platform_system_core/rootdir/init.rc

1387 lines
58 KiB
Text
Raw Normal View History

# Copyright (C) 2012 The Android Open Source Project
#
# IMPORTANT: Do not create world writable files or directories.
# This is a common source of Android security bugs.
#
import /init.environ.rc
import /system/etc/init/hw/init.usb.rc
import /init.${ro.hardware}.rc
import /vendor/etc/init/hw/init.${ro.hardware}.rc
import /system/etc/init/hw/init.usb.configfs.rc
import /system/etc/init/hw/init.${ro.zygote}.rc
# Cgroups are mounted right before early-init using list from /etc/cgroups.json
on early-init
# Disable sysrq from keyboard
write /proc/sys/kernel/sysrq 0
init.rc: disable kernel module autoloading There is a longstanding bug where file-based encryption causes spurious SELinux denials of module_request because it uses the kernel's crypto API, and the crypto API tries to autoload kernel modules. While this sometimes indicate missing kconfig options, it can still happen even if all needed kconfig options are enabled. This is because a crypto algorithm can be a composition like "hmac(sha512)", and the crypto API will first look for the full composition before it instantiates it using the components like "hmac" and "sha512". But often an implementation of the full composition doesn't exist. However, as far as I can tell, Android doesn't actually use kernel module autoloading at all. First, Android never changes /proc/sys/kernel/modprobe from the default of "/sbin/modprobe", yet this isn't where modprobe is located on Android. Android's SELinux policy contains a neverallow rule that ensures that only init (not even vendor_init) can write to this setting, so vendors can't be changing it. Vendors could potentially be setting CONFIG_STATIC_USERMODEHELPER_PATH, which overrides the path of all usermode helpers including modprobe. But this is a relatively new kconfig option, available only in android-4.14 and later. Also, for a vendor to actually do this they'd also need to extend the SELinux policy with a domain_auto_trans rule to allow their usermode helper to be executed by the kernel. Android does increasingly use kernel modules, and GKI (Generic Kernel Image) will require them. However, the modules are actually inserted by userspace by 'init', not autoloaded. It's possible to disable kernel module autoloading completely by setting /proc/sys/kernel/modprobe to an empty string. So, let's do that. This prevents lots of spurious SELinux denials, and allows removing unnecessary rules to allow or dontaudit the module_request permission. Note: when the kernel doesn't have CONFIG_ANDROID_BINDERFS enabled, this change exposes a kernel bug that causes a WARNING in get_fs_type(). To avoid this WARNING, a kernel fix should be applied too -- currently under discussion upstream (https://lkml.kernel.org/r/20200310223731.126894-1-ebiggers@kernel.org). Bug: 130424539 Bug: 132409186 Bug: 144399145 Bug: 146477240 Bug: 148005188 Bug: 149542343 Test: Tested on cuttlefish and coral: - Checked that /proc/sys/kernel/modprobe contains /sbin/modprobe before this change, and the empty string after. - Checked that if all SELinux rules for module_request are removed, there are SELinux denials for module_request before this change but none after. - Ran lsmod both before and after and verified that the list is the same, i.e. checked that this change doesn't break how Android actually loads kernel modules. Change-Id: I4132fe1a491e7b789311afcf693c1f6493fb9dc5
2020-03-11 17:56:15 +01:00
# Android doesn't need kernel module autoloading, and it causes SELinux
# denials. So disable it by setting modprobe to the empty string. Note: to
# explicitly set a sysctl to an empty string, a trailing newline is needed.
write /proc/sys/kernel/modprobe \n
# Set the security context of /adb_keys if present.
restorecon /adb_keys
# Set the security context of /postinstall if present.
restorecon /postinstall
mkdir /acct/uid
# memory.pressure_level used by lmkd
chown root system /dev/memcg/memory.pressure_level
chmod 0040 /dev/memcg/memory.pressure_level
# app mem cgroups, used by activity manager, lmkd and zygote
mkdir /dev/memcg/apps/ 0755 system system
# cgroup for system_server and surfaceflinger
mkdir /dev/memcg/system 0550 system system
# symlink the Android specific /dev/tun to Linux expected /dev/net/tun
mkdir /dev/net 0755 root root
symlink ../tun /dev/net/tun
# set RLIMIT_NICE to allow priorities from 19 to -20
setrlimit nice 40 40
# Allow up to 32K FDs per process
setrlimit nofile 32768 32768
# set RLIMIT_MEMLOCK to 64KB
setrlimit memlock 65536 65536
# Set up linker config subdirectories based on mount namespaces
mkdir /linkerconfig/bootstrap 0755
mkdir /linkerconfig/default 0755
# Disable dm-verity hash prefetching, since it doesn't help performance
# Read more in b/136247322
write /sys/module/dm_verity/parameters/prefetch_cluster 0
# Generate empty ld.config.txt for early executed processes which rely on
# /system/lib libraries.
write /linkerconfig/bootstrap/ld.config.txt \#
write /linkerconfig/default/ld.config.txt \#
chmod 644 /linkerconfig/bootstrap/ld.config.txt
chmod 644 /linkerconfig/default/ld.config.txt
# Mount bootstrap linker configuration as current
mount none /linkerconfig/bootstrap /linkerconfig bind rec
start ueventd
Activate system APEXes early Summary: Boot sequence around apexd is changed to make it possible for pre-apexd processes to use libraries from APEXes. They no longer need to wait for the apexd to finish activating APEXes, which again can be done only after /data/ is mounted. This improves overall boot performance. Detail: This change fixes the problem that processes that are started before apexd (so called pre-apexd processes) can't access libraries that are provided only by the APEXes but are not found in the system partition (e.g. libdexfile_external.so, etc.). Main idea is to activate system APEXes (/system/apex/*.apex) before /data is mounted and then activate the updated APEXes (/data/apex/*.apex) after the /data mount. Detailed boot sequence is as follows. 1) init prepares the bootstrap and default mount namespaces. A tmpfs is mounted on /apex and the propagation type of the mountpoint is set to private. 2) before any other process is started, apexd is started in bootstrap mode. When executed in the mode, apexd only activates APEXes under /system/apex. Note that APEXes activated in this phase are mounted in the bootstrap mount namespace only. 3) other pre-apexd processes are started. They are in the bootstrap mount namespace and thus are provided with the libraries from the system APEXes. 4) /data is mounted. init switches into the default mount namespace and starts apexd as a daemon as usual. 5) apexd scans both /data/apex and /system/apex, and activate latest APEXes from the directories. Note that APEXes activated in this phase are mounted in the default namespaces only and thus are not visible to the pre-apexd processes. Bug: 125549215 Test: m; device boots Change-Id: I21c60d0ebe188fa4f24d6e6861f85ca204843069
