]> git.ipfire.org Git - thirdparty/linux.git/commitdiff
drivers: hv: log when enabling crash_kexec_post_notifiers
authorStephen Brennan <stephen.s.brennan@oracle.com>
Tue, 15 Feb 2022 01:37:35 +0000 (17:37 -0800)
committerWei Liu <wei.liu@kernel.org>
Fri, 18 Feb 2022 13:09:38 +0000 (13:09 +0000)
Recently I went down a rabbit hole looking at a race condition in
panic() on a Hyper-V guest. I assumed, since it was missing from the
command line, that crash_kexec_post_notifiers was disabled. Only after
a rather long reproduction and analysis process did I learn that Hyper-V
actually enables this setting unconditionally.

Users and debuggers alike would like to know when these things happen. I
think it would be good to print a message to the kernel log when this
happens, so that a grep for "crash_kexec_post_notifiers" shows relevant
results.

Signed-off-by: Stephen Brennan <stephen.s.brennan@oracle.com>
Reviewed-by: Michael Kelley <mikelley@microsoft.com>
Link: https://lore.kernel.org/r/20220215013735.358327-1-stephen.s.brennan@oracle.com
Signed-off-by: Wei Liu <wei.liu@kernel.org>
drivers/hv/hv_common.c

index 181d16bbf49d7d6fe13d9bb6195a6b07452fd6b8..c1dd21d0d7ef8d1ba96523b602ba71a48f0963b1 100644 (file)
@@ -79,8 +79,10 @@ int __init hv_common_init(void)
         * calling crash enlightment interface before running kdump
         * kernel.
         */
-       if (ms_hyperv.misc_features & HV_FEATURE_GUEST_CRASH_MSR_AVAILABLE)
+       if (ms_hyperv.misc_features & HV_FEATURE_GUEST_CRASH_MSR_AVAILABLE) {
                crash_kexec_post_notifiers = true;
+               pr_info("Hyper-V: enabling crash_kexec_post_notifiers\n");
+       }
 
        /*
         * Allocate the per-CPU state for the hypercall input arg.