See http://sourceware.org/ml/systemtap/2009-q2/msg00635.html In particular: BUG: scheduling while atomic: stapio/0x00000001/26142, CPU#3 [...] [<ffffffff812a2dea>] cpufreq_unregister_notifier+0x35/0x5c [<ffffffffa02e0a1f>] _stp_kill_time+0xb6/0xbd [stap_246f93f30a500769142af9987624737a_5072] [<ffffffffa02e1749>] probe_1391+0x3c/0xa8 [stap_246f93f30a500769142af9987624737a_5072] [<ffffffffa02e2621>] enter_end_probe+0x14a/0x1e3 [stap_246f93f30a500769142af9987624737a_5072] It seems that cpufreq_unregister_notifier is sleepy, and so shouldn't be called from probe context.
This should be fixed in git now, but I don't have an RT kernel to test with. Please test and confirm with this fix: commit 29e2616aeeb82605a6efe1dbc574b499781eafbe Author: Josh Stone <jistone@redhat.com> Date: Wed May 20 14:46:25 2009 -0700 PR10177: init/kill time in sleepy context only
Hi Josh, I tested your fix with RT kernel. The bug is resolved now. Thanks a lot
Fix confirmed...