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: <7d3f1877-8762-34f4-f1bb-c5c2924c2b77@intel.com>
Date:   Thu, 27 Apr 2023 12:56:52 -0700
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.10.0
Subject: Re: [PATCH v2 3/3] selftests/tdx: Test GetQuote TDX attestation
 feature
Content-Language: en-US
To:     Sathyanarayanan Kuppuswamy 
        <sathyanarayanan.kuppuswamy@linux.intel.com>,
        Dionna Amalie Glaze <dionnaglaze@google.com>
Cc:     Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
        Dave Hansen <dave.hansen@linux.intel.com>, x86@kernel.org,
        Shuah Khan <shuah@kernel.org>,
        Jonathan Corbet <corbet@lwn.net>,
        "H . Peter Anvin" <hpa@zytor.com>,
        "Kirill A . Shutemov" <kirill.shutemov@linux.intel.com>,
        Tony Luck <tony.luck@intel.com>,
        Wander Lairson Costa <wander@redhat.com>,
        Erdem Aktas <erdemaktas@google.com>,
        Chong Cai <chongc@google.com>, Qinkun Bao <qinkun@apache.org>,
        Guorui Yu <GuoRui.Yu@linux.alibaba.com>,
        Du Fan <fan.du@intel.com>, linux-kernel@vger.kernel.org,
        linux-kselftest@vger.kernel.org, linux-doc@vger.kernel.org
References: <20230413034108.1902712-1-sathyanarayanan.kuppuswamy@linux.intel.com>
 <20230413034108.1902712-4-sathyanarayanan.kuppuswamy@linux.intel.com>
 <CAAH4kHb_PAhR5jLFi2KsGku9ALNtjtxb-JO-k1ULK-Mj5mi=pw@mail.gmail.com>
 <a418c2d3-e333-fe3d-63a5-c6e555f81fc7@linux.intel.com>
From:   Dave Hansen <dave.hansen@intel.com>
In-Reply-To: <a418c2d3-e333-fe3d-63a5-c6e555f81fc7@linux.intel.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 4/27/23 12:10, Sathyanarayanan Kuppuswamy wrote:
>> Shouldn't req be zeroed before populating reportdata? We wouldn't want
>> uninitialized memory to leave the guest. I know this is just a test,
> There are only two members in struct tdx_report_req (reportdata and tdreport).
> The reportdata has already been updated here, and the tdreport will be updated
> by the kernel on output. Since TDX_CMD_GET_REPORT0 IOCTL handler uses an
> intermediate kernel buffer to the TDREPORT and copies the generated report back
> to this user buffer, this uninitialized tdreport data never leaves the guest.

Is that really even relevant?

I mean, we could implement the whole thing with get_user_pages() and
then just pass the physical address of the reportdata and tdreport down
to the TDX module.

It doesn't matter either way.  The data is going from guest userspace to
the guest kernel to the TDX module, all of which are trusted.

It's a selftest.  I'd just leave it alone.