From patchwork Tue Aug 13 22:22:36 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Tony Nguyen X-Patchwork-Id: 13762622 X-Patchwork-Delegate: kuba@kernel.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.8]) (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 EDE231AAE10 for ; Tue, 13 Aug 2024 22:22:55 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=192.198.163.8 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587777; cv=none; b=F9x7iui1zldUyBpYij6PUd0GQC49sXtmt4oOEjt0rOp7OrEhZyV/Fhx/P+UCrBHyBEKV6qRv7e1bTNa1/wB+kgFVj4tKbBj1Ws7P30BWJIj9p082yz1Cdd8AAjeO8XoInFk/+sb/Z7qVLCdJpenuqjSQH/iTmx7MQCrGEt86w2g= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587777; c=relaxed/simple; bh=6TAdpQz6PFfeYz4b9vjrb+jwoaa3+l5JaXG5oN52HTE=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=uZq8sIXcJjs4BbwhmAVLaG7/7uq989AB+uWu1QXbb/Ckz80fvkLt3znZRQrWvgVNayVOZTIFIaZXaH+rfLcjbc+OUBEV6kH5d0bXC7XQCb73FjFHMkcWgcYbh8KWqAki47rNVWkQc52eV0WtZqNe0QrTvhWzZKR4Kgkfub8ZzU8= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com; spf=pass smtp.mailfrom=intel.com; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b=NtZl6HD+; arc=none smtp.client-ip=192.198.163.8 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=intel.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b="NtZl6HD+" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1723587776; x=1755123776; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=6TAdpQz6PFfeYz4b9vjrb+jwoaa3+l5JaXG5oN52HTE=; b=NtZl6HD+T95WcJdHipdiHK6tOnESY/GzCGRQukmeeTJo/LfaQKgYxZD2 91jXd82ATxsVpV7Xhs6ctIlwydR5SB4+f2H7wrwHTt+CksmCsQXbvdVwX S0numAp0lKGNGRX7eF7yt3tLlXwxW4TqtDfXkmfG0n6rdylN1nKu6k+8/ 6hS14lNy+TUr7DYgrHi8Eh9/6T4nIi2usGI8yGxd08r/aIes9iJ8DLx89 MbXkAqcMPng7gqcIVCcBkFTkSPC/kBYm9sMJVTZkMISESiGrz9uVrIb1h 0hZcjopKsrcIrBMppL2eiSqA1gJIheVWS7QkhDQsQ3AMc2aEvdVLbwnCC w==; X-CSE-ConnectionGUID: F7/oHgKTSiaIzmK6xvo5wA== X-CSE-MsgGUID: uWBDxZshS+SisRG1AOEsPg== X-IronPort-AV: E=McAfee;i="6700,10204,11163"; a="39287072" X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="39287072" Received: from orviesa007.jf.intel.com ([10.64.159.147]) by fmvoesa102.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 13 Aug 2024 15:22:54 -0700 X-CSE-ConnectionGUID: 4fSM2LicTWqStpmUAx8bGQ== X-CSE-MsgGUID: 3jzEkLz5TmGlBymLnPL1Qg== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="59381048" Received: from anguy11-upstream.jf.intel.com ([10.166.9.133]) by orviesa007.jf.intel.com with ESMTP; 13 Aug 2024 15:22:54 -0700 From: Tony Nguyen To: davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com, edumazet@google.com, netdev@vger.kernel.org Cc: Junfeng Guo , anthony.l.nguyen@intel.com, ahmed.zaki@intel.com, madhu.chittim@intel.com, horms@kernel.org, hkelam@marvell.com, Marcin Szycik , Wojciech Drewek , Qi Zhang , Rafal Romanowski Subject: [PATCH net-next v2 01/13] ice: add parser create and destroy skeleton Date: Tue, 13 Aug 2024 15:22:36 -0700 Message-ID: <20240813222249.3708070-2-anthony.l.nguyen@intel.com> X-Mailer: git-send-email 2.42.0 In-Reply-To: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> References: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Patchwork-Delegate: kuba@kernel.org From: Junfeng Guo Add new parser module which can parse a packet in binary and generate information like ptype, protocol/offset pairs and flags which can be later used to feed the FXP profile creation directly. Add skeleton of the create and destroy APIs: ice_parser_create() ice_parser_destroy() Reviewed-by: Simon Horman Reviewed-by: Marcin Szycik Reviewed-by: Wojciech Drewek Signed-off-by: Qi Zhang Signed-off-by: Junfeng Guo Co-developed-by: Ahmed Zaki Signed-off-by: Ahmed Zaki Tested-by: Rafal Romanowski Signed-off-by: Tony Nguyen --- drivers/net/ethernet/intel/ice/Makefile | 1 + drivers/net/ethernet/intel/ice/ice_common.h | 1 + drivers/net/ethernet/intel/ice/ice_parser.c | 32 +++++++++++++++++++++ drivers/net/ethernet/intel/ice/ice_parser.h | 13 +++++++++ 4 files changed, 47 insertions(+) create mode 100644 drivers/net/ethernet/intel/ice/ice_parser.c create mode 100644 drivers/net/ethernet/intel/ice/ice_parser.h diff --git a/drivers/net/ethernet/intel/ice/Makefile b/drivers/net/ethernet/intel/ice/Makefile index 03500e28ac99..23fa3f7f36ef 100644 --- a/drivers/net/ethernet/intel/ice/Makefile +++ b/drivers/net/ethernet/intel/ice/Makefile @@ -28,6 +28,7 @@ ice-y := ice_main.o \ ice_vlan_mode.o \ ice_flex_pipe.o \ ice_flow.o \ + ice_parser.o \ ice_idc.o \ devlink/devlink.o \ devlink/devlink_port.o \ diff --git a/drivers/net/ethernet/intel/ice/ice_common.h b/drivers/net/ethernet/intel/ice/ice_common.h index 66f29bac783a..27208a60cece 100644 --- a/drivers/net/ethernet/intel/ice/ice_common.h +++ b/drivers/net/ethernet/intel/ice/ice_common.h @@ -10,6 +10,7 @@ #include "ice_type.h" #include "ice_nvm.h" #include "ice_flex_pipe.h" +#include "ice_parser.h" #include #include "ice_switch.h" #include "ice_fdir.h" diff --git a/drivers/net/ethernet/intel/ice/ice_parser.c b/drivers/net/ethernet/intel/ice/ice_parser.c new file mode 100644 index 000000000000..6c50164efae0 --- /dev/null +++ b/drivers/net/ethernet/intel/ice/ice_parser.c @@ -0,0 +1,32 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Copyright (C) 2024 Intel Corporation */ + +#include "ice_common.h" + +/** + * ice_parser_create - create a parser instance + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated parser instance or ERR_PTR + * in case of error. + */ +struct ice_parser *ice_parser_create(struct ice_hw *hw) +{ + struct ice_parser *p; + + p = kzalloc(sizeof(*p), GFP_KERNEL); + if (!p) + return ERR_PTR(-ENOMEM); + + p->hw = hw; + return p; +} + +/** + * ice_parser_destroy - destroy a parser instance + * @psr: pointer to a parser instance + */ +void ice_parser_destroy(struct ice_parser *psr) +{ + kfree(psr); +} diff --git a/drivers/net/ethernet/intel/ice/ice_parser.h b/drivers/net/ethernet/intel/ice/ice_parser.h new file mode 100644 index 000000000000..09ed380eee32 --- /dev/null +++ b/drivers/net/ethernet/intel/ice/ice_parser.h @@ -0,0 +1,13 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* Copyright (C) 2024 Intel Corporation */ + +#ifndef _ICE_PARSER_H_ +#define _ICE_PARSER_H_ + +struct ice_parser { + struct ice_hw *hw; /* pointer to the hardware structure */ +}; + +struct ice_parser *ice_parser_create(struct ice_hw *hw); +void ice_parser_destroy(struct ice_parser *psr); +#endif /* _ICE_PARSER_H_ */ From patchwork Tue Aug 13 22:22:37 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Tony Nguyen X-Patchwork-Id: 13762625 X-Patchwork-Delegate: kuba@kernel.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.8]) (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 03B6F1AAE37 for ; Tue, 13 Aug 2024 22:22:56 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=192.198.163.8 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587780; cv=none; b=TS1pZ7UlpSPTqQcO01g21JfpoyIE9kg2S5XEbloroqovc7fcxF0ZUwSACqwNhZQueGrk1RLprwviVrjEDtuakZCA4VXEhSeFqK1GGSfniLfxdnTy+B42nEVTKv4YtoJnJ93YqSuOzv0kxiMWvYU1J0klaZ+dvWcpWgnrQgpPi1g= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587780; c=relaxed/simple; bh=oZB/+meoF3vWWLmppAhrZmUzx6ZWXj5JF/Ov3ZWgPKE=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=DbTaqMIcMUKWSSnRYQD5wXVSwePChdRt06GEnZ8M2Isf5dHIXJD/z3gk25aGwPm7YaxElJuj5pUII6TA1zFRW8mPZWtltGY+TlwVoNqk+YGeq0uwlwQlKxNRL9sURjv4TST+I86RfzGxxWXfi9VEiw2lHhoFdWXdLhy+8AvyBNs= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com; spf=pass smtp.mailfrom=intel.com; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b=djFvowKd; arc=none smtp.client-ip=192.198.163.8 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=intel.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b="djFvowKd" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1723587777; x=1755123777; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=oZB/+meoF3vWWLmppAhrZmUzx6ZWXj5JF/Ov3ZWgPKE=; b=djFvowKd8OQKsducWMCyHNTNHyIIlu2/8fxiKk8uROuXzBhmXnEryoLR Tv1cVkZhBIH60YEmbQ87QcRWV20cIuYKUwHVUFpBQQU6V6ib0fc3bBIYc 9y/F6fSrs9vuPRa3kZM0f3Mz+vsMt9SiY2gaztQAqij75uNAZcp0NtCfl Jz4KqkqXWDadc8MoUOz0ZoJQRnnlIFcVH8XWZ9UnBSAoM9EtLP8TNbW5O T953UQTF9UIcEYht9tmHK1p1N34AfWXWFQYdA2X5czLVfu+zJNTdOkLEA bAW4OHlA2UqZkCLJZFAPoeibhHhvMqD+c+V5tai71Pu4gJJkSBx0M58BQ w==; X-CSE-ConnectionGUID: pMiz0Ko3RDSNvUe134Dp/Q== X-CSE-MsgGUID: egSVCA2cRti8vVEsvTewkQ== X-IronPort-AV: E=McAfee;i="6700,10204,11163"; a="39287078" X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="39287078" Received: from orviesa007.jf.intel.com ([10.64.159.147]) by fmvoesa102.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 13 Aug 2024 15:22:54 -0700 X-CSE-ConnectionGUID: GcIRIBpAT+G27PX5t1aqFA== X-CSE-MsgGUID: L/qxnXd5QVuE3sYzhKMorw== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="59381051" Received: from anguy11-upstream.jf.intel.com ([10.166.9.133]) by orviesa007.jf.intel.com with ESMTP; 13 Aug 2024 15:22:54 -0700 From: Tony Nguyen To: davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com, edumazet@google.com, netdev@vger.kernel.org Cc: Junfeng Guo , anthony.l.nguyen@intel.com, ahmed.zaki@intel.com, madhu.chittim@intel.com, horms@kernel.org, hkelam@marvell.com, Marcin Szycik , Qi Zhang , Rafal Romanowski Subject: [PATCH net-next v2 02/13] ice: parse and init various DDP parser sections Date: Tue, 13 Aug 2024 15:22:37 -0700 Message-ID: <20240813222249.3708070-3-anthony.l.nguyen@intel.com> X-Mailer: git-send-email 2.42.0 In-Reply-To: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> References: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Patchwork-Delegate: kuba@kernel.org From: Junfeng Guo Parse the following DDP sections: - ICE_SID_RXPARSER_IMEM into an array of struct ice_imem_item - ICE_SID_RXPARSER_METADATA_INIT into an array of struct ice_metainit_item - ICE_SID_RXPARSER_CAM or ICE_SID_RXPARSER_PG_SPILL into an array of struct ice_pg_cam_item - ICE_SID_RXPARSER_NOMATCH_CAM or ICE_SID_RXPARSER_NOMATCH_SPILL into an array of struct ice_pg_nm_cam_item - ICE_SID_RXPARSER_CAM into an array of ice_bst_tcam_item - ICE_SID_LBL_RXPARSER_TMEM into an array of ice_lbl_item - ICE_SID_RXPARSER_MARKER_PTYPE into an array of ice_ptype_mk_tcam_item - ICE_SID_RXPARSER_MARKER_GRP into an array of ice_mk_grp_item - ICE_SID_RXPARSER_PROTO_GRP into an array of ice_proto_grp_item - ICE_SID_RXPARSER_FLAG_REDIR into an array of ice_flg_rd_item - ICE_SID_XLT_KEY_BUILDER_SW, ICE_SID_XLT_KEY_BUILDER_ACL, ICE_SID_XLT_KEY_BUILDER_FD and ICE_SID_XLT_KEY_BUILDER_RSS into struct ice_xlt_kb Reviewed-by: Marcin Szycik Signed-off-by: Qi Zhang Signed-off-by: Junfeng Guo Co-developed-by: Ahmed Zaki Signed-off-by: Ahmed Zaki Tested-by: Rafal Romanowski Reviewed-by: Simon Horman Signed-off-by: Tony Nguyen --- drivers/net/ethernet/intel/ice/ice_ddp.c | 10 +- drivers/net/ethernet/intel/ice/ice_ddp.h | 13 + drivers/net/ethernet/intel/ice/ice_parser.c | 1383 +++++++++++++++++++ drivers/net/ethernet/intel/ice/ice_parser.h | 358 +++++ drivers/net/ethernet/intel/ice/ice_type.h | 1 + 5 files changed, 1760 insertions(+), 5 deletions(-) diff --git a/drivers/net/ethernet/intel/ice/ice_ddp.c b/drivers/net/ethernet/intel/ice/ice_ddp.c index f182179529b7..953262b88a58 100644 --- a/drivers/net/ethernet/intel/ice/ice_ddp.c +++ b/drivers/net/ethernet/intel/ice/ice_ddp.c @@ -289,11 +289,11 @@ void *ice_pkg_enum_section(struct ice_seg *ice_seg, struct ice_pkg_enum *state, * indicates a base offset of 10, and the index for the entry is 2, then * section handler function should set the offset to 10 + 2 = 12. */ -static void *ice_pkg_enum_entry(struct ice_seg *ice_seg, - struct ice_pkg_enum *state, u32 sect_type, - u32 *offset, - void *(*handler)(u32 sect_type, void *section, - u32 index, u32 *offset)) +void *ice_pkg_enum_entry(struct ice_seg *ice_seg, + struct ice_pkg_enum *state, u32 sect_type, + u32 *offset, + void *(*handler)(u32 sect_type, void *section, + u32 index, u32 *offset)) { void *entry; diff --git a/drivers/net/ethernet/intel/ice/ice_ddp.h b/drivers/net/ethernet/intel/ice/ice_ddp.h index 622543f08b43..97f272317475 100644 --- a/drivers/net/ethernet/intel/ice/ice_ddp.h +++ b/drivers/net/ethernet/intel/ice/ice_ddp.h @@ -261,10 +261,17 @@ struct ice_meta_sect { #define ICE_SID_CDID_KEY_BUILDER_RSS 47 #define ICE_SID_CDID_REDIR_RSS 48 +#define ICE_SID_RXPARSER_CAM 50 +#define ICE_SID_RXPARSER_NOMATCH_CAM 51 +#define ICE_SID_RXPARSER_IMEM 52 #define ICE_SID_RXPARSER_MARKER_PTYPE 55 #define ICE_SID_RXPARSER_BOOST_TCAM 56 +#define ICE_SID_RXPARSER_PROTO_GRP 57 #define ICE_SID_RXPARSER_METADATA_INIT 58 #define ICE_SID_TXPARSER_BOOST_TCAM 66 +#define ICE_SID_RXPARSER_MARKER_GRP 72 +#define ICE_SID_RXPARSER_PG_SPILL 76 +#define ICE_SID_RXPARSER_NOMATCH_SPILL 78 #define ICE_SID_XLT0_PE 80 #define ICE_SID_XLT_KEY_BUILDER_PE 81 @@ -276,6 +283,7 @@ struct ice_meta_sect { #define ICE_SID_CDID_KEY_BUILDER_PE 87 #define ICE_SID_CDID_REDIR_PE 88 +#define ICE_SID_RXPARSER_FLAG_REDIR 97 /* Label Metadata section IDs */ #define ICE_SID_LBL_FIRST 0x80000010 #define ICE_SID_LBL_RXPARSER_TMEM 0x80000018 @@ -451,6 +459,11 @@ int ice_update_pkg(struct ice_hw *hw, struct ice_buf *bufs, u32 count); int ice_pkg_buf_reserve_section(struct ice_buf_build *bld, u16 count); u16 ice_pkg_buf_get_active_sections(struct ice_buf_build *bld); +void * +ice_pkg_enum_entry(struct ice_seg *ice_seg, struct ice_pkg_enum *state, + u32 sect_type, u32 *offset, + void *(*handler)(u32 sect_type, void *section, + u32 index, u32 *offset)); void *ice_pkg_enum_section(struct ice_seg *ice_seg, struct ice_pkg_enum *state, u32 sect_type); diff --git a/drivers/net/ethernet/intel/ice/ice_parser.c b/drivers/net/ethernet/intel/ice/ice_parser.c index 6c50164efae0..5fdc87034fc6 100644 --- a/drivers/net/ethernet/intel/ice/ice_parser.c +++ b/drivers/net/ethernet/intel/ice/ice_parser.c @@ -3,6 +3,1271 @@ #include "ice_common.h" +struct ice_pkg_sect_hdr { + __le16 count; + __le16 offset; +}; + +/** + * ice_parser_sect_item_get - parse an item from a section + * @sect_type: section type + * @section: section object + * @index: index of the item to get + * @offset: dummy as prototype of ice_pkg_enum_entry's last parameter + * + * Return: a pointer to the item or NULL. + */ +static void *ice_parser_sect_item_get(u32 sect_type, void *section, + u32 index, u32 __maybe_unused *offset) +{ + size_t data_off = ICE_SEC_DATA_OFFSET; + struct ice_pkg_sect_hdr *hdr; + size_t size; + + if (!section) + return NULL; + + switch (sect_type) { + case ICE_SID_RXPARSER_IMEM: + size = ICE_SID_RXPARSER_IMEM_ENTRY_SIZE; + break; + case ICE_SID_RXPARSER_METADATA_INIT: + size = ICE_SID_RXPARSER_METADATA_INIT_ENTRY_SIZE; + break; + case ICE_SID_RXPARSER_CAM: + size = ICE_SID_RXPARSER_CAM_ENTRY_SIZE; + break; + case ICE_SID_RXPARSER_PG_SPILL: + size = ICE_SID_RXPARSER_PG_SPILL_ENTRY_SIZE; + break; + case ICE_SID_RXPARSER_NOMATCH_CAM: + size = ICE_SID_RXPARSER_NOMATCH_CAM_ENTRY_SIZE; + break; + case ICE_SID_RXPARSER_NOMATCH_SPILL: + size = ICE_SID_RXPARSER_NOMATCH_SPILL_ENTRY_SIZE; + break; + case ICE_SID_RXPARSER_BOOST_TCAM: + size = ICE_SID_RXPARSER_BOOST_TCAM_ENTRY_SIZE; + break; + case ICE_SID_LBL_RXPARSER_TMEM: + data_off = ICE_SEC_LBL_DATA_OFFSET; + size = ICE_SID_LBL_ENTRY_SIZE; + break; + case ICE_SID_RXPARSER_MARKER_PTYPE: + size = ICE_SID_RXPARSER_MARKER_TYPE_ENTRY_SIZE; + break; + case ICE_SID_RXPARSER_MARKER_GRP: + size = ICE_SID_RXPARSER_MARKER_GRP_ENTRY_SIZE; + break; + case ICE_SID_RXPARSER_PROTO_GRP: + size = ICE_SID_RXPARSER_PROTO_GRP_ENTRY_SIZE; + break; + case ICE_SID_RXPARSER_FLAG_REDIR: + size = ICE_SID_RXPARSER_FLAG_REDIR_ENTRY_SIZE; + break; + default: + return NULL; + } + + hdr = section; + if (index >= le16_to_cpu(hdr->count)) + return NULL; + + return section + data_off + index * size; +} + +/** + * ice_parser_create_table - create an item table from a section + * @hw: pointer to the hardware structure + * @sect_type: section type + * @item_size: item size in bytes + * @length: number of items in the table to create + * @parse_item: the function to parse the item + * @no_offset: ignore header offset, calculate index from 0 + * + * Return: a pointer to the allocated table or ERR_PTR. + */ +static void * +ice_parser_create_table(struct ice_hw *hw, u32 sect_type, + u32 item_size, u32 length, + void (*parse_item)(struct ice_hw *hw, u16 idx, + void *item, void *data, + int size), bool no_offset) +{ + struct ice_pkg_enum state = {}; + struct ice_seg *seg = hw->seg; + void *table, *data, *item; + u16 idx = 0; + + if (!seg) + return ERR_PTR(-EINVAL); + + table = kzalloc(item_size * length, GFP_KERNEL); + if (!table) + return ERR_PTR(-ENOMEM); + + do { + data = ice_pkg_enum_entry(seg, &state, sect_type, NULL, + ice_parser_sect_item_get); + seg = NULL; + if (data) { + struct ice_pkg_sect_hdr *hdr = state.sect; + + if (!no_offset) + idx = le16_to_cpu(hdr->offset) + + state.entry_idx; + + item = (void *)((uintptr_t)table + idx * item_size); + parse_item(hw, idx, item, data, item_size); + + if (no_offset) + idx++; + } + } while (data); + + return table; +} + +/*** ICE_SID_RXPARSER_IMEM section ***/ +#define ICE_IM_BM_ALU0 BIT(0) +#define ICE_IM_BM_ALU1 BIT(1) +#define ICE_IM_BM_ALU2 BIT(2) +#define ICE_IM_BM_PG BIT(3) + +/** + * ice_imem_bm_init - parse 4 bits of Boost Main + * @bm: pointer to the Boost Main structure + * @data: Boost Main data to be parsed + */ +static void ice_imem_bm_init(struct ice_bst_main *bm, u8 data) +{ + bm->alu0 = FIELD_GET(ICE_IM_BM_ALU0, data); + bm->alu1 = FIELD_GET(ICE_IM_BM_ALU1, data); + bm->alu2 = FIELD_GET(ICE_IM_BM_ALU2, data); + bm->pg = FIELD_GET(ICE_IM_BM_PG, data); +} + +#define ICE_IM_BKB_PRIO GENMASK(7, 0) +#define ICE_IM_BKB_TSR_CTRL BIT(8) + +/** + * ice_imem_bkb_init - parse 10 bits of Boost Main Build + * @bkb: pointer to the Boost Main Build structure + * @data: Boost Main Build data to be parsed + */ +static void ice_imem_bkb_init(struct ice_bst_keybuilder *bkb, u16 data) +{ + bkb->prio = FIELD_GET(ICE_IM_BKB_PRIO, data); + bkb->tsr_ctrl = FIELD_GET(ICE_IM_BKB_TSR_CTRL, data); +} + +#define ICE_IM_NPKB_OPC GENMASK(1, 0) +#define ICE_IM_NPKB_S_R0 GENMASK(9, 2) +#define ICE_IM_NPKB_L_R1 GENMASK(17, 10) + +/** + * ice_imem_npkb_init - parse 18 bits of Next Protocol Key Build + * @kb: pointer to the Next Protocol Key Build structure + * @data: Next Protocol Key Build data to be parsed + */ +static void ice_imem_npkb_init(struct ice_np_keybuilder *kb, u32 data) +{ + kb->opc = FIELD_GET(ICE_IM_NPKB_OPC, data); + kb->start_reg0 = FIELD_GET(ICE_IM_NPKB_S_R0, data); + kb->len_reg1 = FIELD_GET(ICE_IM_NPKB_L_R1, data); +} + +#define ICE_IM_PGKB_F0_ENA BIT_ULL(0) +#define ICE_IM_PGKB_F0_IDX GENMASK_ULL(6, 1) +#define ICE_IM_PGKB_F1_ENA BIT_ULL(7) +#define ICE_IM_PGKB_F1_IDX GENMASK_ULL(13, 8) +#define ICE_IM_PGKB_F2_ENA BIT_ULL(14) +#define ICE_IM_PGKB_F2_IDX GENMASK_ULL(20, 15) +#define ICE_IM_PGKB_F3_ENA BIT_ULL(21) +#define ICE_IM_PGKB_F3_IDX GENMASK_ULL(27, 22) +#define ICE_IM_PGKB_AR_IDX GENMASK_ULL(34, 28) + +/** + * ice_imem_pgkb_init - parse 35 bits of Parse Graph Key Build + * @kb: pointer to the Parse Graph Key Build structure + * @data: Parse Graph Key Build data to be parsed + */ +static void ice_imem_pgkb_init(struct ice_pg_keybuilder *kb, u64 data) +{ + kb->flag0_ena = FIELD_GET(ICE_IM_PGKB_F0_ENA, data); + kb->flag0_idx = FIELD_GET(ICE_IM_PGKB_F0_IDX, data); + kb->flag1_ena = FIELD_GET(ICE_IM_PGKB_F1_ENA, data); + kb->flag1_idx = FIELD_GET(ICE_IM_PGKB_F1_IDX, data); + kb->flag2_ena = FIELD_GET(ICE_IM_PGKB_F2_ENA, data); + kb->flag2_idx = FIELD_GET(ICE_IM_PGKB_F2_IDX, data); + kb->flag3_ena = FIELD_GET(ICE_IM_PGKB_F3_ENA, data); + kb->flag3_idx = FIELD_GET(ICE_IM_PGKB_F3_IDX, data); + kb->alu_reg_idx = FIELD_GET(ICE_IM_PGKB_AR_IDX, data); +} + +#define ICE_IM_ALU_OPC GENMASK_ULL(5, 0) +#define ICE_IM_ALU_SS GENMASK_ULL(13, 6) +#define ICE_IM_ALU_SL GENMASK_ULL(18, 14) +#define ICE_IM_ALU_SXS BIT_ULL(19) +#define ICE_IM_ALU_SXK GENMASK_ULL(23, 20) +#define ICE_IM_ALU_SRID GENMASK_ULL(30, 24) +#define ICE_IM_ALU_DRID GENMASK_ULL(37, 31) +#define ICE_IM_ALU_INC0 BIT_ULL(38) +#define ICE_IM_ALU_INC1 BIT_ULL(39) +#define ICE_IM_ALU_POO GENMASK_ULL(41, 40) +#define ICE_IM_ALU_PO GENMASK_ULL(49, 42) +#define ICE_IM_ALU_BA_S 50 /* offset for the 2nd 64-bits field */ +#define ICE_IM_ALU_BA GENMASK_ULL(57 - ICE_IM_ALU_BA_S, \ + 50 - ICE_IM_ALU_BA_S) +#define ICE_IM_ALU_IMM GENMASK_ULL(73 - ICE_IM_ALU_BA_S, \ + 58 - ICE_IM_ALU_BA_S) +#define ICE_IM_ALU_DFE BIT_ULL(74 - ICE_IM_ALU_BA_S) +#define ICE_IM_ALU_DS GENMASK_ULL(80 - ICE_IM_ALU_BA_S, \ + 75 - ICE_IM_ALU_BA_S) +#define ICE_IM_ALU_DL GENMASK_ULL(86 - ICE_IM_ALU_BA_S, \ + 81 - ICE_IM_ALU_BA_S) +#define ICE_IM_ALU_FEI BIT_ULL(87 - ICE_IM_ALU_BA_S) +#define ICE_IM_ALU_FSI GENMASK_ULL(95 - ICE_IM_ALU_BA_S, \ + 88 - ICE_IM_ALU_BA_S) + +/** + * ice_imem_alu_init - parse 96 bits of ALU entry + * @alu: pointer to the ALU entry structure + * @data: ALU entry data to be parsed + * @off: offset of the ALU entry data + */ +static void ice_imem_alu_init(struct ice_alu *alu, u8 *data, u8 off) +{ + u64 d64; + u8 idd; + + d64 = *((u64 *)data) >> off; + + alu->opc = FIELD_GET(ICE_IM_ALU_OPC, d64); + alu->src_start = FIELD_GET(ICE_IM_ALU_SS, d64); + alu->src_len = FIELD_GET(ICE_IM_ALU_SL, d64); + alu->shift_xlate_sel = FIELD_GET(ICE_IM_ALU_SXS, d64); + alu->shift_xlate_key = FIELD_GET(ICE_IM_ALU_SXK, d64); + alu->src_reg_id = FIELD_GET(ICE_IM_ALU_SRID, d64); + alu->dst_reg_id = FIELD_GET(ICE_IM_ALU_DRID, d64); + alu->inc0 = FIELD_GET(ICE_IM_ALU_INC0, d64); + alu->inc1 = FIELD_GET(ICE_IM_ALU_INC1, d64); + alu->proto_offset_opc = FIELD_GET(ICE_IM_ALU_POO, d64); + alu->proto_offset = FIELD_GET(ICE_IM_ALU_PO, d64); + + idd = (ICE_IM_ALU_BA_S + off) / BITS_PER_BYTE; + off = (ICE_IM_ALU_BA_S + off) % BITS_PER_BYTE; + d64 = *((u64 *)(&data[idd])) >> off; + + alu->branch_addr = FIELD_GET(ICE_IM_ALU_BA, d64); + alu->imm = FIELD_GET(ICE_IM_ALU_IMM, d64); + alu->dedicate_flags_ena = FIELD_GET(ICE_IM_ALU_DFE, d64); + alu->dst_start = FIELD_GET(ICE_IM_ALU_DS, d64); + alu->dst_len = FIELD_GET(ICE_IM_ALU_DL, d64); + alu->flags_extr_imm = FIELD_GET(ICE_IM_ALU_FEI, d64); + alu->flags_start_imm = FIELD_GET(ICE_IM_ALU_FSI, d64); +} + +#define ICE_IMEM_BM_S 0 +#define ICE_IMEM_BKB_S 4 +#define ICE_IMEM_BKB_IDD (ICE_IMEM_BKB_S / BITS_PER_BYTE) +#define ICE_IMEM_BKB_OFF (ICE_IMEM_BKB_S % BITS_PER_BYTE) +#define ICE_IMEM_PGP GENMASK(15, 14) +#define ICE_IMEM_NPKB_S 16 +#define ICE_IMEM_NPKB_IDD (ICE_IMEM_NPKB_S / BITS_PER_BYTE) +#define ICE_IMEM_NPKB_OFF (ICE_IMEM_NPKB_S % BITS_PER_BYTE) +#define ICE_IMEM_PGKB_S 34 +#define ICE_IMEM_PGKB_IDD (ICE_IMEM_PGKB_S / BITS_PER_BYTE) +#define ICE_IMEM_PGKB_OFF (ICE_IMEM_PGKB_S % BITS_PER_BYTE) +#define ICE_IMEM_ALU0_S 69 +#define ICE_IMEM_ALU0_IDD (ICE_IMEM_ALU0_S / BITS_PER_BYTE) +#define ICE_IMEM_ALU0_OFF (ICE_IMEM_ALU0_S % BITS_PER_BYTE) +#define ICE_IMEM_ALU1_S 165 +#define ICE_IMEM_ALU1_IDD (ICE_IMEM_ALU1_S / BITS_PER_BYTE) +#define ICE_IMEM_ALU1_OFF (ICE_IMEM_ALU1_S % BITS_PER_BYTE) +#define ICE_IMEM_ALU2_S 357 +#define ICE_IMEM_ALU2_IDD (ICE_IMEM_ALU2_S / BITS_PER_BYTE) +#define ICE_IMEM_ALU2_OFF (ICE_IMEM_ALU2_S % BITS_PER_BYTE) + +/** + * ice_imem_parse_item - parse 384 bits of IMEM entry + * @hw: pointer to the hardware structure + * @idx: index of IMEM entry + * @item: item of IMEM entry + * @data: IMEM entry data to be parsed + * @size: size of IMEM entry + */ +static void ice_imem_parse_item(struct ice_hw *hw, u16 idx, void *item, + void *data, int __maybe_unused size) +{ + struct ice_imem_item *ii = item; + u8 *buf = data; + + ii->idx = idx; + + ice_imem_bm_init(&ii->b_m, *(u8 *)buf); + ice_imem_bkb_init(&ii->b_kb, + *((u16 *)(&buf[ICE_IMEM_BKB_IDD])) >> + ICE_IMEM_BKB_OFF); + + ii->pg_prio = FIELD_GET(ICE_IMEM_PGP, *(u16 *)buf); + + ice_imem_npkb_init(&ii->np_kb, + *((u32 *)(&buf[ICE_IMEM_NPKB_IDD])) >> + ICE_IMEM_NPKB_OFF); + ice_imem_pgkb_init(&ii->pg_kb, + *((u64 *)(&buf[ICE_IMEM_PGKB_IDD])) >> + ICE_IMEM_PGKB_OFF); + + ice_imem_alu_init(&ii->alu0, + &buf[ICE_IMEM_ALU0_IDD], + ICE_IMEM_ALU0_OFF); + ice_imem_alu_init(&ii->alu1, + &buf[ICE_IMEM_ALU1_IDD], + ICE_IMEM_ALU1_OFF); + ice_imem_alu_init(&ii->alu2, + &buf[ICE_IMEM_ALU2_IDD], + ICE_IMEM_ALU2_OFF); +} + +/** + * ice_imem_table_get - create an imem table + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated IMEM table. + */ +static struct ice_imem_item *ice_imem_table_get(struct ice_hw *hw) +{ + return ice_parser_create_table(hw, ICE_SID_RXPARSER_IMEM, + sizeof(struct ice_imem_item), + ICE_IMEM_TABLE_SIZE, + ice_imem_parse_item, false); +} + +/*** ICE_SID_RXPARSER_METADATA_INIT section ***/ +#define ICE_MI_TSR GENMASK_ULL(7, 0) +#define ICE_MI_HO GENMASK_ULL(16, 8) +#define ICE_MI_PC GENMASK_ULL(24, 17) +#define ICE_MI_PGRN GENMASK_ULL(35, 25) +#define ICE_MI_CD GENMASK_ULL(38, 36) +#define ICE_MI_GAC BIT_ULL(39) +#define ICE_MI_GADM GENMASK_ULL(44, 40) +#define ICE_MI_GADS GENMASK_ULL(48, 45) +#define ICE_MI_GADL GENMASK_ULL(53, 49) +#define ICE_MI_GAI GENMASK_ULL(59, 56) +#define ICE_MI_GBC BIT_ULL(60) +#define ICE_MI_GBDM_S 61 /* offset for the 2nd 64-bits field */ +#define ICE_MI_GBDM_IDD (ICE_MI_GBDM_S / BITS_PER_BYTE) +#define ICE_MI_GBDM_OFF (ICE_MI_GBDM_S % BITS_PER_BYTE) + +#define ICE_MI_GBDM_GENMASK_ULL(high, low) \ + GENMASK_ULL((high) - ICE_MI_GBDM_S, (low) - ICE_MI_GBDM_S) +#define ICE_MI_GBDM ICE_MI_GBDM_GENMASK_ULL(65, 61) +#define ICE_MI_GBDS ICE_MI_GBDM_GENMASK_ULL(69, 66) +#define ICE_MI_GBDL ICE_MI_GBDM_GENMASK_ULL(74, 70) +#define ICE_MI_GBI ICE_MI_GBDM_GENMASK_ULL(80, 77) +#define ICE_MI_GCC BIT_ULL(81 - ICE_MI_GBDM_S) +#define ICE_MI_GCDM ICE_MI_GBDM_GENMASK_ULL(86, 82) +#define ICE_MI_GCDS ICE_MI_GBDM_GENMASK_ULL(90, 87) +#define ICE_MI_GCDL ICE_MI_GBDM_GENMASK_ULL(95, 91) +#define ICE_MI_GCI ICE_MI_GBDM_GENMASK_ULL(101, 98) +#define ICE_MI_GDC BIT_ULL(102 - ICE_MI_GBDM_S) +#define ICE_MI_GDDM ICE_MI_GBDM_GENMASK_ULL(107, 103) +#define ICE_MI_GDDS ICE_MI_GBDM_GENMASK_ULL(111, 108) +#define ICE_MI_GDDL ICE_MI_GBDM_GENMASK_ULL(116, 112) +#define ICE_MI_GDI ICE_MI_GBDM_GENMASK_ULL(122, 119) +#define ICE_MI_FLAG_S 123 /* offset for the 3rd 64-bits field */ +#define ICE_MI_FLAG_IDD (ICE_MI_FLAG_S / BITS_PER_BYTE) +#define ICE_MI_FLAG_OFF (ICE_MI_FLAG_S % BITS_PER_BYTE) +#define ICE_MI_FLAG GENMASK_ULL(186 - ICE_MI_FLAG_S, \ + 123 - ICE_MI_FLAG_S) + +/** + * ice_metainit_parse_item - parse 192 bits of Metadata Init entry + * @hw: pointer to the hardware structure + * @idx: index of Metadata Init entry + * @item: item of Metadata Init entry + * @data: Metadata Init entry data to be parsed + * @size: size of Metadata Init entry + */ +static void ice_metainit_parse_item(struct ice_hw *hw, u16 idx, void *item, + void *data, int __maybe_unused size) +{ + struct ice_metainit_item *mi = item; + u8 *buf = data; + u64 d64; + + mi->idx = idx; + + d64 = *(u64 *)buf; + + mi->tsr = FIELD_GET(ICE_MI_TSR, d64); + mi->ho = FIELD_GET(ICE_MI_HO, d64); + mi->pc = FIELD_GET(ICE_MI_PC, d64); + mi->pg_rn = FIELD_GET(ICE_MI_PGRN, d64); + mi->cd = FIELD_GET(ICE_MI_CD, d64); + + mi->gpr_a_ctrl = FIELD_GET(ICE_MI_GAC, d64); + mi->gpr_a_data_mdid = FIELD_GET(ICE_MI_GADM, d64); + mi->gpr_a_data_start = FIELD_GET(ICE_MI_GADS, d64); + mi->gpr_a_data_len = FIELD_GET(ICE_MI_GADL, d64); + mi->gpr_a_id = FIELD_GET(ICE_MI_GAI, d64); + + mi->gpr_b_ctrl = FIELD_GET(ICE_MI_GBC, d64); + + d64 = *((u64 *)&buf[ICE_MI_GBDM_IDD]) >> ICE_MI_GBDM_OFF; + + mi->gpr_b_data_mdid = FIELD_GET(ICE_MI_GBDM, d64); + mi->gpr_b_data_start = FIELD_GET(ICE_MI_GBDS, d64); + mi->gpr_b_data_len = FIELD_GET(ICE_MI_GBDL, d64); + mi->gpr_b_id = FIELD_GET(ICE_MI_GBI, d64); + + mi->gpr_c_ctrl = FIELD_GET(ICE_MI_GCC, d64); + mi->gpr_c_data_mdid = FIELD_GET(ICE_MI_GCDM, d64); + mi->gpr_c_data_start = FIELD_GET(ICE_MI_GCDS, d64); + mi->gpr_c_data_len = FIELD_GET(ICE_MI_GCDL, d64); + mi->gpr_c_id = FIELD_GET(ICE_MI_GCI, d64); + + mi->gpr_d_ctrl = FIELD_GET(ICE_MI_GDC, d64); + mi->gpr_d_data_mdid = FIELD_GET(ICE_MI_GDDM, d64); + mi->gpr_d_data_start = FIELD_GET(ICE_MI_GDDS, d64); + mi->gpr_d_data_len = FIELD_GET(ICE_MI_GDDL, d64); + mi->gpr_d_id = FIELD_GET(ICE_MI_GDI, d64); + + d64 = *((u64 *)&buf[ICE_MI_FLAG_IDD]) >> ICE_MI_FLAG_OFF; + + mi->flags = FIELD_GET(ICE_MI_FLAG, d64); +} + +/** + * ice_metainit_table_get - create a metainit table + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated Metadata initialization table. + */ +static struct ice_metainit_item *ice_metainit_table_get(struct ice_hw *hw) +{ + return ice_parser_create_table(hw, ICE_SID_RXPARSER_METADATA_INIT, + sizeof(struct ice_metainit_item), + ICE_METAINIT_TABLE_SIZE, + ice_metainit_parse_item, false); +} + +/*** ICE_SID_RXPARSER_CAM, ICE_SID_RXPARSER_PG_SPILL, + * ICE_SID_RXPARSER_NOMATCH_CAM and ICE_SID_RXPARSER_NOMATCH_CAM + * sections ***/ +#define ICE_PGCA_NN GENMASK_ULL(10, 0) +#define ICE_PGCA_NPC GENMASK_ULL(18, 11) +#define ICE_PGCA_IPG BIT_ULL(19) +#define ICE_PGCA_PID GENMASK_ULL(30, 23) +#define ICE_PGCA_IMG BIT_ULL(31) +#define ICE_PGCA_MID GENMASK_ULL(39, 32) +#define ICE_PGCA_ILR BIT_ULL(40) +#define ICE_PGCA_HOP BIT_ULL(41) +#define ICE_PGCA_HOI GENMASK_ULL(50, 42) + +/** + * ice_pg_cam_action_init - parse 55 bits of Parse Graph CAM Action + * @action: pointer to the Parse Graph CAM Action structure + * @data: Parse Graph CAM Action data to be parsed + */ +static void ice_pg_cam_action_init(struct ice_pg_cam_action *action, u64 data) +{ + action->next_node = FIELD_GET(ICE_PGCA_NN, data); + action->next_pc = FIELD_GET(ICE_PGCA_NPC, data); + action->is_pg = FIELD_GET(ICE_PGCA_IPG, data); + action->proto_id = FIELD_GET(ICE_PGCA_PID, data); + action->is_mg = FIELD_GET(ICE_PGCA_IMG, data); + action->marker_id = FIELD_GET(ICE_PGCA_MID, data); + action->is_last_round = FIELD_GET(ICE_PGCA_ILR, data); + action->ho_polarity = FIELD_GET(ICE_PGCA_HOP, data); + action->ho_inc = FIELD_GET(ICE_PGCA_HOI, data); +} + +#define ICE_PGNCK_VLD BIT_ULL(0) +#define ICE_PGNCK_NID GENMASK_ULL(11, 1) +#define ICE_PGNCK_F0 BIT_ULL(12) +#define ICE_PGNCK_F1 BIT_ULL(13) +#define ICE_PGNCK_F2 BIT_ULL(14) +#define ICE_PGNCK_F3 BIT_ULL(15) +#define ICE_PGNCK_BH BIT_ULL(16) +#define ICE_PGNCK_BI GENMASK_ULL(24, 17) +#define ICE_PGNCK_AR GENMASK_ULL(40, 25) + +/** + * ice_pg_nm_cam_key_init - parse 41 bits of Parse Graph NoMatch CAM Key + * @key: pointer to the Parse Graph NoMatch CAM Key structure + * @data: Parse Graph NoMatch CAM Key data to be parsed + */ +static void ice_pg_nm_cam_key_init(struct ice_pg_nm_cam_key *key, u64 data) +{ + key->valid = FIELD_GET(ICE_PGNCK_VLD, data); + key->node_id = FIELD_GET(ICE_PGNCK_NID, data); + key->flag0 = FIELD_GET(ICE_PGNCK_F0, data); + key->flag1 = FIELD_GET(ICE_PGNCK_F1, data); + key->flag2 = FIELD_GET(ICE_PGNCK_F2, data); + key->flag3 = FIELD_GET(ICE_PGNCK_F3, data); + + if (FIELD_GET(ICE_PGNCK_BH, data)) + key->boost_idx = FIELD_GET(ICE_PGNCK_BI, data); + else + key->boost_idx = 0; + + key->alu_reg = FIELD_GET(ICE_PGNCK_AR, data); +} + +#define ICE_PGCK_VLD BIT_ULL(0) +#define ICE_PGCK_NID GENMASK_ULL(11, 1) +#define ICE_PGCK_F0 BIT_ULL(12) +#define ICE_PGCK_F1 BIT_ULL(13) +#define ICE_PGCK_F2 BIT_ULL(14) +#define ICE_PGCK_F3 BIT_ULL(15) +#define ICE_PGCK_BH BIT_ULL(16) +#define ICE_PGCK_BI GENMASK_ULL(24, 17) +#define ICE_PGCK_AR GENMASK_ULL(40, 25) +#define ICE_PGCK_NPK_S 41 /* offset for the 2nd 64-bits field */ +#define ICE_PGCK_NPK_IDD (ICE_PGCK_NPK_S / BITS_PER_BYTE) +#define ICE_PGCK_NPK_OFF (ICE_PGCK_NPK_S % BITS_PER_BYTE) +#define ICE_PGCK_NPK GENMASK_ULL(72 - ICE_PGCK_NPK_S, \ + 41 - ICE_PGCK_NPK_S) + +/** + * ice_pg_cam_key_init - parse 73 bits of Parse Graph CAM Key + * @key: pointer to the Parse Graph CAM Key structure + * @data: Parse Graph CAM Key data to be parsed + */ +static void ice_pg_cam_key_init(struct ice_pg_cam_key *key, u8 *data) +{ + u64 d64 = *(u64 *)data; + + key->valid = FIELD_GET(ICE_PGCK_VLD, d64); + key->node_id = FIELD_GET(ICE_PGCK_NID, d64); + key->flag0 = FIELD_GET(ICE_PGCK_F0, d64); + key->flag1 = FIELD_GET(ICE_PGCK_F1, d64); + key->flag2 = FIELD_GET(ICE_PGCK_F2, d64); + key->flag3 = FIELD_GET(ICE_PGCK_F3, d64); + + if (FIELD_GET(ICE_PGCK_BH, d64)) + key->boost_idx = FIELD_GET(ICE_PGCK_BI, d64); + else + key->boost_idx = 0; + + key->alu_reg = FIELD_GET(ICE_PGCK_AR, d64); + + d64 = *((u64 *)&data[ICE_PGCK_NPK_IDD]) >> ICE_PGCK_NPK_OFF; + + key->next_proto = FIELD_GET(ICE_PGCK_NPK, d64); +} + +#define ICE_PG_CAM_ACT_S 73 +#define ICE_PG_CAM_ACT_IDD (ICE_PG_CAM_ACT_S / BITS_PER_BYTE) +#define ICE_PG_CAM_ACT_OFF (ICE_PG_CAM_ACT_S % BITS_PER_BYTE) + +/** + * ice_pg_cam_parse_item - parse 128 bits of Parse Graph CAM Entry + * @hw: pointer to the hardware structure + * @idx: index of Parse Graph CAM Entry + * @item: item of Parse Graph CAM Entry + * @data: Parse Graph CAM Entry data to be parsed + * @size: size of Parse Graph CAM Entry + */ +static void ice_pg_cam_parse_item(struct ice_hw *hw, u16 idx, void *item, + void *data, int __maybe_unused size) +{ + struct ice_pg_cam_item *ci = item; + u8 *buf = data; + u64 d64; + + ci->idx = idx; + + ice_pg_cam_key_init(&ci->key, buf); + + d64 = *((u64 *)&buf[ICE_PG_CAM_ACT_IDD]) >> ICE_PG_CAM_ACT_OFF; + ice_pg_cam_action_init(&ci->action, d64); +} + +#define ICE_PG_SP_CAM_KEY_S 56 +#define ICE_PG_SP_CAM_KEY_IDD (ICE_PG_SP_CAM_KEY_S / BITS_PER_BYTE) + +/** + * ice_pg_sp_cam_parse_item - parse 136 bits of Parse Graph Spill CAM Entry + * @hw: pointer to the hardware structure + * @idx: index of Parse Graph Spill CAM Entry + * @item: item of Parse Graph Spill CAM Entry + * @data: Parse Graph Spill CAM Entry data to be parsed + * @size: size of Parse Graph Spill CAM Entry + */ +static void ice_pg_sp_cam_parse_item(struct ice_hw *hw, u16 idx, void *item, + void *data, int __maybe_unused size) +{ + struct ice_pg_cam_item *ci = item; + u8 *buf = data; + u64 d64; + + ci->idx = idx; + + d64 = *(u64 *)buf; + ice_pg_cam_action_init(&ci->action, d64); + + ice_pg_cam_key_init(&ci->key, &buf[ICE_PG_SP_CAM_KEY_IDD]); +} + +#define ICE_PG_NM_CAM_ACT_S 41 +#define ICE_PG_NM_CAM_ACT_IDD (ICE_PG_NM_CAM_ACT_S / BITS_PER_BYTE) +#define ICE_PG_NM_CAM_ACT_OFF (ICE_PG_NM_CAM_ACT_S % BITS_PER_BYTE) + +/** + * ice_pg_nm_cam_parse_item - parse 96 bits of Parse Graph NoMatch CAM Entry + * @hw: pointer to the hardware structure + * @idx: index of Parse Graph NoMatch CAM Entry + * @item: item of Parse Graph NoMatch CAM Entry + * @data: Parse Graph NoMatch CAM Entry data to be parsed + * @size: size of Parse Graph NoMatch CAM Entry + */ +static void ice_pg_nm_cam_parse_item(struct ice_hw *hw, u16 idx, void *item, + void *data, int __maybe_unused size) +{ + struct ice_pg_nm_cam_item *ci = item; + u8 *buf = data; + u64 d64; + + ci->idx = idx; + + d64 = *(u64 *)buf; + ice_pg_nm_cam_key_init(&ci->key, d64); + + d64 = *((u64 *)&buf[ICE_PG_NM_CAM_ACT_IDD]) >> ICE_PG_NM_CAM_ACT_OFF; + ice_pg_cam_action_init(&ci->action, d64); +} + +#define ICE_PG_NM_SP_CAM_ACT_S 56 +#define ICE_PG_NM_SP_CAM_ACT_IDD (ICE_PG_NM_SP_CAM_ACT_S / BITS_PER_BYTE) +#define ICE_PG_NM_SP_CAM_ACT_OFF (ICE_PG_NM_SP_CAM_ACT_S % BITS_PER_BYTE) + +/** + * ice_pg_nm_sp_cam_parse_item - parse 104 bits of Parse Graph NoMatch Spill + * CAM Entry + * @hw: pointer to the hardware structure + * @idx: index of Parse Graph NoMatch Spill CAM Entry + * @item: item of Parse Graph NoMatch Spill CAM Entry + * @data: Parse Graph NoMatch Spill CAM Entry data to be parsed + * @size: size of Parse Graph NoMatch Spill CAM Entry + */ +static void ice_pg_nm_sp_cam_parse_item(struct ice_hw *hw, u16 idx, + void *item, void *data, + int __maybe_unused size) +{ + struct ice_pg_nm_cam_item *ci = item; + u8 *buf = data; + u64 d64; + + ci->idx = idx; + + d64 = *(u64 *)buf; + ice_pg_cam_action_init(&ci->action, d64); + + d64 = *((u64 *)&buf[ICE_PG_NM_SP_CAM_ACT_IDD]) >> + ICE_PG_NM_SP_CAM_ACT_OFF; + ice_pg_nm_cam_key_init(&ci->key, d64); +} + +/** + * ice_pg_cam_table_get - create a parse graph cam table + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated Parse Graph CAM table. + */ +static struct ice_pg_cam_item *ice_pg_cam_table_get(struct ice_hw *hw) +{ + return ice_parser_create_table(hw, ICE_SID_RXPARSER_CAM, + sizeof(struct ice_pg_cam_item), + ICE_PG_CAM_TABLE_SIZE, + ice_pg_cam_parse_item, false); +} + +/** + * ice_pg_sp_cam_table_get - create a parse graph spill cam table + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated Parse Graph Spill CAM table. + */ +static struct ice_pg_cam_item *ice_pg_sp_cam_table_get(struct ice_hw *hw) +{ + return ice_parser_create_table(hw, ICE_SID_RXPARSER_PG_SPILL, + sizeof(struct ice_pg_cam_item), + ICE_PG_SP_CAM_TABLE_SIZE, + ice_pg_sp_cam_parse_item, false); +} + +/** + * ice_pg_nm_cam_table_get - create a parse graph no match cam table + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated Parse Graph No Match CAM table. + */ +static struct ice_pg_nm_cam_item *ice_pg_nm_cam_table_get(struct ice_hw *hw) +{ + return ice_parser_create_table(hw, ICE_SID_RXPARSER_NOMATCH_CAM, + sizeof(struct ice_pg_nm_cam_item), + ICE_PG_NM_CAM_TABLE_SIZE, + ice_pg_nm_cam_parse_item, false); +} + +/** + * ice_pg_nm_sp_cam_table_get - create a parse graph no match spill cam table + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated Parse Graph No Match Spill CAM table. + */ +static struct ice_pg_nm_cam_item *ice_pg_nm_sp_cam_table_get(struct ice_hw *hw) +{ + return ice_parser_create_table(hw, ICE_SID_RXPARSER_NOMATCH_SPILL, + sizeof(struct ice_pg_nm_cam_item), + ICE_PG_NM_SP_CAM_TABLE_SIZE, + ice_pg_nm_sp_cam_parse_item, false); +} + +/*** ICE_SID_RXPARSER_BOOST_TCAM and ICE_SID_LBL_RXPARSER_TMEM sections ***/ +#define ICE_BST_ALU_OPC GENMASK_ULL(5, 0) +#define ICE_BST_ALU_SS GENMASK_ULL(13, 6) +#define ICE_BST_ALU_SL GENMASK_ULL(18, 14) +#define ICE_BST_ALU_SXS BIT_ULL(19) +#define ICE_BST_ALU_SXK GENMASK_ULL(23, 20) +#define ICE_BST_ALU_SRID GENMASK_ULL(30, 24) +#define ICE_BST_ALU_DRID GENMASK_ULL(37, 31) +#define ICE_BST_ALU_INC0 BIT_ULL(38) +#define ICE_BST_ALU_INC1 BIT_ULL(39) +#define ICE_BST_ALU_POO GENMASK_ULL(41, 40) +#define ICE_BST_ALU_PO GENMASK_ULL(49, 42) +#define ICE_BST_ALU_BA_S 50 /* offset for the 2nd 64-bits field */ +#define ICE_BST_ALU_BA GENMASK_ULL(57 - ICE_BST_ALU_BA_S, \ + 50 - ICE_BST_ALU_BA_S) +#define ICE_BST_ALU_IMM GENMASK_ULL(73 - ICE_BST_ALU_BA_S, \ + 58 - ICE_BST_ALU_BA_S) +#define ICE_BST_ALU_DFE BIT_ULL(74 - ICE_BST_ALU_BA_S) +#define ICE_BST_ALU_DS GENMASK_ULL(80 - ICE_BST_ALU_BA_S, \ + 75 - ICE_BST_ALU_BA_S) +#define ICE_BST_ALU_DL GENMASK_ULL(86 - ICE_BST_ALU_BA_S, \ + 81 - ICE_BST_ALU_BA_S) +#define ICE_BST_ALU_FEI BIT_ULL(87 - ICE_BST_ALU_BA_S) +#define ICE_BST_ALU_FSI GENMASK_ULL(95 - ICE_BST_ALU_BA_S, \ + 88 - ICE_BST_ALU_BA_S) + +/** + * ice_bst_alu_init - parse 96 bits of ALU entry + * @alu: pointer to the ALU entry structure + * @data: ALU entry data to be parsed + * @off: offset of the ALU entry data + */ +static void ice_bst_alu_init(struct ice_alu *alu, u8 *data, u8 off) +{ + u64 d64; + u8 idd; + + d64 = *((u64 *)data) >> off; + + alu->opc = FIELD_GET(ICE_BST_ALU_OPC, d64); + alu->src_start = FIELD_GET(ICE_BST_ALU_SS, d64); + alu->src_len = FIELD_GET(ICE_BST_ALU_SL, d64); + alu->shift_xlate_sel = FIELD_GET(ICE_BST_ALU_SXS, d64); + alu->shift_xlate_key = FIELD_GET(ICE_BST_ALU_SXK, d64); + alu->src_reg_id = FIELD_GET(ICE_BST_ALU_SRID, d64); + alu->dst_reg_id = FIELD_GET(ICE_BST_ALU_DRID, d64); + alu->inc0 = FIELD_GET(ICE_BST_ALU_INC0, d64); + alu->inc1 = FIELD_GET(ICE_BST_ALU_INC1, d64); + alu->proto_offset_opc = FIELD_GET(ICE_BST_ALU_POO, d64); + alu->proto_offset = FIELD_GET(ICE_BST_ALU_PO, d64); + + idd = (ICE_BST_ALU_BA_S + off) / BITS_PER_BYTE; + off = (ICE_BST_ALU_BA_S + off) % BITS_PER_BYTE; + d64 = *((u64 *)(&data[idd])) >> off; + + alu->branch_addr = FIELD_GET(ICE_BST_ALU_BA, d64); + alu->imm = FIELD_GET(ICE_BST_ALU_IMM, d64); + alu->dedicate_flags_ena = FIELD_GET(ICE_BST_ALU_DFE, d64); + alu->dst_start = FIELD_GET(ICE_BST_ALU_DS, d64); + alu->dst_len = FIELD_GET(ICE_BST_ALU_DL, d64); + alu->flags_extr_imm = FIELD_GET(ICE_BST_ALU_FEI, d64); + alu->flags_start_imm = FIELD_GET(ICE_BST_ALU_FSI, d64); +} + +#define ICE_BST_PGKB_F0_ENA BIT_ULL(0) +#define ICE_BST_PGKB_F0_IDX GENMASK_ULL(6, 1) +#define ICE_BST_PGKB_F1_ENA BIT_ULL(7) +#define ICE_BST_PGKB_F1_IDX GENMASK_ULL(13, 8) +#define ICE_BST_PGKB_F2_ENA BIT_ULL(14) +#define ICE_BST_PGKB_F2_IDX GENMASK_ULL(20, 15) +#define ICE_BST_PGKB_F3_ENA BIT_ULL(21) +#define ICE_BST_PGKB_F3_IDX GENMASK_ULL(27, 22) +#define ICE_BST_PGKB_AR_IDX GENMASK_ULL(34, 28) + +/** + * ice_bst_pgkb_init - parse 35 bits of Parse Graph Key Build + * @kb: pointer to the Parse Graph Key Build structure + * @data: Parse Graph Key Build data to be parsed + */ +static void ice_bst_pgkb_init(struct ice_pg_keybuilder *kb, u64 data) +{ + kb->flag0_ena = FIELD_GET(ICE_BST_PGKB_F0_ENA, data); + kb->flag0_idx = FIELD_GET(ICE_BST_PGKB_F0_IDX, data); + kb->flag1_ena = FIELD_GET(ICE_BST_PGKB_F1_ENA, data); + kb->flag1_idx = FIELD_GET(ICE_BST_PGKB_F1_IDX, data); + kb->flag2_ena = FIELD_GET(ICE_BST_PGKB_F2_ENA, data); + kb->flag2_idx = FIELD_GET(ICE_BST_PGKB_F2_IDX, data); + kb->flag3_ena = FIELD_GET(ICE_BST_PGKB_F3_ENA, data); + kb->flag3_idx = FIELD_GET(ICE_BST_PGKB_F3_IDX, data); + kb->alu_reg_idx = FIELD_GET(ICE_BST_PGKB_AR_IDX, data); +} + +#define ICE_BST_NPKB_OPC GENMASK(1, 0) +#define ICE_BST_NPKB_S_R0 GENMASK(9, 2) +#define ICE_BST_NPKB_L_R1 GENMASK(17, 10) + +/** + * ice_bst_npkb_init - parse 18 bits of Next Protocol Key Build + * @kb: pointer to the Next Protocol Key Build structure + * @data: Next Protocol Key Build data to be parsed + */ +static void ice_bst_npkb_init(struct ice_np_keybuilder *kb, u32 data) +{ + kb->opc = FIELD_GET(ICE_BST_NPKB_OPC, data); + kb->start_reg0 = FIELD_GET(ICE_BST_NPKB_S_R0, data); + kb->len_reg1 = FIELD_GET(ICE_BST_NPKB_L_R1, data); +} + +#define ICE_BT_KEY_S 32 +#define ICE_BT_KEY_IDD (ICE_BT_KEY_S / BITS_PER_BYTE) +#define ICE_BT_KIV_S 192 +#define ICE_BT_KIV_IDD (ICE_BT_KIV_S / BITS_PER_BYTE) +#define ICE_BT_HIG_S 352 +#define ICE_BT_HIG_IDD (ICE_BT_HIG_S / BITS_PER_BYTE) +#define ICE_BT_PGP_S 360 +#define ICE_BT_PGP_IDD (ICE_BT_PGP_S / BITS_PER_BYTE) +#define ICE_BT_PGP_M GENMASK(361 - ICE_BT_PGP_S, 360 - ICE_BT_PGP_S) +#define ICE_BT_NPKB_S 362 +#define ICE_BT_NPKB_IDD (ICE_BT_NPKB_S / BITS_PER_BYTE) +#define ICE_BT_NPKB_OFF (ICE_BT_NPKB_S % BITS_PER_BYTE) +#define ICE_BT_PGKB_S 380 +#define ICE_BT_PGKB_IDD (ICE_BT_PGKB_S / BITS_PER_BYTE) +#define ICE_BT_PGKB_OFF (ICE_BT_PGKB_S % BITS_PER_BYTE) +#define ICE_BT_ALU0_S 415 +#define ICE_BT_ALU0_IDD (ICE_BT_ALU0_S / BITS_PER_BYTE) +#define ICE_BT_ALU0_OFF (ICE_BT_ALU0_S % BITS_PER_BYTE) +#define ICE_BT_ALU1_S 511 +#define ICE_BT_ALU1_IDD (ICE_BT_ALU1_S / BITS_PER_BYTE) +#define ICE_BT_ALU1_OFF (ICE_BT_ALU1_S % BITS_PER_BYTE) +#define ICE_BT_ALU2_S 607 +#define ICE_BT_ALU2_IDD (ICE_BT_ALU2_S / BITS_PER_BYTE) +#define ICE_BT_ALU2_OFF (ICE_BT_ALU2_S % BITS_PER_BYTE) + +/** + * ice_bst_parse_item - parse 704 bits of Boost TCAM entry + * @hw: pointer to the hardware structure + * @idx: index of Boost TCAM entry + * @item: item of Boost TCAM entry + * @data: Boost TCAM entry data to be parsed + * @size: size of Boost TCAM entry + */ +static void ice_bst_parse_item(struct ice_hw *hw, u16 idx, void *item, + void *data, int __maybe_unused size) +{ + struct ice_bst_tcam_item *ti = item; + u8 *buf = (u8 *)data; + int i; + + ti->addr = *(u16 *)buf; + + for (i = 0; i < ICE_BST_TCAM_KEY_SIZE; i++) { + ti->key[i] = buf[ICE_BT_KEY_IDD + i]; + ti->key_inv[i] = buf[ICE_BT_KIV_IDD + i]; + } + ti->hit_idx_grp = buf[ICE_BT_HIG_IDD]; + ti->pg_prio = buf[ICE_BT_PGP_IDD] & ICE_BT_PGP_M; + + ice_bst_npkb_init(&ti->np_kb, + *((u32 *)(&buf[ICE_BT_NPKB_IDD])) >> + ICE_BT_NPKB_OFF); + ice_bst_pgkb_init(&ti->pg_kb, + *((u64 *)(&buf[ICE_BT_PGKB_IDD])) >> + ICE_BT_PGKB_OFF); + + ice_bst_alu_init(&ti->alu0, &buf[ICE_BT_ALU0_IDD], ICE_BT_ALU0_OFF); + ice_bst_alu_init(&ti->alu1, &buf[ICE_BT_ALU1_IDD], ICE_BT_ALU1_OFF); + ice_bst_alu_init(&ti->alu2, &buf[ICE_BT_ALU2_IDD], ICE_BT_ALU2_OFF); +} + +/** + * ice_bst_tcam_table_get - create a boost tcam table + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated Boost TCAM table. + */ +static struct ice_bst_tcam_item *ice_bst_tcam_table_get(struct ice_hw *hw) +{ + return ice_parser_create_table(hw, ICE_SID_RXPARSER_BOOST_TCAM, + sizeof(struct ice_bst_tcam_item), + ICE_BST_TCAM_TABLE_SIZE, + ice_bst_parse_item, true); +} + +static void ice_parse_lbl_item(struct ice_hw *hw, u16 idx, void *item, + void *data, int size) +{ + memcpy(item, data, size); +} + +/** + * ice_bst_lbl_table_get - create a boost label table + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated Boost label table. + */ +static struct ice_lbl_item *ice_bst_lbl_table_get(struct ice_hw *hw) +{ + return ice_parser_create_table(hw, ICE_SID_LBL_RXPARSER_TMEM, + sizeof(struct ice_lbl_item), + ICE_BST_TCAM_TABLE_SIZE, + ice_parse_lbl_item, true); +} + +/*** ICE_SID_RXPARSER_MARKER_PTYPE section ***/ +static void ice_parse_ptype_mk_tcam_item(struct ice_hw *hw, u16 idx, + void *item, void *data, int size) +{ + memcpy(item, data, size); +} + +/** + * ice_ptype_mk_tcam_table_get - create a ptype marker tcam table + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated Marker PType TCAM table. + */ +static +struct ice_ptype_mk_tcam_item *ice_ptype_mk_tcam_table_get(struct ice_hw *hw) +{ + return ice_parser_create_table(hw, ICE_SID_RXPARSER_MARKER_PTYPE, + sizeof(struct ice_ptype_mk_tcam_item), + ICE_PTYPE_MK_TCAM_TABLE_SIZE, + ice_parse_ptype_mk_tcam_item, true); +} + +/*** ICE_SID_RXPARSER_MARKER_GRP section ***/ +static void ice_mk_grp_parse_item(struct ice_hw *hw, u16 idx, void *item, + void *data, int __maybe_unused size) +{ + struct ice_mk_grp_item *grp = item; + u8 *buf = data; + int i; + + grp->idx = idx; + + for (i = 0; i < ICE_MK_COUNT_PER_GRP; i++) + grp->markers[i] = buf[i]; +} + +/** + * ice_mk_grp_table_get - create a marker group table + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated Marker Group ID table. + */ +static struct ice_mk_grp_item *ice_mk_grp_table_get(struct ice_hw *hw) +{ + return ice_parser_create_table(hw, ICE_SID_RXPARSER_MARKER_GRP, + sizeof(struct ice_mk_grp_item), + ICE_MK_GRP_TABLE_SIZE, + ice_mk_grp_parse_item, false); +} + +/*** ICE_SID_RXPARSER_PROTO_GRP section ***/ +#define ICE_PO_POL BIT(0) +#define ICE_PO_PID GENMASK(8, 1) +#define ICE_PO_OFF GENMASK(21, 12) + +/** + * ice_proto_off_parse - parse 22 bits of Protocol entry + * @po: pointer to the Protocol entry structure + * @data: Protocol entry data to be parsed + */ +static void ice_proto_off_parse(struct ice_proto_off *po, u32 data) +{ + po->polarity = FIELD_GET(ICE_PO_POL, data); + po->proto_id = FIELD_GET(ICE_PO_PID, data); + po->offset = FIELD_GET(ICE_PO_OFF, data); +} + +/** + * ice_proto_grp_parse_item - parse 192 bits of Protocol Group Table entry + * @hw: pointer to the hardware structure + * @idx: index of Protocol Group Table entry + * @item: item of Protocol Group Table entry + * @data: Protocol Group Table entry data to be parsed + * @size: size of Protocol Group Table entry + */ +static void ice_proto_grp_parse_item(struct ice_hw *hw, u16 idx, void *item, + void *data, int __maybe_unused size) +{ + struct ice_proto_grp_item *grp = item; + u8 *buf = (u8 *)data; + u8 idd, off; + u32 d32; + int i; + + grp->idx = idx; + + for (i = 0; i < ICE_PROTO_COUNT_PER_GRP; i++) { + idd = (ICE_PROTO_GRP_ITEM_SIZE * i) / BITS_PER_BYTE; + off = (ICE_PROTO_GRP_ITEM_SIZE * i) % BITS_PER_BYTE; + d32 = *((u32 *)&buf[idd]) >> off; + ice_proto_off_parse(&grp->po[i], d32); + } +} + +/** + * ice_proto_grp_table_get - create a proto group table + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated Protocol Group table. + */ +static struct ice_proto_grp_item *ice_proto_grp_table_get(struct ice_hw *hw) +{ + return ice_parser_create_table(hw, ICE_SID_RXPARSER_PROTO_GRP, + sizeof(struct ice_proto_grp_item), + ICE_PROTO_GRP_TABLE_SIZE, + ice_proto_grp_parse_item, false); +} + +/*** ICE_SID_RXPARSER_FLAG_REDIR section ***/ +#define ICE_FRT_EXPO BIT(0) +#define ICE_FRT_IFID GENMASK(6, 1) + +/** + * ice_flg_rd_parse_item - parse 8 bits of Flag Redirect Table entry + * @hw: pointer to the hardware structure + * @idx: index of Flag Redirect Table entry + * @item: item of Flag Redirect Table entry + * @data: Flag Redirect Table entry data to be parsed + * @size: size of Flag Redirect Table entry + */ +static void ice_flg_rd_parse_item(struct ice_hw *hw, u16 idx, void *item, + void *data, int __maybe_unused size) +{ + struct ice_flg_rd_item *rdi = item; + u8 d8 = *(u8 *)data; + + rdi->idx = idx; + rdi->expose = FIELD_GET(ICE_FRT_EXPO, d8); + rdi->intr_flg_id = FIELD_GET(ICE_FRT_IFID, d8); +} + +/** + * ice_flg_rd_table_get - create a flag redirect table + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated Flags Redirection table. + */ +static struct ice_flg_rd_item *ice_flg_rd_table_get(struct ice_hw *hw) +{ + return ice_parser_create_table(hw, ICE_SID_RXPARSER_FLAG_REDIR, + sizeof(struct ice_flg_rd_item), + ICE_FLG_RD_TABLE_SIZE, + ice_flg_rd_parse_item, false); +} + +/*** ICE_SID_XLT_KEY_BUILDER_SW, ICE_SID_XLT_KEY_BUILDER_ACL, + * ICE_SID_XLT_KEY_BUILDER_FD and ICE_SID_XLT_KEY_BUILDER_RSS + * sections ***/ +#define ICE_XLT_KB_X1AS_S 32 /* offset for the 1st 64-bits field */ +#define ICE_XLT_KB_X1AS_IDD (ICE_XLT_KB_X1AS_S / BITS_PER_BYTE) +#define ICE_XLT_KB_X1AS_OFF (ICE_XLT_KB_X1AS_S % BITS_PER_BYTE) +#define ICE_XLT_KB_X1AS GENMASK_ULL(34 - ICE_XLT_KB_X1AS_S, \ + 32 - ICE_XLT_KB_X1AS_S) +#define ICE_XLT_KB_X2AS GENMASK_ULL(37 - ICE_XLT_KB_X1AS_S, \ + 35 - ICE_XLT_KB_X1AS_S) +#define ICE_XLT_KB_FL00 GENMASK_ULL(46 - ICE_XLT_KB_X1AS_S, \ + 38 - ICE_XLT_KB_X1AS_S) +#define ICE_XLT_KB_FL01 GENMASK_ULL(55 - ICE_XLT_KB_X1AS_S, \ + 47 - ICE_XLT_KB_X1AS_S) +#define ICE_XLT_KB_FL02 GENMASK_ULL(64 - ICE_XLT_KB_X1AS_S, \ + 56 - ICE_XLT_KB_X1AS_S) +#define ICE_XLT_KB_FL03 GENMASK_ULL(73 - ICE_XLT_KB_X1AS_S, \ + 65 - ICE_XLT_KB_X1AS_S) +#define ICE_XLT_KB_FL04 GENMASK_ULL(82 - ICE_XLT_KB_X1AS_S, \ + 74 - ICE_XLT_KB_X1AS_S) +#define ICE_XLT_KB_FL05 GENMASK_ULL(91 - ICE_XLT_KB_X1AS_S, \ + 83 - ICE_XLT_KB_X1AS_S) +#define ICE_XLT_KB_FL06_S 92 /* offset for the 2nd 64-bits field */ +#define ICE_XLT_KB_FL06_IDD (ICE_XLT_KB_FL06_S / BITS_PER_BYTE) +#define ICE_XLT_KB_FL06_OFF (ICE_XLT_KB_FL06_S % BITS_PER_BYTE) +#define ICE_XLT_KB_FL06 GENMASK_ULL(100 - ICE_XLT_KB_FL06_S, \ + 92 - ICE_XLT_KB_FL06_S) +#define ICE_XLT_KB_FL07 GENMASK_ULL(109 - ICE_XLT_KB_FL06_S, \ + 101 - ICE_XLT_KB_FL06_S) +#define ICE_XLT_KB_FL08 GENMASK_ULL(118 - ICE_XLT_KB_FL06_S, \ + 110 - ICE_XLT_KB_FL06_S) +#define ICE_XLT_KB_FL09 GENMASK_ULL(127 - ICE_XLT_KB_FL06_S, \ + 119 - ICE_XLT_KB_FL06_S) +#define ICE_XLT_KB_FL10 GENMASK_ULL(136 - ICE_XLT_KB_FL06_S, \ + 128 - ICE_XLT_KB_FL06_S) +#define ICE_XLT_KB_FL11 GENMASK_ULL(145 - ICE_XLT_KB_FL06_S, \ + 137 - ICE_XLT_KB_FL06_S) +#define ICE_XLT_KB_FL12_S 146 /* offset for the 3rd 64-bits field */ +#define ICE_XLT_KB_FL12_IDD (ICE_XLT_KB_FL12_S / BITS_PER_BYTE) +#define ICE_XLT_KB_FL12_OFF (ICE_XLT_KB_FL12_S % BITS_PER_BYTE) +#define ICE_XLT_KB_FL12 GENMASK_ULL(154 - ICE_XLT_KB_FL12_S, \ + 146 - ICE_XLT_KB_FL12_S) +#define ICE_XLT_KB_FL13 GENMASK_ULL(163 - ICE_XLT_KB_FL12_S, \ + 155 - ICE_XLT_KB_FL12_S) +#define ICE_XLT_KB_FL14 GENMASK_ULL(181 - ICE_XLT_KB_FL12_S, \ + 164 - ICE_XLT_KB_FL12_S) +#define ICE_XLT_KB_X1MS GENMASK_ULL(186 - ICE_XLT_KB_FL12_S, \ + 182 - ICE_XLT_KB_FL12_S) +#define ICE_XLT_KB_X2MS GENMASK_ULL(191 - ICE_XLT_KB_FL12_S, \ + 187 - ICE_XLT_KB_FL12_S) + +/** + * ice_kb_entry_init - parse 192 bits of XLT Key Builder entry + * @entry: pointer to the XLT Key Builder entry structure + * @data: XLT Key Builder entry data to be parsed + */ +static void ice_kb_entry_init(struct ice_xlt_kb_entry *entry, u8 *data) +{ + u8 i = 0; + u64 d64; + + d64 = *((u64 *)&data[ICE_XLT_KB_X1AS_IDD]) >> ICE_XLT_KB_X1AS_OFF; + + entry->xlt1_ad_sel = FIELD_GET(ICE_XLT_KB_X1AS, d64); + entry->xlt2_ad_sel = FIELD_GET(ICE_XLT_KB_X2AS, d64); + + entry->flg0_14_sel[i++] = FIELD_GET(ICE_XLT_KB_FL00, d64); + entry->flg0_14_sel[i++] = FIELD_GET(ICE_XLT_KB_FL01, d64); + entry->flg0_14_sel[i++] = FIELD_GET(ICE_XLT_KB_FL02, d64); + entry->flg0_14_sel[i++] = FIELD_GET(ICE_XLT_KB_FL03, d64); + entry->flg0_14_sel[i++] = FIELD_GET(ICE_XLT_KB_FL04, d64); + entry->flg0_14_sel[i++] = FIELD_GET(ICE_XLT_KB_FL05, d64); + + d64 = *((u64 *)&data[ICE_XLT_KB_FL06_IDD]) >> ICE_XLT_KB_FL06_OFF; + + entry->flg0_14_sel[i++] = FIELD_GET(ICE_XLT_KB_FL06, d64); + entry->flg0_14_sel[i++] = FIELD_GET(ICE_XLT_KB_FL07, d64); + entry->flg0_14_sel[i++] = FIELD_GET(ICE_XLT_KB_FL08, d64); + entry->flg0_14_sel[i++] = FIELD_GET(ICE_XLT_KB_FL09, d64); + entry->flg0_14_sel[i++] = FIELD_GET(ICE_XLT_KB_FL10, d64); + entry->flg0_14_sel[i++] = FIELD_GET(ICE_XLT_KB_FL11, d64); + + d64 = *((u64 *)&data[ICE_XLT_KB_FL12_IDD]) >> ICE_XLT_KB_FL12_OFF; + + entry->flg0_14_sel[i++] = FIELD_GET(ICE_XLT_KB_FL12, d64); + entry->flg0_14_sel[i++] = FIELD_GET(ICE_XLT_KB_FL13, d64); + entry->flg0_14_sel[i] = FIELD_GET(ICE_XLT_KB_FL14, d64); + + entry->xlt1_md_sel = FIELD_GET(ICE_XLT_KB_X1MS, d64); + entry->xlt2_md_sel = FIELD_GET(ICE_XLT_KB_X2MS, d64); +} + +#define ICE_XLT_KB_X1PM_OFF 0 +#define ICE_XLT_KB_X2PM_OFF 1 +#define ICE_XLT_KB_PIPM_OFF 2 +#define ICE_XLT_KB_FL15_OFF 4 +#define ICE_XLT_KB_TBL_OFF 12 + +/** + * ice_parse_kb_data - parse 204 bits of XLT Key Build Table + * @hw: pointer to the hardware structure + * @kb: pointer to the XLT Key Build Table structure + * @data: XLT Key Build Table data to be parsed + */ +static void ice_parse_kb_data(struct ice_hw *hw, struct ice_xlt_kb *kb, + void *data) +{ + u8 *buf = data; + int i; + + kb->xlt1_pm = buf[ICE_XLT_KB_X1PM_OFF]; + kb->xlt2_pm = buf[ICE_XLT_KB_X2PM_OFF]; + kb->prof_id_pm = buf[ICE_XLT_KB_PIPM_OFF]; + + kb->flag15 = *(u64 *)&buf[ICE_XLT_KB_FL15_OFF]; + for (i = 0; i < ICE_XLT_KB_TBL_CNT; i++) + ice_kb_entry_init(&kb->entries[i], + &buf[ICE_XLT_KB_TBL_OFF + + i * ICE_XLT_KB_TBL_ENTRY_SIZE]); +} + +static struct ice_xlt_kb *ice_xlt_kb_get(struct ice_hw *hw, u32 sect_type) +{ + struct ice_pkg_enum state = {}; + struct ice_seg *seg = hw->seg; + struct ice_xlt_kb *kb; + void *data; + + if (!seg) + return ERR_PTR(-EINVAL); + + kb = kzalloc(sizeof(*kb), GFP_KERNEL); + if (!kb) + return ERR_PTR(-ENOMEM); + + data = ice_pkg_enum_section(seg, &state, sect_type); + if (!data) { + ice_debug(hw, ICE_DBG_PARSER, "failed to find section type %d.\n", + sect_type); + kfree(kb); + return ERR_PTR(-EINVAL); + } + + ice_parse_kb_data(hw, kb, data); + + return kb; +} + +/** + * ice_xlt_kb_get_sw - create switch xlt key build + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated Key Builder table for Switch. + */ +static struct ice_xlt_kb *ice_xlt_kb_get_sw(struct ice_hw *hw) +{ + return ice_xlt_kb_get(hw, ICE_SID_XLT_KEY_BUILDER_SW); +} + +/** + * ice_xlt_kb_get_acl - create acl xlt key build + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated Key Builder table for ACL. + */ +static struct ice_xlt_kb *ice_xlt_kb_get_acl(struct ice_hw *hw) +{ + return ice_xlt_kb_get(hw, ICE_SID_XLT_KEY_BUILDER_ACL); +} + +/** + * ice_xlt_kb_get_fd - create fdir xlt key build + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated Key Builder table for Flow Director. + */ +static struct ice_xlt_kb *ice_xlt_kb_get_fd(struct ice_hw *hw) +{ + return ice_xlt_kb_get(hw, ICE_SID_XLT_KEY_BUILDER_FD); +} + +/** + * ice_xlt_kb_get_rss - create rss xlt key build + * @hw: pointer to the hardware structure + * + * Return: a pointer to the allocated Key Builder table for RSS. + */ +static struct ice_xlt_kb *ice_xlt_kb_get_rss(struct ice_hw *hw) +{ + return ice_xlt_kb_get(hw, ICE_SID_XLT_KEY_BUILDER_RSS); +} + +/*** Parser API ***/ /** * ice_parser_create - create a parser instance * @hw: pointer to the hardware structure @@ -13,13 +1278,114 @@ struct ice_parser *ice_parser_create(struct ice_hw *hw) { struct ice_parser *p; + void *err; p = kzalloc(sizeof(*p), GFP_KERNEL); if (!p) return ERR_PTR(-ENOMEM); p->hw = hw; + + p->imem_table = ice_imem_table_get(hw); + if (IS_ERR(p->imem_table)) { + err = p->imem_table; + goto err; + } + + p->mi_table = ice_metainit_table_get(hw); + if (IS_ERR(p->mi_table)) { + err = p->mi_table; + goto err; + } + + p->pg_cam_table = ice_pg_cam_table_get(hw); + if (IS_ERR(p->pg_cam_table)) { + err = p->pg_cam_table; + goto err; + } + + p->pg_sp_cam_table = ice_pg_sp_cam_table_get(hw); + if (IS_ERR(p->pg_sp_cam_table)) { + err = p->pg_sp_cam_table; + goto err; + } + + p->pg_nm_cam_table = ice_pg_nm_cam_table_get(hw); + if (IS_ERR(p->pg_nm_cam_table)) { + err = p->pg_nm_cam_table; + goto err; + } + + p->pg_nm_sp_cam_table = ice_pg_nm_sp_cam_table_get(hw); + if (IS_ERR(p->pg_nm_sp_cam_table)) { + err = p->pg_nm_sp_cam_table; + goto err; + } + + p->bst_tcam_table = ice_bst_tcam_table_get(hw); + if (IS_ERR(p->bst_tcam_table)) { + err = p->bst_tcam_table; + goto err; + } + + p->bst_lbl_table = ice_bst_lbl_table_get(hw); + if (IS_ERR(p->bst_lbl_table)) { + err = p->bst_lbl_table; + goto err; + } + + p->ptype_mk_tcam_table = ice_ptype_mk_tcam_table_get(hw); + if (IS_ERR(p->ptype_mk_tcam_table)) { + err = p->ptype_mk_tcam_table; + goto err; + } + + p->mk_grp_table = ice_mk_grp_table_get(hw); + if (IS_ERR(p->mk_grp_table)) { + err = p->mk_grp_table; + goto err; + } + + p->proto_grp_table = ice_proto_grp_table_get(hw); + if (IS_ERR(p->proto_grp_table)) { + err = p->proto_grp_table; + goto err; + } + + p->flg_rd_table = ice_flg_rd_table_get(hw); + if (IS_ERR(p->flg_rd_table)) { + err = p->flg_rd_table; + goto err; + } + + p->xlt_kb_sw = ice_xlt_kb_get_sw(hw); + if (IS_ERR(p->xlt_kb_sw)) { + err = p->xlt_kb_sw; + goto err; + } + + p->xlt_kb_acl = ice_xlt_kb_get_acl(hw); + if (IS_ERR(p->xlt_kb_acl)) { + err = p->xlt_kb_acl; + goto err; + } + + p->xlt_kb_fd = ice_xlt_kb_get_fd(hw); + if (IS_ERR(p->xlt_kb_fd)) { + err = p->xlt_kb_fd; + goto err; + } + + p->xlt_kb_rss = ice_xlt_kb_get_rss(hw); + if (IS_ERR(p->xlt_kb_rss)) { + err = p->xlt_kb_rss; + goto err; + } + return p; +err: + ice_parser_destroy(p); + return err; } /** @@ -28,5 +1394,22 @@ struct ice_parser *ice_parser_create(struct ice_hw *hw) */ void ice_parser_destroy(struct ice_parser *psr) { + kfree(psr->imem_table); + kfree(psr->mi_table); + kfree(psr->pg_cam_table); + kfree(psr->pg_sp_cam_table); + kfree(psr->pg_nm_cam_table); + kfree(psr->pg_nm_sp_cam_table); + kfree(psr->bst_tcam_table); + kfree(psr->bst_lbl_table); + kfree(psr->ptype_mk_tcam_table); + kfree(psr->mk_grp_table); + kfree(psr->proto_grp_table); + kfree(psr->flg_rd_table); + kfree(psr->xlt_kb_sw); + kfree(psr->xlt_kb_acl); + kfree(psr->xlt_kb_fd); + kfree(psr->xlt_kb_rss); + kfree(psr); } diff --git a/drivers/net/ethernet/intel/ice/ice_parser.h b/drivers/net/ethernet/intel/ice/ice_parser.h index 09ed380eee32..6e8054d59644 100644 --- a/drivers/net/ethernet/intel/ice/ice_parser.h +++ b/drivers/net/ethernet/intel/ice/ice_parser.h @@ -4,8 +4,366 @@ #ifndef _ICE_PARSER_H_ #define _ICE_PARSER_H_ +#define ICE_SEC_DATA_OFFSET 4 +#define ICE_SID_RXPARSER_IMEM_ENTRY_SIZE 48 +#define ICE_SID_RXPARSER_METADATA_INIT_ENTRY_SIZE 24 +#define ICE_SID_RXPARSER_CAM_ENTRY_SIZE 16 +#define ICE_SID_RXPARSER_PG_SPILL_ENTRY_SIZE 17 +#define ICE_SID_RXPARSER_NOMATCH_CAM_ENTRY_SIZE 12 +#define ICE_SID_RXPARSER_NOMATCH_SPILL_ENTRY_SIZE 13 +#define ICE_SID_RXPARSER_BOOST_TCAM_ENTRY_SIZE 88 +#define ICE_SID_RXPARSER_MARKER_TYPE_ENTRY_SIZE 24 +#define ICE_SID_RXPARSER_MARKER_GRP_ENTRY_SIZE 8 +#define ICE_SID_RXPARSER_PROTO_GRP_ENTRY_SIZE 24 +#define ICE_SID_RXPARSER_FLAG_REDIR_ENTRY_SIZE 1 + +#define ICE_SEC_LBL_DATA_OFFSET 2 +#define ICE_SID_LBL_ENTRY_SIZE 66 + +/*** ICE_SID_RXPARSER_IMEM section ***/ +#define ICE_IMEM_TABLE_SIZE 192 + +/* TCAM boost Master; if bit is set, and TCAM hit, TCAM output overrides iMEM + * output. + */ +struct ice_bst_main { + bool alu0; + bool alu1; + bool alu2; + bool pg; +}; + +struct ice_bst_keybuilder { + u8 prio; /* 0-3: PG precedence within ALUs (3 highest) */ + bool tsr_ctrl; /* TCAM Search Register control */ +}; + +/* Next protocol Key builder */ +struct ice_np_keybuilder { + u8 opc; + u8 start_reg0; + u8 len_reg1; +}; + +enum ice_np_keybuilder_opcode { + ICE_NPKB_OPC_EXTRACT = 0, + ICE_NPKB_OPC_BUILD = 1, + ICE_NPKB_OPC_BYPASS = 2, +}; + +/* Parse Graph Key builder */ +struct ice_pg_keybuilder { + bool flag0_ena; + bool flag1_ena; + bool flag2_ena; + bool flag3_ena; + u8 flag0_idx; + u8 flag1_idx; + u8 flag2_idx; + u8 flag3_idx; + u8 alu_reg_idx; +}; + +enum ice_alu_idx { + ICE_ALU0_IDX = 0, + ICE_ALU1_IDX = 1, + ICE_ALU2_IDX = 2, +}; + +enum ice_alu_opcode { + ICE_ALU_PARK = 0, + ICE_ALU_MOV_ADD = 1, + ICE_ALU_ADD = 2, + ICE_ALU_MOV_AND = 4, + ICE_ALU_AND = 5, + ICE_ALU_AND_IMM = 6, + ICE_ALU_MOV_OR = 7, + ICE_ALU_OR = 8, + ICE_ALU_MOV_XOR = 9, + ICE_ALU_XOR = 10, + ICE_ALU_NOP = 11, + ICE_ALU_BR = 12, + ICE_ALU_BREQ = 13, + ICE_ALU_BRNEQ = 14, + ICE_ALU_BRGT = 15, + ICE_ALU_BRLT = 16, + ICE_ALU_BRGEQ = 17, + ICE_ALU_BRLEG = 18, + ICE_ALU_SETEQ = 19, + ICE_ALU_ANDEQ = 20, + ICE_ALU_OREQ = 21, + ICE_ALU_SETNEQ = 22, + ICE_ALU_ANDNEQ = 23, + ICE_ALU_ORNEQ = 24, + ICE_ALU_SETGT = 25, + ICE_ALU_ANDGT = 26, + ICE_ALU_ORGT = 27, + ICE_ALU_SETLT = 28, + ICE_ALU_ANDLT = 29, + ICE_ALU_ORLT = 30, + ICE_ALU_MOV_SUB = 31, + ICE_ALU_SUB = 32, + ICE_ALU_INVALID = 64, +}; + +enum ice_proto_off_opcode { + ICE_PO_OFF_REMAIN = 0, + ICE_PO_OFF_HDR_ADD = 1, + ICE_PO_OFF_HDR_SUB = 2, +}; + +struct ice_alu { + enum ice_alu_opcode opc; + u8 src_start; + u8 src_len; + bool shift_xlate_sel; + u8 shift_xlate_key; + u8 src_reg_id; + u8 dst_reg_id; + bool inc0; + bool inc1; + u8 proto_offset_opc; + u8 proto_offset; + u8 branch_addr; + u16 imm; + bool dedicate_flags_ena; + u8 dst_start; + u8 dst_len; + bool flags_extr_imm; + u8 flags_start_imm; +}; + +/* Parser program code (iMEM) */ +struct ice_imem_item { + u16 idx; + struct ice_bst_main b_m; + struct ice_bst_keybuilder b_kb; + u8 pg_prio; + struct ice_np_keybuilder np_kb; + struct ice_pg_keybuilder pg_kb; + struct ice_alu alu0; + struct ice_alu alu1; + struct ice_alu alu2; +}; + +/*** ICE_SID_RXPARSER_METADATA_INIT section ***/ +#define ICE_METAINIT_TABLE_SIZE 16 + +/* Metadata Initialization item */ +struct ice_metainit_item { + u16 idx; + + u8 tsr; /* TCAM Search key Register */ + u16 ho; /* Header Offset register */ + u16 pc; /* Program Counter register */ + u16 pg_rn; /* Parse Graph Root Node */ + u8 cd; /* Control Domain ID */ + + /* General Purpose Registers */ + bool gpr_a_ctrl; + u8 gpr_a_data_mdid; + u8 gpr_a_data_start; + u8 gpr_a_data_len; + u8 gpr_a_id; + + bool gpr_b_ctrl; + u8 gpr_b_data_mdid; + u8 gpr_b_data_start; + u8 