linux-kernel - Re: [PATCH] genirq: update irq stats from NMI handlers

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <a4ce3800-22f4-72dc-6ff8-75dfed1c377b@marvell.com>
Date:   Thu, 4 Jul 2019 10:36:57 +0000
From:   Shijith Thotton <sthotton@...vell.com>
To:     Julien Thierry <julien.thierry@....com>,
        Thomas Gleixner <tglx@...utronix.de>
CC:     "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-arm-kernel@...ts.infradead.org" 
        <linux-arm-kernel@...ts.infradead.org>,
        Catalin Marinas <catalin.marinas@....com>,
        Will Deacon <will.deacon@....com>,
        "Jayachandran Chandrasekharan Nair" <jnair@...vell.com>,
        Ganapatrao Kulkarni <gkulkarni@...vell.com>,
        Jan Glauber <jglauber@...vell.com>,
        Robert Richter <rrichter@...vell.com>
Subject: Re: [PATCH] genirq: update irq stats from NMI handlers


Hi Julien,

On 7/4/19 12:13 AM, Julien Thierry wrote:
> On 04/07/2019 05:22, Shijith Thotton wrote:
>> The NMI handlers handle_percpu_devid_fasteoi_nmi() and
>> handle_fasteoi_nmi() added by commit 2dcf1fbcad35 ("genirq: Provide NMI
>> handlers") do not update the interrupt counts. Due to that the NMI
>> interrupt count does not show up correctly in /proc/interrupts.
>>
>> Update the functions to fix this. With this change, we can see stats of
>> the perf NMI interrupts on arm64.
>>
>> Fixes: 2dcf1fbcad35 ("genirq: Provide NMI handlers")
>>
>> Signed-off-by: Shijith Thotton <sthotton@...vell.com>
>> ---
>>   kernel/irq/chip.c | 4 ++++
>>   1 file changed, 4 insertions(+)
>>
>> diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c
>> index 29d6c7d070b4..88d1e054c6ea 100644
>> --- a/kernel/irq/chip.c
>> +++ b/kernel/irq/chip.c
>> @@ -748,6 +748,8 @@ void handle_fasteoi_nmi(struct irq_desc *desc)
>>   	unsigned int irq = irq_desc_get_irq(desc);
>>   	irqreturn_t res;
>>   
>> +	kstat_incr_irqs_this_cpu(desc);
>> +
> 
> This needs to be called with the desc lock taken, otherwise we're likely
> to corrupt desc->tot_count.
> But taking the desc lock is something we can't do in NMI context (
> *spin_lock_irq*() won't prevent an NMI from happening).
> 
>>   	trace_irq_handler_entry(irq, action);
>>   	/*
>>   	 * NMIs cannot be shared, there is only one action.
>> @@ -962,6 +964,8 @@ void handle_percpu_devid_fasteoi_nmi(struct irq_desc *desc)
>>   	unsigned int irq = irq_desc_get_irq(desc);
>>   	irqreturn_t res;
>>   
>> +	__kstat_incr_irqs_this_cpu(desc);
>> +
> 
> Looking at handle_percpu_irq(), I think this might be acceptable. But
> does it make sense to only have kstats for percpu NMIs?
> 

It would be better to have stats for both.

handle_fasteoi_nmi() can use __kstat_incr_irqs_this_cpu() if below 
change can be added to kstat_irqs_cpu().

diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c
index a92b33593b8d..9484e88dabc2 100644
--- a/kernel/irq/irqdesc.c
+++ b/kernel/irq/irqdesc.c
@@ -950,6 +950,11 @@ unsigned int kstat_irqs_cpu(unsigned int irq, int cpu)
                         *per_cpu_ptr(desc->kstat_irqs, cpu) : 0;
  }

+static bool irq_is_nmi(struct irq_desc *desc)
+{
+       return desc->istate & IRQS_NMI;
+}
+
  /**
   * kstat_irqs - Get the statistics for an interrupt
   * @irq:       The interrupt number
@@ -967,7 +972,8 @@ unsigned int kstat_irqs(unsigned int irq)
         if (!desc || !desc->kstat_irqs)
                 return 0;
         if (!irq_settings_is_per_cpu_devid(desc) &&
-           !irq_settings_is_per_cpu(desc))
+           !irq_settings_is_per_cpu(desc) &&
+           !irq_is_nmi(desc))
             return desc->tot_count;

         for_each_possible_cpu(cpu)


Thomas,
Please suggest a better way if any.

Thanks,
Shijith