Drivers: hv: vmbus: don't manipulate with clocksources on crash
authorVitaly Kuznetsov <vkuznets@redhat.com>
Thu, 28 Jan 2016 06:29:36 +0000 (22:29 -0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Mon, 8 Feb 2016 05:34:12 +0000 (21:34 -0800)
clocksource_change_rating() involves mutex usage and can't be called
in interrupt context. It also makes sense to avoid doing redundant work
on crash.

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/hv/hv.c

index 1c677d0f16d61e486da4412d06eea7b2a439998e..ccb335f57c8823bdadd79bc03e60140c5d07c9ba 100644 (file)
@@ -295,8 +295,14 @@ void hv_cleanup(void)
         * Cleanup the TSC page based CS.
         */
        if (ms_hyperv.features & HV_X64_MSR_REFERENCE_TSC_AVAILABLE) {
-               clocksource_change_rating(&hyperv_cs_tsc, 10);
-               clocksource_unregister(&hyperv_cs_tsc);
+               /*
+                * Crash can happen in an interrupt context and unregistering
+                * a clocksource is impossible and redundant in this case.
+                */
+               if (!oops_in_progress) {
+                       clocksource_change_rating(&hyperv_cs_tsc, 10);
+                       clocksource_unregister(&hyperv_cs_tsc);
+               }
 
                hypercall_msr.as_uint64 = 0;
                wrmsrl(HV_X64_MSR_REFERENCE_TSC, hypercall_msr.as_uint64);