Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH][venus-for-next-v5.14] media: venus: hfi_cmds: Fix packet
 size calculation
From:   "Gustavo A. R. Silva" <gustavo@embeddedor.com>
To:     Kees Cook <keescook@chromium.org>,
        "Gustavo A. R. Silva" <gustavoars@kernel.org>
Cc:     Stanimir Varbanov <stanimir.varbanov@linaro.org>,
        Andy Gross <agross@kernel.org>,
        Bjorn Andersson <bjorn.andersson@linaro.org>,
        Mauro Carvalho Chehab <mchehab@kernel.org>,
        linux-media@vger.kernel.org, linux-arm-msm@vger.kernel.org,
        linux-kernel@vger.kernel.org, linux-hardening@vger.kernel.org
References: <20210601184616.GA23488@embeddedor>
 <202106021254.39A1561075@keescook>
 <e40f4067-82e2-31ff-0694-375a59f949de@embeddedor.com>
Message-ID: <113fd896-464c-6aef-215a-a53ac6103a62@embeddedor.com>
Date:   Wed, 2 Jun 2021 17:02:25 -0500
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101
 Thunderbird/78.8.1
MIME-Version: 1.0
In-Reply-To: <e40f4067-82e2-31ff-0694-375a59f949de@embeddedor.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk


On 6/2/21 15:34, Gustavo A. R. Silva wrote:
> 
> 
> On 6/2/21 14:55, Kees Cook wrote:
>> On Tue, Jun 01, 2021 at 01:46:16PM -0500, Gustavo A. R. Silva wrote:
>>> Now that a one-element array was replaced with a flexible-array member
>>> in struct hfi_sys_set_property_pkt, use the struct_size() helper to
>>> correctly calculate the packet size.
>>>
>>> Fixes: 701e10b3fd9f ("media: venus: hfi_cmds.h: Replace one-element array with flexible-array member")
>>> Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
>>> ---
>>> BTW... it seems that a similar problem is present in
>>> https://lore.kernel.org/linux-hardening/20210211001044.GA69612@embeddedor/ 
>>> and that is what is causing the regression. I will send v2 of that
>>> patch, shortly. Thanks.
>>>
>>>  drivers/media/platform/qcom/venus/hfi_cmds.c | 8 ++++----
>>>  1 file changed, 4 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/drivers/media/platform/qcom/venus/hfi_cmds.c b/drivers/media/platform/qcom/venus/hfi_cmds.c
>>> index 11a8347e5f5c..c86279e5d6e8 100644
>>> --- a/drivers/media/platform/qcom/venus/hfi_cmds.c
>>> +++ b/drivers/media/platform/qcom/venus/hfi_cmds.c
>>> @@ -27,7 +27,7 @@ void pkt_sys_idle_indicator(struct hfi_sys_set_property_pkt *pkt, u32 enable)
>>>  {
>>>  	struct hfi_enable *hfi = (struct hfi_enable *)&pkt->data[1];
>>>  
>>> -	pkt->hdr.size = sizeof(*pkt) + sizeof(*hfi) + sizeof(u32);
>>> +	pkt->hdr.size = struct_size(pkt, data, 2) + sizeof(*hfi);
>>
>> I think this should be "1" not "2".
>>
>> (i.e. there is a single "data" item, followed by an entire *hfi (which
>> starts immediate after data[0]).
> 
> Yeah; I see your point. Here I just wanted to preserve the exact same size
> as the original code, which turns out has a "benign" off-by-one issue.

And the reason why I say it's a "benign" off-by-one (which is actually off-by-four-bytes)
issue is because of this piece of code:

drivers/media/platform/qcom/venus/hfi_venus.c:
 864 static int venus_sys_set_idle_message(struct venus_hfi_device *hdev,
 865                                       bool enable)
 866 {
 867         struct hfi_sys_set_property_pkt *pkt;
 868         u8 packet[IFACEQ_VAR_SMALL_PKT_SIZE];
 869         int ret;
 870
 871         if (!enable)
 872                 return 0;
 873
 874         pkt = (struct hfi_sys_set_property_pkt *)packet;
 875
 876         pkt_sys_idle_indicator(pkt, enable);
 877
 878         ret = venus_iface_cmdq_write(hdev, pkt, false);
 879         if (ret)
 880                 return ret;
 881
 882         return 0;
 883 }

IFACEQ_VAR_SMALL_PKT_SIZE at line 868 is of size 100, which is greater than

sizeof(*pkt) + sizeof(*hfi) + sizeof(u32); in the original code:

drivers/media/platform/qcom/venus/hfi_cmds.c:
  26 void pkt_sys_idle_indicator(struct hfi_sys_set_property_pkt *pkt, u32 enable)
  27 {
  28         struct hfi_enable *hfi = (struct hfi_enable *)&pkt->data[1];
  29
  30         pkt->hdr.size = sizeof(*pkt) + sizeof(*hfi) + sizeof(u32);
  31         pkt->hdr.pkt_type = HFI_CMD_SYS_SET_PROPERTY;
  32         pkt->num_properties = 1;
  33         pkt->data[0] = HFI_PROPERTY_SYS_IDLE_INDICATOR;
  34         hfi->enable = enable;
  35 }

--
Gustavo

> I'll fix it up and respin.
> 
> Thanks!
> --
> Gustavo