2019-02-22 14:15:25 +01:00
# Run apexd-bootstrap so that APEXes that provide critical libraries
# become available. Note that this is executed as exec_start to ensure that
# the libraries are available to the processes started after this statement.
exec_start apexd-bootstrap
perform_apex_config --bootstrap
# These must already exist by the time boringssl_self_test32 / boringssl_self_test64 run.
mkdir /dev/boringssl 0755 root root
mkdir /dev/boringssl/selftest 0755 root root
# Mount tracefs (with GID=AID_READTRACEFS)
mount tracefs tracefs /sys/kernel/tracing gid=3012
# create sys dirctory
mkdir /dev/sys 0755 system system
mkdir /dev/sys/fs 0755 system system
mkdir /dev/sys/block 0755 system system
# Create location for fs_mgr to store abbreviated output from filesystem
# checker programs.
mkdir /dev/fscklogs 0770 root system
# Create tmpfs for use by the shell user.
mount tmpfs tmpfs /tmp
restorecon /tmp
chown shell shell /tmp
chmod 0771 /tmp
on init
sysclktz 0
# Mix device-specific information into the entropy pool
copy /proc/cmdline /dev/urandom
copy /system/etc/prop.default /dev/urandom
symlink /proc/self/fd/0 /dev/stdin
symlink /proc/self/fd/1 /dev/stdout
symlink /proc/self/fd/2 /dev/stderr
# Create socket dir for ot-daemon
mkdir /dev/socket/ot-daemon 0770 thread_network thread_network
# Create energy-aware scheduler tuning nodes
mkdir /dev/stune/foreground
mkdir /dev/stune/background
mkdir /dev/stune/top-app
mkdir /dev/stune/rt
chown system system /dev/stune
chown system system /dev/stune/foreground
chown system system /dev/stune/background
chown system system /dev/stune/top-app
chown system system /dev/stune/rt
chown system system /dev/stune/tasks
chown system system /dev/stune/foreground/tasks
chown system system /dev/stune/background/tasks
chown system system /dev/stune/top-app/tasks
chown system system /dev/stune/rt/tasks
chown system system /dev/stune/cgroup.procs
chown system system /dev/stune/foreground/cgroup.procs
chown system system /dev/stune/background/cgroup.procs
chown system system /dev/stune/top-app/cgroup.procs
chown system system /dev/stune/rt/cgroup.procs
chmod 0664 /dev/stune/tasks
chmod 0664 /dev/stune/foreground/tasks
chmod 0664 /dev/stune/background/tasks
chmod 0664 /dev/stune/top-app/tasks
chmod 0664 /dev/stune/rt/tasks
chmod 0664 /dev/stune/cgroup.procs
chmod 0664 /dev/stune/foreground/cgroup.procs
chmod 0664 /dev/stune/background/cgroup.procs
chmod 0664 /dev/stune/top-app/cgroup.procs
chmod 0664 /dev/stune/rt/cgroup.procs
# cpuctl hierarchy for devices using utilclamp
mkdir /dev/cpuctl/foreground
mkdir /dev/cpuctl/background
mkdir /dev/cpuctl/top-app
mkdir /dev/cpuctl/rt
mkdir /dev/cpuctl/system
mkdir /dev/cpuctl/system-background
mkdir /dev/cpuctl/dex2oat
chown system system /dev/cpuctl
chown system system /dev/cpuctl/foreground
chown system system /dev/cpuctl/background
chown system system /dev/cpuctl/top-app
chown system system /dev/cpuctl/rt
chown system system /dev/cpuctl/system
chown system system /dev/cpuctl/system-background
chown system system /dev/cpuctl/dex2oat
chown system system /dev/cpuctl/tasks
chown system system /dev/cpuctl/foreground/tasks
chown system system /dev/cpuctl/background/tasks
chown system system /dev/cpuctl/top-app/tasks
chown system system /dev/cpuctl/rt/tasks
chown system system /dev/cpuctl/system/tasks
chown system system /dev/cpuctl/system-background/tasks
chown system system /dev/cpuctl/dex2oat/tasks
chown system system /dev/cpuctl/cgroup.procs
chown system system /dev/cpuctl/foreground/cgroup.procs
chown system system /dev/cpuctl/background/cgroup.procs
chown system system /dev/cpuctl/top-app/cgroup.procs
chown system system /dev/cpuctl/rt/cgroup.procs
chown system system /dev/cpuctl/system/cgroup.procs
chown system system /dev/cpuctl/system-background/cgroup.procs
chown system system /dev/cpuctl/dex2oat/cgroup.procs
chmod 0664 /dev/cpuctl/tasks
chmod 0664 /dev/cpuctl/foreground/tasks
chmod 0664 /dev/cpuctl/background/tasks
chmod 0664 /dev/cpuctl/top-app/tasks
chmod 0664 /dev/cpuctl/rt/tasks
chmod 0664 /dev/cpuctl/system/tasks
chmod 0664 /dev/cpuctl/system-background/tasks
chmod 0664 /dev/cpuctl/dex2oat/tasks
chmod 0664 /dev/cpuctl/cgroup.procs
chmod 0664 /dev/cpuctl/foreground/cgroup.procs
chmod 0664 /dev/cpuctl/background/cgroup.procs
chmod 0664 /dev/cpuctl/top-app/cgroup.procs
chmod 0664 /dev/cpuctl/rt/cgroup.procs
chmod 0664 /dev/cpuctl/system/cgroup.procs
chmod 0664 /dev/cpuctl/system-background/cgroup.procs
chmod 0664 /dev/cpuctl/dex2oat/cgroup.procs
# Create a cpu group for NNAPI HAL processes
mkdir /dev/cpuctl/nnapi-hal
chown system system /dev/cpuctl/nnapi-hal
chown system system /dev/cpuctl/nnapi-hal/tasks
chown system system /dev/cpuctl/nnapi-hal/cgroup.procs
chmod 0664 /dev/cpuctl/nnapi-hal/tasks
chmod 0664 /dev/cpuctl/nnapi-hal/cgroup.procs
write /dev/cpuctl/nnapi-hal/cpu.uclamp.min 1
write /dev/cpuctl/nnapi-hal/cpu.uclamp.latency_sensitive 1
# Create a cpu group for camera daemon processes
mkdir /dev/cpuctl/camera-daemon
chown system system /dev/cpuctl/camera-daemon
chown system system /dev/cpuctl/camera-daemon/tasks
chown system system /dev/cpuctl/camera-daemon/cgroup.procs
chmod 0664 /dev/cpuctl/camera-daemon/tasks
chmod 0664 /dev/cpuctl/camera-daemon/cgroup.procs
# Create an stune group for camera-specific processes
mkdir /dev/stune/camera-daemon
chown system system /dev/stune/camera-daemon
chown system system /dev/stune/camera-daemon/tasks
chown system system /dev/stune/camera-daemon/cgroup.procs
chmod 0664 /dev/stune/camera-daemon/tasks
chmod 0664 /dev/stune/camera-daemon/cgroup.procs
# Create an stune group for NNAPI HAL processes
mkdir /dev/stune/nnapi-hal
chown system system /dev/stune/nnapi-hal
chown system system /dev/stune/nnapi-hal/tasks
chown system system /dev/stune/nnapi-hal/cgroup.procs
chmod 0664 /dev/stune/nnapi-hal/tasks
chmod 0664 /dev/stune/nnapi-hal/cgroup.procs
write /dev/stune/nnapi-hal/schedtune.boost 1
write /dev/stune/nnapi-hal/schedtune.prefer_idle 1
# Create blkio group and apply initial settings.
# This feature needs kernel to support it, and the
# device's init.rc must actually set the correct values.
mkdir /dev/blkio/background
chown system system /dev/blkio
chown system system /dev/blkio/background
chown system system /dev/blkio/tasks
chown system system /dev/blkio/background/tasks
chown system system /dev/blkio/cgroup.procs
chown system system /dev/blkio/background/cgroup.procs
chmod 0664 /dev/blkio/tasks
chmod 0664 /dev/blkio/background/tasks
chmod 0664 /dev/blkio/cgroup.procs
chmod 0664 /dev/blkio/background/cgroup.procs
write /dev/blkio/blkio.weight 1000
write /dev/blkio/background/blkio.weight 200
write /dev/blkio/background/blkio.bfq.weight 10
write /dev/blkio/blkio.group_idle 0
write /dev/blkio/background/blkio.group_idle 0
write /dev/blkio/background/blkio.prio.class restrict-to-be
restorecon_recursive /mnt
mount configfs none /config nodev noexec nosuid
chmod 0770 /config/sdcardfs
chown system package_info /config/sdcardfs
# Mount binderfs
mkdir /dev/binderfs
mount binder binder /dev/binderfs stats=global
chmod 0755 /dev/binderfs
# Mount fusectl
mount fusectl none /sys/fs/fuse/connections
symlink /dev/binderfs/binder /dev/binder
symlink /dev/binderfs/hwbinder /dev/hwbinder
symlink /dev/binderfs/vndbinder /dev/vndbinder
chmod 0666 /dev/binderfs/hwbinder
chmod 0666 /dev/binderfs/binder
chmod 0666 /dev/binderfs/vndbinder
mkdir /mnt/secure 0700 root root
mkdir /mnt/secure/asec 0700 root root
mkdir /mnt/asec 0755 root system
mkdir /mnt/obb 0755 root system
mkdir /mnt/media_rw 0750 root external_storage
mkdir /mnt/user 0755 root root
mkdir /mnt/user/0 0755 root root
mkdir /mnt/user/0/self 0755 root root
mkdir /mnt/user/0/emulated 0755 root root
mkdir /mnt/user/0/emulated/0 0755 root root
# Prepare directories for pass through processes
mkdir /mnt/pass_through 0700 root root
mkdir /mnt/pass_through/0 0710 root media_rw
mkdir /mnt/pass_through/0/self 0710 root media_rw
mkdir /mnt/pass_through/0/emulated 0710 root media_rw
mkdir /mnt/pass_through/0/emulated/0 0710 root media_rw
mkdir /mnt/expand 0771 system system
mkdir /mnt/appfuse 0711 root root
# Storage views to support runtime permissions
mkdir /mnt/runtime 0700 root root
mkdir /mnt/runtime/default 0755 root root
mkdir /mnt/runtime/default/self 0755 root root
mkdir /mnt/runtime/read 0755 root root
mkdir /mnt/runtime/read/self 0755 root root
mkdir /mnt/runtime/write 0755 root root
mkdir /mnt/runtime/write/self 0755 root root
mkdir /mnt/runtime/full 0755 root root
mkdir /mnt/runtime/full/self 0755 root root
# For Pre-reboot Dexopt
mkdir /mnt/pre_reboot_dexopt 0755 artd artd
# Symlink to keep legacy apps working in multi-user world
symlink /storage/self/primary /mnt/sdcard
symlink /mnt/user/0/primary /mnt/runtime/default/self/primary
write /proc/sys/kernel/panic_on_oops 1
write /proc/sys/kernel/hung_task_timeout_secs 0
write /proc/cpu/alignment 4
# scheduler tunables
# Disable auto-scaling of scheduler tunables with hotplug. The tunables
# will vary across devices in unpredictable ways if allowed to scale with
# cpu cores.
write /proc/sys/kernel/sched_tunable_scaling 0
write /proc/sys/kernel/sched_latency_ns 10000000
write /proc/sys/kernel/sched_wakeup_granularity_ns 2000000
write /proc/sys/kernel/sched_child_runs_first 0
write /proc/sys/kernel/randomize_va_space 2
write /proc/sys/vm/mmap_min_addr 32768
write /proc/sys/net/ipv4/ping_group_range "0 2147483647"
write /proc/sys/net/unix/max_dgram_qlen 2400
# Assign reasonable ceiling values for socket rcv/snd buffers.
# These should almost always be overridden by the target per the
# the corresponding technology maximums.
write /proc/sys/net/core/rmem_max 262144
write /proc/sys/net/core/wmem_max 262144
# reflect fwmark from incoming packets onto generated replies
write /proc/sys/net/ipv4/fwmark_reflect 1
write /proc/sys/net/ipv6/fwmark_reflect 1
# set fwmark on accepted sockets
write /proc/sys/net/ipv4/tcp_fwmark_accept 1
# disable icmp redirects
write /proc/sys/net/ipv4/conf/all/accept_redirects 0
write /proc/sys/net/ipv6/conf/all/accept_redirects 0
# /proc/net/fib_trie leaks interface IP addresses
chmod 0400 /proc/net/fib_trie
# sets up initial cpusets for ActivityManager
# this ensures that the cpusets are present and usable, but the device's
# init.rc must actually set the correct cpus
mkdir /dev/cpuset/foreground
copy /dev/cpuset/cpus /dev/cpuset/foreground/cpus
copy /dev/cpuset/mems /dev/cpuset/foreground/mems
mkdir /dev/cpuset/background
copy /dev/cpuset/cpus /dev/cpuset/background/cpus
copy /dev/cpuset/mems /dev/cpuset/background/mems
# system-background is for system tasks that should only run on
# little cores, not on bigs
mkdir /dev/cpuset/system-background
copy /dev/cpuset/cpus /dev/cpuset/system-background/cpus
copy /dev/cpuset/mems /dev/cpuset/system-background/mems
# restricted is for system tasks that are being throttled
# due to screen off.
mkdir /dev/cpuset/restricted
copy /dev/cpuset/cpus /dev/cpuset/restricted/cpus
copy /dev/cpuset/mems /dev/cpuset/restricted/mems
mkdir /dev/cpuset/top-app
copy /dev/cpuset/cpus /dev/cpuset/top-app/cpus
copy /dev/cpuset/mems /dev/cpuset/top-app/mems
# create a cpuset for camera daemon processes
mkdir /dev/cpuset/camera-daemon
copy /dev/cpuset/cpus /dev/cpuset/camera-daemon/cpus
copy /dev/cpuset/mems /dev/cpuset/camera-daemon/mems
# change permissions for all cpusets we'll touch at runtime
chown system system /dev/cpuset
chown system system /dev/cpuset/foreground
chown system system /dev/cpuset/background
chown system system /dev/cpuset/system-background
chown system system /dev/cpuset/top-app
chown system system /dev/cpuset/restricted
chown system system /dev/cpuset/camera-daemon
chown system system /dev/cpuset/tasks
chown system system /dev/cpuset/foreground/tasks
chown system system /dev/cpuset/background/tasks
chown system system /dev/cpuset/system-background/tasks
chown system system /dev/cpuset/top-app/tasks
chown system system /dev/cpuset/restricted/tasks
chown system system /dev/cpuset/camera-daemon/tasks
chown system system /dev/cpuset/cgroup.procs
chown system system /dev/cpuset/foreground/cgroup.procs
chown system system /dev/cpuset/background/cgroup.procs
chown system system /dev/cpuset/system-background/cgroup.procs
chown system system /dev/cpuset/top-app/cgroup.procs
chown system system /dev/cpuset/restricted/cgroup.procs
chown system system /dev/cpuset/camera-daemon/cgroup.procs
# set system-background to 0775 so SurfaceFlinger can touch it
chmod 0775 /dev/cpuset/system-background
chmod 0664 /dev/cpuset/foreground/tasks
chmod 0664 /dev/cpuset/background/tasks
chmod 0664 /dev/cpuset/system-background/tasks
chmod 0664 /dev/cpuset/top-app/tasks
chmod 0664 /dev/cpuset/restricted/tasks
chmod 0664 /dev/cpuset/tasks
chmod 0664 /dev/cpuset/camera-daemon/tasks
chmod 0664 /dev/cpuset/foreground/cgroup.procs
chmod 0664 /dev/cpuset/background/cgroup.procs
chmod 0664 /dev/cpuset/system-background/cgroup.procs
chmod 0664 /dev/cpuset/top-app/cgroup.procs
chmod 0664 /dev/cpuset/restricted/cgroup.procs
chmod 0664 /dev/cpuset/cgroup.procs
chmod 0664 /dev/cpuset/camera-daemon/cgroup.procs
# make the PSI monitor accessible to others
chown system system /proc/pressure/memory
chmod 0664 /proc/pressure/memory
mount bpf bpf /sys/fs/bpf nodev noexec nosuid
# pstore/ramoops previous console log
mount pstore pstore /sys/fs/pstore nodev noexec nosuid
chown system log /sys/fs/pstore
chmod 0550 /sys/fs/pstore
chown system log /sys/fs/pstore/console-ramoops
chmod 0440 /sys/fs/pstore/console-ramoops
chown system log /sys/fs/pstore/console-ramoops-0
chmod 0440 /sys/fs/pstore/console-ramoops-0
chown system log /sys/fs/pstore/pmsg-ramoops-0
chmod 0440 /sys/fs/pstore/pmsg-ramoops-0
# enable armv8_deprecated instruction hooks
write /proc/sys/abi/swp 1
# Linux's execveat() syscall may construct paths containing /dev/fd
# expecting it to point to /proc/self/fd
symlink /proc/self/fd /dev/fd
export DOWNLOAD_CACHE /data/cache
# This allows the ledtrig-transient properties to be created here so
# that they can be chown'd to system:system later on boot
write /sys/class/leds/vibrator/trigger "transient"
# This is used by Bionic to select optimized routines.
write /dev/cpu_variant:${ro.bionic.arch} ${ro.bionic.cpu_variant}
chmod 0444 /dev/cpu_variant:${ro.bionic.arch}
write /dev/cpu_variant:${ro.bionic.2nd_arch} ${ro.bionic.2nd_cpu_variant}
chmod 0444 /dev/cpu_variant:${ro.bionic.2nd_arch}
# Allow system processes to read / write power state.
chown system system /sys/power/state
chown system system /sys/power/wakeup_count
chmod 0660 /sys/power/state
chown radio wakelock /sys/power/wake_lock
chown radio wakelock /sys/power/wake_unlock
chmod 0660 /sys/power/wake_lock
chmod 0660 /sys/power/wake_unlock
# Start logd before any other services run to ensure we capture all of their logs.
start logd
# Start lmkd before any other services run so that it can register them
write /proc/sys/vm/watermark_boost_factor 0
chown root system /sys/module/lowmemorykiller/parameters/adj
chmod 0664 /sys/module/lowmemorykiller/parameters/adj
chown root system /sys/module/lowmemorykiller/parameters/minfree
chmod 0664 /sys/module/lowmemorykiller/parameters/minfree
start lmkd
# Start essential services.
start servicemanager
start hwservicemanager
start vndservicemanager
# Run boringssl self test for each ABI. Any failures trigger reboot to firmware.
import /system/etc/init/hw/init.boringssl.${ro.zygote}.rc
service boringssl_self_test32 /system/bin/boringssl_self_test32
reboot_on_failure reboot,boringssl-self-check-failed
stdio_to_kmsg
# Explicitly specify that boringssl_self_test32 doesn't require any capabilities
capabilities
user nobody
service boringssl_self_test64 /system/bin/boringssl_self_test64
reboot_on_failure reboot,boringssl-self-check-failed
stdio_to_kmsg
# Explicitly specify that boringssl_self_test64 doesn't require any capabilities
capabilities
user nobody
service boringssl_self_test_apex32 /apex/com.android.conscrypt/bin/boringssl_self_test32
reboot_on_failure reboot,boringssl-self-check-failed
stdio_to_kmsg
# Explicitly specify that boringssl_self_test_apex32 doesn't require any capabilities
capabilities
user nobody
service boringssl_self_test_apex64 /apex/com.android.conscrypt/bin/boringssl_self_test64
reboot_on_failure reboot,boringssl-self-check-failed
stdio_to_kmsg
# Explicitly specify that boringssl_self_test_apex64 doesn't require any capabilities
capabilities
user nobody
# Healthd can trigger a full boot from charger mode by signaling this
# property when the power button is held.
on property:sys.boot_from_charger_mode=1
class_stop charger
trigger late-init
# Indicate to fw loaders that the relevant mounts are up.
on firmware_mounts_complete
rm /dev/.booting
# Mount filesystems and start core system services.
on late-init
trigger early-fs
# Mount fstab in init.{$device}.rc by mount_all command. Optional parameter
# '--early' can be specified to skip entries with 'latemount'.
# /system and /vendor must be mounted by the end of the fs stage,
# while /data is optional.
trigger fs
trigger post-fs
# Mount fstab in init.{$device}.rc by mount_all with '--late' parameter
# to only mount entries with 'latemount'. This is needed if '--early' is
# specified in the previous mount_all command on the fs stage.
# With /system mounted and properties form /system + /factory available,
# some services can be started.
trigger late-fs
# Now we can mount /data. File encryption requires keymaster to decrypt
# /data, which in turn can only be loaded when system properties are present.
trigger post-fs-data
# Should be before netd, but after apex, properties and logging is available.
trigger load_bpf_programs
# Now we can start zygote.
trigger zygote-start
# Remove a file to wake up anything waiting for firmware.
trigger firmware_mounts_complete
trigger early-boot
trigger boot
on early-fs
# Once metadata has been mounted, we'll need vold to deal with userdata checkpointing
start vold
on post-fs
exec - system system -- /system/bin/vdc checkpoint markBootAttempt
# Once everything is setup, no need to modify /.
# The bind+remount combination allows this to work in containers.
mount rootfs rootfs / remount bind ro nodev
# Mount default storage into root namespace
mount none /mnt/user/0 /storage bind rec
mount none none /storage slave rec
# Make sure /sys/kernel/debug (if present) is labeled properly
# Note that tracefs may be mounted under debug, so we need to cross filesystems
restorecon --recursive --cross-filesystems /sys/kernel/debug
# We chown/chmod /cache again so because mount is run as root + defaults
chown system cache /cache
chmod 0770 /cache
# We restorecon /cache in case the cache partition has been reset.
restorecon_recursive /cache
# Create /cache/recovery in case it's not there. It'll also fix the odd
# permissions if created by the recovery system.
mkdir /cache/recovery 0770 system cache
# Backup/restore mechanism uses the cache partition
mkdir /cache/backup_stage 0700 system system
mkdir /cache/backup 0700 system system
#change permissions on vmallocinfo so we can grab it from bugreports
chown root log /proc/vmallocinfo
chmod 0440 /proc/vmallocinfo
chown root log /proc/slabinfo
chmod 0440 /proc/slabinfo
chown root log /proc/pagetypeinfo
chmod 0440 /proc/pagetypeinfo
#change permissions on kmsg & sysrq-trigger so bugreports can grab kthread stacks
