1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136

// Copyright 2020 The Fuchsia Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.

#![allow(clippy::let_unit_value)]

use {
    crate::{fidl::FidlServer, metadata::put_metadata_in_happy_state, reboot::wait_and_reboot},
    anyhow::{anyhow, Context, Error},
    config::Config,
    fidl_fuchsia_hardware_power_statecontrol::AdminMarker as PowerStateControlMarker,
    fidl_fuchsia_paver::PaverMarker,
    fidl_fuchsia_update_verify::{BlobfsVerifierMarker, NetstackVerifierMarker},
    fuchsia_async as fasync,
    fuchsia_component::{client::connect_to_protocol, server::ServiceFs},
    fuchsia_inspect::{self as finspect, health::Reporter},
    fuchsia_zircon::{self as zx, HandleBased},
    futures::{channel::oneshot, prelude::*, stream::FuturesUnordered},
    std::{
        sync::Arc,
        time::{Duration, Instant},
    },
    tracing::{error, info, warn},
};

mod config;
mod fidl;
mod metadata;
mod reboot;

// The system persists the reboot reason via a component called last_reboot. If we issue a reboot
// before last_reboot starts, the reboot reason will not persist. Since last_reboot is v1 and the
// system-update-committer is v2, and v2 components start earlier than v1 components, let's try to
// minimize this risk by defining a minimum duration we must wait to issue a reboot. Ideally, reboot
// clients should not have to worry about this. However, given the transition from v1 to v2, for now
// we mitigate this with a timer. This value was determined experimentally on Astro, where there
// seems to be a ~2 second gap between the system-update-committer and last_reboot starting.
const MINIMUM_REBOOT_WAIT: Duration = Duration::from_secs(5);

#[fuchsia::main(logging_tags = ["system-update-committer"])]
pub fn main() -> Result<(), Error> {
    info!("starting system-update-committer");

    let mut executor = fasync::LocalExecutor::new();
    let () = executor.run_singlethreaded(main_inner_async()).map_err(|err| {
        // Use anyhow to print the error chain.
        let err = anyhow!(err);
        error!("error running system-update-committer: {:#}", err);
        err
    })?;

    info!("shutting down system-update-committer");
    Ok(())
}

async fn main_inner_async() -> Result<(), Error> {
    let inspector = finspect::Inspector::default();
    let _inspect_server_task =
        inspect_runtime::publish(&inspector, inspect_runtime::PublishOptions::default());

    let verification_node = inspector.root().create_child("verification");
    let mut health_node = finspect::health::Node::new(inspector.root());

    let verification_node_ref = &verification_node;
    let health_node_ref = &mut health_node;

    let config = Config::load_from_config_data_or_default();
    let reboot_deadline = Instant::now() + MINIMUM_REBOOT_WAIT;

    let paver = connect_to_protocol::<PaverMarker>().context("while connecting to paver")?;
    let (boot_manager, boot_manager_server_end) =
        ::fidl::endpoints::create_proxy().context("while creating BootManager endpoints")?;

    paver
        .find_boot_manager(boot_manager_server_end)
        .context("transport error while calling find_boot_manager()")?;

    let reboot_proxy = connect_to_protocol::<PowerStateControlMarker>()
        .context("while connecting to power state control")?;
    let blobfs_verifier = connect_to_protocol::<BlobfsVerifierMarker>()
        .context("while connecting to blobfs verifier")?;
    let netstack_verifier = connect_to_protocol::<NetstackVerifierMarker>()
        .context("while connecting to netstack verifier")?;

    let futures = FuturesUnordered::new();
    let (p_internal, p_external) = zx::EventPair::create();

    // Keep a copy of the internal pair so that external consumers don't observe EVENTPAIR_PEER_CLOSED.
    let _p_internal_clone =
        p_internal.duplicate_handle(zx::Rights::NONE).context("while duplicating p_internal")?;

    let (unblocker, blocker) = oneshot::channel();

    // Handle putting boot metadata in happy state, rebooting on failure (if necessary), and
    // reporting health to the inspect health node.
    futures.push(
        async move {
            if let Err(e) = put_metadata_in_happy_state(
                &boot_manager,
                &p_internal,
                unblocker,
                &[&blobfs_verifier, &netstack_verifier],
                verification_node_ref,
                &config,
            )
            .await
            {
                let msg = format!(
                    "Failed to put metadata in happy state. Rebooting at {:?} given error {:#} and {:?}",
                    reboot_deadline,
                    anyhow!(e),
                    config
                );
                health_node_ref.set_unhealthy(&msg);
                warn!("{}", msg);
                wait_and_reboot(fasync::Timer::new(reboot_deadline), &reboot_proxy).await;
            } else {
                info!("metadata is in happy state!");
                health_node_ref.set_ok();
            }
        }
        .boxed_local(),
    );

    // Handle ServiceFs and inspect
    let mut fs = ServiceFs::new_local();
    fs.take_and_serve_directory_handle().context("while serving directory handle")?;

    // Handle FIDL.
    let fidl = Arc::new(FidlServer::new(p_external, blocker));
    futures.push(FidlServer::run(fidl, fs).boxed_local());

    let () = futures.collect::<()>().await;

    Ok(())
}