summaryrefslogtreecommitdiff
path: root/scripts/Makefile.thinlto
diff options
context:
space:
mode:
authorAbdurrahman Hussain <abdurrahman@nexthop.ai>2026-05-20 15:42:42 -0700
committerGuenter Roeck <linux@roeck-us.net>2026-06-09 08:23:09 -0700
commit22c1992642ebe95f4ec9f6456cf56ae330d2571a (patch)
treeaaac234d0422d5b82c3c533c19a39e7f0951f997 /scripts/Makefile.thinlto
parente8d55e1f6ebf6d22727fe46ce8714906b3afe23e (diff)
hwmon: (pmbus/adm1266) add rtc debugfs entry
The driver seeds the chip's SET_RTC register once at probe with ktime_get_real_seconds(). Over a long uptime the chip's internal seconds counter drifts away from the host's wall-clock time, so the timestamp embedded in each blackbox record stops being meaningful in wall-clock terms. The datasheet recommends that the host periodically resynchronise the counter to address this; today the driver has no userspace-facing knob for that. Expose SET_RTC via an rtc debugfs file alongside the other adm1266 debugfs entries: read -- returns the chip's current SET_RTC seconds counter, so userspace can observe how far the chip has drifted from host wall-clock without writing anything. write -- the kernel re-reads ktime_get_real_seconds() itself and pushes it to the chip. The write payload is ignored; userspace does not get to supply its own timestamp value, so there is no way for it to push a wrong time into the chip. A small userspace agent (chrony hook, systemd-timesyncd dispatch script, or a periodic cron job) can write to this file to keep the chip's counter aligned with wall-clock across long uptimes. Both the read and write paths take pmbus_lock to serialise against the pmbus_core's own PAGE+register sequences and against the other adm1266 debugfs accessors that already run under the same lock. While at it, drop the now-redundant adm1266_set_rtc() probe-time helper. The new adm1266_rtc_set() callback does exactly the same byte-packing and write; probe just calls adm1266_rtc_set(client, 0) (the ignored @val argument) after pmbus_do_probe() so the pmbus_lock acquired by the new helper has a live mutex to take. Signed-off-by: Abdurrahman Hussain <abdurrahman@nexthop.ai> Assisted-by: Claude-Code:claude-opus-4-7 Assisted-by: sashiko:gemini-3.1-pro-preview Link: https://lore.kernel.org/r/20260520-adm1266-v5-3-c72ef1fac1ea@nexthop.ai Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Diffstat (limited to 'scripts/Makefile.thinlto')
0 files changed, 0 insertions, 0 deletions