HalQueryRealTimeClock function disables the nmi

[Mikel C.]

Hi all.
I've seen that the HalQueryRealTimeClock function (called every hour
in my PC) uses the port 70 (the same used by the nmi interruption), it
puts some values on it, and when the function ends, it always puts an
FF value on this port (even if the value that was in the port before
the function was a different one), disabling the posterior nmi calls.
Is this an Operating system (XP) failure?
Could I avoid this behaviour anyway?
Thanks in advance.
Mikel.

[Mikel C.]

Post by Mikel C.
Hi all.
I've seen that the HalQueryRealTimeClock function (called every hour
in my PC) uses the port 70 (the same used by the nmi interruption), it
puts some values on it, and when the function ends, it always puts an
FF value on this port (even if the value that was in the port before
the function was a different one), disabling the posterior nmi calls.
Is this an Operating system (XP) failure?
Could I avoid this behaviour anyway?
Thanks in advance.
Mikel.

Hi again.
The problem seems to be with the restoration of the value that was in
the port 70. For this issue, the system's function reads the value
stored in the port and later it tries to restore it. The problem with
this is that the port 70 is a write-only port and the read instruction
always returns ff (so that's the value HalQueryRealTimeClock
restores!!)
Is there any way of telling the hal that in my PC, the port 70 is
write-only?
Is there any hal patch that fixes what seems to be an incorrect
behaviour?
Thanks.
Mikel.