| Message ID | 20240110114033.32575-1-quic_luoj@quicinc.com |
|---|---|
| Headers | show |
| Series | net: ethernet: Add qcom PPE driver | expand |
On 1/10/2024 11:44 PM, Simon Horman wrote: > On Wed, Jan 10, 2024 at 01:24:06PM +0100, Krzysztof Kozlowski wrote: >> On 10/01/2024 12:40, Luo Jie wrote: >>> The PPE(packet process engine) hardware block is available in Qualcomm >>> IPQ chipsets that support PPE architecture, such as IPQ9574 and IPQ5332. >>> The PPE includes integrated ethernet MAC and PCS(uniphy), which is used >>> to connect with external PHY devices by PCS. The PPE also includes >>> various packet processing offload capabilities such as routing and >>> briding offload, L2 switch capability, VLAN and tunnel processing >>> offload. >>> >>> This patch series enables support for the PPE driver which intializes >>> and configures the PPE, and provides various services for higher level >>> network drivers in the system such as EDMA (Ethernet DMA) driver or a >>> DSA switch driver for PPE L2 Switch, for Qualcomm IPQ SoCs. >> >> net-next is closed. > > Also, please try to avoid sending patch-sets with more than 15 patches > for net or net-next. > > https://www.kernel.org/doc/html/next/process/maintainer-netdev.html#dividing-work-into-patches Got it, at a later point when this review resumes, we will split the PPE driver patches into two series, one is for PPE switch core feature, another is for MAC/UNIPHY features. Hope this is fine. Thanks for this comment.
On Thu, Jan 11, 2024 at 11:49:53PM +0800, Jie Luo wrote: > > > On 1/11/2024 6:24 AM, Jakub Kicinski wrote: > > On Wed, 10 Jan 2024 19:40:12 +0800 Luo Jie wrote: > > > The PPE(packet process engine) hardware block is available in Qualcomm > > > IPQ chipsets that support PPE architecture, such as IPQ9574 and IPQ5332. > > > > What's the relationship between this driver and QCA8084? > > The PPE (packet processing engine) is the network processing hardware block > in QCOM IPQ SoC. It includes the ethernet MAC and UNIPHY(PCS). This driver > is the base PPE driver which brings up the PPE and handles MAC/UNIPHY > operations. QCA8084 is the external 2.5Gbps 4-port PHY device, which can be > connected with PPE integrated MAC by UNIPHY(PCS). > > Here is the relationship. > PPE integrated MAC --- PPE integrated UNIPHY(PCS) --- (PCS)QCA8084. > > > > > In the last month I see separate changes from you for mdio-ipq4019.c, > > phy/at803x.c and now this driver (none of which got merged, AFAICT.) > > Are you actually the author of this code, or are you just trying > > to upstream bunch of vendor code? > > Yes, Jakub, there are two authors in these patch series, Lei Wei and me. > The patches are already ready for some time, the code has been verified > on the Qualcomm reference design board. These are not downstream drivers > but drivers re-written for upstream. > > > > > Now you're dumping another 10kLoC on the list, and even though this is > > hardly your first posting you're apparently not aware of our most basic > > posting rules: > > https://www.kernel.org/doc/html/next/process/maintainer-netdev.html#tl-dr > > > > The reviewers are getting frustrated. Please, help us help you. > > Stop throwing code at the list and work out a plan with Andrew > > and others on how to get something merged... > > Sorry for trouble caused, will learn about the guidance provided by > the review comments, and follow up on the guidance and have the full > internal review of the patch updates before pushing the patch series. I renew my will of helping in any kind of manner in this, I love the intention for EDMAv2 to have an upstream driver instead of SSDK, hoping in the future to also have the same treatement for EDMAv1 (it's really a pitty to have a support hole with ipq807x not supported) Feel free to send an email or anything, considering this is massive, an extra eye before sending might make things better than reaching (I can already see this) a massive series with at least 20 revision given the complexity of this thing.
On 1/13/2024 1:56 AM, Christian Marangi wrote: > On Thu, Jan 11, 2024 at 11:49:53PM +0800, Jie Luo wrote: >> >> >> On 1/11/2024 6:24 AM, Jakub Kicinski wrote: >>> On Wed, 10 Jan 2024 19:40:12 +0800 Luo Jie wrote: >>>> The PPE(packet process engine) hardware block is available in Qualcomm >>>> IPQ chipsets that support PPE architecture, such as IPQ9574 and IPQ5332. >>> >>> What's the relationship between this driver and QCA8084? >> >> The PPE (packet processing engine) is the network processing hardware block >> in QCOM IPQ SoC. It includes the ethernet MAC and UNIPHY(PCS). This driver >> is the base PPE driver which brings up the PPE and handles MAC/UNIPHY >> operations. QCA8084 is the external 2.5Gbps 4-port PHY device, which can be >> connected with PPE integrated MAC by UNIPHY(PCS). >> >> Here is the relationship. >> PPE integrated MAC --- PPE integrated UNIPHY(PCS) --- (PCS)QCA8084. >> >>> >>> In the last month I see separate changes from you for mdio-ipq4019.c, >>> phy/at803x.c and now this driver (none of which got merged, AFAICT.) >>> Are you actually the author of this code, or are you just trying >>> to upstream bunch of vendor code? >> >> Yes, Jakub, there are two authors in these patch series, Lei Wei and me. >> The patches are already ready for some time, the code has been verified >> on the Qualcomm reference design board. These are not downstream drivers >> but drivers re-written for upstream. >> >>> >>> Now you're dumping another 10kLoC on the list, and even though this is >>> hardly your first posting you're apparently not aware of our most basic >>> posting rules: >>> https://www.kernel.org/doc/html/next/process/maintainer-netdev.html#tl-dr >>> >>> The reviewers are getting frustrated. Please, help us help you. >>> Stop throwing code at the list and work out a plan with Andrew >>> and others on how to get something merged... >> >> Sorry for trouble caused, will learn about the guidance provided by >> the review comments, and follow up on the guidance and have the full >> internal review of the patch updates before pushing the patch series. > > I renew my will of helping in any kind of manner in this, I love the > intention for EDMAv2 to have an upstream driver instead of SSDK, hoping > in the future to also have the same treatement for EDMAv1 (it's really a > pitty to have a support hole with ipq807x not supported) > > Feel free to send an email or anything, considering this is massive, an > extra eye before sending might make things better than reaching (I can > already see this) a massive series with at least 20 revision given the > complexity of this thing. > Thanks Christian for the help. Yes, the EDMAV2 driver will be posted some time after net-next is reopen and after this PPE driver patch series resumes. The EDMAv2 driver will be posted as separate driver series, which depends on this PPE driver. Currently we plan to post the EDMAv2 driver support for IPQ5332 and IPQ9574 firstly. For IPQ807x, it is a driver for an older architecture as you can see, but we will consider this for the future. We will certainly review it internally before publishing it later for upstream review.