gpr_b_data_len; + u8 gpr_b_id; + + bool gpr_c_ctrl; + u8 gpr_c_data_mdid; + u8 gpr_c_data_start; + u8 gpr_c_data_len; + u8 gpr_c_id; + + bool gpr_d_ctrl; + u8 gpr_d_data_mdid; + u8 gpr_d_data_start; + u8 gpr_d_data_len; + u8 gpr_d_id; + + u64 flags; /* Initial value for all flags */ +}; + +/*** ICE_SID_RXPARSER_CAM, ICE_SID_RXPARSER_PG_SPILL, + * ICE_SID_RXPARSER_NOMATCH_CAM and ICE_SID_RXPARSER_NOMATCH_CAM + * sections ***/ +#define ICE_PG_CAM_TABLE_SIZE 2048 +#define ICE_PG_SP_CAM_TABLE_SIZE 128 +#define ICE_PG_NM_CAM_TABLE_SIZE 1024 +#define ICE_PG_NM_SP_CAM_TABLE_SIZE 64 + +struct ice_pg_cam_key { + bool valid; + u16 node_id; /* Node ID of protocol in parse graph */ + bool flag0; + bool flag1; + bool flag2; + bool flag3; + u8 boost_idx; /* Boost TCAM match index */ + u16 alu_reg; + u32 next_proto; /* next Protocol value */ +}; + +struct ice_pg_nm_cam_key { + bool valid; + u16 node_id; + bool flag0; + bool flag1; + bool flag2; + bool flag3; + u8 boost_idx; + u16 alu_reg; +}; + +struct ice_pg_cam_action { + u16 next_node; /* Parser Node ID for the next round */ + u8 next_pc; /* next Program Counter */ + bool is_pg; /* is protocol group */ + u8 proto_id; /* protocol ID or proto group ID */ + bool is_mg; /* is marker group */ + u8 marker_id; /* marker ID or marker group ID */ + bool is_last_round; + bool ho_polarity; /* header offset polarity */ + u16 ho_inc; +}; + +/* Parse Graph item */ +struct ice_pg_cam_item { + u16 idx; + struct ice_pg_cam_key key; + struct ice_pg_cam_action action; +}; + +/* Parse Graph No Match item */ +struct ice_pg_nm_cam_item { + u16 idx; + struct ice_pg_nm_cam_key key; + struct ice_pg_cam_action action; +}; + +/*** ICE_SID_RXPARSER_BOOST_TCAM and ICE_SID_LBL_RXPARSER_TMEM sections ***/ +#define ICE_BST_TCAM_TABLE_SIZE 256 +#define ICE_BST_TCAM_KEY_SIZE 20 +#define ICE_BST_KEY_TCAM_SIZE 19 + +/* Boost TCAM item */ +struct ice_bst_tcam_item { + u16 addr; + u8 key[ICE_BST_TCAM_KEY_SIZE]; + u8 key_inv[ICE_BST_TCAM_KEY_SIZE]; + u8 hit_idx_grp; + u8 pg_prio; + struct ice_np_keybuilder np_kb; + struct ice_pg_keybuilder pg_kb; + struct ice_alu alu0; + struct ice_alu alu1; + struct ice_alu alu2; +}; + +#define ICE_LBL_LEN 64 +struct ice_lbl_item { + u16 idx; + char label[ICE_LBL_LEN]; +}; + +/*** ICE_SID_RXPARSER_MARKER_PTYPE section ***/ +#define ICE_PTYPE_MK_TCAM_TABLE_SIZE 1024 +#define ICE_PTYPE_MK_TCAM_KEY_SIZE 10 + +struct ice_ptype_mk_tcam_item { + u16 address; + u16 ptype; + u8 key[ICE_PTYPE_MK_TCAM_KEY_SIZE]; + u8 key_inv[ICE_PTYPE_MK_TCAM_KEY_SIZE]; +} __packed; + +/*** ICE_SID_RXPARSER_MARKER_GRP section ***/ +#define ICE_MK_GRP_TABLE_SIZE 128 +#define ICE_MK_COUNT_PER_GRP 8 + +/* Marker Group item */ +struct ice_mk_grp_item { + int idx; + u8 markers[ICE_MK_COUNT_PER_GRP]; +}; + +/*** ICE_SID_RXPARSER_PROTO_GRP section ***/ +#define ICE_PROTO_COUNT_PER_GRP 8 +#define ICE_PROTO_GRP_TABLE_SIZE 192 +#define ICE_PROTO_GRP_ITEM_SIZE 22 +struct ice_proto_off { + bool polarity; /* true: positive, false: negative */ + u8 proto_id; + u16 offset; /* 10 bit protocol offset */ +}; + +/* Protocol Group item */ +struct ice_proto_grp_item { + u16 idx; + struct ice_proto_off po[ICE_PROTO_COUNT_PER_GRP]; +}; + +/*** ICE_SID_RXPARSER_FLAG_REDIR section ***/ +#define ICE_FLG_RD_TABLE_SIZE 64 + +/* Flags Redirection item */ +struct ice_flg_rd_item { + u16 idx; + bool expose; + u8 intr_flg_id; /* Internal Flag ID */ +}; + +/*** ICE_SID_XLT_KEY_BUILDER_SW, ICE_SID_XLT_KEY_BUILDER_ACL, + * ICE_SID_XLT_KEY_BUILDER_FD and ICE_SID_XLT_KEY_BUILDER_RSS + * sections ***/ +#define ICE_XLT_KB_FLAG0_14_CNT 15 +#define ICE_XLT_KB_TBL_CNT 8 +#define ICE_XLT_KB_TBL_ENTRY_SIZE 24 + +struct ice_xlt_kb_entry { + u8 xlt1_ad_sel; + u8 xlt2_ad_sel; + u16 flg0_14_sel[ICE_XLT_KB_FLAG0_14_CNT]; + u8 xlt1_md_sel; + u8 xlt2_md_sel; +}; + +/* XLT Key Builder */ +struct ice_xlt_kb { + u8 xlt1_pm; /* XLT1 Partition Mode */ + u8 xlt2_pm; /* XLT2 Partition Mode */ + u8 prof_id_pm; /* Profile ID Partition Mode */ + u64 flag15; + + struct ice_xlt_kb_entry entries[ICE_XLT_KB_TBL_CNT]; +}; + +/*** Parser API ***/ struct ice_parser { struct ice_hw *hw; /* pointer to the hardware structure */ + + struct ice_imem_item *imem_table; + struct ice_metainit_item *mi_table; + + struct ice_pg_cam_item *pg_cam_table; + struct ice_pg_cam_item *pg_sp_cam_table; + struct ice_pg_nm_cam_item *pg_nm_cam_table; + struct ice_pg_nm_cam_item *pg_nm_sp_cam_table; + + struct ice_bst_tcam_item *bst_tcam_table; + struct ice_lbl_item *bst_lbl_table; + struct ice_ptype_mk_tcam_item *ptype_mk_tcam_table; + struct ice_mk_grp_item *mk_grp_table; + struct ice_proto_grp_item *proto_grp_table; + struct ice_flg_rd_item *flg_rd_table; + + struct ice_xlt_kb *xlt_kb_sw; + struct ice_xlt_kb *xlt_kb_acl; + struct ice_xlt_kb *xlt_kb_fd; + struct ice_xlt_kb *xlt_kb_rss; }; struct ice_parser *ice_parser_create(struct ice_hw *hw); diff --git a/drivers/net/ethernet/intel/ice/ice_type.h b/drivers/net/ethernet/intel/ice/ice_type.h index 96037bef3e78..b9e443232335 100644 --- a/drivers/net/ethernet/intel/ice/ice_type.h +++ b/drivers/net/ethernet/intel/ice/ice_type.h @@ -61,6 +61,7 @@ static inline u32 ice_round_to_num(u32 N, u32 R) ICE_DBG_AQ_DESC | \ ICE_DBG_AQ_DESC_BUF | \ ICE_DBG_AQ_CMD) +#define ICE_DBG_PARSER BIT_ULL(28) #define ICE_DBG_USER BIT_ULL(31) From patchwork Tue Aug 13 22:22:38 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Tony Nguyen X-Patchwork-Id: 13762623 X-Patchwork-Delegate: kuba@kernel.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.8]) (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 F30941AAE2B for ; Tue, 13 Aug 2024 22:22:56 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=192.198.163.8 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587779; cv=none; b=QMkE+Ket2o+oF0nP+SSGwAXml2Movwb/i0yCqYwUDz62eIAUL78Egv1W6BzHV9bze6f386WuVtLC9qLE1iElI1QtPxMspW2JCb+04FqvtGUoxRYtuX7O7uIvEwNUDACrbqnntggAFU55fihsygL9RGRjG/CAbLL84ckpRFRDJ0c= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587779; c=relaxed/simple; bh=hLiJO8NAHo4R4rDsWf/y5mNAPjHc2XMf0n6HyxnOBU4=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=NIjxdA97xjcUnOQ3uiLH35T14uOxO43Sh4VGMLtlA60T1n/NNe15meQtO0sCIIs+vMlp65TXLIh9wC3tpdm+u7qsBtlObWJ/H8vdExvkyPR08/E8WerkuwXJNlHrlw1EjEesKqtlOI4OrG2EZm/Je/u/EOwUWkF7U4IwAORCkvQ= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com; spf=pass smtp.mailfrom=intel.com; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b=O0vo2sd3; arc=none smtp.client-ip=192.198.163.8 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=intel.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b="O0vo2sd3" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1723587777; x=1755123777; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=hLiJO8NAHo4R4rDsWf/y5mNAPjHc2XMf0n6HyxnOBU4=; b=O0vo2sd3jGHWLqJTbRAcW9lMOU5yKfKijiy8d+9DmHaNQvWDcWNoVx4z WBeQHVbjvOJVmAUQId5N3+1eYyE2sOhOsH+vpUL3i3HKpGW+sXBOrBM7m dIim4Pzb6AZdusSmD2PVuh3pwAsYqDxKUldvEbEC50pfh0snoxm7KTEc5 9fv/3+D3w8oDEml/YgX8bkFVy6gp1r9vVClHUp6vQ82PFKg2yQt2Q+bh9 ze93grrtTzkED+lxWy5y+H3OtvgMPKMgutgUypdJVefMS4dk5cWMfloxY G4MeH6aNs9HGkXLh3JTGz8DC/mJaKhxhTJGZWCXuT1z68DVrS/Z1L6+i/ w==; X-CSE-ConnectionGUID: UpSHJqWVR0mIlCd/6ZoTng== X-CSE-MsgGUID: dKxpn+2NTjuzfrgFoxEBrA== X-IronPort-AV: E=McAfee;i="6700,10204,11163"; a="39287084" X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="39287084" Received: from orviesa007.jf.intel.com ([10.64.159.147]) by fmvoesa102.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 13 Aug 2024 15:22:55 -0700 X-CSE-ConnectionGUID: S9EXafVARX+og9q8mHnyAA== X-CSE-MsgGUID: uR/WNpGQS+OiLYbuOtRH9A== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="59381055" Received: from anguy11-upstream.jf.intel.com ([10.166.9.133]) by orviesa007.jf.intel.com with ESMTP; 13 Aug 2024 15:22:54 -0700 From: Tony Nguyen To: davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com, edumazet@google.com, netdev@vger.kernel.org Cc: Junfeng Guo , anthony.l.nguyen@intel.com, ahmed.zaki@intel.com, madhu.chittim@intel.com, horms@kernel.org, hkelam@marvell.com, Marcin Szycik , Qi Zhang , Rafal Romanowski Subject: [PATCH net-next v2 03/13] ice: add debugging functions for the parser sections Date: Tue, 13 Aug 2024 15:22:38 -0700 Message-ID: <20240813222249.3708070-4-anthony.l.nguyen@intel.com> X-Mailer: git-send-email 2.42.0 In-Reply-To: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> References: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Patchwork-Delegate: kuba@kernel.org From: Junfeng Guo Add debug for all parser sections. Reviewed-by: Marcin Szycik Signed-off-by: Qi Zhang Signed-off-by: Junfeng Guo Signed-off-by: Ahmed Zaki Tested-by: Rafal Romanowski Signed-off-by: Tony Nguyen --- drivers/net/ethernet/intel/ice/ice_parser.c | 470 ++++++++++++++++++++ 1 file changed, 470 insertions(+) diff --git a/drivers/net/ethernet/intel/ice/ice_parser.c b/drivers/net/ethernet/intel/ice/ice_parser.c index 5fdc87034fc6..23ba1a36742e 100644 --- a/drivers/net/ethernet/intel/ice/ice_parser.c +++ b/drivers/net/ethernet/intel/ice/ice_parser.c @@ -129,6 +129,100 @@ ice_parser_create_table(struct ice_hw *hw, u32 sect_type, } /*** ICE_SID_RXPARSER_IMEM section ***/ +static void ice_imem_bst_bm_dump(struct ice_hw *hw, struct ice_bst_main *bm) +{ + struct device *dev = ice_hw_to_dev(hw); + + dev_info(dev, "boost main:\n"); + dev_info(dev, "\talu0 = %d\n", bm->alu0); + dev_info(dev, "\talu1 = %d\n", bm->alu1); + dev_info(dev, "\talu2 = %d\n", bm->alu2); + dev_info(dev, "\tpg = %d\n", bm->pg); +} + +static void ice_imem_bst_kb_dump(struct ice_hw *hw, + struct ice_bst_keybuilder *kb) +{ + struct device *dev = ice_hw_to_dev(hw); + + dev_info(dev, "boost key builder:\n"); + dev_info(dev, "\tpriority = %d\n", kb->prio); + dev_info(dev, "\ttsr_ctrl = %d\n", kb->tsr_ctrl); +} + +static void ice_imem_np_kb_dump(struct ice_hw *hw, + struct ice_np_keybuilder *kb) +{ + struct device *dev = ice_hw_to_dev(hw); + + dev_info(dev, "next proto key builder:\n"); + dev_info(dev, "\topc = %d\n", kb->opc); + dev_info(dev, "\tstart_or_reg0 = %d\n", kb->start_reg0); + dev_info(dev, "\tlen_or_reg1 = %d\n", kb->len_reg1); +} + +static void ice_imem_pg_kb_dump(struct ice_hw *hw, + struct ice_pg_keybuilder *kb) +{ + struct device *dev = ice_hw_to_dev(hw); + + dev_info(dev, "parse graph key builder:\n"); + dev_info(dev, "\tflag0_ena = %d\n", kb->flag0_ena); + dev_info(dev, "\tflag1_ena = %d\n", kb->flag1_ena); + dev_info(dev, "\tflag2_ena = %d\n", kb->flag2_ena); + dev_info(dev, "\tflag3_ena = %d\n", kb->flag3_ena); + dev_info(dev, "\tflag0_idx = %d\n", kb->flag0_idx); + dev_info(dev, "\tflag1_idx = %d\n", kb->flag1_idx); + dev_info(dev, "\tflag2_idx = %d\n", kb->flag2_idx); + dev_info(dev, "\tflag3_idx = %d\n", kb->flag3_idx); + dev_info(dev, "\talu_reg_idx = %d\n", kb->alu_reg_idx); +} + +static void ice_imem_alu_dump(struct ice_hw *hw, + struct ice_alu *alu, int index) +{ + struct device *dev = ice_hw_to_dev(hw); + + dev_info(dev, "alu%d:\n", index); + dev_info(dev, "\topc = %d\n", alu->opc); + dev_info(dev, "\tsrc_start = %d\n", alu->src_start); + dev_info(dev, "\tsrc_len = %d\n", alu->src_len); + dev_info(dev, "\tshift_xlate_sel = %d\n", alu->shift_xlate_sel); + dev_info(dev, "\tshift_xlate_key = %d\n", alu->shift_xlate_key); + dev_info(dev, "\tsrc_reg_id = %d\n", alu->src_reg_id); + dev_info(dev, "\tdst_reg_id = %d\n", alu->dst_reg_id); + dev_info(dev, "\tinc0 = %d\n", alu->inc0); + dev_info(dev, "\tinc1 = %d\n", alu->inc1); + dev_info(dev, "\tproto_offset_opc = %d\n", alu->proto_offset_opc); + dev_info(dev, "\tproto_offset = %d\n", alu->proto_offset); + dev_info(dev, "\tbranch_addr = %d\n", alu->branch_addr); + dev_info(dev, "\timm = %d\n", alu->imm); + dev_info(dev, "\tdst_start = %d\n", alu->dst_start); + dev_info(dev, "\tdst_len = %d\n", alu->dst_len); + dev_info(dev, "\tflags_extr_imm = %d\n", alu->flags_extr_imm); + dev_info(dev, "\tflags_start_imm= %d\n", alu->flags_start_imm); +} + +/** + * ice_imem_dump - dump an imem item info + * @hw: pointer to the hardware structure + * @item: imem item to dump + */ +static void ice_imem_dump(struct ice_hw *hw, struct ice_imem_item *item) +{ + struct device *dev = ice_hw_to_dev(hw); + + dev_info(dev, "index = %d\n", item->idx); + ice_imem_bst_bm_dump(hw, &item->b_m); + ice_imem_bst_kb_dump(hw, &item->b_kb); + dev_info(dev, "pg priority = %d\n", item->pg_prio); + ice_imem_np_kb_dump(hw, &item->np_kb); + ice_imem_pg_kb_dump(hw, &item->pg_kb); + ice_imem_alu_dump(hw, &item->alu0, 0); + ice_imem_alu_dump(hw, &item->alu1, 1); + ice_imem_alu_dump(hw, &item->alu2, 2); +} + #define ICE_IM_BM_ALU0 BIT(0) #define ICE_IM_BM_ALU1 BIT(1) #define ICE_IM_BM_ALU2 BIT(2) @@ -328,6 +422,9 @@ static void ice_imem_parse_item(struct ice_hw *hw, u16 idx, void *item, ice_imem_alu_init(&ii->alu2, &buf[ICE_IMEM_ALU2_IDD], ICE_IMEM_ALU2_OFF); + + if (hw->debug_mask & ICE_DBG_PARSER) + ice_imem_dump(hw, ii); } /** @@ -345,6 +442,50 @@ static struct ice_imem_item *ice_imem_table_get(struct ice_hw *hw) } /*** ICE_SID_RXPARSER_METADATA_INIT section ***/ +/** + * ice_metainit_dump - dump an metainit item info + * @hw: pointer to the hardware structure + * @item: metainit item to dump + */ +static void ice_metainit_dump(struct ice_hw *hw, struct ice_metainit_item *item) +{ + struct device *dev = ice_hw_to_dev(hw); + + dev_info(dev, "index = %d\n", item->idx); + + dev_info(dev, "tsr = %d\n", item->tsr); + dev_info(dev, "ho = %d\n", item->ho); + dev_info(dev, "pc = %d\n", item->pc); + dev_info(dev, "pg_rn = %d\n", item->pg_rn); + dev_info(dev, "cd = %d\n", item->cd); + + dev_info(dev, "gpr_a_ctrl = %d\n", item->gpr_a_ctrl); + dev_info(dev, "gpr_a_data_mdid = %d\n", item->gpr_a_data_mdid); + dev_info(dev, "gpr_a_data_start = %d\n", item->gpr_a_data_start); + dev_info(dev, "gpr_a_data_len = %d\n", item->gpr_a_data_len); + dev_info(dev, "gpr_a_id = %d\n", item->gpr_a_id); + + dev_info(dev, "gpr_b_ctrl = %d\n", item->gpr_b_ctrl); + dev_info(dev, "gpr_b_data_mdid = %d\n", item->gpr_b_data_mdid); + dev_info(dev, "gpr_b_data_start = %d\n", item->gpr_b_data_start); + dev_info(dev, "gpr_b_data_len = %d\n", item->gpr_b_data_len); + dev_info(dev, "gpr_b_id = %d\n", item->gpr_b_id); + + dev_info(dev, "gpr_c_ctrl = %d\n", item->gpr_c_ctrl); + dev_info(dev, "gpr_c_data_mdid = %d\n", item->gpr_c_data_mdid); + dev_info(dev, "gpr_c_data_start = %d\n", item->gpr_c_data_start); + dev_info(dev, "gpr_c_data_len = %d\n", item->gpr_c_data_len); + dev_info(dev, "gpr_c_id = %d\n", item->gpr_c_id); + + dev_info(dev, "gpr_d_ctrl = %d\n", item->gpr_d_ctrl); + dev_info(dev, "gpr_d_data_mdid = %d\n", item->gpr_d_data_mdid); + dev_info(dev, "gpr_d_data_start = %d\n", item->gpr_d_data_start); + dev_info(dev, "gpr_d_data_len = %d\n", item->gpr_d_data_len); + dev_info(dev, "gpr_d_id = %d\n", item->gpr_d_id); + + dev_info(dev, "flags = 0x%llx\n", (unsigned long long)(item->flags)); +} + #define ICE_MI_TSR GENMASK_ULL(7, 0) #define ICE_MI_HO GENMASK_ULL(16, 8) #define ICE_MI_PC GENMASK_ULL(24, 17) @@ -437,6 +578,9 @@ static void ice_metainit_parse_item(struct ice_hw *hw, u16 idx, void *item, d64 = *((u64 *)&buf[ICE_MI_FLAG_IDD]) >> ICE_MI_FLAG_OFF; mi->flags = FIELD_GET(ICE_MI_FLAG, d64); + + if (hw->debug_mask & ICE_DBG_PARSER) + ice_metainit_dump(hw, mi); } /** @@ -456,6 +600,80 @@ static struct ice_metainit_item *ice_metainit_table_get(struct ice_hw *hw) /*** ICE_SID_RXPARSER_CAM, ICE_SID_RXPARSER_PG_SPILL, * ICE_SID_RXPARSER_NOMATCH_CAM and ICE_SID_RXPARSER_NOMATCH_CAM * sections ***/ +static void ice_pg_cam_key_dump(struct ice_hw *hw, struct ice_pg_cam_key *key) +{ + struct device *dev = ice_hw_to_dev(hw); + + dev_info(dev, "key:\n"); + dev_info(dev, "\tvalid = %d\n", key->valid); + dev_info(dev, "\tnode_id = %d\n", key->node_id); + dev_info(dev, "\tflag0 = %d\n", key->flag0); + dev_info(dev, "\tflag1 = %d\n", key->flag1); + dev_info(dev, "\tflag2 = %d\n", key->flag2); + dev_info(dev, "\tflag3 = %d\n", key->flag3); + dev_info(dev, "\tboost_idx = %d\n", key->boost_idx); + dev_info(dev, "\talu_reg = 0x%04x\n", key->alu_reg); + dev_info(dev, "\tnext_proto = 0x%08x\n", key->next_proto); +} + +static void ice_pg_nm_cam_key_dump(struct ice_hw *hw, + struct ice_pg_nm_cam_key *key) +{ + struct device *dev = ice_hw_to_dev(hw); + + dev_info(dev, "key:\n"); + dev_info(dev, "\tvalid = %d\n", key->valid); + dev_info(dev, "\tnode_id = %d\n", key->node_id); + dev_info(dev, "\tflag0 = %d\n", key->flag0); + dev_info(dev, "\tflag1 = %d\n", key->flag1); + dev_info(dev, "\tflag2 = %d\n", key->flag2); + dev_info(dev, "\tflag3 = %d\n", key->flag3); + dev_info(dev, "\tboost_idx = %d\n", key->boost_idx); + dev_info(dev, "\talu_reg = 0x%04x\n", key->alu_reg); +} + +static void ice_pg_cam_action_dump(struct ice_hw *hw, + struct ice_pg_cam_action *action) +{ + struct device *dev = ice_hw_to_dev(hw); + + dev_info(dev, "action:\n"); + dev_info(dev, "\tnext_node = %d\n", action->next_node); + dev_info(dev, "\tnext_pc = %d\n", action->next_pc); + dev_info(dev, "\tis_pg = %d\n", action->is_pg); + dev_info(dev, "\tproto_id = %d\n", action->proto_id); + dev_info(dev, "\tis_mg = %d\n", action->is_mg); + dev_info(dev, "\tmarker_id = %d\n", action->marker_id); + dev_info(dev, "\tis_last_round = %d\n", action->is_last_round); + dev_info(dev, "\tho_polarity = %d\n", action->ho_polarity); + dev_info(dev, "\tho_inc = %d\n", action->ho_inc); +} + +/** + * ice_pg_cam_dump - dump an parse graph cam info + * @hw: pointer to the hardware structure + * @item: parse graph cam to dump + */ +static void ice_pg_cam_dump(struct ice_hw *hw, struct ice_pg_cam_item *item) +{ + dev_info(ice_hw_to_dev(hw), "index = %d\n", item->idx); + ice_pg_cam_key_dump(hw, &item->key); + ice_pg_cam_action_dump(hw, &item->action); +} + +/** + * ice_pg_nm_cam_dump - dump an parse graph no match cam info + * @hw: pointer to the hardware structure + * @item: parse graph no match cam to dump + */ +static void ice_pg_nm_cam_dump(struct ice_hw *hw, + struct ice_pg_nm_cam_item *item) +{ + dev_info(ice_hw_to_dev(hw), "index = %d\n", item->idx); + ice_pg_nm_cam_key_dump(hw, &item->key); + ice_pg_cam_action_dump(hw, &item->action); +} + #define ICE_PGCA_NN GENMASK_ULL(10, 0) #define ICE_PGCA_NPC GENMASK_ULL(18, 11) #define ICE_PGCA_IPG BIT_ULL(19) @@ -584,6 +802,9 @@ static void ice_pg_cam_parse_item(struct ice_hw *hw, u16 idx, void *item, d64 = *((u64 *)&buf[ICE_PG_CAM_ACT_IDD]) >> ICE_PG_CAM_ACT_OFF; ice_pg_cam_action_init(&ci->action, d64); + + if (hw->debug_mask & ICE_DBG_PARSER) + ice_pg_cam_dump(hw, ci); } #define ICE_PG_SP_CAM_KEY_S 56 @@ -610,6 +831,9 @@ static void ice_pg_sp_cam_parse_item(struct ice_hw *hw, u16 idx, void *item, ice_pg_cam_action_init(&ci->action, d64); ice_pg_cam_key_init(&ci->key, &buf[ICE_PG_SP_CAM_KEY_IDD]); + + if (hw->debug_mask & ICE_DBG_PARSER) + ice_pg_cam_dump(hw, ci); } #define ICE_PG_NM_CAM_ACT_S 41 @@ -638,6 +862,9 @@ static void ice_pg_nm_cam_parse_item(struct ice_hw *hw, u16 idx, void *item, d64 = *((u64 *)&buf[ICE_PG_NM_CAM_ACT_IDD]) >> ICE_PG_NM_CAM_ACT_OFF; ice_pg_cam_action_init(&ci->action, d64); + + if (hw->debug_mask & ICE_DBG_PARSER) + ice_pg_nm_cam_dump(hw, ci); } #define ICE_PG_NM_SP_CAM_ACT_S 56 @@ -669,6 +896,9 @@ static void ice_pg_nm_sp_cam_parse_item(struct ice_hw *hw, u16 idx, d64 = *((u64 *)&buf[ICE_PG_NM_SP_CAM_ACT_IDD]) >> ICE_PG_NM_SP_CAM_ACT_OFF; ice_pg_nm_cam_key_init(&ci->key, d64); + + if (hw->debug_mask & ICE_DBG_PARSER) + ice_pg_nm_cam_dump(hw, ci); } /** @@ -728,6 +958,99 @@ static struct ice_pg_nm_cam_item *ice_pg_nm_sp_cam_table_get(struct ice_hw *hw) } /*** ICE_SID_RXPARSER_BOOST_TCAM and ICE_SID_LBL_RXPARSER_TMEM sections ***/ +static void ice_bst_np_kb_dump(struct ice_hw *hw, struct ice_np_keybuilder *kb) +{ + struct device *dev = ice_hw_to_dev(hw); + + dev_info(dev, "next proto key builder:\n"); + dev_info(dev, "\topc = %d\n", kb->opc); + dev_info(dev, "\tstart_reg0 = %d\n", kb->start_reg0); + dev_info(dev, "\tlen_reg1 = %d\n", kb->len_reg1); +} + +static void ice_bst_pg_kb_dump(struct ice_hw *hw, struct ice_pg_keybuilder *kb) +{ + struct device *dev = ice_hw_to_dev(hw); + + dev_info(dev, "parse graph key builder:\n"); + dev_info(dev, "\tflag0_ena = %d\n", kb->flag0_ena); + dev_info(dev, "\tflag1_ena = %d\n", kb->flag1_ena); + dev_info(dev, "\tflag2_ena = %d\n", kb->flag2_ena); + dev_info(dev, "\tflag3_ena = %d\n", kb->flag3_ena); + dev_info(dev, "\tflag0_idx = %d\n", kb->flag0_idx); + dev_info(dev, "\tflag1_idx = %d\n", kb->flag1_idx); + dev_info(dev, "\tflag2_idx = %d\n", kb->flag2_idx); + dev_info(dev, "\tflag3_idx = %d\n", kb->flag3_idx); + dev_info(dev, "\talu_reg_idx = %d\n", kb->alu_reg_idx); +} + +static void ice_bst_alu_dump(struct ice_hw *hw, struct ice_alu *alu, int idx) +{ + struct device *dev = ice_hw_to_dev(hw); + + dev_info(dev, "alu%d:\n", idx); + dev_info(dev, "\topc = %d\n", alu->opc); + dev_info(dev, "\tsrc_start = %d\n", alu->src_start); + dev_info(dev, "\tsrc_len = %d\n", alu->src_len); + dev_info(dev, "\tshift_xlate_sel = %d\n", alu->shift_xlate_sel); + dev_info(dev, "\tshift_xlate_key = %d\n", alu->shift_xlate_key); + dev_info(dev, "\tsrc_reg_id = %d\n", alu->src_reg_id); + dev_info(dev, "\tdst_reg_id = %d\n", alu->dst_reg_id); + dev_info(dev, "\tinc0 = %d\n", alu->inc0); + dev_info(dev, "\tinc1 = %d\n", alu->inc1); + dev_info(dev, "\tproto_offset_opc = %d\n", alu->proto_offset_opc); + dev_info(dev, "\tproto_offset = %d\n", alu->proto_offset); + dev_info(dev, "\tbranch_addr = %d\n", alu->branch_addr); + dev_info(dev, "\timm = %d\n", alu->imm); + dev_info(dev, "\tdst_start = %d\n", alu->dst_start); + dev_info(dev, "\tdst_len = %d\n", alu->dst_len); + dev_info(dev, "\tflags_extr_imm = %d\n", alu->flags_extr_imm); + dev_info(dev, "\tflags_start_imm= %d\n", alu->flags_start_imm); +} + +/** + * ice_bst_tcam_dump - dump a boost tcam info + * @hw: pointer to the hardware structure + * @item: boost tcam to dump + */ +static void ice_bst_tcam_dump(struct ice_hw *hw, struct ice_bst_tcam_item *item) +{ + struct device *dev = ice_hw_to_dev(hw); + int i; + + dev_info(dev, "addr = %d\n", item->addr); + + dev_info(dev, "key : "); + for (i = 0; i < ICE_BST_TCAM_KEY_SIZE; i++) + dev_info(dev, "%02x ", item->key[i]); + + dev_info(dev, "\n"); + + dev_info(dev, "key_inv: "); + for (i = 0; i < ICE_BST_TCAM_KEY_SIZE; i++) + dev_info(dev, "%02x ", item->key_inv[i]); + + dev_info(dev, "\n"); + + dev_info(dev, "hit_idx_grp = %d\n", item->hit_idx_grp); + dev_info(dev, "pg_prio = %d\n", item->pg_prio); + + ice_bst_np_kb_dump(hw, &item->np_kb); + ice_bst_pg_kb_dump(hw, &item->pg_kb); + + ice_bst_alu_dump(hw, &item->alu0, ICE_ALU0_IDX); + ice_bst_alu_dump(hw, &item->alu1, ICE_ALU1_IDX); + ice_bst_alu_dump(hw, &item->alu2, ICE_ALU2_IDX); +} + +static void ice_lbl_dump(struct ice_hw *hw, struct ice_lbl_item *item) +{ + struct device *dev = ice_hw_to_dev(hw); + + dev_info(dev, "index = %u\n", item->idx); + dev_info(dev, "label = %s\n", item->label); +} + #define ICE_BST_ALU_OPC GENMASK_ULL(5, 0) #define ICE_BST_ALU_SS GENMASK_ULL(13, 6) #define ICE_BST_ALU_SL GENMASK_ULL(18, 14) @@ -894,6 +1217,9 @@ static void ice_bst_parse_item(struct ice_hw *hw, u16 idx, void *item, ice_bst_alu_init(&ti->alu0, &buf[ICE_BT_ALU0_IDD], ICE_BT_ALU0_OFF); ice_bst_alu_init(&ti->alu1, &buf[ICE_BT_ALU1_IDD], ICE_BT_ALU1_OFF); ice_bst_alu_init(&ti->alu2, &buf[ICE_BT_ALU2_IDD], ICE_BT_ALU2_OFF); + + if (hw->debug_mask & ICE_DBG_PARSER) + ice_bst_tcam_dump(hw, ti); } /** @@ -914,6 +1240,9 @@ static void ice_parse_lbl_item(struct ice_hw *hw, u16 idx, void *item, void *data, int size) { memcpy(item, data, size); + + if (hw->debug_mask & ICE_DBG_PARSER) + ice_lbl_dump(hw, (struct ice_lbl_item *)item); } /** @@ -931,10 +1260,41 @@ static struct ice_lbl_item *ice_bst_lbl_table_get(struct ice_hw *hw) } /*** ICE_SID_RXPARSER_MARKER_PTYPE section ***/ +/** + * ice_ptype_mk_tcam_dump - dump an ptype marker tcam info + * @hw: pointer to the hardware structure + * @item: ptype marker tcam to dump + */ +static void ice_ptype_mk_tcam_dump(struct ice_hw *hw, + struct ice_ptype_mk_tcam_item *item) +{ + struct device *dev = ice_hw_to_dev(hw); + int i; + + dev_info(dev, "address = %d\n", item->address); + dev_info(dev, "ptype = %d\n", item->ptype); + + dev_info(dev, "key :"); + for (i = 0; i < ICE_PTYPE_MK_TCAM_KEY_SIZE; i++) + dev_info(dev, "%02x ", item->key[i]); + + dev_info(dev, "\n"); + + dev_info(dev, "key_inv:"); + for (i = 0; i < ICE_PTYPE_MK_TCAM_KEY_SIZE; i++) + dev_info(dev, "%02x ", item->key_inv[i]); + + dev_info(dev, "\n"); +} + static void ice_parse_ptype_mk_tcam_item(struct ice_hw *hw, u16 idx, void *item, void *data, int size) { memcpy(item, data, size); + + if (hw->debug_mask & ICE_DBG_PARSER) + ice_ptype_mk_tcam_dump(hw, + (struct ice_ptype_mk_tcam_item *)item); } /** @@ -953,6 +1313,25 @@ struct ice_ptype_mk_tcam_item *ice_ptype_mk_tcam_table_get(struct ice_hw *hw) } /*** ICE_SID_RXPARSER_MARKER_GRP section ***/ +/** + * ice_mk_grp_dump - dump an marker group item info + * @hw: pointer to the hardware structure + * @item: marker group item to dump + */ +static void ice_mk_grp_dump(struct ice_hw *hw, struct ice_mk_grp_item *item) +{ + struct device *dev = ice_hw_to_dev(hw); + int i; + + dev_info(dev, "index = %d\n", item->idx); + + dev_info(dev, "markers: "); + for (i = 0; i < ICE_MK_COUNT_PER_GRP; i++) + dev_info(dev, "%d ", item->markers[i]); + + dev_info(dev, "\n"); +} + static void ice_mk_grp_parse_item(struct ice_hw *hw, u16 idx, void *item, void *data, int __maybe_unused size) { @@ -964,6 +1343,9 @@ static void ice_mk_grp_parse_item(struct ice_hw *hw, u16 idx, void *item, for (i = 0; i < ICE_MK_COUNT_PER_GRP; i++) grp->markers[i] = buf[i]; + + if (hw->debug_mask & ICE_DBG_PARSER) + ice_mk_grp_dump(hw, grp); } /** @@ -981,6 +1363,33 @@ static struct ice_mk_grp_item *ice_mk_grp_table_get(struct ice_hw *hw) } /*** ICE_SID_RXPARSER_PROTO_GRP section ***/ +static void ice_proto_off_dump(struct ice_hw *hw, + struct ice_proto_off *po, int idx) +{ + struct device *dev = ice_hw_to_dev(hw); + + dev_info(dev, "proto %d\n", idx); + dev_info(dev, "\tpolarity = %d\n", po->polarity); + dev_info(dev, "\tproto_id = %d\n", po->proto_id); + dev_info(dev, "\toffset = %d\n", po->offset); +} + +/** + * ice_proto_grp_dump - dump a proto group item info + * @hw: pointer to the hardware structure + * @item: proto group item to dump + */ +static void ice_proto_grp_dump(struct ice_hw *hw, + struct ice_proto_grp_item *item) +{ + int i; + + dev_info(ice_hw_to_dev(hw), "index = %d\n", item->idx); + + for (i = 0; i < ICE_PROTO_COUNT_PER_GRP; i++) + ice_proto_off_dump(hw, &item->po[i], i); +} + #define ICE_PO_POL BIT(0) #define ICE_PO_PID GENMASK(8, 1) #define ICE_PO_OFF GENMASK(21, 12) @@ -1022,6 +1431,9 @@ static void ice_proto_grp_parse_item(struct ice_hw *hw, u16 idx, void *item, d32 = *((u32 *)&buf[idd]) >> off; ice_proto_off_parse(&grp->po[i], d32); } + + if (hw->debug_mask & ICE_DBG_PARSER) + ice_proto_grp_dump(hw, grp); } /** @@ -1039,6 +1451,20 @@ static struct ice_proto_grp_item *ice_proto_grp_table_get(struct ice_hw *hw) } /*** ICE_SID_RXPARSER_FLAG_REDIR section ***/ +/** + * ice_flg_rd_dump - dump a flag redirect item info + * @hw: pointer to the hardware structure + * @item: flag redirect item to dump + */ +static void ice_flg_rd_dump(struct ice_hw *hw, struct ice_flg_rd_item *item) +{ + struct device *dev = ice_hw_to_dev(hw); + + dev_info(dev, "index = %d\n", item->idx); + dev_info(dev, "expose = %d\n", item->expose); + dev_info(dev, "intr_flg_id = %d\n", item->intr_flg_id); +} + #define ICE_FRT_EXPO BIT(0) #define ICE_FRT_IFID GENMASK(6, 1) @@ -1059,6 +1485,9 @@ static void ice_flg_rd_parse_item(struct ice_hw *hw, u16 idx, void *item, rdi->idx = idx; rdi->expose = FIELD_GET(ICE_FRT_EXPO, d8); rdi->intr_flg_id = FIELD_GET(ICE_FRT_IFID, d8); + + if (hw->debug_mask & ICE_DBG_PARSER) + ice_flg_rd_dump(hw, rdi); } /** @@ -1078,6 +1507,44 @@ static struct ice_flg_rd_item *ice_flg_rd_table_get(struct ice_hw *hw) /*** ICE_SID_XLT_KEY_BUILDER_SW, ICE_SID_XLT_KEY_BUILDER_ACL, * ICE_SID_XLT_KEY_BUILDER_FD and ICE_SID_XLT_KEY_BUILDER_RSS * sections ***/ +static void ice_xlt_kb_entry_dump(struct ice_hw *hw, + struct ice_xlt_kb_entry *entry, int idx) +{ + struct device *dev = ice_hw_to_dev(hw); + int i; + + dev_info(dev, "key builder entry %d\n", idx); + dev_info(dev, "\txlt1_ad_sel = %d\n", entry->xlt1_ad_sel); + dev_info(dev, "\txlt2_ad_sel = %d\n", entry->xlt2_ad_sel); + + for (i = 0; i < ICE_XLT_KB_FLAG0_14_CNT; i++) + dev_info(dev, "\tflg%d_sel = %d\n", i, entry->flg0_14_sel[i]); + + dev_info(dev, "\txlt1_md_sel = %d\n", entry->xlt1_md_sel); + dev_info(dev, "\txlt2_md_sel = %d\n", entry->xlt2_md_sel); +} + +/** + * ice_xlt_kb_dump - dump a xlt key build info + * @hw: pointer to the hardware structure + * @kb: key build to dump + */ +static void ice_xlt_kb_dump(struct ice_hw *hw, struct ice_xlt_kb *kb) +{ + struct device *dev = ice_hw_to_dev(hw); + int i; + + dev_info(dev, "xlt1_pm = %d\n", kb->xlt1_pm); + dev_info(dev, "xlt2_pm = %d\n", kb->xlt2_pm); + dev_info(dev, "prof_id_pm = %d\n", kb->prof_id_pm); + dev_info(dev, "flag15 lo = 0x%08x\n", (u32)kb->flag15); + dev_info(dev, "flag15 hi = 0x%08x\n", + (u32)(kb->flag15 >> (sizeof(u32) * BITS_PER_BYTE))); + + for (i = 0; i < ICE_XLT_KB_TBL_CNT; i++) + ice_xlt_kb_entry_dump(hw, &kb->entries[i], i); +} + #define ICE_XLT_KB_X1AS_S 32 /* offset for the 1st 64-bits field */ #define ICE_XLT_KB_X1AS_IDD (ICE_XLT_KB_X1AS_S / BITS_PER_BYTE) #define ICE_XLT_KB_X1AS_OFF (ICE_XLT_KB_X1AS_S % BITS_PER_BYTE) @@ -1194,6 +1661,9 @@ static void ice_parse_kb_data(struct ice_hw *hw, struct ice_xlt_kb *kb, ice_kb_entry_init(&kb->entries[i], &buf[ICE_XLT_KB_TBL_OFF + i * ICE_XLT_KB_TBL_ENTRY_SIZE]); + + if (hw->debug_mask & ICE_DBG_PARSER) + ice_xlt_kb_dump(hw, kb); } static struct ice_xlt_kb *ice_xlt_kb_get(struct ice_hw *hw, u32 sect_type) From patchwork Tue Aug 13 22:22:39 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Tony Nguyen X-Patchwork-Id: 13762624 X-Patchwork-Delegate: kuba@kernel.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.8]) (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 09A371AB515 for ; Tue, 13 Aug 2024 22:22:57 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=192.198.163.8 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587779; cv=none; b=pBMjC9Ql0W4fKz4zYe3R4lUX3/VqsNvEzfbkFueY05qi5ez9uj4cZhDFmvCHiBHVQByjhQK9ECC1vjboATjQoufGljHYE54Fgdap4KZy2p+xMBN3nGqprel4jsOflbrDjnPIBihDLU+4VnprVvSRMVgms2APGWc6GV8V9jbgjqk= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587779; c=relaxed/simple; bh=KQHJ2SgNY64HGwyqSaa4nSLlYVYOHnD/TPPAnpF+oyg=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=IbWx+ue+7p9aY2Db9xVwRLNEfDExTJyt4HwMN3jy7Ek1NN5woICcUOAO/85jGeFFzmad6Nc2/SbpWi87/OU6QvCIpw+S34QrBDQ4MvdS3ProKP52ACWtsfrhonwoZd9s9TdCujAtopiXJ46QMRy+540CYSFYTU0RAc4R7rk6z5A= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com; spf=pass smtp.mailfrom=intel.com; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b=A3Q7kKW5; arc=none smtp.client-ip=192.198.163.8 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=intel.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b="A3Q7kKW5" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1723587778; x=1755123778; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=KQHJ2SgNY64HGwyqSaa4nSLlYVYOHnD/TPPAnpF+oyg=; b=A3Q7kKW5N03Odg2dCM6L0sSt3oZm5Y9cOzEWUuA9CphDMusjbcn71oOw 5TiL61KKqTePxpv5KaxCcInzTSldwTpACjpAFhw5xNKNH+Qn14vCU6Dk5 SQgVKURZ8WkG0if3VQ/Lo2KoSFdFU5DBNjGLZO8PcdAjbWyDt7iVR4fiQ ehcg4+CUfw5OGJx8+VLh5ifN2slvcvKqINH/QiKaIysKzP/GUM0ImgRVM ZSBncdzKkkrPWyvPPu9Rg3yvj+kWcmGEmXACMNo/iiqzUTH5cyqOXSiXm nu4Qc8dvL81BeQRBwWde5U0nOvN0lH4gCbyW3L7t5neWT80dY4pS0v47B g==; X-CSE-ConnectionGUID: 3r4/6RK6QUuhnoxj6/izNg== X-CSE-MsgGUID: g0J0b0npQoySgs7myf0Jkg== X-IronPort-AV: E=McAfee;i="6700,10204,11163"; a="39287088" X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="39287088" Received: from orviesa007.jf.intel.com ([10.64.159.147]) by fmvoesa102.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 13 Aug 2024 15:22:55 -0700 X-CSE-ConnectionGUID: pvuBccudRied4FG2LhnFBA== X-CSE-MsgGUID: WRlnM207RrGmrkplC5kiNw== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="59381061" Received: from anguy11-upstream.jf.intel.com ([10.166.9.133]) by orviesa007.jf.intel.com with ESMTP; 13 Aug 2024 15:22:54 -0700 From: Tony Nguyen To: davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com, edumazet@google.com, netdev@vger.kernel.org Cc: Junfeng Guo , anthony.l.nguyen@intel.com, ahmed.zaki@intel.com, madhu.chittim@intel.com, horms@kernel.org, hkelam@marvell.com, Aleksandr Loktionov , Wojciech Drewek , Marcin Szycik , Qi Zhang , Rafal Romanowski Subject: [PATCH net-next v2 04/13] ice: add parser internal helper functions Date: Tue, 13 Aug 2024 15:22:39 -0700 Message-ID: <20240813222249.3708070-5-anthony.l.nguyen@intel.com> X-Mailer: git-send-email 2.42.0 In-Reply-To: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> References: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Patchwork-Delegate: kuba@kernel.org From: Junfeng Guo Add the following internal helper functions: - ice_bst_tcam_match(): to perform ternary match on boost TCAM. - ice_pg_cam_match(): to perform parse graph key match in cam table. - ice_pg_nm_cam_match(): to perform parse graph key no match in cam table. - ice_ptype_mk_tcam_match(): to perform ptype markers match in tcam table. - ice_flg_redirect(): to redirect parser flags to packet flags. - ice_xlt_kb_flag_get(): to aggregate 64 bit packet flag into 16 bit key builder flags. Reviewed-by: Simon Horman Reviewed-by: Aleksandr Loktionov Reviewed-by: Wojciech Drewek Reviewed-by: Marcin Szycik Signed-off-by: Qi Zhang Signed-off-by: Junfeng Guo Signed-off-by: Ahmed Zaki Tested-by: Rafal Romanowski Signed-off-by: Tony Nguyen --- drivers/net/ethernet/intel/ice/ice_parser.c | 208 ++++++++++++++++++++ drivers/net/ethernet/intel/ice/ice_parser.h | 51 +++-- 2 files changed, 244 insertions(+), 15 deletions(-) diff --git a/drivers/net/ethernet/intel/ice/ice_parser.c b/drivers/net/ethernet/intel/ice/ice_parser.c index 23ba1a36742e..f44310e94150 100644 --- a/drivers/net/ethernet/intel/ice/ice_parser.c +++ b/drivers/net/ethernet/intel/ice/ice_parser.c @@ -957,6 +957,109 @@ static struct ice_pg_nm_cam_item *ice_pg_nm_sp_cam_table_get(struct ice_hw *hw) ice_pg_nm_sp_cam_parse_item, false); } +static bool __ice_pg_cam_match(struct ice_pg_cam_item *item, + struct ice_pg_cam_key *key) +{ + return (item->key.valid && + !memcmp(&item->key.val, &key->val, sizeof(key->val))); +} + +static bool __ice_pg_nm_cam_match(struct ice_pg_nm_cam_item *item, + struct ice_pg_cam_key *key) +{ + return (item->key.valid && + !memcmp(&item->key.val, &key->val, sizeof(item->key.val))); +} + +/** + * ice_pg_cam_match - search parse graph cam table by key + * @table: parse graph cam table to search + * @size: cam table size + * @key: search key + * + * Return: a pointer to the matching PG CAM item or NULL. + */ +struct ice_pg_cam_item *ice_pg_cam_match(struct ice_pg_cam_item *table, + int size, struct ice_pg_cam_key *key) +{ + int i; + + for (i = 0; i < size; i++) { + struct ice_pg_cam_item *item = &table[i]; + + if (__ice_pg_cam_match(item, key)) + return item; + } + + return NULL; +} + +/** + * ice_pg_nm_cam_match - search parse graph no match cam table by key + * @table: parse graph no match cam table to search + * @size: cam table size + * @key: search key + * + * Return: a pointer to the matching PG No Match CAM item or NULL. + */ +struct ice_pg_nm_cam_item * +ice_pg_nm_cam_match(struct ice_pg_nm_cam_item *table, int size, + struct ice_pg_cam_key *key) +{ + int i; + + for (i = 0; i < size; i++) { + struct ice_pg_nm_cam_item *item = &table[i]; + + if (__ice_pg_nm_cam_match(item, key)) + return item; + } + + return NULL; +} + +/*** Ternary match ***/ +/* Perform a ternary match on a 1-byte pattern (@pat) given @key and @key_inv + * Rules (per bit): + * Key == 0 and Key_inv == 0 : Never match (Don't care) + * Key == 0 and Key_inv == 1 : Match on bit == 1 + * Key == 1 and Key_inv == 0 : Match on bit == 0 + * Key == 1 and Key_inv == 1 : Always match (Don't care) + * + * Return: true if all bits match, false otherwise. + */ +static bool ice_ternary_match_byte(u8 key, u8 key_inv, u8 pat) +{ + u8 bit_key, bit_key_inv, bit_pat; + int i; + + for (i = 0; i < BITS_PER_BYTE; i++) { + bit_key = key & BIT(i); + bit_key_inv = key_inv & BIT(i); + bit_pat = pat & BIT(i); + + if (bit_key != 0 && bit_key_inv != 0) + continue; + + if ((bit_key == 0 && bit_key_inv == 0) || bit_key == bit_pat) + return false; + } + + return true; +} + +static bool ice_ternary_match(const u8 *key, const u8 *key_inv, + const u8 *pat, int len) +{ + int i; + + for (i = 0; i < len; i++) + if (!ice_ternary_match_byte(key[i], key_inv[i], pat[i])) + return false; + + return true; +} + /*** ICE_SID_RXPARSER_BOOST_TCAM and ICE_SID_LBL_RXPARSER_TMEM sections ***/ static void ice_bst_np_kb_dump(struct ice_hw *hw, struct ice_np_keybuilder *kb) { @@ -1259,6 +1362,31 @@ static struct ice_lbl_item *ice_bst_lbl_table_get(struct ice_hw *hw) ice_parse_lbl_item, true); } +/** + * ice_bst_tcam_match - match a pattern on the boost tcam table + * @tcam_table: boost tcam table to search + * @pat: pattern to match + * + * Return: a pointer to the matching Boost TCAM item or NULL. + */ +struct ice_bst_tcam_item * +ice_bst_tcam_match(struct ice_bst_tcam_item *tcam_table, u8 *pat) +{ + int i; + + for (i = 0; i < ICE_BST_TCAM_TABLE_SIZE; i++) { + struct ice_bst_tcam_item *item = &tcam_table[i]; + + if (item->hit_idx_grp == 0) + continue; + if (ice_ternary_match(item->key, item->key_inv, pat, + ICE_BST_TCAM_KEY_SIZE)) + return item; + } + + return NULL; +} + /*** ICE_SID_RXPARSER_MARKER_PTYPE section ***/ /** * ice_ptype_mk_tcam_dump - dump an ptype marker tcam info @@ -1312,6 +1440,30 @@ struct ice_ptype_mk_tcam_item *ice_ptype_mk_tcam_table_get(struct ice_hw *hw) ice_parse_ptype_mk_tcam_item, true); } +/** + * ice_ptype_mk_tcam_match - match a pattern on a ptype marker tcam table + * @table: ptype marker tcam table to search + * @pat: pattern to match + * @len: length of the pattern + * + * Return: a pointer to the matching Marker PType item or NULL. + */ +struct ice_ptype_mk_tcam_item * +ice_ptype_mk_tcam_match(struct ice_ptype_mk_tcam_item *table, + u8 *pat, int len) +{ + int i; + + for (i = 0; i < ICE_PTYPE_MK_TCAM_TABLE_SIZE; i++) { + struct ice_ptype_mk_tcam_item *item = &table[i]; + + if (ice_ternary_match(item->key, item->key_inv, pat, len)) + return item; + } + + return NULL; +} + /*** ICE_SID_RXPARSER_MARKER_GRP section ***/ /** * ice_mk_grp_dump - dump an marker group item info @@ -1504,6 +1656,31 @@ static struct ice_flg_rd_item *ice_flg_rd_table_get(struct ice_hw *hw) ice_flg_rd_parse_item, false); } +/** + * ice_flg_redirect - redirect a parser flag to packet flag + * @table: flag redirect table + * @psr_flg: parser flag to redirect + * + * Return: flag or 0 if @psr_flag = 0. + */ +u64 ice_flg_redirect(struct ice_flg_rd_item *table, u64 psr_flg) +{ + u64 flg = 0; + int i; + + for (i = 0; i < ICE_FLG_RDT_SIZE; i++) { + struct ice_flg_rd_item *item = &table[i]; + + if (!item->expose) + continue; + + if (psr_flg & BIT(item->intr_flg_id)) + flg |= BIT(i); + } + + return flg; +} + /*** ICE_SID_XLT_KEY_BUILDER_SW, ICE_SID_XLT_KEY_BUILDER_ACL, * ICE_SID_XLT_KEY_BUILDER_FD and ICE_SID_XLT_KEY_BUILDER_RSS * sections ***/ @@ -1737,6 +1914,37 @@ static struct ice_xlt_kb *ice_xlt_kb_get_rss(struct ice_hw *hw) return ice_xlt_kb_get(hw, ICE_SID_XLT_KEY_BUILDER_RSS); } +#define ICE_XLT_KB_MASK GENMASK_ULL(5, 0) + +/** + * ice_xlt_kb_flag_get - aggregate 64 bits packet flag into 16 bits xlt flag + * @kb: xlt key build + * @pkt_flag: 64 bits packet flag + * + * Return: XLT flag or 0 if @pkt_flag = 0. + */ +u16 ice_xlt_kb_flag_get(struct ice_xlt_kb *kb, u64 pkt_flag) +{ + struct ice_xlt_kb_entry *entry = &kb->entries[0]; + u16 flag = 0; + int i; + + /* check flag 15 */ + if (kb->flag15 & pkt_flag) + flag = BIT(ICE_XLT_KB_FLAG0_14_CNT); + + /* check flag 0 - 14 */ + for (i = 0; i < ICE_XLT_KB_FLAG0_14_CNT; i++) { + /* only check first entry */ + u16 idx = entry->flg0_14_sel[i] & ICE_XLT_KB_MASK; + + if (pkt_flag & BIT(idx)) + flag |= (u16)BIT(i); + } + + return flag; +} + /*** Parser API ***/ /** * ice_parser_create - create a parser instance diff --git a/drivers/net/ethernet/intel/ice/ice_parser.h b/drivers/net/ethernet/intel/ice/ice_parser.h index 6e8054d59644..1d2453890d8e 100644 --- a/drivers/net/ethernet/intel/ice/ice_parser.h +++ b/drivers/net/ethernet/intel/ice/ice_parser.h @@ -197,25 +197,29 @@ struct ice_metainit_item { struct ice_pg_cam_key { bool valid; - u16 node_id; /* Node ID of protocol in parse graph */ - bool flag0; - bool flag1; - bool flag2; - bool flag3; - u8 boost_idx; /* Boost TCAM match index */ - u16 alu_reg; - u32 next_proto; /* next Protocol value */ + struct_group_attr(val, __packed, + u16 node_id; /* Node ID of protocol in parse graph */ + bool flag0; + bool flag1; + bool flag2; + bool flag3; + u8 boost_idx; /* Boost TCAM match index */ + u16 alu_reg; + u32 next_proto; /* next Protocol value (must be last) */ + ); }; struct ice_pg_nm_cam_key { bool valid; - u16 node_id; - bool flag0; - bool flag1; - bool flag2; - bool flag3; - u8 boost_idx; - u16 alu_reg; + struct_group_attr(val, __packed, + u16 node_id; + bool flag0; + bool flag1; + bool flag2; + bool flag3; + u8 boost_idx; + u16 alu_reg; + ); }; struct ice_pg_cam_action { @@ -244,6 +248,12 @@ struct ice_pg_nm_cam_item { struct ice_pg_cam_action action; }; +struct ice_pg_cam_item *ice_pg_cam_match(struct ice_pg_cam_item *table, + int size, struct ice_pg_cam_key *key); +struct ice_pg_nm_cam_item * +ice_pg_nm_cam_match(struct ice_pg_nm_cam_item *table, int size, + struct ice_pg_cam_key *key); + /*** ICE_SID_RXPARSER_BOOST_TCAM and ICE_SID_LBL_RXPARSER_TMEM sections ***/ #define ICE_BST_TCAM_TABLE_SIZE 256 #define ICE_BST_TCAM_KEY_SIZE 20 @@ -269,6 +279,9 @@ struct ice_lbl_item { char label[ICE_LBL_LEN]; }; +struct ice_bst_tcam_item * +ice_bst_tcam_match(struct ice_bst_tcam_item *tcam_table, u8 *pat); + /*** ICE_SID_RXPARSER_MARKER_PTYPE section ***/ #define ICE_PTYPE_MK_TCAM_TABLE_SIZE 1024 #define ICE_PTYPE_MK_TCAM_KEY_SIZE 10 @@ -280,6 +293,9 @@ struct ice_ptype_mk_tcam_item { u8 key_inv[ICE_PTYPE_MK_TCAM_KEY_SIZE]; } __packed; +struct ice_ptype_mk_tcam_item * +ice_ptype_mk_tcam_match(struct ice_ptype_mk_tcam_item *table, + u8 *pat, int len); /*** ICE_SID_RXPARSER_MARKER_GRP section ***/ #define ICE_MK_GRP_TABLE_SIZE 128 #define ICE_MK_COUNT_PER_GRP 8 @@ -308,6 +324,7 @@ struct ice_proto_grp_item { /*** ICE_SID_RXPARSER_FLAG_REDIR section ***/ #define ICE_FLG_RD_TABLE_SIZE 64 +#define ICE_FLG_RDT_SIZE 64 /* Flags Redirection item */ struct ice_flg_rd_item { @@ -316,6 +333,8 @@ struct ice_flg_rd_item { u8 intr_flg_id; /* Internal Flag ID */ }; +u64 ice_flg_redirect(struct ice_flg_rd_item *table, u64 psr_flg); + /*** ICE_SID_XLT_KEY_BUILDER_SW, ICE_SID_XLT_KEY_BUILDER_ACL, * ICE_SID_XLT_KEY_BUILDER_FD and ICE_SID_XLT_KEY_BUILDER_RSS * sections ***/ @@ -341,6 +360,8 @@ struct ice_xlt_kb { struct ice_xlt_kb_entry entries[ICE_XLT_KB_TBL_CNT]; }; +u16 ice_xlt_kb_flag_get(struct ice_xlt_kb *kb, u64 pkt_flag); + /*** Parser API ***/ struct ice_parser { struct ice_hw *hw; /* pointer to the hardware structure */ From patchwork Tue Aug 13 22:22:40 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Tony Nguyen X-Patchwork-Id: 13762627 X-Patchwork-Delegate: kuba@kernel.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.8]) (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 535A01AB522 for ; Tue, 13 Aug 2024 22:22:59 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=192.198.163.8 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587781; cv=none; b=dD5FBL6UEQzU1h1eXY2e1gYW9/l7WM3Jh2dQXnbCu7AtRpM2iqEmF2UD5eHXvgyVqmiJJpuLBO0DrFx55quTnz3jUb6BysRiMBFogacGBN/9a2meWq2rjmonosIz9duv5f6cERgxkd1mGIz5E+72W4V6PAif+zl5MnRz0X0hKLA= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587781; c=relaxed/simple; bh=M4mqbdRBIJ8n7qAvDeyK7zuB+wM+6Uhst4kVJqNucSs=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=bI42KU5JTVrg+65WfZRMwWffDYxdZBBoqfVKP2Ut28S/nN2XXmlidFx5cpTfuRviBrHI1U13WaRTQV5KnIs03+Acs52GUw4mowfl8kr8H3DUlV1wXalyCh3JZe4V9yJB9ABSeBNwGYI1dvfFQucXzgQcGcAJ2Oe4IZKvrjcNmyU= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com; spf=pass smtp.mailfrom=intel.com; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b=j/GeelwV; arc=none smtp.client-ip=192.198.163.8 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=intel.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b="j/GeelwV" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1723587779; x=1755123779; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=M4mqbdRBIJ8n7qAvDeyK7zuB+wM+6Uhst4kVJqNucSs=; b=j/GeelwVKpJA8q4BXnPk/QvG2R0aUH0Iw34iwrZTf8ZlVPME9xVc4g/M MQYY2SsxMtDapuVei6CvDWsAc+5a2Wp5TzwN4mCPNjsVUeUWL9cWsh/Hf c+yrWxm8+ZoUcM3VuyCsfMt9uWgOjjMVO/mx2c9a6ceYDZAwXvIlr0EXC LYOxEgyaA5hKYUcVAAOGqZqB6O7ZHMdvAHC9fz3teuQQjJNmqMn6aKzhp 6jTRdwsxS0ZW2fREECtajiEpmQ5i3EC9l2mugybZOxkf7h8uuTzmgZgTH t+43VKO2lu3oXg1z0b8RRMbtG8fKqiKyOtUagsEqizsPe9ss4xrgbRxBt w==; X-CSE-ConnectionGUID: aI4oOOjWRm+C/3Vg5HjOvw== X-CSE-MsgGUID: rU477yjOQ8CfixHllHO/hQ== X-IronPort-AV: E=McAfee;i="6700,10204,11163"; a="39287094" X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="39287094" Received: from orviesa007.jf.intel.com ([10.64.159.147]) by fmvoesa102.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 13 Aug 2024 15:22:55 -0700 X-CSE-ConnectionGUID: csQsjvvcSAyR2aLL0mvbSg== X-CSE-MsgGUID: b7jUi66zRI2HbNkoxHc87A== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="59381065" Received: from anguy11-upstream.jf.intel.com ([10.166.9.133]) by orviesa007.jf.intel.com with ESMTP; 13 Aug 2024 15:22:54 -0700 From: Tony Nguyen To: davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com, edumazet@google.com, netdev@vger.kernel.org Cc: Junfeng Guo , anthony.l.nguyen@intel.com, ahmed.zaki@intel.com, madhu.chittim@intel.com, horms@kernel.org, hkelam@marvell.com, Marcin Szycik , Qi Zhang , Rafal Romanowski Subject: [PATCH net-next v2 05/13] ice: add parser execution main loop Date: Tue, 13 Aug 2024 15:22:40 -0700 Message-ID: <20240813222249.3708070-6-anthony.l.nguyen@intel.com> X-Mailer: git-send-email 2.42.0 In-Reply-To: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> References: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Patchwork-Delegate: kuba@kernel.org From: Junfeng Guo Implement the core work of the runtime parser via: - ice_parser_rt_execute() - ice_parser_rt_reset() - ice_parser_rt_pkt_buf_set() Reviewed-by: Marcin Szycik Signed-off-by: Qi Zhang Signed-off-by: Junfeng Guo Signed-off-by: Ahmed Zaki Tested-by: Rafal Romanowski Reviewed-by: Simon Horman Signed-off-by: Tony Nguyen --- drivers/net/ethernet/intel/ice/Makefile | 1 + drivers/net/ethernet/intel/ice/ice_parser.c | 41 + drivers/net/ethernet/intel/ice/ice_parser.h | 91 ++ .../net/ethernet/intel/ice/ice_parser_rt.c | 861 ++++++++++++++++++ 4 files changed, 994 insertions(+) create mode 100644 drivers/net/ethernet/intel/ice/ice_parser_rt.c diff --git a/drivers/net/ethernet/intel/ice/Makefile b/drivers/net/ethernet/intel/ice/Makefile index 23fa3f7f36ef..b4f6fa4ba13d 100644 --- a/drivers/net/ethernet/intel/ice/Makefile +++ b/drivers/net/ethernet/intel/ice/Makefile @@ -29,6 +29,7 @@ ice-y := ice_main.o \ ice_flex_pipe.o \ ice_flow.o \ ice_parser.o \ + ice_parser_rt.o \ ice_idc.o \ devlink/devlink.o \ devlink/devlink_port.o \ diff --git a/drivers/net/ethernet/intel/ice/ice_parser.c b/drivers/net/ethernet/intel/ice/ice_parser.c index f44310e94150..e40dc31a5601 100644 --- a/drivers/net/ethernet/intel/ice/ice_parser.c +++ b/drivers/net/ethernet/intel/ice/ice_parser.c @@ -1963,6 +1963,7 @@ struct ice_parser *ice_parser_create(struct ice_hw *hw) return ERR_PTR(-ENOMEM); p->hw = hw; + p->rt.psr = p; p->imem_table = ice_imem_table_get(hw); if (IS_ERR(p->imem_table)) { @@ -2091,3 +2092,43 @@ void ice_parser_destroy(struct ice_parser *psr) kfree(psr); } + +/** + * ice_parser_run - parse on a packet in binary and return the result + * @psr: pointer to a parser instance + * @pkt_buf: packet data + * @pkt_len: packet length + * @rslt: input/output parameter to save parser result. + * + * Return: 0 on success or errno. + */ +int ice_parser_run(struct ice_parser *psr, const u8 *pkt_buf, + int pkt_len, struct ice_parser_result *rslt) +{ + ice_parser_rt_reset(&psr->rt); + ice_parser_rt_pktbuf_set(&psr->rt, pkt_buf, pkt_len); + + return ice_parser_rt_execute(&psr->rt, rslt); +} + +/** + * ice_parser_result_dump - dump a parser result info + * @hw: pointer to the hardware structure + * @rslt: parser result info to dump + */ +void ice_parser_result_dump(struct ice_hw *hw, struct ice_parser_result *rslt) +{ + struct device *dev = ice_hw_to_dev(hw); + int i; + + dev_info(dev, "ptype = %d\n", rslt->ptype); + for (i = 0; i < rslt->po_num; i++) + dev_info(dev, "proto = %d, offset = %d\n", + rslt->po[i].proto_id, rslt->po[i].offset); + + dev_info(dev, "flags_psr = 0x%016llx\n", rslt->flags_psr); + dev_info(dev, "flags_pkt = 0x%016llx\n", rslt->flags_pkt); + dev_info(dev, "flags_sw = 0x%04x\n", rslt->flags_sw); + dev_info(dev, "flags_fd = 0x%04x\n", rslt->flags_fd); + dev_info(dev, "flags_rss = 0x%04x\n", rslt->flags_rss); +} diff --git a/drivers/net/ethernet/intel/ice/ice_parser.h b/drivers/net/ethernet/intel/ice/ice_parser.h index 1d2453890d8e..ef08ccbe4874 100644 --- a/drivers/net/ethernet/intel/ice/ice_parser.h +++ b/drivers/net/ethernet/intel/ice/ice_parser.h @@ -363,6 +363,92 @@ struct ice_xlt_kb { u16 ice_xlt_kb_flag_get(struct ice_xlt_kb *kb, u64 pkt_flag); /*** Parser API ***/ +#define ICE_GPR_HV_IDX 64 +#define ICE_GPR_HV_SIZE 32 +#define ICE_GPR_ERR_IDX 84 +#define ICE_GPR_FLG_IDX 104 +#define ICE_GPR_FLG_SIZE 16 + +#define ICE_GPR_TSR_IDX 108 /* TSR: TCAM Search Register */ +#define ICE_GPR_NN_IDX 109 /* NN: Next Parsing Cycle Node ID */ +#define ICE_GPR_HO_IDX 110 /* HO: Next Parsing Cycle hdr Offset */ +#define ICE_GPR_NP_IDX 111 /* NP: Next Parsing Cycle */ + +#define ICE_PARSER_MAX_PKT_LEN 504 +#define ICE_PARSER_PKT_REV 32 +#define ICE_PARSER_GPR_NUM 128 +#define ICE_PARSER_FLG_NUM 64 +#define ICE_PARSER_ERR_NUM 16 +#define ICE_BST_KEY_SIZE 10 +#define ICE_MARKER_ID_SIZE 9 +#define ICE_MARKER_MAX_SIZE \ + (ICE_MARKER_ID_SIZE * BITS_PER_BYTE - 1) +#define ICE_MARKER_ID_NUM 8 +#define ICE_PO_PAIR_SIZE 256 + +struct ice_gpr_pu { + /* array of flags to indicate if GRP needs to be updated */ + bool gpr_val_upd[ICE_PARSER_GPR_NUM]; + u16 gpr_val[ICE_PARSER_GPR_NUM]; + u64 flg_msk; + u64 flg_val; + u16 err_msk; + u16 err_val; +}; + +enum ice_pg_prio { + ICE_PG_P0 = 0, + ICE_PG_P1 = 1, + ICE_PG_P2 = 2, + ICE_PG_P3 = 3, +}; + +struct ice_parser_rt { + struct ice_parser *psr; + u16 gpr[ICE_PARSER_GPR_NUM]; + u8 pkt_buf[ICE_PARSER_MAX_PKT_LEN + ICE_PARSER_PKT_REV]; + u16 pkt_len; + u16 po; + u8 bst_key[ICE_BST_KEY_SIZE]; + struct ice_pg_cam_key pg_key; + struct ice_alu *alu0; + struct ice_alu *alu1; + struct ice_alu *alu2; + struct ice_pg_cam_action *action; + u8 pg_prio; + struct ice_gpr_pu pu; + u8 markers[ICE_MARKER_ID_SIZE]; + bool protocols[ICE_PO_PAIR_SIZE]; + u16 offsets[ICE_PO_PAIR_SIZE]; +}; + +struct ice_parser_proto_off { + u8 proto_id; /* hardware protocol ID */ + u16 offset; /* offset from the start of the protocol header */ +}; + +#define ICE_PARSER_PROTO_OFF_PAIR_SIZE 16 +#define ICE_PARSER_FLAG_PSR_SIZE 8 + +struct ice_parser_result { + u16 ptype; /* 16 bits hardware PTYPE */ + /* array of protocol and header offset pairs */ + struct ice_parser_proto_off po[ICE_PARSER_PROTO_OFF_PAIR_SIZE]; + int po_num; /* # of protocol-offset pairs must <= 16 */ + u64 flags_psr; /* parser flags */ + u64 flags_pkt; /* packet flags */ + u16 flags_sw; /* key builder flags for SW */ + u16 flags_acl; /* key builder flags for ACL */ + u16 flags_fd; /* key builder flags for FD */ + u16 flags_rss; /* key builder flags for RSS */ +}; + +void ice_parser_rt_reset(struct ice_parser_rt *rt); +void ice_parser_rt_pktbuf_set(struct ice_parser_rt *rt, const u8 *pkt_buf, + int pkt_len); +int ice_parser_rt_execute(struct ice_parser_rt *rt, + struct ice_parser_result *rslt); + struct ice_parser { struct ice_hw *hw; /* pointer to the hardware structure */ @@ -385,8 +471,13 @@ struct ice_parser { struct ice_xlt_kb *xlt_kb_acl; struct ice_xlt_kb *xlt_kb_fd; struct ice_xlt_kb *xlt_kb_rss; + + struct ice_parser_rt rt; }; struct ice_parser *ice_parser_create(struct ice_hw *hw); void ice_parser_destroy(struct ice_parser *psr); +int ice_parser_run(struct ice_parser *psr, const u8 *pkt_buf, + int pkt_len, struct ice_parser_result *rslt); +void ice_parser_result_dump(struct ice_hw *hw, struct ice_parser_result *rslt); #endif /* _ICE_PARSER_H_ */ diff --git a/drivers/net/ethernet/intel/ice/ice_parser_rt.c b/drivers/net/ethernet/intel/ice/ice_parser_rt.c new file mode 100644 index 000000000000..d5bcc266b01e --- /dev/null +++ b/drivers/net/ethernet/intel/ice/ice_parser_rt.c @@ -0,0 +1,861 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Copyright (C) 2024 Intel Corporation */ + +#include "ice_common.h" + +static void ice_rt_tsr_set(struct ice_parser_rt *rt, u16 tsr) +{ + rt->gpr[ICE_GPR_TSR_IDX] = tsr; +} + +static void ice_rt_ho_set(struct ice_parser_rt *rt, u16 ho) +{ + rt->gpr[ICE_GPR_HO_IDX] = ho; + memcpy(&rt->gpr[ICE_GPR_HV_IDX], &rt->pkt_buf[ho], ICE_GPR_HV_SIZE); +} + +static void ice_rt_np_set(struct ice_parser_rt *rt, u16 pc) +{ + rt->gpr[ICE_GPR_NP_IDX] = pc; +} + +static void ice_rt_nn_set(struct ice_parser_rt *rt, u16 node) +{ + rt->gpr[ICE_GPR_NN_IDX] = node; +} + +static void +ice_rt_flag_set(struct ice_parser_rt *rt, unsigned int idx, bool set) +{ + struct ice_hw *hw = rt->psr->hw; + unsigned int word, id; + + word = idx / ICE_GPR_FLG_SIZE; + id = idx % ICE_GPR_FLG_SIZE; + + if (set) { + rt->gpr[ICE_GPR_FLG_IDX + word] |= (u16)BIT(id); + ice_debug(hw, ICE_DBG_PARSER, "Set parser flag %u\n", idx); + } else { + rt->gpr[ICE_GPR_FLG_IDX + word] &= ~(u16)BIT(id); + ice_debug(hw, ICE_DBG_PARSER, "Clear parser flag %u\n", idx); + } +} + +static void ice_rt_gpr_set(struct ice_parser_rt *rt, int idx, u16 val) +{ + struct ice_hw *hw = rt->psr->hw; + + if (idx == ICE_GPR_HO_IDX) + ice_rt_ho_set(rt, val); + else + rt->gpr[idx] = val; + + ice_debug(hw, ICE_DBG_PARSER, "Set GPR %d value %d\n", idx, val); +} + +static void ice_rt_err_set(struct ice_parser_rt *rt, unsigned int idx, bool set) +{ + struct ice_hw *hw = rt->psr->hw; + + if (set) { + rt->gpr[ICE_GPR_ERR_IDX] |= (u16)BIT(idx); + ice_debug(hw, ICE_DBG_PARSER, "Set parser error %u\n", idx); + } else { + rt->gpr[ICE_GPR_ERR_IDX] &= ~(u16)BIT(idx); + ice_debug(hw, ICE_DBG_PARSER, "Reset parser error %u\n", idx); + } +} + +/** + * ice_parser_rt_reset - reset the parser runtime + * @rt: pointer to the parser runtime + */ +void ice_parser_rt_reset(struct ice_parser_rt *rt) +{ + struct ice_parser *psr = rt->psr; + struct ice_metainit_item *mi; + unsigned int i; + + mi = &psr->mi_table[0]; + + memset(rt, 0, sizeof(*rt)); + rt->psr = psr; + + ice_rt_tsr_set(rt, mi->tsr); + ice_rt_ho_set(rt, mi->ho); + ice_rt_np_set(rt, mi->pc); + ice_rt_nn_set(rt, mi->pg_rn); + + for (i = 0; i < ICE_PARSER_FLG_NUM; i++) { + if (mi->flags & BIT(i)) + ice_rt_flag_set(rt, i, true); + } +} + +/** + * ice_parser_rt_pktbuf_set - set a packet into parser runtime + * @rt: pointer to the parser runtime + * @pkt_buf: buffer with packet data + * @pkt_len: packet buffer length + */ +void ice_parser_rt_pktbuf_set(struct ice_parser_rt *rt, const u8 *pkt_buf, + int pkt_len) +{ + int len = min(ICE_PARSER_MAX_PKT_LEN, pkt_len); + u16 ho = rt->gpr[ICE_GPR_HO_IDX]; + + memcpy(rt->pkt_buf, pkt_buf, len); + rt->pkt_len = pkt_len; + + memcpy(&rt->gpr[ICE_GPR_HV_IDX], &rt->pkt_buf[ho], ICE_GPR_HV_SIZE); +} + +static void ice_bst_key_init(struct ice_parser_rt *rt, + struct ice_imem_item *imem) +{ + u8 tsr = (u8)rt->gpr[ICE_GPR_TSR_IDX]; + u16 ho = rt->gpr[ICE_GPR_HO_IDX]; + u8 *key = rt->bst_key; + int idd, i; + + idd = ICE_BST_TCAM_KEY_SIZE - 1; + if (imem->b_kb.tsr_ctrl) + key[idd] = tsr; + else + key[idd] = imem->b_kb.prio; + + idd = ICE_BST_KEY_TCAM_SIZE - 1; + for (i = idd; i >= 0; i--) { + int j; + + j = ho + idd - i; + if (j < ICE_PARSER_MAX_PKT_LEN) + key[i] = rt->pkt_buf[ho + idd - i]; + else + key[i] = 0; + } + + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Generated Boost TCAM Key:\n"); + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "%02X %02X %02X %02X %02X %02X %02X %02X %02X %02X\n", + key[0], key[1], key[2], key[3], key[4], + key[5], key[6], key[7], key[8], key[9]); + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "\n"); +} + +static u16 ice_bit_rev_u16(u16 v, int len) +{ + return bitrev16(v) >> (BITS_PER_TYPE(v) - len); +} + +static u32 ice_bit_rev_u32(u32 v, int len) +{ + return bitrev32(v) >> (BITS_PER_TYPE(v) - len); +} + +static u32 ice_hv_bit_sel(struct ice_parser_rt *rt, int start, int len) +{ + int offset; + u32 buf[2]; + u64 val; + + offset = ICE_GPR_HV_IDX + (start / BITS_PER_TYPE(u16)); + + memcpy(buf, &rt->gpr[offset], sizeof(buf)); + + buf[0] = bitrev8x4(buf[0]); + buf[1] = bitrev8x4(buf[1]); + + val = *(u64 *)buf; + val >>= start % BITS_PER_TYPE(u16); + + return ice_bit_rev_u32(val, len); +} + +static u32 ice_pk_build(struct ice_parser_rt *rt, + struct ice_np_keybuilder *kb) +{ + if (kb->opc == ICE_NPKB_OPC_EXTRACT) + return ice_hv_bit_sel(rt, kb->start_reg0, kb->len_reg1); + else if (kb->opc == ICE_NPKB_OPC_BUILD) + return rt->gpr[kb->start_reg0] | + ((u32)rt->gpr[kb->len_reg1] << BITS_PER_TYPE(u16)); + else if (kb->opc == ICE_NPKB_OPC_BYPASS) + return 0; + + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Unsupported OP Code %u\n", + kb->opc); + return U32_MAX; +} + +static bool ice_flag_get(struct ice_parser_rt *rt, unsigned int index) +{ + int word = index / ICE_GPR_FLG_SIZE; + int id = index % ICE_GPR_FLG_SIZE; + + return !!(rt->gpr[ICE_GPR_FLG_IDX + word] & (u16)BIT(id)); +} + +static int ice_imem_pgk_init(struct ice_parser_rt *rt, + struct ice_imem_item *imem) +{ + memset(&rt->pg_key, 0, sizeof(rt->pg_key)); + rt->pg_key.next_proto = ice_pk_build(rt, &imem->np_kb); + if (rt->pg_key.next_proto == U32_MAX) + return -EINVAL; + + if (imem->pg_kb.flag0_ena) + rt->pg_key.flag0 = ice_flag_get(rt, imem->pg_kb.flag0_idx); + if (imem->pg_kb.flag1_ena) + rt->pg_key.flag1 = ice_flag_get(rt, imem->pg_kb.flag1_idx); + if (imem->pg_kb.flag2_ena) + rt->pg_key.flag2 = ice_flag_get(rt, imem->pg_kb.flag2_idx); + if (imem->pg_kb.flag3_ena) + rt->pg_key.flag3 = ice_flag_get(rt, imem->pg_kb.flag3_idx); + + rt->pg_key.alu_reg = rt->gpr[imem->pg_kb.alu_reg_idx]; + rt->pg_key.node_id = rt->gpr[ICE_GPR_NN_IDX]; + + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Generate Parse Graph Key: node_id(%d), flag0-3(%d,%d,%d,%d), boost_idx(%d), alu_reg(0x%04x), next_proto(0x%08x)\n", + rt->pg_key.node_id, + rt->pg_key.flag0, + rt->pg_key.flag1, + rt->pg_key.flag2, + rt->pg_key.flag3, + rt->pg_key.boost_idx, + rt->pg_key.alu_reg, + rt->pg_key.next_proto); + + return 0; +} + +static void ice_imem_alu0_set(struct ice_parser_rt *rt, + struct ice_imem_item *imem) +{ + rt->alu0 = &imem->alu0; + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Load ALU0 from imem pc %d\n", + imem->idx); +} + +static void ice_imem_alu1_set(struct ice_parser_rt *rt, + struct ice_imem_item *imem) +{ + rt->alu1 = &imem->alu1; + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Load ALU1 from imem pc %d\n", + imem->idx); +} + +static void ice_imem_alu2_set(struct ice_parser_rt *rt, + struct ice_imem_item *imem) +{ + rt->alu2 = &imem->alu2; + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Load ALU2 from imem pc %d\n", + imem->idx); +} + +static void ice_imem_pgp_set(struct ice_parser_rt *rt, + struct ice_imem_item *imem) +{ + rt->pg_prio = imem->pg_prio; + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Load PG priority %d from imem pc %d\n", + rt->pg_prio, imem->idx); +} + +static int ice_bst_pgk_init(struct ice_parser_rt *rt, + struct ice_bst_tcam_item *bst) +{ + memset(&rt->pg_key, 0, sizeof(rt->pg_key)); + rt->pg_key.boost_idx = bst->hit_idx_grp; + rt->pg_key.next_proto = ice_pk_build(rt, &bst->np_kb); + if (rt->pg_key.next_proto == U32_MAX) + return -EINVAL; + + if (bst->pg_kb.flag0_ena) + rt->pg_key.flag0 = ice_flag_get(rt, bst->pg_kb.flag0_idx); + if (bst->pg_kb.flag1_ena) + rt->pg_key.flag1 = ice_flag_get(rt, bst->pg_kb.flag1_idx); + if (bst->pg_kb.flag2_ena) + rt->pg_key.flag2 = ice_flag_get(rt, bst->pg_kb.flag2_idx); + if (bst->pg_kb.flag3_ena) + rt->pg_key.flag3 = ice_flag_get(rt, bst->pg_kb.flag3_idx); + + rt->pg_key.alu_reg = rt->gpr[bst->pg_kb.alu_reg_idx]; + rt->pg_key.node_id = rt->gpr[ICE_GPR_NN_IDX]; + + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Generate Parse Graph Key: node_id(%d), flag0-3(%d,%d,%d,%d), boost_idx(%d), alu_reg(0x%04x), next_proto(0x%08x)\n", + rt->pg_key.node_id, + rt->pg_key.flag0, + rt->pg_key.flag1, + rt->pg_key.flag2, + rt->pg_key.flag3, + rt->pg_key.boost_idx, + rt->pg_key.alu_reg, + rt->pg_key.next_proto); + + return 0; +} + +static void ice_bst_alu0_set(struct ice_parser_rt *rt, + struct ice_bst_tcam_item *bst) +{ + rt->alu0 = &bst->alu0; + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Load ALU0 from boost address %d\n", + bst->addr); +} + +static void ice_bst_alu1_set(struct ice_parser_rt *rt, + struct ice_bst_tcam_item *bst) +{ + rt->alu1 = &bst->alu1; + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Load ALU1 from boost address %d\n", + bst->addr); +} + +static void ice_bst_alu2_set(struct ice_parser_rt *rt, + struct ice_bst_tcam_item *bst) +{ + rt->alu2 = &bst->alu2; + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Load ALU2 from boost address %d\n", + bst->addr); +} + +static void ice_bst_pgp_set(struct ice_parser_rt *rt, + struct ice_bst_tcam_item *bst) +{ + rt->pg_prio = bst->pg_prio; + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Load PG priority %d from boost address %d\n", + rt->pg_prio, bst->addr); +} + +static struct ice_pg_cam_item *ice_rt_pg_cam_match(struct ice_parser_rt *rt) +{ + struct ice_parser *psr = rt->psr; + struct ice_pg_cam_item *item; + + item = ice_pg_cam_match(psr->pg_cam_table, ICE_PG_CAM_TABLE_SIZE, + &rt->pg_key); + if (!item) + item = ice_pg_cam_match(psr->pg_sp_cam_table, + ICE_PG_SP_CAM_TABLE_SIZE, &rt->pg_key); + return item; +} + +static +struct ice_pg_nm_cam_item *ice_rt_pg_nm_cam_match(struct ice_parser_rt *rt) +{ + struct ice_parser *psr = rt->psr; + struct ice_pg_nm_cam_item *item; + + item = ice_pg_nm_cam_match(psr->pg_nm_cam_table, + ICE_PG_NM_CAM_TABLE_SIZE, &rt->pg_key); + + if (!item) + item = ice_pg_nm_cam_match(psr->pg_nm_sp_cam_table, + ICE_PG_NM_SP_CAM_TABLE_SIZE, + &rt->pg_key); + return item; +} + +static void ice_gpr_add(struct ice_parser_rt *rt, int idx, u16 val) +{ + rt->pu.gpr_val_upd[idx] = true; + rt->pu.gpr_val[idx] = val; + + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Pending update for register %d value %d\n", + idx, val); +} + +static void ice_pg_exe(struct ice_parser_rt *rt) +{ + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Executing ParseGraph action ...\n"); + + ice_gpr_add(rt, ICE_GPR_NP_IDX, rt->action->next_pc); + ice_gpr_add(rt, ICE_GPR_NN_IDX, rt->action->next_node); + + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Executing ParseGraph action done.\n"); +} + +static void ice_flg_add(struct ice_parser_rt *rt, int idx, bool val) +{ + rt->pu.flg_msk |= BIT(idx); + if (val) + rt->pu.flg_val |= BIT(idx); + else + rt->pu.flg_val &= ~BIT(idx); + + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Pending update for flag %d value %d\n", + idx, val); +} + +static void ice_flg_update(struct ice_parser_rt *rt, struct ice_alu *alu) +{ + u32 hv_bit_sel; + int i; + + if (!alu->dedicate_flags_ena) + return; + + if (alu->flags_extr_imm) { + for (i = 0; i < alu->dst_len; i++) + ice_flg_add(rt, alu->dst_start + i, + !!(alu->flags_start_imm & BIT(i))); + } else { + for (i = 0; i < alu->dst_len; i++) { + hv_bit_sel = ice_hv_bit_sel(rt, + alu->flags_start_imm + i, + 1); + ice_flg_add(rt, alu->dst_start + i, !!hv_bit_sel); + } + } +} + +static void ice_po_update(struct ice_parser_rt *rt, struct ice_alu *alu) +{ + if (alu->proto_offset_opc == ICE_PO_OFF_HDR_ADD) + rt->po = (u16)(rt->gpr[ICE_GPR_HO_IDX] + alu->proto_offset); + else if (alu->proto_offset_opc == ICE_PO_OFF_HDR_SUB) + rt->po = (u16)(rt->gpr[ICE_GPR_HO_IDX] - alu->proto_offset); + else if (alu->proto_offset_opc == ICE_PO_OFF_REMAIN) + rt->po = rt->gpr[ICE_GPR_HO_IDX]; + + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Update Protocol Offset = %d\n", + rt->po); +} + +static u16 ice_reg_bit_sel(struct ice_parser_rt *rt, int reg_idx, + int start, int len) +{ + int offset; + u32 val; + + offset = ICE_GPR_HV_IDX + (start / BITS_PER_TYPE(u16)); + + memcpy(&val, &rt->gpr[offset], sizeof(val)); + + val = bitrev8x4(val); + val >>= start % BITS_PER_TYPE(u16); + + return ice_bit_rev_u16(val, len); +} + +static void ice_err_add(struct ice_parser_rt *rt, int idx, bool val) +{ + rt->pu.err_msk |= (u16)BIT(idx); + if (val) + rt->pu.flg_val |= (u64)BIT_ULL(idx); + else + rt->pu.flg_val &= ~(u64)BIT_ULL(idx); + + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Pending update for error %d value %d\n", + idx, val); +} + +static void ice_dst_reg_bit_set(struct ice_parser_rt *rt, struct ice_alu *alu, + bool val) +{ + u16 flg_idx; + + if (alu->dedicate_flags_ena) { + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "DedicatedFlagsEnable should not be enabled in opcode %d\n", + alu->opc); + return; + } + + if (alu->dst_reg_id == ICE_GPR_ERR_IDX) { + if (alu->dst_start >= ICE_PARSER_ERR_NUM) { + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Invalid error %d\n", + alu->dst_start); + return; + } + ice_err_add(rt, alu->dst_start, val); + } else if (alu->dst_reg_id >= ICE_GPR_FLG_IDX) { + flg_idx = (u16)(((alu->dst_reg_id - ICE_GPR_FLG_IDX) << 4) + + alu->dst_start); + + if (flg_idx >= ICE_PARSER_FLG_NUM) { + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Invalid flag %d\n", + flg_idx); + return; + } + ice_flg_add(rt, flg_idx, val); + } else { + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Unexpected Dest Register Bit set, RegisterID %d Start %d\n", + alu->dst_reg_id, alu->dst_start); + } +} + +static void ice_alu_exe(struct ice_parser_rt *rt, struct ice_alu *alu) +{ + u16 dst, src, shift, imm; + + if (alu->shift_xlate_sel) { + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "shift_xlate_sel != 0 is not expected\n"); + return; + } + + ice_po_update(rt, alu); + ice_flg_update(rt, alu); + + dst = rt->gpr[alu->dst_reg_id]; + src = ice_reg_bit_sel(rt, alu->src_reg_id, + alu->src_start, alu->src_len); + shift = alu->shift_xlate_key; + imm = alu->imm; + + switch (alu->opc) { + case ICE_ALU_PARK: + break; + case ICE_ALU_MOV_ADD: + dst = (src << shift) + imm; + ice_gpr_add(rt, alu->dst_reg_id, dst); + break; + case ICE_ALU_ADD: + dst += (src << shift) + imm; + ice_gpr_add(rt, alu->dst_reg_id, dst); + break; + case ICE_ALU_ORLT: + if (src < imm) + ice_dst_reg_bit_set(rt, alu, true); + ice_gpr_add(rt, ICE_GPR_NP_IDX, alu->branch_addr); + break; + case ICE_ALU_OREQ: + if (src == imm) + ice_dst_reg_bit_set(rt, alu, true); + ice_gpr_add(rt, ICE_GPR_NP_IDX, alu->branch_addr); + break; + case ICE_ALU_SETEQ: + ice_dst_reg_bit_set(rt, alu, src == imm); + ice_gpr_add(rt, ICE_GPR_NP_IDX, alu->branch_addr); + break; + case ICE_ALU_MOV_XOR: + dst = (src << shift) ^ imm; + ice_gpr_add(rt, alu->dst_reg_id, dst); + break; + default: + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Unsupported ALU instruction %d\n", + alu->opc); + break; + } +} + +static void ice_alu0_exe(struct ice_parser_rt *rt) +{ + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Executing ALU0 ...\n"); + ice_alu_exe(rt, rt->alu0); + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Executing ALU0 done.\n"); +} + +static void ice_alu1_exe(struct ice_parser_rt *rt) +{ + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Executing ALU1 ...\n"); + ice_alu_exe(rt, rt->alu1); + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Executing ALU1 done.\n"); +} + +static void ice_alu2_exe(struct ice_parser_rt *rt) +{ + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Executing ALU2 ...\n"); + ice_alu_exe(rt, rt->alu2); + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Executing ALU2 done.\n"); +} + +static void ice_pu_exe(struct ice_parser_rt *rt) +{ + struct ice_gpr_pu *pu = &rt->pu; + unsigned int i; + + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Updating Registers ...\n"); + + for (i = 0; i < ICE_PARSER_GPR_NUM; i++) { + if (pu->gpr_val_upd[i]) + ice_rt_gpr_set(rt, i, pu->gpr_val[i]); + } + + for (i = 0; i < ICE_PARSER_FLG_NUM; i++) { + if (pu->flg_msk & BIT(i)) + ice_rt_flag_set(rt, i, pu->flg_val & BIT(i)); + } + + for (i = 0; i < ICE_PARSER_ERR_NUM; i++) { + if (pu->err_msk & BIT(i)) + ice_rt_err_set(rt, i, pu->err_val & BIT(i)); + } + + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Updating Registers done.\n"); +} + +static void ice_alu_pg_exe(struct ice_parser_rt *rt) +{ + memset(&rt->pu, 0, sizeof(rt->pu)); + + switch (rt->pg_prio) { + case (ICE_PG_P0): + ice_pg_exe(rt); + ice_alu0_exe(rt); + ice_alu1_exe(rt); + ice_alu2_exe(rt); + break; + case (ICE_PG_P1): + ice_alu0_exe(rt); + ice_pg_exe(rt); + ice_alu1_exe(rt); + ice_alu2_exe(rt); + break; + case (ICE_PG_P2): + ice_alu0_exe(rt); + ice_alu1_exe(rt); + ice_pg_exe(rt); + ice_alu2_exe(rt); + break; + case (ICE_PG_P3): + ice_alu0_exe(rt); + ice_alu1_exe(rt); + ice_alu2_exe(rt); + ice_pg_exe(rt); + break; + } + + ice_pu_exe(rt); + + if (rt->action->ho_inc == 0) + return; + + if (rt->action->ho_polarity) + ice_rt_ho_set(rt, rt->gpr[ICE_GPR_HO_IDX] + rt->action->ho_inc); + else + ice_rt_ho_set(rt, rt->gpr[ICE_GPR_HO_IDX] - rt->action->ho_inc); +} + +static void ice_proto_off_update(struct ice_parser_rt *rt) +{ + struct ice_parser *psr = rt->psr; + + if (rt->action->is_pg) { + struct ice_proto_grp_item *proto_grp = + &psr->proto_grp_table[rt->action->proto_id]; + u16 po; + int i; + + for (i = 0; i < ICE_PROTO_COUNT_PER_GRP; i++) { + struct ice_proto_off *entry = &proto_grp->po[i]; + + if (entry->proto_id == U8_MAX) + break; + + if (!entry->polarity) + po = rt->po + entry->offset; + else + po = rt->po - entry->offset; + + rt->protocols[entry->proto_id] = true; + rt->offsets[entry->proto_id] = po; + + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Set Protocol %d at offset %d\n", + entry->proto_id, po); + } + } else { + rt->protocols[rt->action->proto_id] = true; + rt->offsets[rt->action->proto_id] = rt->po; + + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Set Protocol %d at offset %d\n", + rt->action->proto_id, rt->po); + } +} + +static void ice_marker_set(struct ice_parser_rt *rt, int idx) +{ + unsigned int byte = idx / BITS_PER_BYTE; + unsigned int bit = idx % BITS_PER_BYTE; + + rt->markers[byte] |= (u8)BIT(bit); +} + +static void ice_marker_update(struct ice_parser_rt *rt) +{ + struct ice_parser *psr = rt->psr; + + if (rt->action->is_mg) { + struct ice_mk_grp_item *mk_grp = + &psr->mk_grp_table[rt->action->marker_id]; + int i; + + for (i = 0; i < ICE_MARKER_ID_NUM; i++) { + u8 marker = mk_grp->markers[i]; + + if (marker == ICE_MARKER_MAX_SIZE) + break; + + ice_marker_set(rt, marker); + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Set Marker %d\n", + marker); + } + } else { + if (rt->action->marker_id != ICE_MARKER_MAX_SIZE) + ice_marker_set(rt, rt->action->marker_id); + + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Set Marker %d\n", + rt->action->marker_id); + } +} + +static u16 ice_ptype_resolve(struct ice_parser_rt *rt) +{ + struct ice_ptype_mk_tcam_item *item; + struct ice_parser *psr = rt->psr; + + item = ice_ptype_mk_tcam_match(psr->ptype_mk_tcam_table, + rt->markers, ICE_MARKER_ID_SIZE); + if (item) + return item->ptype; + + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Could not resolve PTYPE\n"); + return U16_MAX; +} + +static void ice_proto_off_resolve(struct ice_parser_rt *rt, + struct ice_parser_result *rslt) +{ + int i; + + for (i = 0; i < ICE_PO_PAIR_SIZE - 1; i++) { + if (rt->protocols[i]) { + rslt->po[rslt->po_num].proto_id = (u8)i; + rslt->po[rslt->po_num].offset = rt->offsets[i]; + rslt->po_num++; + } + } +} + +static void ice_result_resolve(struct ice_parser_rt *rt, + struct ice_parser_result *rslt) +{ + struct ice_parser *psr = rt->psr; + + memset(rslt, 0, sizeof(*rslt)); + + memcpy(&rslt->flags_psr, &rt->gpr[ICE_GPR_FLG_IDX], + ICE_PARSER_FLAG_PSR_SIZE); + rslt->flags_pkt = ice_flg_redirect(psr->flg_rd_table, rslt->flags_psr); + rslt->flags_sw = ice_xlt_kb_flag_get(psr->xlt_kb_sw, rslt->flags_pkt); + rslt->flags_fd = ice_xlt_kb_flag_get(psr->xlt_kb_fd, rslt->flags_pkt); + rslt->flags_rss = ice_xlt_kb_flag_get(psr->xlt_kb_rss, rslt->flags_pkt); + + ice_proto_off_resolve(rt, rslt); + rslt->ptype = ice_ptype_resolve(rt); +} + +/** + * ice_parser_rt_execute - parser execution routine + * @rt: pointer to the parser runtime + * @rslt: input/output parameter to save parser result + * + * Return: 0 on success or errno. + */ +int ice_parser_rt_execute(struct ice_parser_rt *rt, + struct ice_parser_result *rslt) +{ + struct ice_pg_nm_cam_item *pg_nm_cam; + struct ice_parser *psr = rt->psr; + struct ice_pg_cam_item *pg_cam; + int status = 0; + u16 node; + u16 pc; + + node = rt->gpr[ICE_GPR_NN_IDX]; + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Start with Node: %u\n", node); + + while (true) { + struct ice_bst_tcam_item *bst; + struct ice_imem_item *imem; + + pc = rt->gpr[ICE_GPR_NP_IDX]; + imem = &psr->imem_table[pc]; + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Load imem at pc: %u\n", + pc); + + ice_bst_key_init(rt, imem); + bst = ice_bst_tcam_match(psr->bst_tcam_table, rt->bst_key); + if (!bst) { + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "No Boost TCAM Match\n"); + status = ice_imem_pgk_init(rt, imem); + if (status) + break; + ice_imem_alu0_set(rt, imem); + ice_imem_alu1_set(rt, imem); + ice_imem_alu2_set(rt, imem); + ice_imem_pgp_set(rt, imem); + } else { + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Boost TCAM Match address: %u\n", + bst->addr); + if (imem->b_m.pg) { + status = ice_bst_pgk_init(rt, bst); + if (status) + break; + ice_bst_pgp_set(rt, bst); + } else { + status = ice_imem_pgk_init(rt, imem); + if (status) + break; + ice_imem_pgp_set(rt, imem); + } + + if (imem->b_m.alu0) + ice_bst_alu0_set(rt, bst); + else + ice_imem_alu0_set(rt, imem); + + if (imem->b_m.alu1) + ice_bst_alu1_set(rt, bst); + else + ice_imem_alu1_set(rt, imem); + + if (imem->b_m.alu2) + ice_bst_alu2_set(rt, bst); + else + ice_imem_alu2_set(rt, imem); + } + + rt->action = NULL; + pg_cam = ice_rt_pg_cam_match(rt); + if (!pg_cam) { + pg_nm_cam = ice_rt_pg_nm_cam_match(rt); + if (pg_nm_cam) { + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Match ParseGraph Nomatch CAM Address %u\n", + pg_nm_cam->idx); + rt->action = &pg_nm_cam->action; + } + } else { + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Match ParseGraph CAM Address %u\n", + pg_cam->idx); + rt->action = &pg_cam->action; + } + + if (!rt->action) { + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Failed to match ParseGraph CAM, stop parsing.\n"); + status = -EINVAL; + break; + } + + ice_alu_pg_exe(rt); + ice_marker_update(rt); + ice_proto_off_update(rt); + + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Go to node %u\n", + rt->action->next_node); + + if (rt->action->is_last_round) { + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Last Round in ParseGraph Action, stop parsing.\n"); + break; + } + + if (rt->gpr[ICE_GPR_HO_IDX] >= rt->pkt_len) { + ice_debug(rt->psr->hw, ICE_DBG_PARSER, "Header Offset (%u) is larger than packet len (%u), stop parsing\n", + rt->gpr[ICE_GPR_HO_IDX], rt->pkt_len); + break; + } + } + + ice_result_resolve(rt, rslt); + + return status; +} From patchwork Tue Aug 13 22:22:41 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Tony Nguyen X-Patchwork-Id: 13762626 X-Patchwork-Delegate: kuba@kernel.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.8]) (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 EBF311AB52C for ; Tue, 13 Aug 2024 22:22:59 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=192.198.163.8 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587781; cv=none; b=WtiBBszWk5GmxWkJAeakB4IWR+l32G7tzb4MA+/R4EBRoToVl0bRBgzo37QPlaZ4k81CMz42lnrUxRjR3cRvbmlyzySVht1D7YbVcbHSxr/TP2drubH7d0kaKkq2dCF6INk4WTZ34bGYO1SBQI+QVT//1Dsra+sVNh4pkVZkjnQ= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587781; c=relaxed/simple; bh=GmJ7q9kXA6KefAJkZ3kjWEVL7/foG+4VwOE3lEruv1A=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=TJ3Cp/+VjyyVglxt6ZF1Qun4pFkj126T7XpXhJlbEjIALRcP9tB7noKpKrejcig3x+VstCayVNHuSg0vp4lU9tB5//RagtRGEO5pB6OFXADbFBWvobnjOe6otRggXJePGSuzXKKqQXBDVeX7R2/o67rGjufQMo39eU+9I2bziTI= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com; spf=pass smtp.mailfrom=intel.com; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b=PYRLou0d; arc=none smtp.client-ip=192.198.163.8 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=intel.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b="PYRLou0d" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1723587780; x=1755123780; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=GmJ7q9kXA6KefAJkZ3kjWEVL7/foG+4VwOE3lEruv1A=; b=PYRLou0dz92xuxeCDh+4+bfWXnyYsB507NRs9d5JQEQE8eKEq2ia0HV3 aMafwyeX8h/f0mVc6oVTbW8cfrH1/obMn/YQt/uAy3XnCncCAmUaR+6pT KDgC2vPwascPElV/0o4yjQEAiusIReHG8XypvMm1SxT+Kf453ZW1+k1eX LFF1qFRrd+VjObZHLcuV+sAlDo/1KcRp1/PUgXmaczxKVYqFAUpQDBfjb f6flM2ra3lCYALwvnHHC9uND6cN+f2mg6DIqQrXX+Jp2VdMQRU0LT2Nyn CQhIXjP1IyWAoVxW5EIb93NjExIDHBDrWAJJk7achR6+GLixorPM/iTrY g==; X-CSE-ConnectionGUID: d9YN8slVTuub8Zasukg0cA== X-CSE-MsgGUID: XaxUqzDqS+qqSmeEBOfEtQ== X-IronPort-AV: E=McAfee;i="6700,10204,11163"; a="39287101" X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="39287101" Received: from orviesa007.jf.intel.com ([10.64.159.147]) by fmvoesa102.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 13 Aug 2024 15:22:55 -0700 X-CSE-ConnectionGUID: +aBQQLefS9a1zYx7sCxzqA== X-CSE-MsgGUID: hRem7M/mSGGU1EzzGadyTQ== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="59381068" Received: from anguy11-upstream.jf.intel.com ([10.166.9.133]) by orviesa007.jf.intel.com with ESMTP; 13 Aug 2024 15:22:54 -0700 From: Tony Nguyen To: davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com, edumazet@google.com, netdev@vger.kernel.org Cc: Junfeng Guo , anthony.l.nguyen@intel.com, ahmed.zaki@intel.com, madhu.chittim@intel.com, horms@kernel.org, hkelam@marvell.com, Marcin Szycik , Qi Zhang , Rafal Romanowski Subject: [PATCH net-next v2 06/13] ice: support turning on/off the parser's double vlan mode Date: Tue, 13 Aug 2024 15:22:41 -0700 Message-ID: <20240813222249.3708070-7-anthony.l.nguyen@intel.com> X-Mailer: git-send-email 2.42.0 In-Reply-To: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> References: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Patchwork-Delegate: kuba@kernel.org From: Junfeng Guo Add API ice_parser_dvm_set() to support turning on/off the parser's double vlan mode. Reviewed-by: Marcin Szycik Signed-off-by: Qi Zhang Signed-off-by: Junfeng Guo Co-developed-by: Ahmed Zaki Signed-off-by: Ahmed Zaki Tested-by: Rafal Romanowski Reviewed-by: Simon Horman Signed-off-by: Tony Nguyen --- drivers/net/ethernet/intel/ice/ice_parser.c | 78 ++++++++++++++++++++- drivers/net/ethernet/intel/ice/ice_parser.h | 18 +++++ 2 files changed, 93 insertions(+), 3 deletions(-) diff --git a/drivers/net/ethernet/intel/ice/ice_parser.c b/drivers/net/ethernet/intel/ice/ice_parser.c index e40dc31a5601..f032b907ccd2 100644 --- a/drivers/net/ethernet/intel/ice/ice_parser.c +++ b/drivers/net/ethernet/intel/ice/ice_parser.c @@ -597,6 +597,32 @@ static struct ice_metainit_item *ice_metainit_table_get(struct ice_hw *hw) ice_metainit_parse_item, false); } +/** + * ice_bst_tcam_search - find a TCAM item with specific type + * @tcam_table: the TCAM table + * @lbl_table: the lbl table to search + * @type: the type we need to match against + * @start: start searching from this index + * + * Return: a pointer to the matching BOOST TCAM item or NULL. + */ +struct ice_bst_tcam_item * +ice_bst_tcam_search(struct ice_bst_tcam_item *tcam_table, + struct ice_lbl_item *lbl_table, + enum ice_lbl_type type, u16 *start) +{ + u16 i = *start; + + for (; i < ICE_BST_TCAM_TABLE_SIZE; i++) { + if (lbl_table[i].type == type) { + *start = i; + return &tcam_table[lbl_table[i].idx]; + } + } + + return NULL; +} + /*** ICE_SID_RXPARSER_CAM, ICE_SID_RXPARSER_PG_SPILL, * ICE_SID_RXPARSER_NOMATCH_CAM and ICE_SID_RXPARSER_NOMATCH_CAM * sections ***/ @@ -1151,6 +1177,7 @@ static void ice_lbl_dump(struct ice_hw *hw, struct ice_lbl_item *item) struct device *dev = ice_hw_to_dev(hw); dev_info(dev, "index = %u\n", item->idx); + dev_info(dev, "type = %u\n", item->type); dev_info(dev, "label = %s\n", item->label); } @@ -1340,12 +1367,21 @@ static struct ice_bst_tcam_item *ice_bst_tcam_table_get(struct ice_hw *hw) } static void ice_parse_lbl_item(struct ice_hw *hw, u16 idx, void *item, - void *data, int size) + void *data, int __maybe_unused size) { - memcpy(item, data, size); + struct ice_lbl_item *lbl_item = item; + struct ice_lbl_item *lbl_data = data; + + lbl_item->idx = lbl_data->idx; + memcpy(lbl_item->label, lbl_data->label, sizeof(lbl_item->label)); + + if (strstarts(lbl_item->label, ICE_LBL_BST_DVM)) + lbl_item->type = ICE_LBL_BST_TYPE_DVM; + else if (strstarts(lbl_item->label, ICE_LBL_BST_SVM)) + lbl_item->type = ICE_LBL_BST_TYPE_SVM; if (hw->debug_mask & ICE_DBG_PARSER) - ice_lbl_dump(hw, (struct ice_lbl_item *)item); + ice_lbl_dump(hw, lbl_item); } /** @@ -2132,3 +2168,39 @@ void ice_parser_result_dump(struct ice_hw *hw, struct ice_parser_result *rslt) dev_info(dev, "flags_fd = 0x%04x\n", rslt->flags_fd); dev_info(dev, "flags_rss = 0x%04x\n", rslt->flags_rss); } + +#define ICE_BT_VLD_KEY 0xFF +#define ICE_BT_INV_KEY 0xFE + +static void ice_bst_dvm_set(struct ice_parser *psr, enum ice_lbl_type type, + bool on) +{ + u16 i = 0; + + while (true) { + struct ice_bst_tcam_item *item; + u8 key; + + item = ice_bst_tcam_search(psr->bst_tcam_table, + psr->bst_lbl_table, + type, &i); + if (!item) + break; + + key = on ? ICE_BT_VLD_KEY : ICE_BT_INV_KEY; + item->key[ICE_BT_VM_OFF] = key; + item->key_inv[ICE_BT_VM_OFF] = key; + i++; + } +} + +/** + * ice_parser_dvm_set - configure double vlan mode for parser + * @psr: pointer to a parser instance + * @on: true to turn on; false to turn off + */ +void ice_parser_dvm_set(struct ice_parser *psr, bool on) +{ + ice_bst_dvm_set(psr, ICE_LBL_BST_TYPE_DVM, on); + ice_bst_dvm_set(psr, ICE_LBL_BST_TYPE_SVM, !on); +} diff --git a/drivers/net/ethernet/intel/ice/ice_parser.h b/drivers/net/ethernet/intel/ice/ice_parser.h index ef08ccbe4874..92319c60b388 100644 --- a/drivers/net/ethernet/intel/ice/ice_parser.h +++ b/drivers/net/ethernet/intel/ice/ice_parser.h @@ -274,13 +274,29 @@ struct ice_bst_tcam_item { }; #define ICE_LBL_LEN 64 +#define ICE_LBL_BST_DVM "BOOST_MAC_VLAN_DVM" +#define ICE_LBL_BST_SVM "BOOST_MAC_VLAN_SVM" + +enum ice_lbl_type { + ICE_LBL_BST_TYPE_UNKNOWN, + ICE_LBL_BST_TYPE_DVM, + ICE_LBL_BST_TYPE_SVM, +}; + struct ice_lbl_item { u16 idx; char label[ICE_LBL_LEN]; + + /* must be at the end, not part of the DDP section */ + enum ice_lbl_type type; }; struct ice_bst_tcam_item * ice_bst_tcam_match(struct ice_bst_tcam_item *tcam_table, u8 *pat); +struct ice_bst_tcam_item * +ice_bst_tcam_search(struct ice_bst_tcam_item *tcam_table, + struct ice_lbl_item *lbl_table, + enum ice_lbl_type type, u16 *start); /*** ICE_SID_RXPARSER_MARKER_PTYPE section ***/ #define ICE_PTYPE_MK_TCAM_TABLE_SIZE 1024 @@ -429,6 +445,7 @@ struct ice_parser_proto_off { #define ICE_PARSER_PROTO_OFF_PAIR_SIZE 16 #define ICE_PARSER_FLAG_PSR_SIZE 8 +#define ICE_BT_VM_OFF 0 struct ice_parser_result { u16 ptype; /* 16 bits hardware PTYPE */ @@ -477,6 +494,7 @@ struct ice_parser { struct ice_parser *ice_parser_create(struct ice_hw *hw); void ice_parser_destroy(struct ice_parser *psr); +void ice_parser_dvm_set(struct ice_parser *psr, bool on); int ice_parser_run(struct ice_parser *psr, const u8 *pkt_buf, int pkt_len, struct ice_parser_result *rslt); void ice_parser_result_dump(struct ice_hw *hw, struct ice_parser_result *rslt); From patchwork Tue Aug 13 22:22:42 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Tony Nguyen X-Patchwork-Id: 13762628 X-Patchwork-Delegate: kuba@kernel.