Starting with v4.18, the Linux kernel provides the rseq
syscall which is a mechanism for fast userspace task
synchronization.
Starting with v2.35 glibc uses the new syscall, if it
exists, to gain some performance improvements, so we
need to update the policy files to allow it.
Even on older kernels where rseq is not supported,
glibc will still probe for its existence by expecting
an -ENOSYS response.
BUG=b:235960683
TEST=Local builds against glibc 2.35
Change-Id: I704f2fbf2b058c3a4c3269c7441c3a7324012f8a
Reviewed-on: https://chromium-review.googlesource.com/c/crosvm/crosvm/+/3763901
Commit-Queue: Manoj Gupta <manojgupta@chromium.org>
Owners-Override: Dominick Ng <dominickn@google.com>
Reviewed-by: Allen Webb <allenwebb@google.com>
Reviewed-by: Dennis Kempin <denniskempin@google.com>
Tested-by: Manoj Gupta <manojgupta@chromium.org>
This was already allowed in common_device.policy for arm platforms, and
it can be called via future/async code that is used by many devices. Add
it to x86_64 and aarch64 common policies as well.
BUG=b:220908060
TEST=Start Crostini with virtio-snd on hatch and trogdo
Change-Id: I9bf8f56ebed0cfe6d5e17e45226cac73db3d6d37
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3482026
Reviewed-by: Dmitry Torokhov <dtor@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
The panic handler uses getcwd and readlink to print out the executable
name in the backtrace. Allow these for all devices so that panics
actually work instead of crashing the process.
BUG=None
TEST=intentionally panic crosvm on kevin and check /var/log/messages
Change-Id: If64a752a6f0b1f2f6bdd6663ce77078305f38171
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3309201
Reviewed-by: Dennis Kempin <denniskempin@google.com>
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Daniel Verkamp <dverkamp@chromium.org>
The sched_yield system call is somehow called by the code the rust
compiler generates and not directly by the author's implementation. That
along with the fact that it won't get called on every run makes it very
easy to miss when adding a new device (that happened with virtio-snd).
Since that call is quite harmless (it could be argued minijail shouldn't
even block it in the first place) it makes sense to allow it for all
devices.
BUG=b/201306350
Change-Id: I9895da6c8060ae83053474ed9e4472ea2cd8d3e3
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3248126
Auto-Submit: Jorge Moreira Broche <jemoreira@google.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Jorge Moreira Broche <jemoreira@google.com>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Kernels before 5.10 had known bugs in the io_uring implementation.
Don't use io_uring when we detect this. Also skip all the io_uring
tests in this case.
BUG=none
TEST=cargo test
Change-Id: I5fd6203ad25a6fb85ff28f1a6ddb0181f836ad89
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/3006309
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Keiichi Watanabe <keiichiw@chromium.org>
Reviewed-by: Woody Chow <woodychow@google.com>
The libminijail version in AOSP complains when there are multiple entries for
the same system call, which was the case for virtio-fs's policy.
BUG=b/185811304
Change-Id: I389c07c86e7d79f16e4f47a893abad598033352a
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2837307
Commit-Queue: Jorge Moreira Broche <jemoreira@google.com>
Tested-by: Jorge Moreira Broche <jemoreira@google.com>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Fds created via dup don't share file descriptor flags with the original
fd, which means that they don't have the FD_CLOEXEC flag set. Use
fcntl(F_DUPFD_CLOEXEC) so that this flag gets set for the duplicated fds
as well.
BUG=none
TEST=unit tests
Change-Id: Ib471cf40acac1eacf72969ba45247f50b349ed58
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2809687
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Reviewed-by: Zach Reizner <zachr@chromium.org>
Starting with 2.32 glibc nanosleep() was refactored to use the
clock_nanosleep syscall so various software will fail unless
the new syscall is allowed. We can't just drop the old nanosleep
syscall because it will break glibc 2.27 which is still used.
See glibc commits:
807edded25 nptl: Refactor thrd_sleep in terms of clock_nanosleep
3537ecb49c Refactor nanosleep in terms of clock_nanosleep
79a547b162 nptl: Move nanosleep implementation to libc
This is a bulk edit done with the following command:
git grep -rl 'nanosleep: 1' | xargs sed -i \
'/^nanosleep: 1/a clock_nanosleep: 1'
BUG=chromium:1171287
TEST=Local builds and booting on kevin/64/eve/minnie.
