[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <7d8e6c35-624e-4e39-b92e-9786db6514d6@altera.com>
Date: Fri, 31 Oct 2025 16:37:14 +0800
From: "Ng, Adrian Ho Yin" <adrianhoyin.ng@...era.com>
To: Krzysztof Kozlowski <krzk@...nel.org>, alexandre.belloni@...tlin.com,
Frank.Li@....com, wsa+renesas@...g-engineering.com, robh@...nel.org,
krzk+dt@...nel.org, conor+dt@...nel.org, dinguyen@...nel.org,
linux-i3c@...ts.infradead.org, devicetree@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: Re: [RFC PATCH v2 0/3] i3c: dw: Add Altera SoCFPGA runtime PM disable
quirk
On 10/31/2025 4:32 PM, Krzysztof Kozlowski wrote:
> On 31/10/2025 09:27, adrianhoyin.ng@...era.com wrote:
>> From: Adrian Ng Ho Yin <adrianhoyin.ng@...era.com>
>>
>> This patchset adds support for an Altera SoCFPGA-specific quirk
>> in the Synopsys DesignWare I3C master driver.
>>
>> While running the I3C compliance test suite on the Altera Agilex5 SoCFPGA,
>> the I3C bus was observed to hang when a slave device issued an IBI after
>> the Dynamic Address Assignment (DAA) process completed.
>>
>> This issue occurs because the controller enters a suspended state after
>> DAA due to runtime PM being enabled. When suspended, the controller stops
>> driving the SCL line. As a result, an IBI transfer cannot complete, leaving
>> the SDA line stuck low and the bus in a hung state.
>>
>> To address this issue, a new compatible string,
>> "altr,socfpga-dw-i3c-master", is introduced to identify the
>> SoCFPGA variant. When this compatible string is matched, a new
>> quirk (DW_I3C_DISABLE_RUNTIME_PM_QUIRK) disables all runtime PM
>> operations in the driver, ensuring the controller remains active
>> to handle IBI transactions reliably.
>>
>> ---
>> changelog:
>> v1->v2:
>> *Add new compatible string in dw i3c dt binding.
>> *Added new compatible string for altr socfpga platform.
>> *Remove Kconfig that disables runtime PM added in v1.
>> *Update implementation to disable runtime PM via compatible string
>> match
>
>
> Why is this still RFC? Please always provide reason in changelog and be
> aware that some maintainers - including me - skip or just look briefly
> through RFC patches since they are simply not ready.
>
> Also remember that versioning *continuous* when dropping non-RFC.
>
Noted. Will resend and drop the RFC tag and update reason in the
changelog.>
> Best regards,
> Krzysztof
Thank You,
Adrian
Powered by blists - more mailing lists