[iwl-next,01/12] libeth: add cacheline / struct alignment helpers

Message ID	20240528134846.148890-2-aleksander.lobakin@intel.com (mailing list archive)
State	Awaiting Upstream
Delegated to:	Netdev Maintainers
Headers	show Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.15]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 413BE16F282; Tue, 28 May 2024 13:49:21 +0000 (UTC) From: Alexander Lobakin <aleksander.lobakin@intel.com> To: intel-wired-lan@lists.osuosl.org Cc: Alexander Lobakin <aleksander.lobakin@intel.com>, Tony Nguyen <anthony.l.nguyen@intel.com>, "David S. Miller" <davem@davemloft.net>, Eric Dumazet <edumazet@google.com>, Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>, Mina Almasry <almasrymina@google.com>, nex.sw.ncis.osdt.itp.upstreaming@intel.com, netdev@vger.kernel.org, linux-kernel@vger.kernel.org Subject: [PATCH iwl-next 01/12] libeth: add cacheline / struct alignment helpers Date: Tue, 28 May 2024 15:48:35 +0200 Message-ID: <20240528134846.148890-2-aleksander.lobakin@intel.com> In-Reply-To: <20240528134846.148890-1-aleksander.lobakin@intel.com> References: <20240528134846.148890-1-aleksander.lobakin@intel.com> Precedence: bulk MIME-Version: 1.0 Content-Transfer-Encoding: 8bit
Series	idpf: XDP chapter I: convert Rx to libeth \| expand [iwl-next,00/12] idpf: XDP chapter I: convert Rx to libeth [iwl-next,01/12] libeth: add cacheline / struct alignment helpers [iwl-next,02/12] idpf: stop using macros for accessing queue descriptors [iwl-next,03/12] idpf: split &idpf_queue into 4 strictly-typed queue structures [iwl-next,04/12] idpf: avoid bloating &idpf_q_vector with big %NR_CPUS [iwl-next,05/12] idpf: strictly assert cachelines of queue and queue vector structures [iwl-next,06/12] idpf: merge singleq and splitq &net_device_ops [iwl-next,07/12] idpf: compile singleq code only under default-n CONFIG_IDPF_SINGLEQ [iwl-next,08/12] idpf: reuse libeth's definitions of parsed ptype structures [iwl-next,09/12] idpf: remove legacy Page Pool Ethtool stats [iwl-next,10/12] libeth: support different types of buffers for Rx [iwl-next,11/12] idpf: convert header split mode to libeth + napi_build_skb() [iwl-next,12/12] idpf: use libeth Rx buffer management for payload buffer

Context	Check	Description
netdev/tree_selection	success	Guessing tree name failed - patch did not apply, async

