dmesg: get the boot time from CLOCK_BOOTTIME on Linux instead of a utmp record by jayhemnani9910 · Pull Request #672 · uutils/util-linux

jayhemnani9910 · 2026-10-22T05:18:58Z

dmesg panics for every time format that needs the boot time when utmp has no boot record. That is the case on systems without /run/utmp, such as recent Ubuntu and most containers:

$ dmesg -K k.kmsg --time-format=ctime
thread 'main' panicked at src/uu/dmesg/src/time_formatter.rs:149:54:
called `Option::unwrap()` on a `None` value

The same panic hits --time-format=iso and reltime, --since and --until. The tests don't catch it because they build with fixed-boot-time.

On Linux the boot time is now the current time minus CLOCK_BOOTTIME, so no utmp is needed. macOS and FreeBSD read kern.boottime, and other Unix systems keep the utmp lookup. When no boot time can be found, dmesg exits 1 with cannot get the system boot time instead of panicking. uu_dmesg already used uucore::libc without enabling uucore's libc feature, so the crate only built next to another tool that enabled it. The feature is now enabled in its Cargo.toml.

New unit test boot_time_matches_proc_uptime checks the boot time against /proc/uptime. It needs a build without fixed-boot-time, so CI now also runs cargo test -p uu_dmesg; on macOS that runs boot_time_comes_from_kern_boottime. The uptime test fails on the old code on a system without a utmp boot record. cargo clippy -p uu_dmesg --all-targets is clean for armv7-linux-androideabi, x86_64-apple-darwin, x86_64-unknown-freebsd and i686-unknown-freebsd. cargo test for the tools that build here (all but lsipc, lslocks and lsns, which need libsmartcols): 232 passed, same as main. clippy and fmt are clean. With a kmsg file, --time-format=ctime, iso and reltime print the same times as util-linux 2.41.3 (iso differs by 1 µs of rounding).

…mp record

Copilot

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing. Review the Copilot code review billing settings.

sylvestre · 2026-10-22T21:38:44Z

+#[cfg(not(any(target_os = "openbsd", target_os = "linux", target_os = "android")))]
 fn boot_time() -> DateTime<FixedOffset> {
    *BOOT_TIME.get_or_init(|| boot_time_from_utmpx().unwrap())
 }


this still panics on macOS/freebsd without a boot record, no?

Yes. In efc8d5d macOS and FreeBSD read kern.boottime with sysctl instead of utmp. When no boot time can be found, on any system, dmesg now exits 1 with dmesg: cannot get the system boot time instead of panicking. With --since or --until it fails before reading any records.

sylvestre · 2026-10-22T21:38:47Z

+    };
+    // SAFETY: `uptime` is a valid timespec for clock_gettime to write.
+    unsafe { libc::clock_gettime(libc::CLOCK_BOOTTIME, &mut uptime) };
+    let uptime = TimeDelta::new(uptime.tv_sec, uptime.tv_nsec as u32).unwrap_or_default();


tv_sec is i32 on 32-bit android, does this build there?

It didn't: cargo check --target armv7-linux-androideabi failed with expected i64, found i32 on tv_sec. efc8d5d converts it when the pointer width isn't 64, as uucore's uptime does. cargo clippy -p uu_dmesg --all-targets is now clean for armv7-linux-androideabi, x86_64-apple-darwin, x86_64-unknown-freebsd and i686-unknown-freebsd.

sylvestre · 2026-10-22T21:38:49Z

+        tv_nsec: 0,
+    };
+    // SAFETY: `uptime` is a valid timespec for clock_gettime to write.
+    unsafe { libc::clock_gettime(libc::CLOCK_BOOTTIME, &mut uptime) };


please check the return value
if it fails, the boot time silently becomes now

Done in efc8d5d: if clock_gettime returns nonzero there is no boot time, so dmesg reports the error instead of using now.

sylvestre · 2026-10-22T21:38:59Z

+    use super::*;
+
+    #[test]
+    fn boot_time_matches_proc_uptime() {


this test never runs in CI, cargo test enables fixed-boot-time
could you please add a cargo test -p uu_dmesg step?

Added cargo test -p uu_dmesg to the test job in efc8d5d, after the workspace cargo test, so it runs on all three OSes. On macOS it runs the new boot_time_comes_from_kern_boottime.

…cking without a boot time

Copilot

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing. Review the Copilot code review billing settings.

sylvestre · 2026-10-24T12:29:10Z

+    if res != 0 {
+        return None;
+    }
+    #[cfg(target_pointer_width = "64")]


same 4 lines as in the linux version, could be a small helper, no?

Done in 440d4b8: time_t_secs does the conversion for both. A plain .into() covers 32 and 64-bit time_t, with useless_conversion allowed for the 64-bit case. cargo clippy -p uu_dmesg --all-targets -- -D warnings is clean for x86_64 Linux, macOS and FreeBSD, i686 FreeBSD and armv7 Android.

Copilot

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota.

An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing. Review the Copilot code review billing settings.

dmesg: get the boot time from CLOCK_BOOTTIME on Linux instead of a ut…

cc7e8bb

…mp record

Copilot AI balanced review requested due to automatic review settings October 22, 2026 05:18

Copilot AI reviewed Oct 22, 2026

sylvestre reviewed Oct 22, 2026

View reviewed changes

dmesg: read kern.boottime on macOS and FreeBSD, error instead of pani…

efc8d5d

…cking without a boot time

Copilot AI balanced review requested due to automatic review settings October 23, 2026 07:08

Copilot AI reviewed Oct 23, 2026

sylvestre reviewed Oct 24, 2026

View reviewed changes

dmesg: share the time_t to i64 conversion

440d4b8

Copilot AI balanced review requested due to automatic review settings October 24, 2026 18:59

Copilot AI reviewed Oct 24, 2026

Conversation

jayhemnani9910 commented Oct 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

jayhemnani9910 commented Oct 22, 2026 •

edited

Loading