chown root system /proc/kmsg
chmod 0440 /proc/kmsg
chown root system /proc/sysrq-trigger
chmod 0220 /proc/sysrq-trigger
chown system log /proc/last_kmsg
chmod 0440 /proc/last_kmsg
# make the selinux kernel policy world-readable
chmod 0444 /sys/fs/selinux/policy
# create the lost+found directories, so as to enforce our permissions
mkdir /cache/lost+found 0770 root root
restorecon_recursive /metadata
mkdir /metadata/vold
chmod 0700 /metadata/vold
mkdir /metadata/password_slots 0771 root system
mkdir /metadata/bootstat 0750 system log
mkdir /metadata/ota 0750 root system
mkdir /metadata/ota/snapshots 0750 root system
mkdir /metadata/userspacereboot 0770 root system
mkdir /metadata/watchdog 0770 root system
mkdir /metadata/apex 0700 root system
mkdir /metadata/apex/sessions 0700 root system
# On some devices we see a weird behaviour in which /metadata/apex doesn't
# have a correct label. To workaround this bug, explicitly call restorecon
# on /metadata/apex. For most of the boot sequences /metadata/apex will
# already have a correct selinux label, meaning that this call will be a
# no-op.
restorecon_recursive /metadata/apex
mkdir /metadata/staged-install 0770 root system
mkdir /metadata/aconfig 0775 root system
mkdir /metadata/aconfig/flags 0770 root system
mkdir /metadata/aconfig/maps 0775 root system
mkdir /metadata/aconfig/boot 0775 root system
mkdir /metadata/aconfig_test_missions 0775 root system
exec_start aconfigd-platform-init
on late-fs
# Ensure that tracefs has the correct permissions.
# This does not work correctly if it is called in post-fs.
chmod 0755 /sys/kernel/tracing
chmod 0755 /sys/kernel/debug/tracing
# HALs required before storage encryption can get unlocked (FBE)
class_start early_hal
# Load trusted keys from dm-verity protected partitions
exec -- /system/bin/fsverity_init --load-verified-keys
# Only enable the bootreceiver tracing instance for kernels 5.10 and above.
on late-fs && property:ro.kernel.version=4.19
setprop bootreceiver.enable 0
on late-fs && property:ro.kernel.version=5.4
setprop bootreceiver.enable 0
on late-fs
# Bootreceiver tracing instance is enabled by default.
setprop bootreceiver.enable ${bootreceiver.enable:-1}
on property:ro.product.cpu.abilist64=* && property:bootreceiver.enable=1
# Set up a tracing instance for system_server to monitor error_report_end events.
# These are sent by kernel tools like KASAN and KFENCE when a memory corruption
# is detected. This is only needed for 64-bit systems.
mkdir /sys/kernel/tracing/instances/bootreceiver 0700 system system
restorecon_recursive /sys/kernel/tracing/instances/bootreceiver
write /sys/kernel/tracing/instances/bootreceiver/buffer_size_kb 1
write /sys/kernel/tracing/instances/bootreceiver/trace_options disable_on_free
write /sys/kernel/tracing/instances/bootreceiver/events/error_report/error_report_end/enable 1
on post-fs-data
Support for stopping/starting post-data-mount class subsets. On devices that use FDE and APEX at the same time, we need to bring up a minimal framework to be able to mount the /data partition. During this period, a tmpfs /data filesystem is created, which doesn't contain any of the updated APEXEs. As a consequence, all those processes will be using the APEXes from the /system partition. This is obviously not desired, as APEXes in /system may be old and/or contain security issues. Additionally, it would create a difference between FBE and FDE devices at runtime. Ideally, we restart all processes that have started after we created the tmpfs /data. We can't (re)start based on class names alone, because some classes (eg 'hal') contain services that are required to start apexd itself and that shouldn't be killed (eg the graphics HAL). To address this, keep track of which processes are started after /data is mounted, with a new 'mark_post_data' keyword. Additionally, create 'class_reset_post_data', which resets all services in the class that were created after the initial /data mount, and 'class_start_post_data', which starts all services in the class that were started after /data was mounted. On a device with FBE, these keywords wouldn't be used; on a device with FDE, we'd use them to bring down the right processes after the user has entered the correct secret, and restart them. Bug: 118485723 Test: manually verified process list Change-Id: I16adb776dacf1dd1feeaff9e60639b99899905eb
2019-04-23 16:26:01 +02:00
mark_post_data
# Start checkpoint before we touch data
exec - system system -- /system/bin/vdc checkpoint prepareCheckpoint
# We chown/chmod /data again so because mount is run as root + defaults
chown system system /data
chmod 0771 /data
# We restorecon /data in case the userdata partition has been reset.
restorecon /data
# Make sure we have the device encryption key.
installkey /data
# Start bootcharting as soon as possible after the data partition is
# mounted to collect more data.
mkdir /data/bootchart 0755 shell shell encryption=Require
bootchart start
# Avoid predictable entropy pool. Carry over entropy from previous boot.
copy /data/system/entropy.dat /dev/urandom
mkdir /data/vendor 0771 root root encryption=Require
mkdir /data/vendor/hardware 0771 root root
# Start tombstoned early to be able to store tombstones.
mkdir /data/anr 0775 system system encryption=Require
mkdir /data/tombstones 0775 system system encryption=Require
mkdir /data/vendor/tombstones 0771 root root
mkdir /data/vendor/tombstones/wifi 0771 wifi wifi
start tombstoned
# Make sure that apexd is started in the default namespace
enter_default_mount_ns
# set up keystore directory structure first so that we can end early boot
# and start apexd
mkdir /data/misc 01771 system misc encryption=Require
mkdir /data/misc/keystore 0700 keystore keystore
# work around b/183668221
restorecon /data/misc /data/misc/keystore
# Boot level 30
# odsign signing keys have MAX_BOOT_LEVEL=30
# This is currently the earliest boot level, but we start at 30
# to leave room for earlier levels.
setprop keystore.boot_level 30
# Now that /data is mounted and we have created /data/misc/keystore,
# we can tell keystore to stop allowing use of early-boot keys,
# and access its database for the first time to support creation and
# use of MAX_BOOT_LEVEL keys.
exec - system system -- /system/bin/vdc keymaster earlyBootEnded
# Multi-installed APEXes are selected using persist props.
# Load persist properties and override properties (if enabled) from /data,
# before starting apexd.
# /data/property should be created before `load_persist_props`
mkdir /data/property 0700 root root encryption=Require
load_persist_props
start logd
start logd-reinit
# Some existing vendor rc files use 'on load_persist_props_action' to know
# when persist props are ready. These are difficult to change due to GRF,
# so continue triggering this action here even though props are already loaded
# by the 'load_persist_props' call above.
trigger load_persist_props_action
Activate system APEXes early Summary: Boot sequence around apexd is changed to make it possible for pre-apexd processes to use libraries from APEXes. They no longer need to wait for the apexd to finish activating APEXes, which again can be done only after /data/ is mounted. This improves overall boot performance. Detail: This change fixes the problem that processes that are started before apexd (so called pre-apexd processes) can't access libraries that are provided only by the APEXes but are not found in the system partition (e.g. libdexfile_external.so, etc.). Main idea is to activate system APEXes (/system/apex/*.apex) before /data is mounted and then activate the updated APEXes (/data/apex/*.apex) after the /data mount. Detailed boot sequence is as follows. 1) init prepares the bootstrap and default mount namespaces. A tmpfs is mounted on /apex and the propagation type of the mountpoint is set to private. 2) before any other process is started, apexd is started in bootstrap mode. When executed in the mode, apexd only activates APEXes under /system/apex. Note that APEXes activated in this phase are mounted in the bootstrap mount namespace only. 3) other pre-apexd processes are started. They are in the bootstrap mount namespace and thus are provided with the libraries from the system APEXes. 4) /data is mounted. init switches into the default mount namespace and starts apexd as a daemon as usual. 5) apexd scans both /data/apex and /system/apex, and activate latest APEXes from the directories. Note that APEXes activated in this phase are mounted in the default namespaces only and thus are not visible to the pre-apexd processes. Bug: 125549215 Test: m; device boots Change-Id: I21c60d0ebe188fa4f24d6e6861f85ca204843069
2019-02-22 14:15:25 +01:00
# /data/apex is now available. Start apexd to scan and activate APEXes.
init.rc: fix (re)starting apexd on FDE devices On first boot, FDE devices hang on the command 'wait_for_prop apexd.status activated'. This is because apexd was already started with the tmpfs /data, then was stopped by vold.decrypt=trigger_shutdown_framework. Then when apexd is started again with the real /data, it sees that apexd.status="ready" already, so it doesn't consider itself to be starting from scratch again. So it doesn't move apexd.status back to "activated" as expected. Fix the above by resetting apexd.status to its initial value of the empty string before trying to start apexd in the post-fs-data trigger. Note that this also takes care of the userspace reboot case which was previously handled in the userspace-reboot-requested trigger. Also, FDE devices hang at the same place on non-first boots with default encryption (i.e., when no PIN is set) because apexd is still running after having been started with the tmpfs /data. This is because vold.decrypt=trigger_shutdown_framework isn't run in that case, but rather vold manually kills processes that have open files on /data -- which doesn't include apexd. But, apexd should be restarted too. Fix that by using 'restart apexd' rather than 'start apexd'. Note that these changes are needed even though FDE devices don't support updatable APEXes, as apexd is needed regardless. This is one of a set of changes that is needed to get FDE working again so that devices that launched with FDE can be upgraded to Android 12. Bug: 186165644 Test: Tested FDE on Cuttlefish. Also tested userspace reboot (with FBE) Change-Id: I4fa57cf15d77b64d1167eaf966347d2a9d6a9b72
