie;
|
|
int pid; // SPAWNED: the child, or negative errno
|
|
int status; // EXITED: wait status
|
|
};
|
|
|
|
// The largest message, which is what a reader must have room for. PID 1 reads
|
|
// into one of these on its own stack and never anywhere else.
|
|
union schinit_msg {
|
|
unsigned int op;
|
|
struct schinit_hello hello;
|
|
struct schinit_spawn spawn;
|
|
struct schinit_reply reply;
|
|
};
|
|
|
|
// The descriptor the supervisor inherits. Fixed rather than passed, because
|
|
// the supervisor is exec'd by PID 1 and there is nowhere to pass it -- and a
|
|
// well-known number is one less thing for the two to disagree about.
|
|
#define SCHINIT_SOCK_FD 3
|
|
|
|
#endif // SCHINIT_H
|