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: <ad2d9610-61c0-4719-9df1-0116ef317d8a@intel.com>
Date:   Tue, 18 Jul 2023 14:36:53 -0700
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.11.0
Subject: Re: [PATCH] x86/sgx: fix a NULL pointer
Content-Language: en-US
To:     Haitao Huang <haitao.huang@linux.intel.com>,
        Jarkko Sakkinen <jarkko@kernel.org>,
        dave.hansen@linux.intel.com, linux-kernel@vger.kernel.org,
        linux-sgx@vger.kernel.org, Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
        x86@kernel.org, "H. Peter Anvin" <hpa@zytor.com>
Cc:     kai.huang@intel.com, reinette.chatre@intel.com,
        kristen@linux.intel.com, seanjc@google.com, stable@vger.kernel.org
References: <CU4OBQ8MQ2LK.2GRBPLQGVTZ3@seitikki>
 <20230717202938.94989-1-haitao.huang@linux.intel.com>
 <dfb1f233-aebd-50cf-8704-e83b91ee110a@intel.com>
 <op.18ah5mn3wjvjmi@hhuan26-mobl.amr.corp.intel.com>
 <b5779418-e2a4-ca7a-866f-97e49cd272cb@intel.com>
 <op.18aontlmwjvjmi@hhuan26-mobl.amr.corp.intel.com>
 <eb1aea6f-3688-f871-2335-ff911a51ef52@intel.com>
 <op.18aqz7sbwjvjmi@hhuan26-mobl.amr.corp.intel.com>
From:   Dave Hansen <dave.hansen@intel.com>
In-Reply-To: <op.18aqz7sbwjvjmi@hhuan26-mobl.amr.corp.intel.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 7/18/23 14:22, Haitao Huang wrote:
> I agree this is the race. But for this to happen, that is at #1 you have
> only one non-SECS page left so #3 can happen. That means it is already
> high pressure 

I think our definitions of memory pressure differ.

Pressure is raised by allocations and dropped by reclaim.  This
raise->drop cycle is (or should be) time-limited and can't take forever.
 The reclaim either works in a short period of time or something dies.
If allocations are transient, pressure is transient.

Let's say a pressure blip (a one-time event) comes along and pages out
that second-to-last page.  That's pretty low pressure.  Years pass.  The
enclave never gets run.  Nothing pages the second-to-last page back in.
A second pressure blip comes along.  The SECS page gets paged out.

That's two pressure blips in, say 10 years.  Is that "high pressure"?

> because reclaimer has swapped all other non-SECS.
> In my example of two enclaves of 5 non-EPC pages. #3 won't happen
> because you don't reach #1 with only one non-SECS left.