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: <b1476c7d-c514-227f-2735-092b5afb7d3a@intel.com>
Date:   Mon, 18 Jul 2022 11:52:54 -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: [tip: x86/urgent] x86/sgx: Set active memcg prior to shmem
 allocation
Content-Language: en-US
To:     Borislav Petkov <bp@alien8.de>,
        Kristen Carlson Accardi <kristen@linux.intel.com>
Cc:     linux-tip-commits@vger.kernel.org,
        Dave Hansen <dave.hansen@linux.intel.com>,
        Shakeel Butt <shakeelb@google.com>,
        Roman Gushchin <roman.gushchin@linux.dev>, x86@kernel.org,
        linux-kernel@vger.kernel.org
References: <20220520174248.4918-1-kristen@linux.intel.com>
 <165419442842.4207.2566961916839377924.tip-bot2@tip-bot2>
 <YtUs3MKLzFg+rqEV@zn.tnic>
From:   Dave Hansen <dave.hansen@intel.com>
In-Reply-To: <YtUs3MKLzFg+rqEV@zn.tnic>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 7/18/22 02:50, Borislav Petkov wrote:
>> -int sgx_encl_get_backing(struct sgx_encl *encl, unsigned long page_index,
>> -			 struct sgx_backing *backing);
>> +int sgx_encl_lookup_backing(struct sgx_encl *encl, unsigned long page_index,
>> +			    struct sgx_backing *backing);
>> +int sgx_encl_alloc_backing(struct sgx_encl *encl, unsigned long page_index,
>> +			   struct sgx_backing *backing);
>>  void sgx_encl_put_backing(struct sgx_backing *backing);
> So this is making the sgx_encl_get_backing() thing static but its
> counterpart sgx_encl_put_backing() is not and is still called by other
> places.
> 
> Perhaps something wrong with the layering or is this on purpose?

All three of the:

	sgx_encl_get_backing()
	sgx_encl_lookup_backing()
	sgx_encl_alloc_backing()

functions take a reference on the backing storage that must be dropped
with sgx_encl_put_backing().  The "lookup" and "alloc" were added
because there really are two different users:

	1. I want to *create* backing storage space (alloc)
	2. I want to find *existing* backing storage (lookup)

and those two logical uses need different cgroup accounting semantics.

As a further cleanup, it would probably be nice to explicitly document
that "lookup" and "alloc" also require a subsequent "put".  It would
also be nice to change sgx_encl_get_backing() to
__sgx_encl_get_backing() to make it clear that it's an internal thing.

So, I think the _code_ is OK as-is, but it could use some more love.