linux-kernel - Re: [PATCH v3] i2c: i2c-qcom-geni: Parse Error correctly in i2c GSI mode

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]

Message-ID: <9dbe987a-fdd1-4bec-b350-5936abf69b1b@quicinc.com>
Date: Fri, 8 Mar 2024 11:13:18 +0530
From: Mukesh Kumar Savaliya <quic_msavaliy@...cinc.com>
To: Andi Shyti <andi.shyti@...nel.org>
CC: <konrad.dybcio@...aro.org>, <andersson@...nel.org>, <vkoul@...nel.org>,
        <wsa@...nel.org>, <linux-arm-msm@...r.kernel.org>,
        <dmaengine@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
        <linux-i2c@...r.kernel.org>, <quic_vdadhani@...cinc.com>
Subject: Re: [PATCH v3] i2c: i2c-qcom-geni: Parse Error correctly in i2c GSI
 mode



On 3/8/2024 4:28 AM, Andi Shyti wrote:
> Hi Mukesh,
> 
> On Fri, Mar 08, 2024 at 02:25:39AM +0530, Mukesh Kumar Savaliya wrote:
>> I2C driver currently reports "DMA txn failed" error even though it's
>> NACK OR BUS_PROTO OR ARB_LOST. Detect NACK error when no device ACKs
>> on the bus instead of generic transfer failure which doesn't give any
>> specific clue.
>>
>> Make Changes inside i2c driver callback handler function
>> i2c_gpi_cb_result() to parse these errors and make sure GSI driver
>> stores the error status during error interrupt.
> 
> funny note: this is half "imperative mood" :-)
> 
> A real imperative would be "Pares the error inside
> i2c_gpi_cb_result()... blah blah blah". No need to resend.
> 

  :-) Thanks !

>> Fixes: d8703554f4de ("i2c: qcom-geni: Add support for GPI DMA")
> 
> I still don't understand what's the fix here. You are making a
> generic DMA error to be more specific... where is the bug? What
> exactly is broken now?
> 
This is about being particular while reporting specific error.
Like i mentioned, instead of generic DMA transfer error, it should be 
particular error 1) NACK 2) BUT_PROTO 3)ARB_LOST.
Ofcourse when data transfer via DMA fails, it can be considered as
DMA Txfer fail.
In summary so far driver was considering all failure as txfer failure,
but i2c has errors which are kind of response/condition on the bus.

Sorry if it confusing still, but please let me know if anything required 
to be updated in  commit log which can bring clarity.

> Besides, keep in mind, that commits with fixes tags get
> backported to older kernels (this one dates back to 5.18) and you
> should also Cc the stable mailing list:
> 
> Cc: <stable@...r.kernel.org> # v5.18+

Sure, will add into CC. was waiting for reviewed-by tag.

> 
> Thanks,
> Andi