[PATCH 1/9] x86, xsave: warn on #NM exceptions caused by the kernel

From: Hans Rosenfeld <hans.rosenfeld@amd.com>
To: <hpa@zytor.com>
Cc: <tglx@linutronix.de>, <mingo@elte.hu>,
	<suresh.b.siddha@intel.com>, <eranian@google.com>,
	<brgerst@gmail.com>, <robert.richter@amd.com>,
	<Andreas.Herrmann3@amd.com>, <x86@kernel.org>,
	<linux-kernel@vger.kernel.org>,
	Hans Rosenfeld <hans.rosenfeld@amd.com>
Subject: [PATCH 1/9] x86, xsave: warn on #NM exceptions caused by the kernel
Date: Tue, 29 Nov 2011 13:41:20 +0100	[thread overview]
Message-ID: <1322570488-21798-2-git-send-email-hans.rosenfeld@amd.com> (raw)
In-Reply-To: <1322570488-21798-1-git-send-email-hans.rosenfeld@amd.com>

The kernel code handling the FPU states should handle the TS bit in such
a way that #NM exceptions cannot happen inside the kernel. Any other
kernel code using FPU features should use kernel_fpu_begin() and
kernel_fpu_end(), which handles the TS bit and disallows preemption.

So if a #NM exception ever comes from kernel mode, it would indicate a
serious bug. Trapping this with WARN_ON_ONCE() could prove helpful in
finding and eliminating such bugs.

Signed-off-by: Hans Rosenfeld <hans.rosenfeld@amd.com>
---
 arch/x86/kernel/traps.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c
index 6913369..953d5f6 100644
--- a/arch/x86/kernel/traps.c
+++ b/arch/x86/kernel/traps.c
@@ -777,6 +777,8 @@ EXPORT_SYMBOL_GPL(math_state_restore);
 dotraplinkage void __kprobes
 do_device_not_available(struct pt_regs *regs, long error_code)
 {
+	WARN_ON_ONCE(!user_mode_vm(regs));
+
 #ifdef CONFIG_MATH_EMULATION
 	if (read_cr0() & X86_CR0_EM) {
 		struct math_emu_info info = { };
-- 
1.7.5.4