2021-04-28 19:15:38 +02:00
#
# To handle userspace reboots, make sure that apexd is started cleanly here
# (set apexd.status="") and that it is restarted if it's already running.
#
# /data/apex uses encryption=None because direct I/O support is needed on
# APEX files, but some devices don't support direct I/O on encrypted files.
# Also, APEXes are public information, similar to the system image.
# /data/apex/decompressed and /data/apex/ota_reserved override this setting;
# they are encrypted so that files in them can be hard-linked into
# /data/rollback which is encrypted.
mkdir /data/apex 0755 root system encryption=None
mkdir /data/apex/active 0755 root system
Activate system APEXes early Summary: Boot sequence around apexd is changed to make it possible for pre-apexd processes to use libraries from APEXes. They no longer need to wait for the apexd to finish activating APEXes, which again can be done only after /data/ is mounted. This improves overall boot performance. Detail: This change fixes the problem that processes that are started before apexd (so called pre-apexd processes) can't access libraries that are provided only by the APEXes but are not found in the system partition (e.g. libdexfile_external.so, etc.). Main idea is to activate system APEXes (/system/apex/*.apex) before /data is mounted and then activate the updated APEXes (/data/apex/*.apex) after the /data mount. Detailed boot sequence is as follows. 1) init prepares the bootstrap and default mount namespaces. A tmpfs is mounted on /apex and the propagation type of the mountpoint is set to private. 2) before any other process is started, apexd is started in bootstrap mode. When executed in the mode, apexd only activates APEXes under /system/apex. Note that APEXes activated in this phase are mounted in the bootstrap mount namespace only. 3) other pre-apexd processes are started. They are in the bootstrap mount namespace and thus are provided with the libraries from the system APEXes. 4) /data is mounted. init switches into the default mount namespace and starts apexd as a daemon as usual. 5) apexd scans both /data/apex and /system/apex, and activate latest APEXes from the directories. Note that APEXes activated in this phase are mounted in the default namespaces only and thus are not visible to the pre-apexd processes. Bug: 125549215 Test: m; device boots Change-Id: I21c60d0ebe188fa4f24d6e6861f85ca204843069
2019-02-22 14:15:25 +01:00
mkdir /data/apex/backup 0700 root system
mkdir /data/apex/decompressed 0755 root system encryption=Require
mkdir /data/apex/hashtree 0700 root system
Activate system APEXes early Summary: Boot sequence around apexd is changed to make it possible for pre-apexd processes to use libraries from APEXes. They no longer need to wait for the apexd to finish activating APEXes, which again can be done only after /data/ is mounted. This improves overall boot performance. Detail: This change fixes the problem that processes that are started before apexd (so called pre-apexd processes) can't access libraries that are provided only by the APEXes but are not found in the system partition (e.g. libdexfile_external.so, etc.). Main idea is to activate system APEXes (/system/apex/*.apex) before /data is mounted and then activate the updated APEXes (/data/apex/*.apex) after the /data mount. Detailed boot sequence is as follows. 1) init prepares the bootstrap and default mount namespaces. A tmpfs is mounted on /apex and the propagation type of the mountpoint is set to private. 2) before any other process is started, apexd is started in bootstrap mode. When executed in the mode, apexd only activates APEXes under /system/apex. Note that APEXes activated in this phase are mounted in the bootstrap mount namespace only. 3) other pre-apexd processes are started. They are in the bootstrap mount namespace and thus are provided with the libraries from the system APEXes. 4) /data is mounted. init switches into the default mount namespace and starts apexd as a daemon as usual. 5) apexd scans both /data/apex and /system/apex, and activate latest APEXes from the directories. Note that APEXes activated in this phase are mounted in the default namespaces only and thus are not visible to the pre-apexd processes. Bug: 125549215 Test: m; device boots Change-Id: I21c60d0ebe188fa4f24d6e6861f85ca204843069
2019-02-22 14:15:25 +01:00
mkdir /data/apex/sessions 0700 root system
mkdir /data/app-staging 0751 system system encryption=DeleteIfNecessary
mkdir /data/apex/ota_reserved 0700 root system encryption=Require
init.rc: fix (re)starting apexd on FDE devices On first boot, FDE devices hang on the command 'wait_for_prop apexd.status activated'. This is because apexd was already started with the tmpfs /data, then was stopped by vold.decrypt=trigger_shutdown_framework. Then when apexd is started again with the real /data, it sees that apexd.status="ready" already, so it doesn't consider itself to be starting from scratch again. So it doesn't move apexd.status back to "activated" as expected. Fix the above by resetting apexd.status to its initial value of the empty string before trying to start apexd in the post-fs-data trigger. Note that this also takes care of the userspace reboot case which was previously handled in the userspace-reboot-requested trigger. Also, FDE devices hang at the same place on non-first boots with default encryption (i.e., when no PIN is set) because apexd is still running after having been started with the tmpfs /data. This is because vold.decrypt=trigger_shutdown_framework isn't run in that case, but rather vold manually kills processes that have open files on /data -- which doesn't include apexd. But, apexd should be restarted too. Fix that by using 'restart apexd' rather than 'start apexd'. Note that these changes are needed even though FDE devices don't support updatable APEXes, as apexd is needed regardless. This is one of a set of changes that is needed to get FDE working again so that devices that launched with FDE can be upgraded to Android 12. Bug: 186165644 Test: Tested FDE on Cuttlefish. Also tested userspace reboot (with FBE) Change-Id: I4fa57cf15d77b64d1167eaf966347d2a9d6a9b72
2021-04-28 19:15:38 +02:00
setprop apexd.status ""
restart apexd
# create rest of basic filesystem structure
mkdir /data/misc/recovery 0770 system log
copy /data/misc/recovery/ro.build.fingerprint /data/misc/recovery/ro.build.fingerprint.1
chmod 0440 /data/misc/recovery/ro.build.fingerprint.1
chown system log /data/misc/recovery/ro.build.fingerprint.1
write /data/misc/recovery/ro.build.fingerprint ${ro.build.fingerprint}
chmod 0440 /data/misc/recovery/ro.build.fingerprint
chown system log /data/misc/recovery/ro.build.fingerprint
mkdir /data/misc/recovery/proc 0770 system log
copy /data/misc/recovery/proc/version /data/misc/recovery/proc/version.1
chmod 0440 /data/misc/recovery/proc/version.1
chown system log /data/misc/recovery/proc/version.1
copy /proc/version /data/misc/recovery/proc/version
chmod 0440 /data/misc/recovery/proc/version
chown system log /data/misc/recovery/proc/version
mkdir /data/misc/bluedroid 02770 bluetooth bluetooth
# Fix the access permissions and group ownership for 'bt_config.conf'
chmod 0660 /data/misc/bluedroid/bt_config.conf
chown bluetooth bluetooth /data/misc/bluedroid/bt_config.conf
mkdir /data/misc/bluetooth 0770 bluetooth bluetooth
mkdir /data/misc/bluetooth/logs 0770 bluetooth bluetooth
mkdir /data/misc/nfc 0770 nfc nfc
mkdir /data/misc/nfc/logs 0770 nfc nfc
mkdir /data/misc/credstore 0700 credstore credstore
mkdir /data/misc/gatekeeper 0700 system system
mkdir /data/misc/keychain 0771 system system
mkdir /data/misc/net 0750 root shell
mkdir /data/misc/radio 0770 system radio
mkdir /data/misc/sms 0770 system radio
mkdir /data/misc/carrierid 0770 system radio
mkdir /data/misc/apns 0770 system radio
mkdir /data/misc/emergencynumberdb 0770 system radio
mkdir /data/misc/network_watchlist 0774 system system
mkdir /data/misc/telephonyconfig 0770 system radio
mkdir /data/misc/textclassifier 0771 system system
mkdir /data/misc/vpn 0770 system vpn
mkdir /data/misc/shared_relro 0771 shared_relro shared_relro
mkdir /data/misc/systemkeys 0700 system system
mkdir /data/misc/wifi 0770 wifi wifi
mkdir /data/misc/wifi/sockets 0770 wifi wifi
mkdir /data/misc/wifi/wpa_supplicant 0770 wifi wifi
mkdir /data/misc/ethernet 0770 system system
mkdir /data/misc/dhcp 0770 dhcp dhcp
mkdir /data/misc/user 0771 root root
# give system access to wpa_supplicant.conf for backup and restore
chmod 0660 /data/misc/wifi/wpa_supplicant.conf
mkdir /data/local 0751 root root encryption=Require
mkdir /data/misc/media 0700 media media
mkdir /data/misc/audioserver 0700 audioserver audioserver
mkdir /data/misc/cameraserver 0700 cameraserver cameraserver
mkdir /data/misc/vold 0700 root root
mkdir /data/misc/boottrace 0771 system shell
mkdir /data/misc/update_engine 0700 root root
mkdir /data/misc/update_engine_log 02750 root log
mkdir /data/misc/trace 0700 root root
# create location to store surface and window trace files
mkdir /data/misc/wmtrace 0700 system system
# create location to store accessibility trace files
mkdir /data/misc/a11ytrace 0700 system system
# profile file layout
mkdir /data/misc/profiles 0771 system system
mkdir /data/misc/profiles/cur 0771 system system
mkdir /data/misc/profiles/ref 0771 system system
mkdir /data/misc/profman 0770 system shell
mkdir /data/misc/gcov 0770 root root
mkdir /data/misc/installd 0700 root root
mkdir /data/misc/apexdata 0711 root root
mkdir /data/misc/apexrollback 0700 root root
mkdir /data/misc/appcompat/ 0700 system system
mkdir /data/misc/uprobestats-configs/ 0777 uprobestats uprobestats
mkdir /data/misc/snapshotctl_log 0755 root root
# create location to store pre-reboot information
mkdir /data/misc/prereboot 0700 system system
# directory used for on-device refresh metrics file.
mkdir /data/misc/odrefresh 0777 system system
# directory used for on-device signing key blob
mkdir /data/misc/odsign 0710 root system
# directory used for odsign metrics
mkdir /data/misc/odsign/metrics 0770 root system
# directory used for connectivity blob store.
mkdir /data/misc/connectivityblobdb 0770 system system
# Directory for VirtualizationService temporary image files.
# Delete any stale files owned by the old virtualizationservice uid (b/230056726).
chmod 0770 /data/misc/virtualizationservice
exec - virtualizationservice system -- /bin/rm -rf /data/misc/virtualizationservice