Change-Id: I975535078d88200f52319c7eea3a4c7ebf299933
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2735575
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Manoj Gupta <manojgupta@chromium.org>
Commit-Queue: Manoj Gupta <manojgupta@chromium.org>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
The newly added async primitives allow for increasing the separation of
the various tasks performed by balloon. Breaking each task in to an
asynchronous function.
BUG=chromium:901139
TEST=Boot crosvm, run 'crosvm balloon' to set the balloon size, check
'vmstat' inside the VM to verify the free memory is affected by the
balloon growing and shrinking.
run crosvm balloon_stats command and ensure that stats are reported
correctly.
Change-Id: I0ae2be5eb8e4be65b2eb74de90888357af6ecfd4
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1993163
Tested-by: Keiichi Watanabe <keiichiw@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Commit-Queue: Keiichi Watanabe <keiichiw@chromium.org>
Reviewed-by: Chirantan Ekbote <chirantan@chromium.org>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Add a frequency file that teaches the seccomp compiler to weight the
comparison tree in favor of the most frequenctly called syscalls.
This frequency file was created by running strace against vm_conciege's
pid (e.g., "strace -p PID -ff -e raw=all -o /tmp/strace") when
performing a start and stop of a VM, deleting the trace files that
weren't for a crosvm process, passing the files to minijail's
tools/generate_seccomp_policy.py (using the -frequency option), and
combining the results of the frequency file. I rounded the #s to the
nearest multiple of 5 and only retained the syscalls that had at least
10 calls.
BUG=None
TEST=Local build and deploy. Verified that crostini VM still boots and
shuts down properly. Used scmp_bpf_disasm to disassemble a few bpf
files before and after this change to confirm that with the frequency
file the first comparision is "jge 2" (to quickly whitelist syscalls
0 and 1 ['read' and 'write']) instead of a comparison around the middle
of the range of syscall numbers that are used.
Change-Id: Icace2b5cdbcae6e51cfd67a3034a1a17fdb6d59e
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/2005793
Commit-Queue: Matt Delco <delco@chromium.org>
Commit-Queue: Stephen Barber <smbarber@chromium.org>
Tested-by: Matt Delco <delco@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Auto-Submit: Matt Delco <delco@chromium.org>
Reviewed-by: Stephen Barber <smbarber@chromium.org>
Minijail's policy compiler complains when there's multiple
unconditional rules for a syscall. In most cases the rules
are redundant to common_device.policy. I don't know what
to do about the intentionally contradictory rules for open
and openat, other than to remove then from the common device
policy and add it to all the others.
BUG=None
TEST=Ran compile_seccomp_policy.py until it stopped
complaining.
Change-Id: I6813dd1e0b39e975415662bd7de74c25a1be9eb3
Signed-off-by: Matt Delco <delco@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1918607
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Dylan Reid <dgreid@chromium.org>
The gettid syscall is used in some corners of glibc and it is a fairly
harmless syscall (we already give getpid), so this change moves it to
the common policy.
TEST=None
BUG=chromium:996938
Change-Id: I129644273f2f02fe917255c7157c48b99c329045
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1952565
Tested-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Auto-Submit: Zach Reizner <zachr@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Zach Reizner <zachr@chromium.org>
Using syslog from glibc will use some syscalls we haven't seen before,
leading to the process getting killed. This change fixes that.
TEST=use syslog from C
BUG=chromium:988082
Change-Id: I4cfb317a8faf70188995487f4fa844229683d6d1
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/crosvm/+/1721616
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
Commit-Queue: Zach Reizner <zachr@chromium.org>
Tested-by: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
The advantage of seqpacket is that they are connection oriented. A
listener can be created that accepts new connections, useful for the
path based VM control sockets. Previously, the only bidirectional
sockets in crosvm were either stream based or made using socketpair.
This change also whitelists sendmsg and recvmsg for the common device
policy.
TEST=cargo test
BUG=chromium:848187
Change-Id: I83fd46f54bce105a7730632cd013b5e7047db22b
Reviewed-on: https://chromium-review.googlesource.com/1470917
Commit-Ready: Zach Reizner <zachr@chromium.org>
Tested-by: kokoro <noreply+kokoro@google.com>
Tested-by: Zach Reizner <zachr@chromium.org>
Reviewed-by: Daniel Verkamp <dverkamp@chromium.org>
