From: Selvarasu Ganesan <selvarasu.g@samsung.com>
To: Krzysztof Kozlowski <krzk@kernel.org>,
peter.griffin@linaro.org, alim.akhtar@samsung.com,
gregkh@linuxfoundation.org, robh@kernel.org, conor+dt@kernel.org,
Thinh.Nguyen@synopsys.com, linux-arm-kernel@lists.infradead.org,
linux-samsung-soc@vger.kernel.org, linux-usb@vger.kernel.org,
devicetree@vger.kernel.org, linux-kernel@vger.kernel.org
Cc: jh0801.jung@samsung.com, dh10.jung@samsung.com,
akash.m5@samsung.com, hongpooh.kim@samsung.com,
eomji.oh@samsung.com, shijie.cai@samsung.com,
muhammed.ali@samsung.com, thiagu.r@samsung.com
Subject: Re: [PATCH 2/3] dt-bindings: usb: Introduce samsung,snps-dwc3
Date: Mon, 7 Sep 2026 11:34:38 +0530 [thread overview]
Message-ID: <e4bafeea-4548-41ef-b5e2-98ade2864e74@samsung.com> (raw)
In-Reply-To: <ad1d8c01-a810-443a-9ad5-721d6d0b68b1@kernel.org>
On 9/5/2026 12:52 PM, Krzysztof Kozlowski wrote:
> On 03/09/2026 09:12, Selvarasu Ganesan wrote:
>> The Samsung Exynos8855 USB glue is not separate from the Synopsys DWC3
>> core, using a flattened model where the DWC3 controller is directly
>> addressable rather than nested as a child node of a glue layer.
>>
>> Describe the Exynos USB block as a single node, referencing the split
>> out Synopsys DWC3 core properties via snps,dwc3-common.yaml. The new
>> binding follows the same approach as qcom,snps-dwc3.
>>
>> A generic fallback compatible "samsung,snps-dwc3" is introduced to
>> describe this binding, selected by the validator based on the presence
>> of the fallback string in the compatible list. Future Exynos SoCs
> This is AI slop text. Over longish useless sentence describing diff
> without actual information WHY.
>
> Don't send AI generated slop messages.
Sorry for the misleading info on this commit message. We were trying to
explain why a separate flattened representation is being introduced with
using generic fallback string, and follow the style of some other DWC3
flattened refactors (like the Qualcomm :
https://lore.kernel.org/all/20250414-dwc3-refactor-v7-2-f015b358722d@oss.qualcomm.com)
to justify the separate binding, but it became too long. We'll simplify
it and keep only the technical reasoning.
>
>> utilizing the same flattened model can be added by extending the
>> compatible enum with per SoC clock and property constraints.
>>
>> Co-developed-by: Muhammed Ali <muhammed.ali@samsung.com>
>> Signed-off-by: Muhammed Ali <muhammed.ali@samsung.com>
>> Signed-off-by: Selvarasu Ganesan <selvarasu.g@samsung.com>
>> ---
>>
>> Krzysztof Kozlowski is added as a maintainer of this YAML binding, as
>> he is the ARM/Samsung SoC architecture maintainer (ARM/SAMSUNG S3C,
>> S5P AND EXYNOS ARM ARCHITECTURES) who already maintains Samsung Exynos
>> DT bindings (e.g. samsung,exynos-dwc3.yaml) and reviews all
>> Samsung related binding changes. This follows the same convention used
>> by existing Samsung Exynos DT binding schemas.
>> ---
>> .../bindings/usb/samsung,snps-dwc3.yaml | 98 +++++++++++++++++++
>> 1 file changed, 98 insertions(+)
>> create mode 100644 Documentation/devicetree/bindings/usb/samsung,snps-dwc3.yaml
>>
>> diff --git a/Documentation/devicetree/bindings/usb/samsung,snps-dwc3.yaml b/Documentation/devicetree/bindings/usb/samsung,snps-dwc3.yaml
>> new file mode 100644
>> index 000000000000..480308c16f37
>> --- /dev/null
>> +++ b/Documentation/devicetree/bindings/usb/samsung,snps-dwc3.yaml
>> @@ -0,0 +1,98 @@
>> +# SPDX-License-Identifier: (GPL-2.0 OR BSD-2-Clause)
>> +%YAML 1.2
>> +---
>> +$id: https://protect2.fireeye.com/v1/url?k=1a7dfa9a-7bf6efdd-1a7c71d5-000babff99aa-643f51fc0561c489&q=1&e=c6c89336-1f3a-4d30-a005-615c2b8d9c6f&u=http%3A%2F%2Fdevicetree.org%2Fschemas%2Fusb%2Fsamsung%2Csnps-dwc3.yaml%23
>> +$schema: https://protect2.fireeye.com/v1/url?k=515f3c48-30d4290f-515eb707-000babff99aa-9a67f9f3980bd8f3&q=1&e=c6c89336-1f3a-4d30-a005-615c2b8d9c6f&u=http%3A%2F%2Fdevicetree.org%2Fmeta-schemas%2Fcore.yaml%23
>> +
>> +title: Samsung Exynos SoC DWC3 USB Controller
>> +
>> +maintainers:
>> + - Krzysztof Kozlowski <krzk@kernel.org>
>> + - Selvarasu Ganesan <selvarasu.g@samsung.com>
>> +
>> +description:
>> + Describes the DWC3 USB controller block implemented on Samsung Exynos SoCs.
