https://bugzilla.novell.com/show_bug.cgi?id=333043#c49
Thomas Renninger changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEEDINFO |ASSIGNED
Info Provider|sontek@gmail.com |
--- Comment #49 from Thomas Renninger 2007-11-14 01:58:03 MST ---
Thanks to coolo, I think I now know what is going on:
the passive trip point provided by BIOS is a kind of dummy..:
The real mechanism to prevent the machine for a critical shutdown is done
through processor events and lower frequency through _PPC.
So you should look out for an acpi event (via acpi_listen) like:
CPU 00000080 00000004 (limit frequency to the lowest, if 4 freqs are
provided)
CPU 00000080 00000000 (allow all freqs again)
This also explains Robert's comment #43 (I unset the private flag for it..),
where he states that it happened that the CPU got limited to the lowest
frequency which does not come from the passive cooling interface, but from
simply lowering the freq to the lowest by BIOS.
The mechanism itself works. I tried with one process producing CPU load and
could not get the machine to shut down.
I expect that if you do a "make -j4" if you compile a kernel or you have
several CPU intensive tasks, it takes too long until the thread to lower the
CPU is scheduled.
So I expect we have priority/schedule issue here.
Maybe Len already knows about changes in this area, if not I am going to dig
for a solution...
--
Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.