motiejus/zig
1
Fork 0
Code Releases Activity

std.child_process: use std.io.poll for collectOutput

Browse Source
This commit is contained in:
Jonathan Marler
2023-02-28 14:10:44 -07:00
committed by Andrew Kelley
parent 4f58a80735
commit 138e8b162a
1 changed files with 30 additions and 178 deletions
Download Patch File Download Diff File Expand all files Collapse all files

208
lib/std/child_process.zig
View File

@@ -197,6 +197,19 @@ pub const ChildProcess = struct {
stderr: []u8,
};
fn fifoToOwnedArrayList(fifo: *std.io.PollFifo) std.ArrayList(u8) {
if (fifo.head > 0) {
std.mem.copy(u8, fifo.buf[0..fifo.count], fifo.buf[fifo.head .. fifo.head + fifo.count]);
}
const result = std.ArrayList(u8){
.items = fifo.buf[0..fifo.count],
.capacity = fifo.buf.len,
.allocator = fifo.allocator,
};
fifo.* = std.io.PollFifo.init(fifo.allocator);
return result;
}
/// Collect the output from the process's stdout and stderr. Will return once all output
/// has been collected. This does not mean that the process has ended. `wait` should still
/// be called to wait for and clean up the process.
@@ -210,189 +223,28 @@ pub const ChildProcess = struct {
) !void {
debug.assert(child.stdout_behavior == .Pipe);
debug.assert(child.stderr_behavior == .Pipe);
if (builtin.os.tag == .windows) {
try collectOutputWindows(child, stdout, stderr, max_output_bytes);
} else {
try collectOutputPosix(child, stdout, stderr, max_output_bytes);
}
}
fn collectOutputPosix(
child: ChildProcess,
stdout: *std.ArrayList(u8),
stderr: *std.ArrayList(u8),
max_output_bytes: usize,
) !void {
var poll_fds = [_]os.pollfd{
.{ .fd = child.stdout.?.handle, .events = os.POLL.IN, .revents = undefined },
.{ .fd = child.stderr.?.handle, .events = os.POLL.IN, .revents = undefined },
};
// we could make this work with multiple allocators but YAGNI
if (stdout.allocator.ptr != stderr.allocator.ptr or
stdout.allocator.vtable != stderr.allocator.vtable)
@panic("ChildProcess.collectOutput only supports 1 allocator");
var dead_fds: usize = 0;
// We ask for ensureTotalCapacity with this much extra space. This has more of an
// effect on small reads because once the reads start to get larger the amount
// of space an ArrayList will allocate grows exponentially.
const bump_amt = 512;
var poller = std.io.poll(stdout.allocator, enum { stdout, stderr }, .{
.stdout = child.stdout.?,
.stderr = child.stderr.?,
});
defer poller.deinit();
const err_mask = os.POLL.ERR | os.POLL.NVAL | os.POLL.HUP;
while (dead_fds < poll_fds.len) {
const events = try os.poll(&poll_fds, std.math.maxInt(i32));
if (events == 0) continue;
var remove_stdout = false;
var remove_stderr = false;
// Try reading whatever is available before checking the error
// conditions.
// It's still possible to read after a POLL.HUP is received, always
// check if there's some data waiting to be read first.
if (poll_fds[0].revents & os.POLL.IN != 0) {
// stdout is ready.
const new_capacity = std.math.min(stdout.items.len + bump_amt, max_output_bytes);
try stdout.ensureTotalCapacity(new_capacity);
const buf = stdout.unusedCapacitySlice();
if (buf.len == 0) return error.StdoutStreamTooLong;
const nread = try os.read(poll_fds[0].fd, buf);
stdout.items.len += nread;
// Remove the fd when the EOF condition is met.
remove_stdout = nread == 0;
} else {
remove_stdout = poll_fds[0].revents & err_mask != 0;
}
if (poll_fds[1].revents & os.POLL.IN != 0) {
// stderr is ready.
const new_capacity = std.math.min(stderr.items.len + bump_amt, max_output_bytes);
try stderr.ensureTotalCapacity(new_capacity);
const buf = stderr.unusedCapacitySlice();
if (buf.len == 0) return error.StderrStreamTooLong;
const nread = try os.read(poll_fds[1].fd, buf);
stderr.items.len += nread;
// Remove the fd when the EOF condition is met.
remove_stderr = nread == 0;
} else {
remove_stderr = poll_fds[1].revents & err_mask != 0;
}
// Exclude the fds that signaled an error.
