Received-SPF: pass (google.com: best guess record for domain of linux-crypto-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Mon, 21 Oct 2019 17:57:31 -0700 (PDT)
From:   David Rientjes <rientjes@google.com>
To:     "Singh, Brijesh" <brijesh.singh@amd.com>
cc:     "Kalra, Ashish" <Ashish.Kalra@amd.com>,
        "Lendacky, Thomas" <Thomas.Lendacky@amd.com>,
        "Hook, Gary" <Gary.Hook@amd.com>,
        "herbert@gondor.apana.org.au" <herbert@gondor.apana.org.au>,
        "davem@davemloft.net" <davem@davemloft.net>,
        "gregkh@linuxfoundation.org" <gregkh@linuxfoundation.org>,
        "tglx@linutronix.de" <tglx@linutronix.de>,
        "allison@lohutok.net" <allison@lohutok.net>,
        "info@metux.net" <info@metux.net>,
        "yamada.masahiro@socionext.com" <yamada.masahiro@socionext.com>,
        "linux-crypto@vger.kernel.org" <linux-crypto@vger.kernel.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "kvm@vger.kernel.org" <kvm@vger.kernel.org>
Subject: Re: [PATCH] crypto: ccp - Retry SEV INIT command in case of integrity
 check failure.
In-Reply-To: <29887804-ecab-ae83-8d3f-52ea83e44b4e@amd.com>
Message-ID: <alpine.DEB.2.21.1910211754550.152056@chino.kir.corp.google.com>
References: <20191017223459.64281-1-Ashish.Kalra@amd.com> <alpine.DEB.2.21.1910190156210.140416@chino.kir.corp.google.com> <29887804-ecab-ae83-8d3f-52ea83e44b4e@amd.com>
User-Agent: Alpine 2.21 (DEB 202 2017-01-01)
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Sender: linux-crypto-owner@vger.kernel.org
Precedence: bulk

On Mon, 21 Oct 2019, Singh, Brijesh wrote:

> >> From: Ashish Kalra <ashish.kalra@amd.com>
> >>
> >> SEV INIT command loads the SEV related persistent data from NVS
> >> and initializes the platform context. The firmware validates the
> >> persistent state. If validation fails, the firmware will reset
> >> the persisent state and return an integrity check failure status.
> >>
> >> At this point, a subsequent INIT command should succeed, so retry
> >> the command. The INIT command retry is only done during driver
> >> initialization.
> >>
> >> Additional enums along with SEV_RET_SECURE_DATA_INVALID are added
> >> to sev_ret_code to maintain continuity and relevance of enum values.
> >>
> >> Signed-off-by: Ashish Kalra <ashish.kalra@amd.com>
> >> ---
> >>   drivers/crypto/ccp/psp-dev.c | 12 ++++++++++++
> >>   include/uapi/linux/psp-sev.h |  3 +++
> >>   2 files changed, 15 insertions(+)
> >>
> >> diff --git a/drivers/crypto/ccp/psp-dev.c b/drivers/crypto/ccp/psp-dev.c
> >> index 6b17d179ef8a..f9318d4482f2 100644
> >> --- a/drivers/crypto/ccp/psp-dev.c
> >> +++ b/drivers/crypto/ccp/psp-dev.c
> >> @@ -1064,6 +1064,18 @@ void psp_pci_init(void)
> >>   
> >>   	/* Initialize the platform */
> >>   	rc = sev_platform_init(&error);
> >> +	if (rc && (error == SEV_RET_SECURE_DATA_INVALID)) {
> >> +		/*
> >> +		 * INIT command returned an integrity check failure
> >> +		 * status code, meaning that firmware load and
> >> +		 * validation of SEV related persistent data has
> >> +		 * failed and persistent state has been erased.
> >> +		 * Retrying INIT command here should succeed.
> >> +		 */
> >> +		dev_dbg(sp->dev, "SEV: retrying INIT command");
> >> +		rc = sev_platform_init(&error);
> >> +	}
> >> +
> >>   	if (rc) {
> >>   		dev_err(sp->dev, "SEV: failed to INIT error %#x\n", error);
> >>   		return;
> > 
> > Curious why this isn't done in __sev_platform_init_locked() since
> > sev_platform_init() can be called when loading the kvm module and the same
> > init failure can happen that way.
> > 
> 
> The FW initialization (aka PLATFORM_INIT) is called in the following
> code paths:
> 
> 1. During system boot up
> 
> and
> 
> 2. After the platform reset command is issued
> 
> The patch takes care of #1. Based on the spec, platform reset command
> should erase the persistent data and the PLATFORM_INIT should *not* fail
> with SEV_RET_SECURE_DATA_INVALID error code. So, I am not able to see
> any  strong reason to move the retry code in
> __sev_platform_init_locked().
> 

Hmm, is the sev_platform_init() call in sev_guest_init() intended to do 
SEV_CMD_INIT only after platform reset?  I was under the impression it was 
done in case any previous init failed.