> Huh? Did you even look at existing bindings?
Yes, We checked samsung,exynos-dwc3.yaml. Exynos8855 could use it, but
that binding uses a parent and child representation. This patch
introduces a flattened model where the USB vendor block and SNPS dwc3
block are in a single node. The goal is to use this for Exynos8855 and
eventually migrate the other samsung USB implementations from the old
parent and child model to this one. We'll make this clearer and shorter
in the commit message.
>
>> +
>> +select:
>> + properties:
>> + compatible:
>> + contains:
>> + const: samsung,snps-dwc3
>> + required:
>> + - compatible
>> +
>> +properties:
>> + compatible:
>> + items:
>> + - enum:
>> + - samsung,exynos8855-dwc3
>> + - const: samsung,snps-dwc3
> There is no snps device from Samsung, neither DWC3. The name of device
> is completely wrong. snps is a vendor, not a device.
>
> Anyway, don't use generic fallbacks.
Understood. The fallback was intended to cover upcoming SoCs (8865,
9955, 9965) and the migration of older SoCs without adding a unique
string for every project, similar to the qcom,snps-dwc3 approach.
We agree that the naming was incorrect since "snps" is the vendor. We
can switch to per SoC compatible strings, but would a corrected fallback
name like samsung,exynos-dwc3-common or samsung,exynos-snps-dwc3 be
acceptable instead?
>
>
>> +
>> + reg:
>> + maxItems: 1
>> +
>> + power-domains:
>> + maxItems: 1
>> +
>> + interrupts:
>> + maxItems: 1
>> +
>> + clocks:
>> + description: |
>> + Several clocks are used, depending on the variant. Typical ones are:
>> + - ref: Reference clock for SOF/ITP generation.
>> + - bus: Bus (APB) clock for register access.
>> + minItems: 1
>> + maxItems: 4
>> +
>> + clock-names:
>> + minItems: 1
>> + maxItems: 4
> You have only one device, why is this flexible?
Exynos8855 only needs two clocks, but maxItems: 4 was added to support
other samsung USB implementations that use up to four. Since we plan to
migrate those to the flattened representation later, we wanted the
binding to cover those requirements as well. Example required 4 clocks
for gs101-dwusb3 (ref:
Documentation/devicetree/bindings/usb/samsung,exynos-dwc3.yaml)
>
>> +
>> +required:
>> + - compatible
>> + - reg
>> + - clocks
>> + - clock-names
>> + - interrupts
>> +
>> +allOf:
>> + - $ref: snps,dwc3-common.yaml#
>> +
>> + - if:
>> + properties:
>> + compatible:
>> + contains:
>> + const: samsung,exynos8855-dwc3
>> + then:
>> + properties:
>> + clocks:
>> + minItems: 2
>> + maxItems: 2
>> + clock-names:
>> + items:
>> + - const: ref
>> + - const: bus
> Why do you need all this? Look at existing code, don't duplicate. And
> for sure do not introduce one more DWC3 binding claiming to be for
> Samsung devices and completely ignoring existing ones.
We understand the concern about adding another binding. The goal is to
migrate all samsung USB implementations to this flattened model, and
then deprecate samsung,exynos-dwc3.yaml to avoid duplication.
Thanks,
Selva
>
> Best regards,
> Krzysztof
next prev parent reply other threads:[~2026-09-07 6:04 UTC|newest]
Thread overview: 10+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <CGME20260903071311epcas5p24c547d69b769818834ae2b7285cb541b@epcas5p2.samsung.com>
2026-09-03 7:12 ` [PATCH 0/3] Add USB DWC3 support for Samsung Exynos8855 Selvarasu Ganesan
2026-09-03 7:12 ` [PATCH 1/3] MAINTAINERS: add Samsung USB DT binding to ARM/SAMSUNG entry Selvarasu Ganesan
2026-09-03 7:12 ` [PATCH 2/3] dt-bindings: usb: Introduce samsung,snps-dwc3 Selvarasu Ganesan
2026-09-05 7:22 ` Krzysztof Kozlowski
2026-09-07 6:04 ` Selvarasu Ganesan [this message]
2026-09-07 7:07 ` Krzysztof Kozlowski
2026-09-07 8:50 ` Selvarasu Ganesan
2026-09-03 7:12 ` [PATCH 3/3] usb: dwc3: Add Exynos8855 to the generic platform driver Selvarasu Ganesan
2026-09-05 0:15 ` Thinh Nguyen
2026-09-05 7:19 ` Krzysztof Kozlowski
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=e4bafeea-4548-41ef-b5e2-98ade2864e74@samsung.com \
--to=selvarasu.g@samsung.com \
--cc=Thinh.Nguyen@synopsys.com \
--cc=akash.m5@samsung.com \
--cc=alim.akhtar@samsung.com \
--cc=conor+dt@kernel.org \
--cc=devicetree@vger.kernel.org \
--cc=dh10.jung@samsung.com \
--cc=eomji.oh@samsung.com \
--cc=gregkh@linuxfoundation.org \
--cc=hongpooh.kim@samsung.com \
--cc=jh0801.jung@samsung.com \
--cc=krzk@kernel.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-samsung-soc@vger.kernel.org \
--cc=linux-usb@vger.kernel.org \
--cc=muhammed.ali@samsung.com \
--cc=peter.griffin@linaro.org \
--cc=robh@kernel.org \
--cc=shijie.cai@samsung.com \
--cc=thiagu.r@samsung.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox