Date:   Wed, 10 Apr 2019 05:35:46 -0700
From:   Fenghua Yu <fenghua.yu@intel.com>
To:     Thomas Gleixner <tglx@linutronix.de>
Cc:     Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
        H Peter Anvin <hpa@zytor.com>,
        Dave Hansen <dave.hansen@intel.com>,
        Paolo Bonzini <pbonzini@redhat.com>,
        Ashok Raj <ashok.raj@intel.com>,
        Peter Zijlstra <peterz@infradead.org>,
        Kalle Valo <kvalo@codeaurora.org>,
        Xiaoyao Li <xiaoyao.li@intel.com>,
        Michael Chan <michael.chan@broadcom.com>,
        Ravi V Shankar <ravi.v.shankar@intel.com>,
        linux-kernel <linux-kernel@vger.kernel.org>,
        x86 <x86@kernel.org>, linux-wireless@vger.kernel.org,
        netdev@vger.kernel.org, kvm@vger.kernel.org
Subject: Re: [PATCH v6 13/20] x86/split_lock: Enable split lock detection by
 default
Message-ID: <20190410123546.GA214455@romley-ivt3.sc.intel.com>
References: <1554326526-172295-1-git-send-email-fenghua.yu@intel.com>
 <1554326526-172295-14-git-send-email-fenghua.yu@intel.com>
 <alpine.DEB.2.21.1904041932500.1802@nanos.tec.linutronix.de>
 <20190410000229.GA209676@romley-ivt3.sc.intel.com>
 <alpine.DEB.2.21.1904100827530.2020@nanos.tec.linutronix.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <alpine.DEB.2.21.1904100827530.2020@nanos.tec.linutronix.de>
User-Agent: Mutt/1.5.23 (2014-03-12)
Sender: linux-wireless-owner@vger.kernel.org
Precedence: bulk

On Wed, Apr 10, 2019 at 08:31:31AM +0200, Thomas Gleixner wrote:
> On Tue, 9 Apr 2019, Fenghua Yu wrote:
> > On Thu, Apr 04, 2019 at 08:07:57PM +0200, Thomas Gleixner wrote:
> > > On Wed, 3 Apr 2019, Fenghua Yu wrote:
> > > >  static void early_init_intel(struct cpuinfo_x86 *c)
> > > >  {
> > > >  	u64 misc_enable;
> > > >  
> > > > +	init_split_lock_detect(c);
> > > 
> > > so we have in early boot:
> > > 
> > > 	early_cpu_init()
> > > 	  early_identify_cpu()
> > > 	    this_cpu->c_early_init(c)
> > > 	      early_init_intel() {
> > > 	        init_split_lock_detect();
> > > 	      }	
> > >             ....
> > >             cpu_set_core_cap_bits(c)
> > > 	       set(FEATURE_SPLIT_LOCK)
> > > 
> > > I don't have to understand how init_split_lock_detect() will magically see
> > > the feature bit which gets set afterwards, right? 
> > 
> > early_init_intel() is called twice on the boot CPU. Besides it's called
> > in earl_cpu_init(), it's also called in:
> > 	identify_boot_cpu()
> > 		identify_cpu()
> > 			init_intel()
> > 				early_init_intel()
> > 					init_split_lock_detect();
> > 
> > It's true that init_split_lock_detect() doesn't see the feature bit when
> > it's called for the first time in early_cpu_init(). But it sees the feature
> > bit when it's called for the second time in identify_boot_cpu().
> 
> That's hideous, really. 
> 
> > So is init_split_lock_detect() in the right place?
> 
> You're not seriously asking that?
> 
> It's obviously not the right place. We are not placing calls at random
> points just because they happen to work by chance.

Is it OK to put init_split_lock_detect(c) after early_init_intel() in
init_intel()? X86_FEATURE_SPLIT_LOCK_DETECT is available now and
init_split_lock_detec() is called only once on each CPU.

@@ -746,6 +749,8 @@ static void init_intel(struct cpuinfo_x86 *c)
 {
        early_init_intel(c);
 
+       init_split_lock_detect(c);
+
        intel_workarounds(c);

Thanks.

-Fenghua