[v1,2/8] bus: mhi: core: Add range check for channel id received in event ring

Commit Message

Bhaumik Bhatt April 27, 2020, 6:32 p.m. UTC

From: Hemant Kumar <[e-mail]>

MHI data completion handler function reads channel id from event
ring element. Value is under the control of MHI devices and can be
any value between 0 and 255. In order to prevent out of bound access
add a bound check against the max channel supported by controller
and skip processing of that event ring element.

Signed-off-by: Hemant Kumar <[e-mail]>
---
 drivers/bus/mhi/core/main.c | 3 +++
 1 file changed, 3 insertions(+)

Comments

Jeffrey Hugo April 27, 2020, 6:50 p.m. UTC | #1

On 4/27/2020 12:32 PM, Bhaumik Bhatt wrote:
> From: Hemant Kumar <[e-mail]>
> 
> MHI data completion handler function reads channel id from event
> ring element. Value is under the control of MHI devices and can be
> any value between 0 and 255. In order to prevent out of bound access
> add a bound check against the max channel supported by controller
> and skip processing of that event ring element.
> 
> Signed-off-by: Hemant Kumar <[e-mail]>

I believe your SOB is needed here after Hemant's per section 11 of 
Documentation/process/submitting-patches.rst

> ---
>   drivers/bus/mhi/core/main.c | 3 +++
>   1 file changed, 3 insertions(+)
> 
> diff --git a/drivers/bus/mhi/core/main.c b/drivers/bus/mhi/core/main.c
> index 23154f1..ba8afa7 100644
> --- a/drivers/bus/mhi/core/main.c
> +++ b/drivers/bus/mhi/core/main.c
> @@ -827,6 +827,9 @@ int mhi_process_data_event_ring(struct mhi_controller *mhi_cntrl,
>   		enum mhi_pkt_type type = MHI_TRE_GET_EV_TYPE(local_rp);
>   
>   		chan = MHI_TRE_GET_EV_CHID(local_rp);
> +		if (WARN_ON(chan >= mhi_cntrl->max_chan))
> +			continue;
> +
>   		mhi_chan = &mhi_cntrl->mhi_chan[chan];
>   
>   		if (likely(type == MHI_PKT_TYPE_TX_EVENT)) {
> 

How does this work?

I understand the need for the range check, however I looks like this 
change doesn't do proper handling.

Since all you do is "continue", we'll remain stuck in the while loop 
this exists in, continuously "processing" the same event, failing the 
same check, and spamming the kernel log with the WARN_ON output until 
the end of time.

What am I missing?

What is the behavior you want?  Do you want to just drop/ignore the 
invalid event, or issue a reset of the device because it is clearly 
misbehaving?

Hemant Kumar April 28, 2020, 1:28 a.m. UTC | #2

On 4/27/20 11:50 AM, Jeffrey Hugo wrote:
> On 4/27/2020 12:32 PM, Bhaumik Bhatt wrote:
>> From: Hemant Kumar <[e-mail]>
>>
>> MHI data completion handler function reads channel id from event
>> ring element. Value is under the control of MHI devices and can be
>> any value between 0 and 255. In order to prevent out of bound access
>> add a bound check against the max channel supported by controller
>> and skip processing of that event ring element.
>>
>> Signed-off-by: Hemant Kumar <[e-mail]>
> 
> I believe your SOB is needed here after Hemant's per section 11 of 
> Documentation/process/submitting-patches.rst
> 
>> ---
>>   drivers/bus/mhi/core/main.c | 3 +++
>>   1 file changed, 3 insertions(+)
>>
>> diff --git a/drivers/bus/mhi/core/main.c b/drivers/bus/mhi/core/main.c
>> index 23154f1..ba8afa7 100644
>> --- a/drivers/bus/mhi/core/main.c
>> +++ b/drivers/bus/mhi/core/main.c
>> @@ -827,6 +827,9 @@ int mhi_process_data_event_ring(struct 
>> mhi_controller *mhi_cntrl,
>>           enum mhi_pkt_type type = MHI_TRE_GET_EV_TYPE(local_rp);
>>           chan = MHI_TRE_GET_EV_CHID(local_rp);
>> +        if (WARN_ON(chan >= mhi_cntrl->max_chan))
>> +            continue;
>> +
>>           mhi_chan = &mhi_cntrl->mhi_chan[chan];
>>           if (likely(type == MHI_PKT_TYPE_TX_EVENT)) {
>>
> 
> How does this work?
> 
> I understand the need for the range check, however I looks like this 
> change doesn't do proper handling.
> 
> Since all you do is "continue", we'll remain stuck in the while loop 
> this exists in, continuously "processing" the same event, failing the 
> same check, and spamming the kernel log with the WARN_ON output until 
> the end of time.
> 
> What am I missing?
Idea was to drop the event and continue, but continue statement is 
wrong here, we need to recycle the event and continue after that. Will 
fix in V2.
> 
> What is the behavior you want?  Do you want to just drop/ignore the 
> invalid event, or issue a reset of the device because it is clearly 
> misbehaving?
>

Patch

diff --git a/drivers/bus/mhi/core/main.c b/drivers/bus/mhi/core/main.c
index 23154f1..ba8afa7 100644
--- a/drivers/bus/mhi/core/main.c
+++ b/drivers/bus/mhi/core/main.c
@@ -827,6 +827,9 @@  int mhi_process_data_event_ring(struct mhi_controller *mhi_cntrl,
 		enum mhi_pkt_type type = MHI_TRE_GET_EV_TYPE(local_rp);
 
 		chan = MHI_TRE_GET_EV_CHID(local_rp);
+		if (WARN_ON(chan >= mhi_cntrl->max_chan))
+			continue;
+
 		mhi_chan = &mhi_cntrl->mhi_chan[chan];
 
 		if (likely(type == MHI_PKT_TYPE_TX_EVENT)) {