diff mbox series

x86/nmi: Fix "in NMI handler" check

Message ID 20240207165237.1048837-1-leitao@debian.org
State New
Headers
Series x86/nmi: Fix "in NMI handler" check |

Commit Message

Breno Leitao Feb. 7, 2024, 4:52 p.m. UTC 
  Commit 344da544f177 ("x86/nmi: Print reasons why backtrace NMIs are
ignored") creates a super nice framework to diagnose NMIs.

Every time nmi_exc() is called, it increments a per_cpu counter
(nsp->idt_nmi_seq). At its exit, it also increments the same counter.
Looking at this counter, you can see how many times that function was
called (dividing by 2), and, if the function is still being executed, by
checking the idt_nmi_seq's last bit.

On the check side (nmi_backtrace_stall_check()), that variable is
queried to check if the NMI is still being executed, but, there is a
mistake in the bitwise operation. That code wants to check if the last
bit of the idt_nmi_seq is set or not, but, does the opposite, and check
for all the other bits, which will always be true after the first
exc_nmi() executed successfully.

This appends the misleading string to the dump "(CPU currently in NMI
handler function)"

Fix it by checking the last bit, and if it is set, append the string.

Fixes: 344da544f177 ("x86/nmi: Print reasons why backtrace NMIs are ignored")
Signed-off-by: Breno Leitao <leitao@debian.org>
---
 arch/x86/kernel/nmi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Paul E. McKenney Feb. 7, 2024, 6:30 p.m. UTC | #1 
On Wed, Feb 07, 2024 at 08:52:35AM -0800, Breno Leitao wrote:
> Commit 344da544f177 ("x86/nmi: Print reasons why backtrace NMIs are
> ignored") creates a super nice framework to diagnose NMIs.
> 
> Every time nmi_exc() is called, it increments a per_cpu counter
> (nsp->idt_nmi_seq). At its exit, it also increments the same counter.
> Looking at this counter, you can see how many times that function was
> called (dividing by 2), and, if the function is still being executed, by
> checking the idt_nmi_seq's last bit.
> 
> On the check side (nmi_backtrace_stall_check()), that variable is
> queried to check if the NMI is still being executed, but, there is a
> mistake in the bitwise operation. That code wants to check if the last
> bit of the idt_nmi_seq is set or not, but, does the opposite, and check
> for all the other bits, which will always be true after the first
> exc_nmi() executed successfully.
> 
> This appends the misleading string to the dump "(CPU currently in NMI
> handler function)"
> 
> Fix it by checking the last bit, and if it is set, append the string.
> 
> Fixes: 344da544f177 ("x86/nmi: Print reasons why backtrace NMIs are ignored")
> Signed-off-by: Breno Leitao <leitao@debian.org>

If someone else is taking this:

Reviewed-by: Paul E. McKenney <paulmck@kernel.org>

(I am queueing it for testing in any case.)

> ---
>  arch/x86/kernel/nmi.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/arch/x86/kernel/nmi.c b/arch/x86/kernel/nmi.c
> index 17e955ab69fe..6e738ad474dc 100644
> --- a/arch/x86/kernel/nmi.c
> +++ b/arch/x86/kernel/nmi.c
> @@ -639,7 +639,7 @@ void nmi_backtrace_stall_check(const struct cpumask *btp)
>  			msgp = nmi_check_stall_msg[idx];
>  			if (nsp->idt_ignored_snap != READ_ONCE(nsp->idt_ignored) && (idx & 0x1))
>  				modp = ", but OK because ignore_nmis was set";
> -			if (nmi_seq & ~0x1)
> +			if (nmi_seq & 0x1)
>  				msghp = " (CPU currently in NMI handler function)";
>  			else if (nsp->idt_nmi_seq_snap + 1 == nmi_seq)
>  				msghp = " (CPU exited one NMI handler function)";
> -- 
> 2.39.3
>
Xin Li (Intel) Feb. 7, 2024, 6:44 p.m. UTC | #2 
On 2/7/2024 10:30 AM, Paul E. McKenney wrote:
> On Wed, Feb 07, 2024 at 08:52:35AM -0800, Breno Leitao wrote:
>> Commit 344da544f177 ("x86/nmi: Print reasons why backtrace NMIs are
>> ignored") creates a super nice framework to diagnose NMIs.
>>
>> Every time nmi_exc() is called, it increments a per_cpu counter
>> (nsp->idt_nmi_seq). At its exit, it also increments the same counter.
>> Looking at this counter, you can see how many times that function was
>> called (dividing by 2), and, if the function is still being executed, by
>> checking the idt_nmi_seq's last bit.
>>
>> On the check side (nmi_backtrace_stall_check()), that variable is
>> queried to check if the NMI is still being executed, but, there is a
>> mistake in the bitwise operation. That code wants to check if the last
>> bit of the idt_nmi_seq is set or not, but, does the opposite, and check
>> for all the other bits, which will always be true after the first
>> exc_nmi() executed successfully.
>>
>> This appends the misleading string to the dump "(CPU currently in NMI
>> handler function)"
>>
>> Fix it by checking the last bit, and if it is set, append the string.
>>
>> Fixes: 344da544f177 ("x86/nmi: Print reasons why backtrace NMIs are ignored")
>> Signed-off-by: Breno Leitao <leitao@debian.org>
> 
> If someone else is taking this:
> 
> Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
> 
> (I am queueing it for testing in any case.)

