Hi all,
It seems that a recent commit to OpenBIOS to increase the L11 timer
frequency from 10Hz to 100Hz (see
http://git.qemu.org/?p=openbios.git;a=commit;h=aad027388a5f0044781a95b184572ff0c38bbe37)
caused HelenOS boot to become extremely unreliable - it would tend to
hang most times at "Booting the kernel" instead of switching to the
splash screen and continuing to boot.
With a bit of detective work, Artyom managed to trace it back to this
particular commit which simply increases the timer frequency from 10Hz
to 100Hz to match the OBP timer frequency. I'm wondering if maybe this
is because HelenOS gets confused if the L11 timer softint happens to be
asserted when booting the kernel which is considerably more likely with
an increase in timer frequency?
I've just posted a patch to the OpenBIOS list to revert the frequency
increase at
http://www.openfirmware.info/pipermail/openbios/2014-August/008496.html
at the expense of the timer resolution but I thought it may be useful to
raise this here just in case it's caused by a bug somewhere in the
interrupt handling as none of Linux/FreeBSD/Solaris seem to have this
problem.
FWIW another alternative solution could be for OpenBIOS to add timer
clearup functionality to SUNW,set-trap-table similar to OBP and so if
HelenOS were to call that method instead of directly altering %tba, then
it could become the responsibility of the PROM instead. Thoughts?
ATB,
Mark.
_______________________________________________
HelenOS-devel mailing list
[email protected]
http://lists.modry.cz/listinfo/helenos-devel