mkdir /data/misc/virtualizationservice 0771 system system
# /data/preloads uses encryption=None because it only contains preloaded
# files that are public information, similar to the system image.
mkdir /data/preloads 0775 system system encryption=None
# For security reasons, /data/local/tmp should always be empty.
# Do not place files or directories in /data/local/tmp
mkdir /data/local/tmp 0771 shell shell
mkdir /data/local/traces 0777 shell shell
mkdir /data/app-private 0771 system system encryption=Require
mkdir /data/app-ephemeral 0771 system system encryption=Require
mkdir /data/app-asec 0700 root root encryption=Require
mkdir /data/app-lib 0771 system system encryption=Require
mkdir /data/app 0771 system system encryption=Require
# Create directory for app metadata files
mkdir /data/app-metadata 0700 system system encryption=Require
# create directory for updated font files.
mkdir /data/fonts/ 0771 root root encryption=Require
mkdir /data/fonts/files 0771 system system
mkdir /data/fonts/config 0770 system system
# Create directories to push tests to for each linker namespace.
# Create the subdirectories in case the first test is run as root
# so it doesn't end up owned by root.
# Set directories to be executable by any process so that debuggerd,
# aka crash_dump, can read any executables/shared libraries.
mkdir /data/local/tests 0701 shell shell
mkdir /data/local/tests/product 0701 shell shell
mkdir /data/local/tests/system 0701 shell shell
mkdir /data/local/tests/unrestricted 0701 shell shell
mkdir /data/local/tests/vendor 0701 shell shell
# create dalvik-cache, so as to enforce our permissions
mkdir /data/dalvik-cache 0771 root root encryption=Require
# create the A/B OTA directory, so as to enforce our permissions
mkdir /data/ota 0771 root root encryption=Require
# create the OTA package directory. It will be accessed by GmsCore (cache
# group), update_engine and update_verifier.
mkdir /data/ota_package 0770 system cache encryption=Require
# create resource-cache and double-check the perms
mkdir /data/resource-cache 0771 system system encryption=Require
chown system system /data/resource-cache
chmod 0771 /data/resource-cache
# Ensure that lost+found exists and has the correct permissions. Linux
# filesystems expect this directory to exist; it's where the fsck tool puts
# any recovered files that weren't present in any directory. It must be
# unencrypted, as fsck must be able to write to it.
mkdir /data/lost+found 0770 root root encryption=None
# create directory for DRM plug-ins - give drm the read/write access to
# the following directory.
mkdir /data/drm 0770 drm drm encryption=Require
# create directory for MediaDrm plug-ins - give drm the read/write access to
# the following directory.
mkdir /data/mediadrm 0770 mediadrm mediadrm encryption=Require
# NFC: create data/nfc for nv storage
mkdir /data/nfc 0770 nfc nfc encryption=Require
mkdir /data/nfc/param 0770 nfc nfc
# Create all remaining /data root dirs so that they are made through init
# and get proper encryption policy installed
mkdir /data/backup 0700 system system encryption=Require
mkdir /data/ss 0700 system system encryption=Require
mkdir /data/system 0775 system system encryption=Require
mkdir /data/system/environ 0700 system system
# b/183861600 attempt to fix selinux label before running derive_classpath service
restorecon /data/system/environ
mkdir /data/system/dropbox 0700 system system
mkdir /data/system/heapdump 0700 system system
mkdir /data/system/users 0775 system system
# Mkdir and set SELinux security contexts for shutdown-checkpoints.
# TODO(b/270286197): remove these after couple releases.
mkdir /data/system/shutdown-checkpoints 0700 system system
restorecon_recursive /data/system/shutdown-checkpoints
# Create the parent directories of the user CE and DE storage directories.
# These parent directories must use encryption=None, since each of their
# subdirectories uses a different encryption policy (a per-user one), and
# encryption policies apply recursively. These directories should never
# contain any subdirectories other than the per-user ones. /data/media/obb
# is an exception that exists for legacy reasons.
#
# Don't use any write mode bits (0222) for any of these directories, since
# the only process that should write to them directly is vold (since it
# needs to set up file-based encryption on the subdirectories), which runs
# as root with CAP_DAC_OVERRIDE. This is also fully enforced via the
# SELinux policy. But we also set the DAC file modes accordingly, to try to
# minimize differences in behavior if SELinux is set to permissive mode.
mkdir /data/media 0550 media_rw media_rw encryption=None
mkdir /data/misc_ce 0551 system misc encryption=None
mkdir /data/misc_de 0551 system misc encryption=None
mkdir /data/system_ce 0550 system system encryption=None
mkdir /data/system_de 0550 system system encryption=None
mkdir /data/user 0511 system system encryption=None
mkdir /data/user_de 0511 system system encryption=None
mkdir /data/vendor_ce 0551 root root encryption=None
mkdir /data/vendor_de 0551 root root encryption=None
# Similar to the top-level CE and DE directories, /data/storage_area must
# itself be unencrypted, since it contains encrypted directories.
mkdir /data/storage_area 0551 root root encryption=None
# Set the casefold flag on /data/media. For upgrades, a restorecon can be
# needed first to relabel the directory from media_rw_data_file.
restorecon /data/media
exec - media_rw media_rw -- /system/bin/chattr +F /data/media
# A tmpfs directory, which will contain all apps and sdk sandbox CE and DE
# data directory that bind mount from the original source.
mount tmpfs tmpfs /data_mirror nodev noexec nosuid mode=0700,uid=0,gid=1000
restorecon /data_mirror
mkdir /data_mirror/data_ce 0700 root root
mkdir /data_mirror/data_de 0700 root root
mkdir /data_mirror/misc_ce 0700 root root
mkdir /data_mirror/misc_de 0700 root root
mkdir /data_mirror/storage_area 0700 root root
# Create CE and DE data directory for default volume
# Not needed for storage_area directory, since this is
# not supported for non-default volumes and the path
# does not include the volume ID
mkdir /data_mirror/data_ce/null 0700 root root
mkdir /data_mirror/data_de/null 0700 root root
mkdir /data_mirror/misc_ce/null 0700 root root
mkdir /data_mirror/misc_de/null 0700 root root
# Bind mount CE and DE data directory to mirror's default volume directory.
# Note that because the /data mount has the "shared" propagation type, the
# later bind mount of /data/data onto /data/user/0 will automatically
# propagate to /data_mirror/data_ce/null/0 as well.
mount none /data/user /data_mirror/data_ce/null bind rec
mount none /data/user_de /data_mirror/data_de/null bind rec
mount none /data/misc_ce /data_mirror/misc_ce/null bind rec
mount none /data/misc_de /data_mirror/misc_de/null bind rec
# Also bind mount for the storage area directory (minus the volume ID)
mount none /data/storage_area /data_mirror/storage_area bind rec
# Create mirror directory for jit profiles
mkdir /data_mirror/cur_profiles 0700 root root
mount none /data/misc/profiles/cur /data_mirror/cur_profiles bind rec
mkdir /data_mirror/ref_profiles 0700 root root
mount none /data/misc/profiles/ref /data_mirror/ref_profiles bind rec
mkdir /data/cache 0770 system cache encryption=Require
mkdir /data/cache/recovery 0770 system cache
mkdir /data/cache/backup_stage 0700 system system
mkdir /data/cache/backup 0700 system system
# Delete these if need be, per b/139193659
mkdir /data/rollback 0700 system system encryption=DeleteIfNecessary
mkdir /data/rollback-observer 0700 system system encryption=DeleteIfNecessary
mkdir /data/rollback-history 0700 system system encryption=DeleteIfNecessary
# Create root dir for Incremental Service
mkdir /data/incremental 0771 system system encryption=Require
# Create directories for statsd
mkdir /data/misc/stats-active-metric/ 0770 statsd system
mkdir /data/misc/stats-data/ 0770 statsd system
mkdir /data/misc/stats-data/restricted-data 0770 statsd system
mkdir /data/misc/stats-metadata/ 0770 statsd system
mkdir /data/misc/stats-service/ 0770 statsd system
mkdir /data/misc/train-info/ 0770 statsd system
Activate system APEXes early Summary: Boot sequence around apexd is changed to make it possible for pre-apexd processes to use libraries from APEXes. They no longer need to wait for the apexd to finish activating APEXes, which again can be done only after /data/ is mounted. This improves overall boot performance. Detail: This change fixes the problem that processes that are started before apexd (so called pre-apexd processes) can't access libraries that are provided only by the APEXes but are not found in the system partition (e.g. libdexfile_external.so, etc.). Main idea is to activate system APEXes (/system/apex/*.apex) before /data is mounted and then activate the updated APEXes (/data/apex/*.apex) after the /data mount. Detailed boot sequence is as follows. 1) init prepares the bootstrap and default mount namespaces. A tmpfs is mounted on /apex and the propagation type of the mountpoint is set to private. 2) before any other process is started, apexd is started in bootstrap mode. When executed in the mode, apexd only activates APEXes under /system/apex. Note that APEXes activated in this phase are mounted in the bootstrap mount namespace only. 3) other pre-apexd processes are started. They are in the bootstrap mount namespace and thus are provided with the libraries from the system APEXes. 4) /data is mounted. init switches into the default mount namespace and starts apexd as a daemon as usual. 5) apexd scans both /data/apex and /system/apex, and activate latest APEXes from the directories. Note that APEXes activated in this phase are mounted in the default namespaces only and thus are not visible to the pre-apexd processes. Bug: 125549215 Test: m; device boots Change-Id: I21c60d0ebe188fa4f24d6e6861f85ca204843069
2019-02-22 14:15:25 +01:00
# Wait for apexd to finish activating APEXes before starting more processes.
wait_for_prop apexd.status activated
perform_apex_config
exec_start aconfigd-mainline-init
start aconfigd
# Create directories for boot animation.
mkdir /data/misc/bootanim 0755 system system
exec_start derive_sdk
Activate system APEXes early Summary: Boot sequence around apexd is changed to make it possible for pre-apexd processes to use libraries from APEXes. They no longer need to wait for the apexd to finish activating APEXes, which again can be done only after /data/ is mounted. This improves overall boot performance. Detail: This change fixes the problem that processes that are started before apexd (so called pre-apexd processes) can't access libraries that are provided only by the APEXes but are not found in the system partition (e.g. libdexfile_external.so, etc.). Main idea is to activate system APEXes (/system/apex/*.apex) before /data is mounted and then activate the updated APEXes (/data/apex/*.apex) after the /data mount. Detailed boot sequence is as follows. 1) init prepares the bootstrap and default mount namespaces. A tmpfs is mounted on /apex and the propagation type of the mountpoint is set to private. 2) before any other process is started, apexd is started in bootstrap mode. When executed in the mode, apexd only activates APEXes under /system/apex. Note that APEXes activated in this phase are mounted in the bootstrap mount namespace only. 3) other pre-apexd processes are started. They are in the bootstrap mount namespace and thus are provided with the libraries from the system APEXes. 4) /data is mounted. init switches into the default mount namespace and starts apexd as a daemon as usual. 5) apexd scans both /data/apex and /system/apex, and activate latest APEXes from the directories. Note that APEXes activated in this phase are mounted in the default namespaces only and thus are not visible to the pre-apexd processes. Bug: 125549215 Test: m; device boots Change-Id: I21c60d0ebe188fa4f24d6e6861f85ca204843069