Does this fix need to be backported?

Commit 344da544f177 has a date on Dec 16 2022.

> 
>> ---
>>   arch/x86/kernel/nmi.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/arch/x86/kernel/nmi.c b/arch/x86/kernel/nmi.c
>> index 17e955ab69fe..6e738ad474dc 100644
>> --- a/arch/x86/kernel/nmi.c
>> +++ b/arch/x86/kernel/nmi.c
>> @@ -639,7 +639,7 @@ void nmi_backtrace_stall_check(const struct cpumask *btp)
>>   			msgp = nmi_check_stall_msg[idx];
>>   			if (nsp->idt_ignored_snap != READ_ONCE(nsp->idt_ignored) && (idx & 0x1))
>>   				modp = ", but OK because ignore_nmis was set";
>> -			if (nmi_seq & ~0x1)
>> +			if (nmi_seq & 0x1)
>>   				msghp = " (CPU currently in NMI handler function)";
>>   			else if (nsp->idt_nmi_seq_snap + 1 == nmi_seq)
>>   				msghp = " (CPU exited one NMI handler function)";
>> -- 
>> 2.39.3
>>
>
Breno Leitao Feb. 8, 2024, 2:59 p.m. UTC | #3 
On Wed, Feb 07, 2024 at 10:44:57AM -0800, Xin Li wrote:
> On 2/7/2024 10:30 AM, Paul E. McKenney wrote:
> > On Wed, Feb 07, 2024 at 08:52:35AM -0800, Breno Leitao wrote:
> > > Commit 344da544f177 ("x86/nmi: Print reasons why backtrace NMIs are
> > > ignored") creates a super nice framework to diagnose NMIs.
> > > 
> > > Every time nmi_exc() is called, it increments a per_cpu counter
> > > (nsp->idt_nmi_seq). At its exit, it also increments the same counter.
> > > Looking at this counter, you can see how many times that function was
> > > called (dividing by 2), and, if the function is still being executed, by
> > > checking the idt_nmi_seq's last bit.
> > > 
> > > On the check side (nmi_backtrace_stall_check()), that variable is
> > > queried to check if the NMI is still being executed, but, there is a
> > > mistake in the bitwise operation. That code wants to check if the last
> > > bit of the idt_nmi_seq is set or not, but, does the opposite, and check
> > > for all the other bits, which will always be true after the first
> > > exc_nmi() executed successfully.
> > > 
> > > This appends the misleading string to the dump "(CPU currently in NMI
> > > handler function)"
> > > 
> > > Fix it by checking the last bit, and if it is set, append the string.
> > > 
> > > Fixes: 344da544f177 ("x86/nmi: Print reasons why backtrace NMIs are ignored")
> > > Signed-off-by: Breno Leitao <leitao@debian.org>
> > 
> > If someone else is taking this:
> > 
> > Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
> > 
> > (I am queueing it for testing in any case.)
> 
> Does this fix need to be backported?
> 
> Commit 344da544f177 has a date on Dec 16 2022.

I would say so, if users are using this detection mechanism and want to
trust the messages being displayed by the kernel.
diff mbox series

Patch

diff --git a/arch/x86/kernel/nmi.c b/arch/x86/kernel/nmi.c
index 17e955ab69fe..6e738ad474dc 100644
--- a/arch/x86/kernel/nmi.c
+++ b/arch/x86/kernel/nmi.c
@@ -639,7 +639,7 @@  void nmi_backtrace_stall_check(const struct cpumask *btp)
 			msgp = nmi_check_stall_msg[idx];
 			if (nsp->idt_ignored_snap != READ_ONCE(nsp->idt_ignored) && (idx & 0x1))
 				modp = ", but OK because ignore_nmis was set";
-			if (nmi_seq & ~0x1)
+			if (nmi_seq & 0x1)
 				msghp = " (CPU currently in NMI handler function)";
 			else if (nsp->idt_nmi_seq_snap + 1 == nmi_seq)
 				msghp = " (CPU exited one NMI handler function)";