Re: [tip:x86/debug] x86, NMI: Add NMI IPI selftest

[Don Zickus <dzickus@redhat.com>]

On Tue, Dec 06, 2011 at 10:06:12AM +0100, Ingo Molnar wrote:
> 
> * tip-bot for Don Zickus <dzickus@redhat.com> wrote:
> 
> > Commit-ID:  99e8b9ca90d688c3ac7d3a141b701c9694a93925
> > Gitweb:     http://git.kernel.org/tip/99e8b9ca90d688c3ac7d3a141b701c9694a93925
> > Author:     Don Zickus <dzickus@redhat.com>
> > AuthorDate: Thu, 13 Oct 2011 15:14:26 -0400
> > Committer:  Ingo Molnar <mingo@elte.hu>
> > CommitDate: Mon, 5 Dec 2011 12:00:16 +0100
> > 
> > x86, NMI: Add NMI IPI selftest
> 
> The NMI test-suite has caught some badness:
> 
> [    0.179992] calling  tracer_alloc_buffers+0x0/0x224 @ 1
> [    0.183417] initcall tracer_alloc_buffers+0x0/0x224 returned 0 after 0 usecs
> [    0.186658] calling  init_trace_printk+0x0/0xf @ 1
> [    0.189992] initcall init_trace_printk+0x0/0xf returned 0 after 0 usecs
> [    0.193324] Brought up 1 CPUs
> [    0.196657] ----------------
> [    0.199989] | NMI testsuite:
> [    0.203322] --------------------
> [    0.206655]   remote IPI:
> [    0.209991] ------------[ cut here ]------------
> [    0.213325] WARNING: at arch/x86/kernel/apic/ipi.c:109 default_send_IPI_mask_logical+0xdb/0x120()
> [    0.216655] Hardware name: System Product Name
> [    0.219987] empty IPI mask
> [    0.223321] Modules linked in:
> [    0.227015] Pid: 1, comm: swapper/0 Tainted: G        W    3.2.0-rc4-tip+ #163124
> [    0.229987] Call Trace:
> [    0.233326]  [<c102e2a2>] warn_slowpath_common+0x72/0xa0
> [    0.236656]  [<c101e04b>] ? default_send_IPI_mask_logical+0xdb/0x120
> [    0.239989]  [<c101e04b>] ? default_send_IPI_mask_logical+0xdb/0x120
> [    0.243321]  [<c102e373>] warn_slowpath_fmt+0x33/0x40
> [    0.246655]  [<c101e04b>] default_send_IPI_mask_logical+0xdb/0x120
> [    0.249988]  [<c1024025>] test_nmi_ipi.constprop.2+0x55/0xb0
> [    0.253320]  [<c10240c8>] remote_ipi+0x48/0x80
> [    0.256654]  [<c102417d>] dotest.constprop.3+0xd/0x90
> [    0.259987]  [<c1024265>] nmi_selftest+0x65/0x120
> [    0.263320]  [<c14d4e73>] native_smp_cpus_done+0x15/0xff
> [    0.266653]  [<c14ddf3f>] smp_init+0xdb/0xe2
> [    0.269988]  [<c14cb7bb>] ? start_kernel+0x386/0x386
> [    0.273320]  [<c14cb81d>] kernel_init+0x62/0x118
> [    0.276653]  [<c138bb3e>] kernel_thread_helper+0x6/0x10
> 
> That's an empty IPI mask. NOTE the boot parameters i used:

I'm on paternity leave, so I hacked this up for now and compiled tested
this.  On Monday I can dive into this further if the patch below doesn't
solve it.

----8<----
From: Don Zickus <dzickus@redhat.com>
Date: Tue, 6 Dec 2011 13:03:18 -0500
Subject: [PATCH] x86, NMI: NMI-selftest should handle UP case properly

If no remote cpus are online, then just quietly skip the remote IPI
test for now.

Signed-off-by: Don Zickus <dzickus@redhat.com>
---
 arch/x86/kernel/nmi_selftest.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/arch/x86/kernel/nmi_selftest.c b/arch/x86/kernel/nmi_selftest.c
index 572adb6..1e42a23 100644
--- a/arch/x86/kernel/nmi_selftest.c
+++ b/arch/x86/kernel/nmi_selftest.c
@@ -90,7 +90,8 @@ static void remote_ipi(void)
 {
 	cpumask_copy(to_cpumask(nmi_ipi_mask), cpu_online_mask);
 	cpumask_clear_cpu(smp_processor_id(), to_cpumask(nmi_ipi_mask));
-	test_nmi_ipi(to_cpumask(nmi_ipi_mask));
+	if (!cpumask_empty(nmi_ipi_mask))
+		test_nmi_ipi(to_cpumask(nmi_ipi_mask));
 }
 
 static void local_ipi(void)
-- 
1.7.6.4