if (remove_stdout) {
poll_fds[0].fd = -1;
dead_fds += 1;
}
if (remove_stderr) {
poll_fds[1].fd = -1;
dead_fds += 1;
}
}
}
const WindowsAsyncReadResult = enum {
pending,
closed,
full,
};
fn windowsAsyncRead(
handle: windows.HANDLE,
overlapped: *windows.OVERLAPPED,
buf: *std.ArrayList(u8),
bump_amt: usize,
max_output_bytes: usize,
) !WindowsAsyncReadResult {
while (true) {
const new_capacity = std.math.min(buf.items.len + bump_amt, max_output_bytes);
try buf.ensureTotalCapacity(new_capacity);
const next_buf = buf.unusedCapacitySlice();
if (next_buf.len == 0) return .full;
var read_bytes: u32 = undefined;
const read_result = windows.kernel32.ReadFile(handle, next_buf.ptr, math.cast(u32, next_buf.len) orelse maxInt(u32), &read_bytes, overlapped);
if (read_result == 0) return switch (windows.kernel32.GetLastError()) {
.IO_PENDING => .pending,
.BROKEN_PIPE => .closed,
else => |err| windows.unexpectedError(err),
};
buf.items.len += read_bytes;
}
}
fn collectOutputWindows(child: ChildProcess, stdout: *std.ArrayList(u8), stderr: *std.ArrayList(u8), max_output_bytes: usize) !void {
const bump_amt = 512;
const outs = [_]*std.ArrayList(u8){
stdout,
stderr,
};
const handles = [_]windows.HANDLE{
child.stdout.?.handle,
child.stderr.?.handle,
};
var overlapped = [_]windows.OVERLAPPED{
mem.zeroes(windows.OVERLAPPED),
mem.zeroes(windows.OVERLAPPED),
};
var wait_objects: [2]windows.HANDLE = undefined;
var wait_object_count: u2 = 0;
// we need to cancel all pending IO before returning so our OVERLAPPED values don't go out of scope
defer for (wait_objects[0..wait_object_count]) |o| {
_ = windows.kernel32.CancelIo(o);
};
// Windows Async IO requires an initial call to ReadFile before waiting on the handle
for ([_]u1{ 0, 1 }) |i| {
switch (try windowsAsyncRead(handles[i], &overlapped[i], outs[i], bump_amt, max_output_bytes)) {
.pending => {
wait_objects[wait_object_count] = handles[i];
wait_object_count += 1;
},
.closed => {}, // don't add to the wait_objects list
.full => return if (i == 0) error.StdoutStreamTooLong else error.StderrStreamTooLong,
}
while (!poller.done()) {
try poller.poll();
if (poller.fifo(.stdout).count > max_output_bytes)
return error.StdoutStreamTooLong;
if (poller.fifo(.stderr).count > max_output_bytes)
return error.StderrStreamTooLong;
}
while (wait_object_count > 0) {
const status = windows.kernel32.WaitForMultipleObjects(wait_object_count, &wait_objects, 0, windows.INFINITE);
if (status == windows.WAIT_FAILED) {
switch (windows.kernel32.GetLastError()) {
else => |err| return windows.unexpectedError(err),
}
}
if (status < windows.WAIT_OBJECT_0 or status > windows.WAIT_OBJECT_0 + wait_object_count - 1)
unreachable;
const wait_idx = status - windows.WAIT_OBJECT_0;
// this extra `i` index is needed to map the wait handle back to the stdout or stderr
// values since the wait_idx can change which handle it corresponds with
const i: u1 = if (wait_objects[wait_idx] == handles[0]) 0 else 1;
// remove completed event from the wait list
wait_object_count -= 1;
if (wait_idx == 0)
wait_objects[0] = wait_objects[1];
var read_bytes: u32 = undefined;
if (windows.kernel32.GetOverlappedResult(handles[i], &overlapped[i], &read_bytes, 0) == 0) {
switch (windows.kernel32.GetLastError()) {
.BROKEN_PIPE => continue,
else => |err| return windows.unexpectedError(err),
}
}
outs[i].items.len += read_bytes;
switch (try windowsAsyncRead(handles[i], &overlapped[i], outs[i], bump_amt, max_output_bytes)) {
.pending => {
wait_objects[wait_object_count] = handles[i];
wait_object_count += 1;
},
.closed => {}, // don't add to the wait_objects list
.full => return if (i == 0) error.StdoutStreamTooLong else error.StderrStreamTooLong,
}
}
stdout.* = fifoToOwnedArrayList(poller.fifo(.stdout));
stderr.* = fifoToOwnedArrayList(poller.fifo(.stderr));
}
/// Spawns a child process, waits for it, collecting stdout and stderr, and then returns.