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: <283d7660-000a-81af-1d6b-4fdedcb6dfd8@intel.com>
Date:   Fri, 8 Jul 2022 08:13:13 -0700
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.9.1
Subject: Re: [PATCH] x86/mm/tlb: ignore f->new_tlb_gen when zero
Content-Language: en-US
To:     David Hildenbrand <david@redhat.com>,
        Nadav Amit <nadav.amit@gmail.com>,
        linux-kernel@vger.kernel.org, Hugh Dickins <hughd@google.com>,
        Thomas Gleixner <tglx@linutronix.de>
Cc:     Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
        x86@kernel.org, Linux MM <linux-mm@kvack.org>,
        Nadav Amit <namit@vmware.com>,
        Dave Hansen <dave.hansen@linux.intel.com>,
        Peter Zijlstra <peterz@infradead.org>,
        Andy Lutomirski <luto@kernel.org>
References: <20220708003053.158480-1-namit@vmware.com>
 <b573a9c6-2ee0-20dc-1f28-84d3a81d40a2@redhat.com>
From:   Dave Hansen <dave.hansen@intel.com>
In-Reply-To: <b573a9c6-2ee0-20dc-1f28-84d3a81d40a2@redhat.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 7/8/22 04:40, David Hildenbrand wrote:
>> diff --git a/arch/x86/mm/tlb.c b/arch/x86/mm/tlb.c
>> index d9314cc8b81f..d81b4084bb8a 100644
>> --- a/arch/x86/mm/tlb.c
>> +++ b/arch/x86/mm/tlb.c
>> @@ -771,14 +771,14 @@ static void flush_tlb_func(void *info)
>>  		return;
>>  	}
>>  
>> -	if (f->new_tlb_gen <= local_tlb_gen) {
>> +	if (unlikely(f->new_tlb_gen != 0 && f->new_tlb_gen <= local_tlb_gen)) {
>>  		/*
>>  		 * The TLB is already up to date in respect to f->new_tlb_gen.
>>  		 * While the core might be still behind mm_tlb_gen, checking
>>  		 * mm_tlb_gen unnecessarily would have negative caching effects
>>  		 * so avoid it.
>>  		 */
>> -		return;
>> +		goto done;
> Does this affect the performance numbers from aa44284960d5 ("x86/mm/tlb:
> Avoid reading mm_tlb_gen when possible")?

It depends on how many batched flushes that workload had.  From the
looks of it, they're all one page:

	madvise(addr + i, pgsize, MADV_DONTNEED);

so there shouldn't be *much* batching in play.  But, it wouldn't hurt to
re-run them in either case.