2019-02-22 14:15:25 +01:00
init_user0
# Set SELinux security contexts on upgrade or policy update.
restorecon --recursive --skip-ce /data
# Define and export *CLASSPATH variables
# Must start before 'odsign', as odsign depends on *CLASSPATH variables
exec_start derive_classpath
load_exports /data/system/environ/classpath
# Start ART's oneshot boot service to propagate boot experiment flags to
# dalvik.vm.*. This needs to be done before odsign since odrefresh uses and
# validates those properties against the signed cache-info.xml.
exec_start art_boot
# Start the on-device signing daemon, and wait for it to finish, to ensure
# ART artifacts are generated if needed.
# Must start after 'derive_classpath' to have *CLASSPATH variables set.
start odsign
# Wait for odsign to be done with the key.
wait_for_prop odsign.key.done 1
# Bump the boot level to 1000000000; this prevents further on-device signing.
# This is a special value that shuts down the thread which listens for
# further updates.
setprop keystore.boot_level 1000000000
# Allow apexd to snapshot and restore device encrypted apex data in the case
# of a rollback. This should be done immediately after DE_user data keys
# are loaded. APEXes should not access this data until this has been
# completed and apexd.status becomes "ready".
exec_start apexd-snapshotde
# sys.memfd_use set to false by default, which keeps it disabled
# until it is confirmed that apps and vendor processes don't make
# IOCTLs on ashmem fds any more.
setprop sys.use_memfd false
# Set fscklog permission
chown root system /dev/fscklogs/log
chmod 0770 /dev/fscklogs/log
# Enable FUSE by default
setprop persist.sys.fuse true
# Update dm-verity state and set partition.*.verified properties.
verity_update_state
# It is recommended to put unnecessary data/ initialization from post-fs-data
# to start-zygote in device's init.rc to unblock zygote start.
on zygote-start
wait_for_prop odsign.verification.done 1
# A/B update verifier that marks a successful boot.
exec_start update_verifier
start statsd
start netd
start zygote
start zygote_secondary
on boot && property:ro.config.low_ram=true
# Tweak background writeout
write /proc/sys/vm/dirty_expire_centisecs 200
write /proc/sys/vm/dirty_background_ratio 5
on boot && property:suspend.disable_sync_on_suspend=true
write /sys/power/sync_on_suspend 0
on boot
# basic network init
ifup lo
hostname localhost
domainname localdomain
# IPsec SA default expiration length
write /proc/sys/net/core/xfrm_acq_expires 3600
# Memory management. Basic kernel parameters, and allow the high
# level system server to be able to adjust the kernel OOM driver
# parameters to match how it is managing things.
write /proc/sys/vm/overcommit_memory 1
write /proc/sys/vm/min_free_order_shift 4
# System server manages zram writeback
chown root system /sys/block/zram0/idle
chmod 0664 /sys/block/zram0/idle
chown root system /sys/block/zram0/writeback
chmod 0664 /sys/block/zram0/writeback
# to access F2FS sysfs on dm-<num> directly
mkdir /dev/sys/fs/by-name 0755 system system
symlink /sys/fs/f2fs/${dev.mnt.dev.data} /dev/sys/fs/by-name/userdata
# dev.mnt.dev.data=dm-N, dev.mnt.blk.data=sdaN/mmcblk0pN, dev.mnt.rootdisk.data=sda/mmcblk0, or
# dev.mnt.dev.data=sdaN/mmcblk0pN, dev.mnt.blk.data=sdaN/mmcblk0pN, dev.mnt.rootdisk.data=sda/mmcblk0
mkdir /dev/sys/block/by-name 0755 system system
symlink /sys/class/block/${dev.mnt.dev.data} /dev/sys/block/by-name/userdata
symlink /sys/class/block/${dev.mnt.rootdisk.data} /dev/sys/block/by-name/rootdisk
# F2FS tuning. Set cp_interval larger than dirty_expire_centisecs, 30 secs,
# to avoid power consumption when system becomes mostly idle. Be careful
# to make it too large, since it may bring userdata loss, if they
# are not aware of using fsync()/sync() to prepare sudden power-cut.
write /dev/sys/fs/by-name/userdata/cp_interval 200
write /dev/sys/fs/by-name/userdata/gc_urgent_sleep_time 50
write /dev/sys/fs/by-name/userdata/iostat_period_ms 1000
write /dev/sys/fs/by-name/userdata/iostat_enable 1
# set readahead multiplier for POSIX_FADV_SEQUENTIAL files
write /dev/sys/fs/by-name/userdata/seq_file_ra_mul 128
# limit discard size to 128MB in order to avoid long IO latency
# for filesystem tuning first (dm or sda)
# this requires enabling selinux entry for sda/mmcblk0 in vendor side
write /dev/sys/block/by-name/userdata/queue/discard_max_bytes 134217728
write /dev/sys/block/by-name/rootdisk/queue/discard_max_bytes 134217728
# Permissions for System Server and daemons.
chown system system /sys/power/autosleep
chown system system /sys/devices/system/cpu/cpufreq/interactive/timer_rate
chmod 0660 /sys/devices/system/cpu/cpufreq/interactive/timer_rate
chown system system /sys/devices/system/cpu/cpufreq/interactive/timer_slack
chmod 0660 /sys/devices/system/cpu/cpufreq/interactive/timer_slack
chown system system /sys/devices/system/cpu/cpufreq/interactive/min_sample_time
chmod 0660 /sys/devices/system/cpu/cpufreq/interactive/min_sample_time
chown system system /sys/devices/system/cpu/cpufreq/interactive/hispeed_freq
chmod 0660 /sys/devices/system/cpu/cpufreq/interactive/hispeed_freq
chown system system /sys/devices/system/cpu/cpufreq/interactive/target_loads
chmod 0660 /sys/devices/system/cpu/cpufreq/interactive/target_loads
chown system system /sys/devices/system/cpu/cpufreq/interactive/go_hispeed_load
chmod 0660 /sys/devices/system/cpu/cpufreq/interactive/go_hispeed_load
chown system system /sys/devices/system/cpu/cpufreq/interactive/above_hispeed_delay
chmod 0660 /sys/devices/system/cpu/cpufreq/interactive/above_hispeed_delay
chown system system /sys/devices/system/cpu/cpufreq/interactive/boost
chmod 0660 /sys/devices/system/cpu/cpufreq/interactive/boost
chown system system /sys/devices/system/cpu/cpufreq/interactive/boostpulse
chown system system /sys/devices/system/cpu/cpufreq/interactive/input_boost
chmod 0660 /sys/devices/system/cpu/cpufreq/interactive/input_boost
chown system system /sys/devices/system/cpu/cpufreq/interactive/boostpulse_duration
chmod 0660 /sys/devices/system/cpu/cpufreq/interactive/boostpulse_duration
chown system system /sys/devices/system/cpu/cpufreq/interactive/io_is_busy
chmod 0660 /sys/devices/system/cpu/cpufreq/interactive/io_is_busy
chown system system /sys/class/leds/vibrator/trigger
chown system system /sys/class/leds/vibrator/activate
chown system system /sys/class/leds/vibrator/brightness
chown system system /sys/class/leds/vibrator/duration
chown system system /sys/class/leds/vibrator/state
chown system system /sys/class/timed_output/vibrator/enable
chown system system /sys/class/leds/keyboard-backlight/brightness
chown system system /sys/class/leds/lcd-backlight/brightness
chown system system /sys/class/leds/button-backlight/brightness
chown system system /sys/class/leds/jogball-backlight/brightness
chown system system /sys/class/leds/red/brightness
chown system system /sys/class/leds/green/brightness
chown system system /sys/class/leds/blue/brightness
chown system system /sys/class/leds/red/device/grpfreq
chown system system /sys/class/leds/red/device/grppwm
chown system system /sys/class/leds/red/device/blink
chown system system /sys/module/sco/parameters/disable_esco
chown system system /sys/kernel/ipv4/tcp_wmem_min
chown system system /sys/kernel/ipv4/tcp_wmem_def
chown system system /sys/kernel/ipv4/tcp_wmem_max
chown system system /sys/kernel/ipv4/tcp_rmem_min
chown system system /sys/kernel/ipv4/tcp_rmem_def
chown system system /sys/kernel/ipv4/tcp_rmem_max
chown root radio /proc/cmdline
chown root system /proc/bootconfig
# Define default initial receive window size in segments.
setprop net.tcp_def_init_rwnd 60
# Start standard binderized HAL daemons
class_start hal
class_start core
on nonencrypted
class_start main
class_start late_start
on property:sys.init_log_level=*
loglevel ${sys.init_log_level}
on charger
class_start charger
on property:sys.boot_completed=1
bootchart stop
# Setup per_boot directory so other .rc could start to use it on boot_completed
exec - system system -- /bin/rm -rf /data/per_boot
mkdir /data/per_boot 0700 system system encryption=Require key=per_boot_ref
# system server cannot write to /proc/sys files,
# and chown/chmod does not work for /proc/sys/ entries.
# So proxy writes through init.
on property:sys.sysctl.extra_free_kbytes=*
exec -- /system/bin/extra_free_kbytes.sh ${sys.sysctl.extra_free_kbytes}
# Allow users to drop caches
on property:perf.drop_caches=3
write /proc/sys/vm/drop_caches 3
setprop perf.drop_caches 0
# "tcp_default_init_rwnd" Is too long!
on property:net.tcp_def_init_rwnd=*
write /proc/sys/net/ipv4/tcp_default_init_rwnd ${net.tcp_def_init_rwnd}
init: add builtin check for perf_event LSM hooks Historically, the syscall was controlled by a system-wide perf_event_paranoid sysctl, which is not flexible enough to allow only specific processes to use the syscall. However, SELinux support for the syscall has been upstreamed recently[1] (and is being backported to Android R release common kernels). [1] https://github.com/torvalds/linux/commit/da97e18458fb42d7c00fac5fd1c56a3896ec666e As the presence of these hooks is not guaranteed on all Android R platforms (since we support upgrades while keeping an older kernel), we need to test for the feature dynamically. The LSM hooks themselves have no way of being detected directly, so we instead test for their effects, so we perform several syscalls, and look for a specific success/failure combination, corresponding to the platform's SELinux policy. If hooks are detected, perf_event_paranoid is set to -1 (unrestricted), as the SELinux policy is then sufficient to control access. This is done within init for several reasons: * CAP_SYS_ADMIN side-steps perf_event_paranoid, so the tests can be done if non-root users aren't allowed to use the syscall (the default). * init is already the setter of the paranoid value (see init.rc), which is also a privileged operation. * the test itself is simple (couple of syscalls), so having a dedicated test binary/domain felt excessive. I decided to go through a new sysprop (set by a builtin test in second-stage init), and keeping the actuation in init.rc. We can change it to an immediate write to the paranoid value if a use-case comes up that requires the decision to be made earlier in the init sequence. Bug: 137092007 Change-Id: Ib13a31fee896f17a28910d993df57168a83a4b3d
