Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <b050f00b-6c3a-a0d9-a3c1-175a724faf1c@igalia.com>
Date:   Thu, 1 Sep 2022 15:46:17 -0300
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.12.0
Subject: Re: [PATCH V3] firmware: google: Test spinlock on panic path to avoid
 lockups
Content-Language: en-US
To:     Greg KH <gregkh@linuxfoundation.org>
Cc:     evgreen@chromium.org, arnd@arndb.de, linux-efi@vger.kernel.org,
        linux-kernel@vger.kernel.org, kernel@gpiccoli.net, ardb@kernel.org,
        davidgow@google.com, jwerner@chromium.org,
        Petr Mladek <pmladek@suse.com>
References: <20220819155059.451674-1-gpiccoli@igalia.com>
 <YxDVPqVkdgQbAIvY@kroah.com>
 <f89cd87c-7d1c-d8e6-ed95-6876f0201872@igalia.com>
 <YxDX9+p+58q2sip2@kroah.com>
 <6bc5dbc3-2cdd-5cb8-1632-11de2008a85a@igalia.com>
 <YxDhiSDs4YcUrqV5@kroah.com>
 <85683284-db85-7e3a-57bd-750e1c204e3e@igalia.com>
 <YxD56RTI9v/P2QOL@kroah.com>
From:   "Guilherme G. Piccoli" <gpiccoli@igalia.com>
In-Reply-To: <YxD56RTI9v/P2QOL@kroah.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 01/09/2022 15:28, Greg KH wrote:
> [...]
>> I honestly didn't understand exactly what you're suggesting Greg...
>> Mind clarifying?
> 
> Something like this totally untested code:
> 
> diff --git a/drivers/firmware/google/gsmi.c b/drivers/firmware/google/gsmi.c
> index adaa492c3d2d..6ad41b22671c 100644
> --- a/drivers/firmware/google/gsmi.c
> +++ b/drivers/firmware/google/gsmi.c
> @@ -19,6 +19,7 @@
>  #include <linux/dma-mapping.h>
>  #include <linux/fs.h>
>  #include <linux/slab.h>
> +#include <linux/panic.h>
>  #include <linux/panic_notifier.h>
>  #include <linux/ioctl.h>
>  #include <linux/acpi.h>
> @@ -611,6 +612,11 @@ static const struct attribute *gsmi_attrs[] = {
>  	NULL,
>  };
>  
> +static bool panic_in_progress(void)
> +{
> +	return unlikely(atomic_read(&panic_cpu) != PANIC_CPU_INVALID);
> +}
> +
>  static int gsmi_shutdown_reason(int reason)
>  {
>  	struct gsmi_log_entry_type_1 entry = {
> @@ -629,7 +635,8 @@ static int gsmi_shutdown_reason(int reason)
>  	if (saved_reason & (1 << reason))
>  		return 0;
>  
> -	spin_lock_irqsave(&gsmi_dev.lock, flags);
> +	if (!panic_in_progress())
> +		spin_lock_irqsave(&gsmi_dev.lock, flags);
>  
>  	saved_reason |= (1 << reason);
>  
> @@ -644,7 +651,8 @@ static int gsmi_shutdown_reason(int reason)
>  
>  	rc = gsmi_exec(GSMI_CALLBACK, GSMI_CMD_SET_EVENT_LOG);
>  
> -	spin_unlock_irqrestore(&gsmi_dev.lock, flags);
> +	if (!panic_in_progress())
> +		spin_unlock_irqrestore(&gsmi_dev.lock, flags);
>  
>  	if (rc < 0)
>  		printk(KERN_ERR "gsmi: Log Shutdown Reason failed\n");
> 
> 
>

Thanks! Personally, I feel the approach a bit more complex than mine,
and...racy!
Imagine CPU0 runs your tests, right after the if (!panic_in_progress())
is done, spinlock is taken and boom - panic on CPU1. This would cause
the same issue...

My approach is zero racy, since it checks if spinlock was taken in a
moment that the machine is like a no-SMP, only a single CPU running...


> That being said, are you sure spinlocks are still held in the panic
> notifier?  What about the call to bust_spinlocks() that is called in
> panic() already?  Wouldn't that have already dropped whatever you were
> worried about here?

This function is very weird. Basically, the call of "bust_spinlocks(1);"
in panic effectively means "++oops_in_progress;" IIUC.
So, I still think we can have lockups in panic notifiers with locks
previously taken =)

Cheers,


Guilherme