e6c91556 - libeth: add Rx buffer management

Thu, 18 Apr 2024 11:36:13 +0000

libeth: add Rx buffer managementAdd a couple intuitive helpers to hide Rx buffer implementation detailsin the library and not multiplicate it between drivers. The settings aresorta optimized for 100G+ NICs, but nothing really HW-specific here.Use the new page_pool_dev_alloc() to dynamically switch betweensplit-page and full-page modes depending on MTU, page size, requiredheadroom etc. For example, on x86_64 with the default driver settingseach page is shared between 2 buffers. Turning on XDP (not in thisseries) -> increasing headroom requirement pushes truesize out of 2048boundary, leading to that each buffer starts getting a full page.The "ceiling" limit is %PAGE_SIZE, as only order-0 pages are used toavoid compound overhead. For the above architecture, this means maximumlinear frame size of 3712 w/o XDP.Not that &libeth_buf_queue is not a complete queue/ring structure fornow, rather a shim, but eventually the libeth-enabled drivers will moveto it, with iavf being the first one.Signed-off-by: Alexander Lobakin Signed-off-by: Tony Nguyen List of files: /linux-6.15/drivers/net/ethernet/intel/libeth/Kconfig

306ec721 - net: intel: introduce {, Intel} Ethernet common library

Thu, 18 Apr 2024 11:36:07 +0000

net: intel: introduce {, Intel} Ethernet common libraryNot a secret there's a ton of code duplication between two and more Intelethernet modules.Before introducing new changes, which would need to be copied over again,start decoupling the already existing duplicate functionality into a newmodule, which will be shared between several Intel Ethernet drivers.Add the lookup table which converts 8/10-bit hardware packet type intoa parsed bitfield structure for easy checking packet format parameters,such as payload level, IP version, etc. This is currently used by i40e,ice and iavf and it's all the same in all three drivers.The only difference introduced in this implementation is that instead ofdefining a 256 (or 1024 in case of ice) element array, add unlikely()condition to limit the input to 154 (current maximum non-reserved packettype). There's no reason to waste 600 (or even 3600) bytes only to nothurt very unlikely exception packets.The hash computation function now takes payload level directly as apkt_hash_type. There's a couple cases when non-IP ptypes are marked asL3 payload and in the previous versions their hash level would be 2, not3. But skb_set_hash() only sees difference between L4 and non-L4, thusthis won't change anything at all.The module is behind the hidden Kconfig symbol, which the drivers willselect when needed. The exports are behind 'LIBIE' namespace to limitthe scope of the functions.Not that non-HW-specific symbols will live in yet another module,libeth. This is done to easily distinguish pretty generic code readyfor reusing by any other vendor and/or for moving the layer up fromthe code useful in Intel's 1-100G drivers only.Signed-off-by: Alexander Lobakin Signed-off-by: Tony Nguyen List of files: /linux-6.15/drivers/net/ethernet/intel/libeth/Kconfig

Changes in Kconfig

e6c91556 - libeth: add Rx buffer management

306ec721 - net: intel: introduce {, Intel} Ethernet common library