My Netra t1 AC 200 has an UltraSPARC IIe processor. However, the us2e_cpufreq driver does not work properly on it. Loading the us2e_cpufreq module sometimes crashes the entire system; the console only spews endless binary garbage and the system is completely unresponsive until powercycled. (The LOM unit still works, fortunately.) This problem is intermittent, I can trigger it some of the time but not always. However, even if the module does load without crashing the system, any attempt to change the governor from the default "performance" to "ondemand" immediately causes a watchdog reset: seraphim ~ # modprobe us2e_cpufreq seraphim ~ # cpufreq-info cpufrequtils 005: cpufreq-info (C) Dominik Brodowski 2004-2006 Report errors and bugs to cpufreq@vger.kernel.org, please. analyzing CPU 0: driver: UltraSPARC-IIe CPUs which need to switch frequency at the same time: 0 hardware limits: 62.5 MHz - 500 MHz available cpufreq governors: performance current policy: frequency should be within 62.5 MHz and 500 MHz. The governor "performance" may decide which speed to use within this range. current CPU frequency is 500 MHz (asserted by call to hardware). seraphim ~ # cpufreq-set -g ondemand seraphim ~ # Watchdog Reset Externally Initiated Reset ok
(switched to email. Please respond via emailed reply-to-all, not via the bugzilla web interface). On Mon, 8 Jun 2009 19:42:33 GMT bugzilla-daemon@bugzilla.kernel.org wrote: > http://bugzilla.kernel.org/show_bug.cgi?id=13489 > > Summary: us2e_cpufreq does not work on Netra t1 200 > Product: Power Management > Version: 2.5 > Kernel Version: 2.6.29 > Platform: All > OS/Version: Linux > Tree: Mainline > Status: NEW > Severity: normal > Priority: P1 > Component: cpufreq > AssignedTo: cpufreq@vger.kernel.org > ReportedBy: seraph@xs4all.nl > Regression: No > > > My Netra t1 AC 200 has an UltraSPARC IIe processor. However, the us2e_cpufreq > driver does not work properly on it. > > Loading the us2e_cpufreq module sometimes crashes the entire system; the > console only spews endless binary garbage and the system is completely > unresponsive until powercycled. (The LOM unit still works, fortunately.) This > problem is intermittent, I can trigger it some of the time but not always. > > However, even if the module does load without crashing the system, any > attempt > to change the governor from the default "performance" to "ondemand" > immediately > causes a watchdog reset: > > seraphim ~ # modprobe us2e_cpufreq > seraphim ~ # cpufreq-info > cpufrequtils 005: cpufreq-info (C) Dominik Brodowski 2004-2006 > Report errors and bugs to cpufreq@vger.kernel.org, please. > analyzing CPU 0: > driver: UltraSPARC-IIe > CPUs which need to switch frequency at the same time: 0 > hardware limits: 62.5 MHz - 500 MHz > available cpufreq governors: performance > current policy: frequency should be within 62.5 MHz and 500 MHz. > The governor "performance" may decide which speed to use > within this range. > current CPU frequency is 500 MHz (asserted by call to hardware). > > seraphim ~ # cpufreq-set -g ondemand > seraphim ~ # > Watchdog Reset > Externally Initiated Reset > ok
On Mon, 8 Jun 2009 15:56:54 -0700 Andrew Morton <akpm@linux-foundation.org> wrote: > (switched to email. Please respond via emailed reply-to-all, not via the > bugzilla web interface). As you wish. Now, what can I do to help resolve this problem?
On Tue, 9 Jun 2009 20:29:04 +0200 Jos van der Ende <seraph@xs4all.nl> wrote: > On Mon, 8 Jun 2009 15:56:54 -0700 > Andrew Morton <akpm@linux-foundation.org> wrote: > > > (switched to email. Please respond via emailed reply-to-all, not via the > > bugzilla web interface). > > As you wish. Well. It's not my wishes, really. Some developemnt teams prefer to work via bugzilla, others via email. I have a rolodex in my head knowing which is which and I try to route bug reports appropriately. > Now, what can I do to help resolve this problem? Wait a bit, hope that a sparc developer takes a look at reproducing this. If nothing has happened in a week or so, retest 2.6.30 and send us a status update
It's great that kernel bugzilla is back. Jos van der Ende, can you please verify if the problem still exists in the latest upstream kernel?
Wow, been a while since I reported this one. I've just tested with a 3.1.6 kernel with Gentoo patches. Yes, the problem still exists; loading the us2e-cpufreq mod caused an immediate hard reset. I'll try with vanilla 3.2.1 as soon as I have some more time, but I have no reason to assume the result will be any different.
Hi Jos, could you check whether this bug still exist in the latest upstream kernel 3.9-rc7 or stable 3.8?
I've tested with vanilla 3.8.7, the problem is still the same. Any attempt to activate frequency scaling either hangs the system entirely with random garbage on its console, or triggers a watchdog reset. This is only on the Netra t1 200, I have two Blade 100s that also use us2e_cpufreq and they both work fine with it.
re-assigned to David S. Miller, the sparc us2e cpufreq driver author.