2020-01-14 23:02:53 +01:00
# perf_event_open syscall security:
# Newer kernels have the ability to control the use of the syscall via SELinux
# hooks. init tests for this, and sets sys_init.perf_lsm_hooks to 1 if the
# kernel has the hooks. In this case, the system-wide perf_event_paranoid
# sysctl is set to -1 (unrestricted use), and the SELinux policy is used for
# controlling access. On older kernels, the paranoid value is the only means of
# controlling access. It is normally 3 (allow only root), but the shell user
# can lower it to 1 (allowing thread-scoped pofiling) via security.perf_harden.
on load_bpf_programs && property:sys.init.perf_lsm_hooks=1
init: add builtin check for perf_event LSM hooks Historically, the syscall was controlled by a system-wide perf_event_paranoid sysctl, which is not flexible enough to allow only specific processes to use the syscall. However, SELinux support for the syscall has been upstreamed recently[1] (and is being backported to Android R release common kernels). [1] https://github.com/torvalds/linux/commit/da97e18458fb42d7c00fac5fd1c56a3896ec666e As the presence of these hooks is not guaranteed on all Android R platforms (since we support upgrades while keeping an older kernel), we need to test for the feature dynamically. The LSM hooks themselves have no way of being detected directly, so we instead test for their effects, so we perform several syscalls, and look for a specific success/failure combination, corresponding to the platform's SELinux policy. If hooks are detected, perf_event_paranoid is set to -1 (unrestricted), as the SELinux policy is then sufficient to control access. This is done within init for several reasons: * CAP_SYS_ADMIN side-steps perf_event_paranoid, so the tests can be done if non-root users aren't allowed to use the syscall (the default). * init is already the setter of the paranoid value (see init.rc), which is also a privileged operation. * the test itself is simple (couple of syscalls), so having a dedicated test binary/domain felt excessive. I decided to go through a new sysprop (set by a builtin test in second-stage init), and keeping the actuation in init.rc. We can change it to an immediate write to the paranoid value if a use-case comes up that requires the decision to be made earlier in the init sequence. Bug: 137092007 Change-Id: Ib13a31fee896f17a28910d993df57168a83a4b3d
2020-01-14 23:02:53 +01:00
write /proc/sys/kernel/perf_event_paranoid -1
on property:security.perf_harden=0 && property:sys.init.perf_lsm_hooks=""
write /proc/sys/kernel/perf_event_paranoid 1
init: add builtin check for perf_event LSM hooks Historically, the syscall was controlled by a system-wide perf_event_paranoid sysctl, which is not flexible enough to allow only specific processes to use the syscall. However, SELinux support for the syscall has been upstreamed recently[1] (and is being backported to Android R release common kernels). [1] https://github.com/torvalds/linux/commit/da97e18458fb42d7c00fac5fd1c56a3896ec666e As the presence of these hooks is not guaranteed on all Android R platforms (since we support upgrades while keeping an older kernel), we need to test for the feature dynamically. The LSM hooks themselves have no way of being detected directly, so we instead test for their effects, so we perform several syscalls, and look for a specific success/failure combination, corresponding to the platform's SELinux policy. If hooks are detected, perf_event_paranoid is set to -1 (unrestricted), as the SELinux policy is then sufficient to control access. This is done within init for several reasons: * CAP_SYS_ADMIN side-steps perf_event_paranoid, so the tests can be done if non-root users aren't allowed to use the syscall (the default). * init is already the setter of the paranoid value (see init.rc), which is also a privileged operation. * the test itself is simple (couple of syscalls), so having a dedicated test binary/domain felt excessive. I decided to go through a new sysprop (set by a builtin test in second-stage init), and keeping the actuation in init.rc. We can change it to an immediate write to the paranoid value if a use-case comes up that requires the decision to be made earlier in the init sequence. Bug: 137092007 Change-Id: Ib13a31fee896f17a28910d993df57168a83a4b3d
2020-01-14 23:02:53 +01:00
on property:security.perf_harden=1 && property:sys.init.perf_lsm_hooks=""
write /proc/sys/kernel/perf_event_paranoid 3
# Additionally, simpleperf profiler uses debug.* and security.perf_harden
# sysprops to be able to indirectly set these sysctls.
on property:security.perf_harden=0
write /proc/sys/kernel/perf_event_max_sample_rate ${debug.perf_event_max_sample_rate:-100000}
write /proc/sys/kernel/perf_cpu_time_max_percent ${debug.perf_cpu_time_max_percent:-25}
write /proc/sys/kernel/perf_event_mlock_kb ${debug.perf_event_mlock_kb:-516}
init: add builtin check for perf_event LSM hooks Historically, the syscall was controlled by a system-wide perf_event_paranoid sysctl, which is not flexible enough to allow only specific processes to use the syscall. However, SELinux support for the syscall has been upstreamed recently[1] (and is being backported to Android R release common kernels). [1] https://github.com/torvalds/linux/commit/da97e18458fb42d7c00fac5fd1c56a3896ec666e As the presence of these hooks is not guaranteed on all Android R platforms (since we support upgrades while keeping an older kernel), we need to test for the feature dynamically. The LSM hooks themselves have no way of being detected directly, so we instead test for their effects, so we perform several syscalls, and look for a specific success/failure combination, corresponding to the platform's SELinux policy. If hooks are detected, perf_event_paranoid is set to -1 (unrestricted), as the SELinux policy is then sufficient to control access. This is done within init for several reasons: * CAP_SYS_ADMIN side-steps perf_event_paranoid, so the tests can be done if non-root users aren't allowed to use the syscall (the default). * init is already the setter of the paranoid value (see init.rc), which is also a privileged operation. * the test itself is simple (couple of syscalls), so having a dedicated test binary/domain felt excessive. I decided to go through a new sysprop (set by a builtin test in second-stage init), and keeping the actuation in init.rc. We can change it to an immediate write to the paranoid value if a use-case comes up that requires the decision to be made earlier in the init sequence. Bug: 137092007 Change-Id: Ib13a31fee896f17a28910d993df57168a83a4b3d
2020-01-14 23:02:53 +01:00
# Default values.
on property:security.perf_harden=1
init: add builtin check for perf_event LSM hooks Historically, the syscall was controlled by a system-wide perf_event_paranoid sysctl, which is not flexible enough to allow only specific processes to use the syscall. However, SELinux support for the syscall has been upstreamed recently[1] (and is being backported to Android R release common kernels). [1] https://github.com/torvalds/linux/commit/da97e18458fb42d7c00fac5fd1c56a3896ec666e As the presence of these hooks is not guaranteed on all Android R platforms (since we support upgrades while keeping an older kernel), we need to test for the feature dynamically. The LSM hooks themselves have no way of being detected directly, so we instead test for their effects, so we perform several syscalls, and look for a specific success/failure combination, corresponding to the platform's SELinux policy. If hooks are detected, perf_event_paranoid is set to -1 (unrestricted), as the SELinux policy is then sufficient to control access. This is done within init for several reasons: * CAP_SYS_ADMIN side-steps perf_event_paranoid, so the tests can be done if non-root users aren't allowed to use the syscall (the default). * init is already the setter of the paranoid value (see init.rc), which is also a privileged operation. * the test itself is simple (couple of syscalls), so having a dedicated test binary/domain felt excessive. I decided to go through a new sysprop (set by a builtin test in second-stage init), and keeping the actuation in init.rc. We can change it to an immediate write to the paranoid value if a use-case comes up that requires the decision to be made earlier in the init sequence. Bug: 137092007 Change-Id: Ib13a31fee896f17a28910d993df57168a83a4b3d
2020-01-14 23:02:53 +01:00
write /proc/sys/kernel/perf_event_max_sample_rate 100000
write /proc/sys/kernel/perf_cpu_time_max_percent 25
write /proc/sys/kernel/perf_event_mlock_kb 516
# This property can be set only on userdebug/eng. See neverallow rule in
# /system/sepolicy/private/property.te .
on property:security.lower_kptr_restrict=1
write /proc/sys/kernel/kptr_restrict 0
on property:security.lower_kptr_restrict=0
write /proc/sys/kernel/kptr_restrict 2
# on shutdown
# In device's init.rc, this trigger can be used to do device-specific actions
# before shutdown. e.g disable watchdog and mask error handling
## Daemon processes to be run by init.
##
service ueventd /system/bin/ueventd
class core
critical
seclabel u:r:ueventd:s0
user root
shutdown critical
service console /system/bin/sh
class core
console
disabled
user shell
group shell log readproc
seclabel u:r:shell:s0
setenv HOSTNAME console
shutdown critical
on property:ro.debuggable=1
# Give writes to the same group for the trace folder on debug builds,
# it's further protected by selinux policy.
# The folder is used to store method traces.
chmod 0773 /data/misc/trace
# Give writes and reads to anyone for the window trace folder on debug builds,
# it's further protected by selinux policy.
chmod 0777 /data/misc/wmtrace
# Give reads to anyone for the accessibility trace folder on debug builds.
chmod 0775 /data/misc/a11ytrace
on init && property:ro.debuggable=1
start console
on userspace-reboot-requested
# TODO(b/135984674): reset all necessary properties here.
setprop sys.boot_completed ""
setprop dev.bootcomplete ""
setprop sys.init.updatable_crashing ""
setprop sys.init.updatable_crashing_process_name ""
setprop sys.user.0.ce_available ""
setprop sys.shutdown.requested ""
setprop service.bootanim.exit ""
setprop service.bootanim.progress ""
on userspace-reboot-fs-remount
# Make sure that vold is running.
# This is mostly a precaution measure in case vold for some reason wasn't running when
# userspace reboot was initiated.
start vold
exec - system system -- /system/bin/vdc checkpoint resetCheckpoint
exec - system system -- /system/bin/vdc checkpoint markBootAttempt
# Unmount /data_mirror mounts in the reverse order of corresponding mounts.
umount /data_mirror/data_ce/null/0
umount /data_mirror/data_ce/null
umount /data_mirror/data_de/null
umount /data_mirror/storage_area/0
umount /data_mirror/storage_area
umount /data_mirror/cur_profiles
umount /data_mirror/ref_profiles
umount /data_mirror
remount_userdata
start bootanim
on userspace-reboot-resume
trigger userspace-reboot-fs-remount
trigger post-fs-data
trigger zygote-start
trigger early-boot
trigger boot
on property:sys.boot_completed=1 && property:sys.init.userspace_reboot.in_progress=1
setprop sys.init.userspace_reboot.in_progress ""
# Multi-Gen LRU Experiment
on property:persist.device_config.mglru_native.lru_gen_config=none
write /sys/kernel/mm/lru_gen/enabled 0
on property:persist.device_config.mglru_native.lru_gen_config=core
write /sys/kernel/mm/lru_gen/enabled 1
on property:persist.device_config.mglru_native.lru_gen_config=core_and_mm_walk
write /sys/kernel/mm/lru_gen/enabled 3
on property:persist.device_config.mglru_native.lru_gen_config=core_and_nonleaf_young
write /sys/kernel/mm/lru_gen/enabled 5
on property:persist.device_config.mglru_native.lru_gen_config=all
write /sys/kernel/mm/lru_gen/enabled 7
# Allow other processes to run `snapshotctl` through `init`. This requires
# `set_prop` permission on `snapshotctl_prop`.
on property:sys.snapshotctl.map=requested
# "root" is needed to talk to gsid and pass its check on uid.
# "system" is needed to write to "/dev/socket/snapuserd" to talk to
# snapuserd.
exec - root root system -- /system/bin/snapshotctl map
setprop sys.snapshotctl.map "finished"
on property:sys.snapshotctl.unmap=requested
exec - root root system -- /system/bin/snapshotctl unmap
setprop sys.snapshotctl.unmap "finished"