netdev - Re: [PATCH v2 1/2] bus: mhi: host: Import mux

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date: Fri, 14 Jun 2024 08:31:03 -0600
From: Jeffrey Hugo <quic_jhugo@...cinc.com>
To: Loic Poulain <loic.poulain@...aro.org>,
        Manivannan Sadhasivam
	<manivannan.sadhasivam@...aro.org>
CC: Slark Xiao <slark_xiao@....com>, <ryazanov.s.a@...il.com>,
        <johannes@...solutions.net>, <netdev@...r.kernel.org>,
        <mhi@...ts.linux.dev>, <linux-arm-msm@...r.kernel.org>,
        <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v2 1/2] bus: mhi: host: Import mux_id item

On 6/14/2024 4:17 AM, Loic Poulain wrote:
> On Wed, 12 Jun 2024 at 16:51, Manivannan Sadhasivam
> <manivannan.sadhasivam@...aro.org> wrote:
>>
>> On Wed, Jun 12, 2024 at 08:19:13AM -0600, Jeffrey Hugo wrote:
>>> On 6/12/2024 3:46 AM, Manivannan Sadhasivam wrote:
>>>> On Wed, Jun 12, 2024 at 05:38:42PM +0800, Slark Xiao wrote:
>>>>
>>>> Subject could be improved:
>>>>
>>>> bus: mhi: host: Add configurable mux_id for MBIM mode
>>>>
>>>>> For SDX72 MBIM mode, it starts data mux id from 112 instead of 0.
>>>>> This would lead to device can't ping outside successfully.
>>>>> Also MBIM side would report "bad packet session (112)".
>>>>> So we add a default mux_id value for SDX72. And this value
>>>>> would be transferred to wwan mbim side.
>>>>>
>>>>> Signed-off-by: Slark Xiao <slark_xiao@....com>
>>>>> ---
>>>>>    drivers/bus/mhi/host/pci_generic.c | 3 +++
>>>>>    include/linux/mhi.h                | 2 ++
>>>>>    2 files changed, 5 insertions(+)
>>>>>
>>>>> diff --git a/drivers/bus/mhi/host/pci_generic.c b/drivers/bus/mhi/host/pci_generic.c
>>>>> index 0b483c7c76a1..9e9adf8320d2 100644
>>>>> --- a/drivers/bus/mhi/host/pci_generic.c
>>>>> +++ b/drivers/bus/mhi/host/pci_generic.c
>>>>> @@ -53,6 +53,7 @@ struct mhi_pci_dev_info {
>>>>>            unsigned int dma_data_width;
>>>>>            unsigned int mru_default;
>>>>>            bool sideband_wake;
>>>>> + unsigned int mux_id;
>>>>>    };
>>>>>    #define MHI_CHANNEL_CONFIG_UL(ch_num, ch_name, el_count, ev_ring) \
>>>>> @@ -469,6 +470,7 @@ static const struct mhi_pci_dev_info mhi_foxconn_sdx72_info = {
>>>>>            .dma_data_width = 32,
>>>>>            .mru_default = 32768,
>>>>>            .sideband_wake = false,
>>>>> + .mux_id = 112,
>>>>>    };
>>>>>    static const struct mhi_channel_config mhi_mv3x_channels[] = {
>>>>> @@ -1035,6 +1037,7 @@ static int mhi_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id)
>>>>>            mhi_cntrl->runtime_get = mhi_pci_runtime_get;
>>>>>            mhi_cntrl->runtime_put = mhi_pci_runtime_put;
>>>>>            mhi_cntrl->mru = info->mru_default;
>>>>> + mhi_cntrl->link_id = info->mux_id;
>>>>
>>>> Again, 'link_id' is just a WWAN term. Use 'mux_id' here also.
>>>
>>> Does this really belong in MHI?  If this was DT, I don't think we would put
>>> this value in DT, but rather have the driver (MBIM) detect the device and
>>> code in the required value.
>>>
>>
>> I believe this is a modem value rather than MHI. But I was OK with keeping it in
>> MHI driver since we kind of keep modem specific config.
>>
>> But if WWAN can detect the device and apply the config, I'm all over it.
> 
> That would require at least some information from the MHI bus for the
> MBIM driver
> to make a decision, such as a generic device ID, or quirk flags...

I don't see why.

The "simple" way to do it would be to have the controller define a 
different channel name, and then have the MBIM driver probe on that. 
The MBIM driver could attach driver data saying that it needs to have a 
specific mux_id.

Or, with zero MHI/Controller changes, the MBIM driver could parse the 
mhi_device struct, get to the struct device, for the underlying device, 
and extract the PCIe Device ID and match that to a white list of known 
devices that need this property.

I guess if the controller could attach a private void * to the 
mhi_device that is opaque to MHI, but allows MBIM to make a decision, 
that would be ok.  Such a mechanism would be generic, and extensible to 
other usecases of the same "class".

-Jeff