diff --git a/scripts/kernel-doc b/scripts/kernel-doc index 95a59ac78f82..d0cf9a2d82de 100755 --- a/scripts/kernel-doc +++ b/scripts/kernel-doc @@ -1155,6 +1155,7 @@ sub dump_struct($$) { $members =~ s/\bstruct_group_attr\s*$([^,]*,){2}/STRUCT_GROUP(/gos; $members =~ s/\bstruct_group_tagged\s*\(([^,]*),([^,]*),/struct $1 $2; STRUCT_GROUP(/gos; $members =~ s/\b__struct_group\s*\(([^,]*,){3}/STRUCT_GROUP(/gos; + $members =~ s/\blibeth_cacheline_group\s*\(([^,]*,)/struct { } $1; STRUCT_GROUP(/gos; $members =~ s/\bSTRUCT_GROUP(\(((?:(?>[^)(]+)|(?1))*)$)[^;]*;/$2/gos; my $args = qr{([^,)]+)}; diff --git a/include/net/libeth/cache.h b/include/net/libeth/cache.h new file mode 100644 index 000000000000..5579240913d2 --- /dev/null +++ b/include/net/libeth/cache.h @@ -0,0 +1,100 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* Copyright (C) 2024 Intel Corporation */ + +#ifndef __LIBETH_CACHE_H +#define __LIBETH_CACHE_H + +#include <linux/cache.h> + +/* ``__aligned_largest`` is architecture-dependent. Get the actual alignment */ +#define ___LIBETH_LARGEST_ALIGN \ + sizeof(struct { long __UNIQUE_ID(long_); } __aligned_largest) +#define __LIBETH_LARGEST_ALIGN \ + (___LIBETH_LARGEST_ALIGN > SMP_CACHE_BYTES ? \ + ___LIBETH_LARGEST_ALIGN : SMP_CACHE_BYTES) +#define __LIBETH_LARGEST_ALIGNED(sz) \ + ALIGN(sz, __LIBETH_LARGEST_ALIGN) + +#define __libeth_cacheline_group_begin(grp) \ + __cacheline_group_begin(grp) __aligned(__LIBETH_LARGEST_ALIGN) +#define __libeth_cacheline_group_end(grp) \ + __cacheline_group_end(grp) __aligned(sizeof(long)) + +/** + * libeth_cacheline_group - declare a cacheline-aligned field group + * @grp: name of the group (usually 'read_mostly', 'read_write', or 'cold') + * @...: struct fields inside the group + * + * Note that the whole group is cacheline-aligned, but the end marker is + * aligned to long, so that you pass the (almost) actual field size sum to + * the assertion macros below instead of CL-aligned values. + * Each cacheline group must be described in struct's kernel-doc. + */ +#define libeth_cacheline_group(grp, ...) \ + struct_group(grp, \ + __libeth_cacheline_group_begin(grp); \ + __VA_ARGS__ \ + __libeth_cacheline_group_end(grp); \ + ) + +/** + * libeth_cacheline_group_assert - make sure cacheline group size is expected + * @type: type of the structure containing the group + * @grp: group name inside the struct + * @sz: expected group size + */ +#if defined(CONFIG_64BIT) && SMP_CACHE_BYTES == 64 +#define libeth_cacheline_group_assert(type, grp, sz) \ + static_assert(offsetof(type, __cacheline_group_end__##grp) - \ + offsetofend(type, __cacheline_group_begin__##grp) == \ + (sz)) +#define __libeth_cacheline_struct_assert(type, sz) \ + static_assert(sizeof(type) == (sz)) +#else /* !CONFIG_64BIT || SMP_CACHE_BYTES != 64 */ +#define libeth_cacheline_group_assert(type, grp, sz) \ + static_assert(offsetof(type, __cacheline_group_end__##grp) - \ + offsetofend(type, __cacheline_group_begin__##grp) <= \ + (sz)) +#define __libeth_cacheline_struct_assert(type, sz) \ + static_assert(sizeof(type) <= (sz)) +#endif /* !CONFIG_64BIT || SMP_CACHE_BYTES != 64 */ + +#define __libeth_cls1(sz1) \ + __LIBETH_LARGEST_ALIGNED(sz1) +#define __libeth_cls2(sz1, sz2) \ + (__LIBETH_LARGEST_ALIGNED(sz1) + __LIBETH_LARGEST_ALIGNED(sz2)) +#define __libeth_cls3(sz1, sz2, sz3) \ + (__LIBETH_LARGEST_ALIGNED(sz1) + __LIBETH_LARGEST_ALIGNED(sz2) + \ + __LIBETH_LARGEST_ALIGNED(sz3)) +#define __libeth_cls(...) \ + CONCATENATE(__libeth_cls, COUNT_ARGS(__VA_ARGS__))(__VA_ARGS__) + +/** + * libeth_cacheline_struct_assert - make sure CL-based struct size is expected + * @type: type of the struct + * @...: from 1 to 3 CL group sizes (read-mostly, read-write, cold) + * + * When a struct contains several CL groups, it's difficult to predict its size + * on different architectures. The macro instead takes sizes of all of the + * groups the structure contains and generates the final struct size. + */ +#define libeth_cacheline_struct_assert(type, ...) \ + __libeth_cacheline_struct_assert(type, __libeth_cls(__VA_ARGS__)); \ + static_assert(__alignof(type) >= __LIBETH_LARGEST_ALIGN) + +/** + * libeth_cacheline_set_assert - make sure CL-based struct layout is expected + * @type: type of the struct + * @ro: expected size of the read-mostly group + * @rw: expected size of the read-write group + * @c: expected size of the cold group + * + * Check that each group size is expected and then do final struct size check. + */ +#define libeth_cacheline_set_assert(type, ro, rw, c) \ + libeth_cacheline_group_assert(type, read_mostly, ro); \ + libeth_cacheline_group_assert(type, read_write, rw); \ + libeth_cacheline_group_assert(type, cold, c); \ + libeth_cacheline_struct_assert(type, ro, rw, c) + +#endif /* __LIBETH_CACHE_H */

[iwl-next,01/12] libeth: add cacheline / struct alignment helpers

Checks

Commit Message

Comments

Patch