You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Linux likes to turn off the TWD clock. One way to get around this is to not use the TWD. However, the TWD is currently used as the 24 MHz counter for OSC16M calibration. So some alternative must be found before the TWD can be disabled.
Everything except A64/H5 has a "timestamp status" MMIO range with a free-running 64-bit counter running at 24 MHz. There is also a latched 64-bit counter in the R_CPUCFG space. This latched counter could be used if the calibration procedure was flipped, to delay using OSC16M and count OSC24M cycles (whereas now it delays using OSC24M and counts OSC16M cycles).
There are other timers available, but one concern is their availability during sleep. See #198.
The text was updated successfully, but these errors were encountered:
Now that a system counter driver is available on all supported
platforms, use that instead of the trusted watchdog.
Closes: #199
Signed-off-by: Samuel Holland <[email protected]>
Type of issue
Enhancement
Description
Linux likes to turn off the TWD clock. One way to get around this is to not use the TWD. However, the TWD is currently used as the 24 MHz counter for OSC16M calibration. So some alternative must be found before the TWD can be disabled.
Everything except A64/H5 has a "timestamp status" MMIO range with a free-running 64-bit counter running at 24 MHz. There is also a latched 64-bit counter in the R_CPUCFG space. This latched counter could be used if the calibration procedure was flipped, to delay using OSC16M and count OSC24M cycles (whereas now it delays using OSC24M and counts OSC16M cycles).
There are other timers available, but one concern is their availability during sleep. See #198.
The text was updated successfully, but these errors were encountered: