This is an automated email from the ASF dual-hosted git repository.
GUIDINGLI pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/nuttx.git
The following commit(s) were added to refs/heads/master by this push:
new ab195942333 arch/x86_64/intel64: Fix the precision and the SMP step of
the TSC RTC.
ab195942333 is described below
commit ab195942333886ce5e8c21463acca5d2e75b3951
Author: Marco Casaroli <[email protected]>
AuthorDate: Tue Nov 17 14:03:34 2026 +0100
arch/x86_64/intel64: Fix the precision and the SMP step of the TSC RTC.
rtc_read() converted TSC cycles to nanoseconds with
(tsc / (freq / 1000000)) * 1000. freq / 1000000 drops the part of the
frequency below 1 MHz, so the clock runs fast, and the result has a step
of 1 us. Calculate the seconds and the remainder separately, as
up_perf_convert() does.
When the TSC value was smaller than the last value, rtc_read() added
0xffffffffffffffff - last to it, for a wrap that the 64-bit TSC does not
do. On SMP a CPU with a TSC that is a little behind gets there, and the
time then goes forward by about 2^64 cycles. Keep the last value
instead, so that the time does not go back.
Assisted-by: Claude Code:claude-opus-5-5
Signed-off-by: Marco Casaroli <[email protected]>
---
arch/x86_64/src/intel64/intel64_rtc.c | 10 +++++++---
1 file changed, 7 insertions(+), 3 deletions(-)
diff --git a/arch/x86_64/src/intel64/intel64_rtc.c
b/arch/x86_64/src/intel64/intel64_rtc.c
index 8d01d648a9e..52ace1fbad7 100644
--- a/arch/x86_64/src/intel64/intel64_rtc.c
+++ b/arch/x86_64/src/intel64/intel64_rtc.c
@@ -100,17 +100,21 @@ static unsigned long rtc_read(void)
uint64_t tmr = rdtscp();
irqstate_t flags;
+ /* The 64-bit TSC does not wrap. A smaller value comes from a CPU whose
+ * TSC is behind the TSC of the CPU that read last: keep the last value.
+ */
+
flags = spin_lock_irqsave(&g_rtc_lock);
if (tmr < rtc_last)
{
- tmr += (0xffffffffffffffffull - rtc_last);
+ tmr = rtc_last;
}
rtc_last = tmr;
spin_unlock_irqrestore(&g_rtc_lock, flags);
- tmr = (tmr / (g_x86_64_timer_freq / 1000000ul)) * 1000l;
- return tmr;
+ return tmr / g_x86_64_timer_freq * NS_PER_SEC +
+ tmr % g_x86_64_timer_freq * NS_PER_SEC / g_x86_64_timer_freq;
}
/****************************************************************************