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.8]) (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 82ADD1AB53A for ; Tue, 13 Aug 2024 22:23:00 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=192.198.163.8 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587782; cv=none; b=BDYmrUElmvanPATRKEAtSE9r2MATBCvJU7oxNJPWw3hQRr+21MKVct2w3b9u1c50tnyP2E2DXhM/E9z3Rvj1KIJ6T26h65Hzi8FEJ5KQQTUs/Qcss6wd5yRTzgCrED9l3VxrxerrH6kKEceBvyHxcY8UO66rFqkwNSTwxidrfuo= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587782; c=relaxed/simple; bh=dyt+1gbCLpb7f6PE/DSbxhmkVo+EG2dGI9Y907+XO0Q=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=XbwPxBnDhWu9tNLmNJU3yyjfYqT4yiKyOlPiCQdqQJZ3U7Z8KfTvQ/4YsLAjfXk3/ROhbL9naCRimFv3db6uayE5I6dsyriOOpgIti/uECSTZi0hGJFcIo1Ek+XvnDPP/cyJSG0LsrLzc9RVkHcu+G9zV6RMrGkOJthZHFTMt28= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com; spf=pass smtp.mailfrom=intel.com; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b=BBeDBW4J; arc=none smtp.client-ip=192.198.163.8 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=intel.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b="BBeDBW4J" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1723587781; x=1755123781; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=dyt+1gbCLpb7f6PE/DSbxhmkVo+EG2dGI9Y907+XO0Q=; b=BBeDBW4JWyOpNATdqEJF8rbneJRk6IoF7GvGwACkcFB5sxKrpiPSeslZ KRUJ1gEEMBD4rmVAjK3fYsqxGi7peB9y/b0axKQwqyE2ZwPPntCP9ar30 VaQWQcInxcLSNa0tTdYEfm93e7xCQ1zj6KfH7d+PQ6EubZtJAUNz7cwEi egKDHSCFnBAmTpGEXSCHOBMwLmRwGxB5jSGOvIIl4lATCsp11nbrYjodR aS8xST3POQD7welG2CXrfVC3YeNfV0jqyCGNZlhiPbJ1DG9WFgnOywAry lfU38crEA8pfeC4iAGkfVCScDj4eERq8B0+i4i9zobqIGoWe0v/vMYEVm w==; X-CSE-ConnectionGUID: htLeTU5kSkuJhsjcW0VK9g== X-CSE-MsgGUID: wcooxVhgT3qUh0fsbC1AmQ== X-IronPort-AV: E=McAfee;i="6700,10204,11163"; a="39287107" X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="39287107" Received: from orviesa007.jf.intel.com ([10.64.159.147]) by fmvoesa102.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 13 Aug 2024 15:22:55 -0700 X-CSE-ConnectionGUID: wChALdaKQECmFpYtB3Ob6w== X-CSE-MsgGUID: p3Nu+PZBRvG1v3AWtF7S8Q== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="59381074" Received: from anguy11-upstream.jf.intel.com ([10.166.9.133]) by orviesa007.jf.intel.com with ESMTP; 13 Aug 2024 15:22:54 -0700 From: Tony Nguyen To: davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com, edumazet@google.com, netdev@vger.kernel.org Cc: Junfeng Guo , anthony.l.nguyen@intel.com, ahmed.zaki@intel.com, madhu.chittim@intel.com, horms@kernel.org, hkelam@marvell.com, Marcin Szycik , Qi Zhang , Rafal Romanowski Subject: [PATCH net-next v2 07/13] ice: add UDP tunnels support to the parser Date: Tue, 13 Aug 2024 15:22:42 -0700 Message-ID: <20240813222249.3708070-8-anthony.l.nguyen@intel.com> X-Mailer: git-send-email 2.42.0 In-Reply-To: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> References: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Patchwork-Delegate: kuba@kernel.org From: Junfeng Guo Add support for the vxlan, geneve, ecpri UDP tunnels through the following APIs: - ice_parser_vxlan_tunnel_set() - ice_parser_geneve_tunnel_set() - ice_parser_ecpri_tunnel_set() Reviewed-by: Simon Horman Reviewed-by: Marcin Szycik Signed-off-by: Qi Zhang Signed-off-by: Junfeng Guo Signed-off-by: Ahmed Zaki Tested-by: Rafal Romanowski Signed-off-by: Tony Nguyen --- drivers/net/ethernet/intel/ice/ice_parser.c | 98 +++++++++++++++++++++ drivers/net/ethernet/intel/ice/ice_parser.h | 13 +++ 2 files changed, 111 insertions(+) diff --git a/drivers/net/ethernet/intel/ice/ice_parser.c b/drivers/net/ethernet/intel/ice/ice_parser.c index f032b907ccd2..ee89208bf21d 100644 --- a/drivers/net/ethernet/intel/ice/ice_parser.c +++ b/drivers/net/ethernet/intel/ice/ice_parser.c @@ -1379,6 +1379,12 @@ static void ice_parse_lbl_item(struct ice_hw *hw, u16 idx, void *item, lbl_item->type = ICE_LBL_BST_TYPE_DVM; else if (strstarts(lbl_item->label, ICE_LBL_BST_SVM)) lbl_item->type = ICE_LBL_BST_TYPE_SVM; + else if (strstarts(lbl_item->label, ICE_LBL_TNL_VXLAN)) + lbl_item->type = ICE_LBL_BST_TYPE_VXLAN; + else if (strstarts(lbl_item->label, ICE_LBL_TNL_GENEVE)) + lbl_item->type = ICE_LBL_BST_TYPE_GENEVE; + else if (strstarts(lbl_item->label, ICE_LBL_TNL_UDP_ECPRI)) + lbl_item->type = ICE_LBL_BST_TYPE_UDP_ECPRI; if (hw->debug_mask & ICE_DBG_PARSER) ice_lbl_dump(hw, lbl_item); @@ -2204,3 +2210,95 @@ void ice_parser_dvm_set(struct ice_parser *psr, bool on) ice_bst_dvm_set(psr, ICE_LBL_BST_TYPE_DVM, on); ice_bst_dvm_set(psr, ICE_LBL_BST_TYPE_SVM, !on); } + +static int ice_tunnel_port_set(struct ice_parser *psr, enum ice_lbl_type type, + u16 udp_port, bool on) +{ + u8 *buf = (u8 *)&udp_port; + u16 i = 0; + + while (true) { + struct ice_bst_tcam_item *item; + + item = ice_bst_tcam_search(psr->bst_tcam_table, + psr->bst_lbl_table, + type, &i); + if (!item) + break; + + /* found empty slot to add */ + if (on && item->key[ICE_BT_TUN_PORT_OFF_H] == ICE_BT_INV_KEY && + item->key_inv[ICE_BT_TUN_PORT_OFF_H] == ICE_BT_INV_KEY) { + item->key_inv[ICE_BT_TUN_PORT_OFF_L] = + buf[ICE_UDP_PORT_OFF_L]; + item->key_inv[ICE_BT_TUN_PORT_OFF_H] = + buf[ICE_UDP_PORT_OFF_H]; + + item->key[ICE_BT_TUN_PORT_OFF_L] = + ICE_BT_VLD_KEY - buf[ICE_UDP_PORT_OFF_L]; + item->key[ICE_BT_TUN_PORT_OFF_H] = + ICE_BT_VLD_KEY - buf[ICE_UDP_PORT_OFF_H]; + + return 0; + /* found a matched slot to delete */ + } else if (!on && + (item->key_inv[ICE_BT_TUN_PORT_OFF_L] == + buf[ICE_UDP_PORT_OFF_L] || + item->key_inv[ICE_BT_TUN_PORT_OFF_H] == + buf[ICE_UDP_PORT_OFF_H])) { + item->key_inv[ICE_BT_TUN_PORT_OFF_L] = ICE_BT_VLD_KEY; + item->key_inv[ICE_BT_TUN_PORT_OFF_H] = ICE_BT_INV_KEY; + + item->key[ICE_BT_TUN_PORT_OFF_L] = ICE_BT_VLD_KEY; + item->key[ICE_BT_TUN_PORT_OFF_H] = ICE_BT_INV_KEY; + + return 0; + } + i++; + } + + return -EINVAL; +} + +/** + * ice_parser_vxlan_tunnel_set - configure vxlan tunnel for parser + * @psr: pointer to a parser instance + * @udp_port: vxlan tunnel port in UDP header + * @on: true to turn on; false to turn off + * + * Return: 0 on success or errno on failure. + */ +int ice_parser_vxlan_tunnel_set(struct ice_parser *psr, + u16 udp_port, bool on) +{ + return ice_tunnel_port_set(psr, ICE_LBL_BST_TYPE_VXLAN, udp_port, on); +} + +/** + * ice_parser_geneve_tunnel_set - configure geneve tunnel for parser + * @psr: pointer to a parser instance + * @udp_port: geneve tunnel port in UDP header + * @on: true to turn on; false to turn off + * + * Return: 0 on success or errno on failure. + */ +int ice_parser_geneve_tunnel_set(struct ice_parser *psr, + u16 udp_port, bool on) +{ + return ice_tunnel_port_set(psr, ICE_LBL_BST_TYPE_GENEVE, udp_port, on); +} + +/** + * ice_parser_ecpri_tunnel_set - configure ecpri tunnel for parser + * @psr: pointer to a parser instance + * @udp_port: ecpri tunnel port in UDP header + * @on: true to turn on; false to turn off + * + * Return: 0 on success or errno on failure. + */ +int ice_parser_ecpri_tunnel_set(struct ice_parser *psr, + u16 udp_port, bool on) +{ + return ice_tunnel_port_set(psr, ICE_LBL_BST_TYPE_UDP_ECPRI, + udp_port, on); +} diff --git a/drivers/net/ethernet/intel/ice/ice_parser.h b/drivers/net/ethernet/intel/ice/ice_parser.h index 92319c60b388..f9500ddf1567 100644 --- a/drivers/net/ethernet/intel/ice/ice_parser.h +++ b/drivers/net/ethernet/intel/ice/ice_parser.h @@ -276,11 +276,17 @@ struct ice_bst_tcam_item { #define ICE_LBL_LEN 64 #define ICE_LBL_BST_DVM "BOOST_MAC_VLAN_DVM" #define ICE_LBL_BST_SVM "BOOST_MAC_VLAN_SVM" +#define ICE_LBL_TNL_VXLAN "TNL_VXLAN" +#define ICE_LBL_TNL_GENEVE "TNL_GENEVE" +#define ICE_LBL_TNL_UDP_ECPRI "TNL_UDP_ECPRI" enum ice_lbl_type { ICE_LBL_BST_TYPE_UNKNOWN, ICE_LBL_BST_TYPE_DVM, ICE_LBL_BST_TYPE_SVM, + ICE_LBL_BST_TYPE_VXLAN, + ICE_LBL_BST_TYPE_GENEVE, + ICE_LBL_BST_TYPE_UDP_ECPRI, }; struct ice_lbl_item { @@ -445,7 +451,11 @@ struct ice_parser_proto_off { #define ICE_PARSER_PROTO_OFF_PAIR_SIZE 16 #define ICE_PARSER_FLAG_PSR_SIZE 8 +#define ICE_BT_TUN_PORT_OFF_H 16 +#define ICE_BT_TUN_PORT_OFF_L 15 #define ICE_BT_VM_OFF 0 +#define ICE_UDP_PORT_OFF_H 1 +#define ICE_UDP_PORT_OFF_L 0 struct ice_parser_result { u16 ptype; /* 16 bits hardware PTYPE */ @@ -495,6 +505,9 @@ struct ice_parser { struct ice_parser *ice_parser_create(struct ice_hw *hw); void ice_parser_destroy(struct ice_parser *psr); void ice_parser_dvm_set(struct ice_parser *psr, bool on); +int ice_parser_vxlan_tunnel_set(struct ice_parser *psr, u16 udp_port, bool on); +int ice_parser_geneve_tunnel_set(struct ice_parser *psr, u16 udp_port, bool on); +int ice_parser_ecpri_tunnel_set(struct ice_parser *psr, u16 udp_port, bool on); int ice_parser_run(struct ice_parser *psr, const u8 *pkt_buf, int pkt_len, struct ice_parser_result *rslt); void ice_parser_result_dump(struct ice_hw *hw, struct ice_parser_result *rslt); From patchwork Tue Aug 13 22:22:43 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Tony Nguyen X-Patchwork-Id: 13762629 X-Patchwork-Delegate: kuba@kernel.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.8]) (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 D47A71ABECC for ; Tue, 13 Aug 2024 22:23:01 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=192.198.163.8 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587783; cv=none; b=CPGAPi06TsJdj9ofKVnrgDaoOfPX8Mkb/tHavlrJsDI3VLnLLOlIL4ME0aU18kTAFfiQRIj4KuJ4/wmMU3zs0QY9M45AiuMO37OJU9dCLwKKf+TDw9fRhOgrxP1lWEjo87HwhnN/KZg10f45F3l8mfSAlUdNJE8kMFnLID8aDRg= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587783; c=relaxed/simple; bh=VK2QhGJX0yR719kO+CwmqZsW81/UyMvX7DPuzT8Z/EE=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=ul5+PbUoVb+UOYZwDfXP8Xxl52GPbhj8tcJiv9kx8WLCvRfE3bSDCW5sTJNgRHvDQRBm8HKzaTgFl5FVhQ3Bye4g5mG6IwtCWUg8BynxmrWXl3TQD5i/D1YUDTmzFlVUOtnA/nBblj4P4jbbV3xj4875k3E2jEv0x/eTygoC4YA= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com; spf=pass smtp.mailfrom=intel.com; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b=IZLAHNQD; arc=none smtp.client-ip=192.198.163.8 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=intel.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b="IZLAHNQD" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1723587782; x=1755123782; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=VK2QhGJX0yR719kO+CwmqZsW81/UyMvX7DPuzT8Z/EE=; b=IZLAHNQDrwzaYlhoNyskM1bCrP1oCVR14+AFGK6CRVZTePIDnhBsvo8L m7Do5FRiNclJhUVeSSuRZtpfsURxo6xKbrDA/7E02Xp3npMTwb+icMLpf 1A5pfhpA1vVxY82FrNvcNCZJrbjthS6luu2Rj3q1FXAQsnT8krMcb/pTh qGUwuDKy9qLyrpaabBp76964ebwCe0vPcSrKAyHPSFl3nDVuumvW7jjia Pm290Earfn+AXw8YQWE4pK8S1shjVk4MapMRvkj2wtpiCjyxOaTlYf59A 1uoNhsy1/iEKwNh8T/RRU+BJyRzMYLyeFADWNsDUQApo05yUd82MUlNWg Q==; X-CSE-ConnectionGUID: G5xjqvMKRvOlPSZ1odKgdw== X-CSE-MsgGUID: 8J8M81YTRVCCuLo4y776WA== X-IronPort-AV: E=McAfee;i="6700,10204,11163"; a="39287115" X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="39287115" Received: from orviesa007.jf.intel.com ([10.64.159.147]) by fmvoesa102.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 13 Aug 2024 15:22:55 -0700 X-CSE-ConnectionGUID: oaNxiuJ2SYWfwdI0f3Rh/g== X-CSE-MsgGUID: k43RzraxTBS5dr1UwC22kQ== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="59381078" Received: from anguy11-upstream.jf.intel.com ([10.166.9.133]) by orviesa007.jf.intel.com with ESMTP; 13 Aug 2024 15:22:55 -0700 From: Tony Nguyen To: davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com, edumazet@google.com, netdev@vger.kernel.org Cc: Junfeng Guo , anthony.l.nguyen@intel.com, ahmed.zaki@intel.com, madhu.chittim@intel.com, horms@kernel.org, hkelam@marvell.com, Marcin Szycik , Qi Zhang , Rafal Romanowski Subject: [PATCH net-next v2 08/13] ice: add API for parser profile initialization Date: Tue, 13 Aug 2024 15:22:43 -0700 Message-ID: <20240813222249.3708070-9-anthony.l.nguyen@intel.com> X-Mailer: git-send-email 2.42.0 In-Reply-To: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> References: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Patchwork-Delegate: kuba@kernel.org From: Junfeng Guo Add API ice_parser_profile_init() to init a parser profile based on a parser result and a mask buffer. The ice_parser_profile struct is used by the low level FXP engine to create HW profile/field vectors. Reviewed-by: Marcin Szycik Signed-off-by: Qi Zhang Signed-off-by: Junfeng Guo Signed-off-by: Ahmed Zaki Tested-by: Rafal Romanowski Reviewed-by: Simon Horman Signed-off-by: Tony Nguyen --- drivers/net/ethernet/intel/ice/ice_parser.c | 130 +++++++++++++++++++- drivers/net/ethernet/intel/ice/ice_parser.h | 26 ++++ 2 files changed, 154 insertions(+), 2 deletions(-) diff --git a/drivers/net/ethernet/intel/ice/ice_parser.c b/drivers/net/ethernet/intel/ice/ice_parser.c index ee89208bf21d..664beb64f557 100644 --- a/drivers/net/ethernet/intel/ice/ice_parser.c +++ b/drivers/net/ethernet/intel/ice/ice_parser.c @@ -2243,9 +2243,9 @@ static int ice_tunnel_port_set(struct ice_parser *psr, enum ice_lbl_type type, /* found a matched slot to delete */ } else if (!on && (item->key_inv[ICE_BT_TUN_PORT_OFF_L] == - buf[ICE_UDP_PORT_OFF_L] || + buf[ICE_UDP_PORT_OFF_L] || item->key_inv[ICE_BT_TUN_PORT_OFF_H] == - buf[ICE_UDP_PORT_OFF_H])) { + buf[ICE_UDP_PORT_OFF_H])) { item->key_inv[ICE_BT_TUN_PORT_OFF_L] = ICE_BT_VLD_KEY; item->key_inv[ICE_BT_TUN_PORT_OFF_H] = ICE_BT_INV_KEY; @@ -2302,3 +2302,129 @@ int ice_parser_ecpri_tunnel_set(struct ice_parser *psr, return ice_tunnel_port_set(psr, ICE_LBL_BST_TYPE_UDP_ECPRI, udp_port, on); } + +/** + * ice_nearest_proto_id - find nearest protocol ID + * @rslt: pointer to a parser result instance + * @offset: a min value for the protocol offset + * @proto_id: the protocol ID (output) + * @proto_off: the protocol offset (output) + * + * From the protocols in @rslt, find the nearest protocol that has offset + * larger than @offset. + * + * Return: if true, the protocol's ID and offset + */ +static bool ice_nearest_proto_id(struct ice_parser_result *rslt, u16 offset, + u8 *proto_id, u16 *proto_off) +{ + u16 dist = U16_MAX; + u8 proto = 0; + int i; + + for (i = 0; i < rslt->po_num; i++) { + if (offset < rslt->po[i].offset) + continue; + if (offset - rslt->po[i].offset < dist) { + proto = rslt->po[i].proto_id; + dist = offset - rslt->po[i].offset; + } + } + + if (dist % 2) + return false; + + *proto_id = proto; + *proto_off = dist; + + return true; +} + +/* default flag mask to cover GTP_EH_PDU, GTP_EH_PDU_LINK and TUN2 + * In future, the flag masks should learn from DDP + */ +#define ICE_KEYBUILD_FLAG_MASK_DEFAULT_SW 0x4002 +#define ICE_KEYBUILD_FLAG_MASK_DEFAULT_ACL 0x0000 +#define ICE_KEYBUILD_FLAG_MASK_DEFAULT_FD 0x6080 +#define ICE_KEYBUILD_FLAG_MASK_DEFAULT_RSS 0x6010 + +/** + * ice_parser_profile_init - initialize a FXP profile based on parser result + * @rslt: a instance of a parser result + * @pkt_buf: packet data buffer + * @msk_buf: packet mask buffer + * @buf_len: packet length + * @blk: FXP pipeline stage + * @prof: input/output parameter to save the profile + * + * Return: 0 on success or errno on failure. + */ +int ice_parser_profile_init(struct ice_parser_result *rslt, + const u8 *pkt_buf, const u8 *msk_buf, + int buf_len, enum ice_block blk, + struct ice_parser_profile *prof) +{ + u8 proto_id = U8_MAX; + u16 proto_off = 0; + u16 off; + + memset(prof, 0, sizeof(*prof)); + set_bit(rslt->ptype, prof->ptypes); + if (blk == ICE_BLK_SW) { + prof->flags = rslt->flags_sw; + prof->flags_msk = ICE_KEYBUILD_FLAG_MASK_DEFAULT_SW; + } else if (blk == ICE_BLK_ACL) { + prof->flags = rslt->flags_acl; + prof->flags_msk = ICE_KEYBUILD_FLAG_MASK_DEFAULT_ACL; + } else if (blk == ICE_BLK_FD) { + prof->flags = rslt->flags_fd; + prof->flags_msk = ICE_KEYBUILD_FLAG_MASK_DEFAULT_FD; + } else if (blk == ICE_BLK_RSS) { + prof->flags = rslt->flags_rss; + prof->flags_msk = ICE_KEYBUILD_FLAG_MASK_DEFAULT_RSS; + } else { + return -EINVAL; + } + + for (off = 0; off < buf_len - 1; off++) { + if (msk_buf[off] == 0 && msk_buf[off + 1] == 0) + continue; + if (!ice_nearest_proto_id(rslt, off, &proto_id, &proto_off)) + continue; + if (prof->fv_num >= ICE_PARSER_FV_MAX) + return -EINVAL; + + prof->fv[prof->fv_num].proto_id = proto_id; + prof->fv[prof->fv_num].offset = proto_off; + prof->fv[prof->fv_num].spec = *(const u16 *)&pkt_buf[off]; + prof->fv[prof->fv_num].msk = *(const u16 *)&msk_buf[off]; + prof->fv_num++; + } + + return 0; +} + +/** + * ice_parser_profile_dump - dump an FXP profile info + * @hw: pointer to the hardware structure + * @prof: profile info to dump + */ +void ice_parser_profile_dump(struct ice_hw *hw, + struct ice_parser_profile *prof) +{ + struct device *dev = ice_hw_to_dev(hw); + u16 i; + + dev_info(dev, "ptypes:\n"); + for (i = 0; i < ICE_FLOW_PTYPE_MAX; i++) + if (test_bit(i, prof->ptypes)) + dev_info(dev, "\t%u\n", i); + + for (i = 0; i < prof->fv_num; i++) + dev_info(dev, "proto = %u, offset = %2u, spec = 0x%04x, mask = 0x%04x\n", + prof->fv[i].proto_id, prof->fv[i].offset, + prof->fv[i].spec, prof->fv[i].msk); + + dev_info(dev, "flags = 0x%04x\n", prof->flags); + dev_info(dev, "flags_msk = 0x%04x\n", prof->flags_msk); +} diff --git a/drivers/net/ethernet/intel/ice/ice_parser.h b/drivers/net/ethernet/intel/ice/ice_parser.h index f9500ddf1567..6509d807627c 100644 --- a/drivers/net/ethernet/intel/ice/ice_parser.h +++ b/drivers/net/ethernet/intel/ice/ice_parser.h @@ -451,6 +451,8 @@ struct ice_parser_proto_off { #define ICE_PARSER_PROTO_OFF_PAIR_SIZE 16 #define ICE_PARSER_FLAG_PSR_SIZE 8 +#define ICE_PARSER_FV_SIZE 48 +#define ICE_PARSER_FV_MAX 24 #define ICE_BT_TUN_PORT_OFF_H 16 #define ICE_BT_TUN_PORT_OFF_L 15 #define ICE_BT_VM_OFF 0 @@ -511,4 +513,28 @@ int ice_parser_ecpri_tunnel_set(struct ice_parser *psr, u16 udp_port, bool on); int ice_parser_run(struct ice_parser *psr, const u8 *pkt_buf, int pkt_len, struct ice_parser_result *rslt); void ice_parser_result_dump(struct ice_hw *hw, struct ice_parser_result *rslt); + +struct ice_parser_fv { + u8 proto_id; /* hardware protocol ID */ + u16 offset; /* offset from the start of the protocol header */ + u16 spec; /* pattern to match */ + u16 msk; /* pattern mask */ +}; + +struct ice_parser_profile { + /* array of field vectors */ + struct ice_parser_fv fv[ICE_PARSER_FV_SIZE]; + int fv_num; /* # of field vectors must <= 48 */ + u16 flags; /* key builder flags */ + u16 flags_msk; /* key builder flag mask */ + + DECLARE_BITMAP(ptypes, ICE_FLOW_PTYPE_MAX); /* PTYPE bitmap */ +}; + +int ice_parser_profile_init(struct ice_parser_result *rslt, + const u8 *pkt_buf, const u8 *msk_buf, + int buf_len, enum ice_block blk, + struct ice_parser_profile *prof); +void ice_parser_profile_dump(struct ice_hw *hw, + struct ice_parser_profile *prof); #endif /* _ICE_PARSER_H_ */ From patchwork Tue Aug 13 22:22:44 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Tony Nguyen X-Patchwork-Id: 13762630 X-Patchwork-Delegate: kuba@kernel.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.8]) (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 DD4581ABECF for ; Tue, 13 Aug 2024 22:23:01 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=192.198.163.8 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587783; cv=none; b=X9yDDTfhX+7Z690JDMSAYN7Heue90pojCwWfcacyKXl/2IkGbwwhxc0ft9EzIjHpqAdCh6xjIZu1CHx1cX1sIdEQkBd2z4P6Mu3In9WouvBZ6tNh8iqgMRs+iaRSg67pXeFjuhkXpFpT2idX2Jgvo+Rcqvo5O2CLzMii7puREk4= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587783; c=relaxed/simple; bh=CcQgzfTQFOOFWMKkzuwilAdxpY11f/Y1xW1+W48wJd8=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=gDipYZ3lYU4Cc360RPiXgI3m6JfND+tJsncbg56A7/GSQDdsbCU4sOVqdX20QgKO4QoAYsL2JHuxuv7GBHW8dE8ejFs3RbEnqr9nox4KWTNfjk8HUvUIK7rbh89Cf7RbbzCrUZ5+w4qE7iNFcH3vceVGoqNYv6yjZc9j7b2aUl8= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com; spf=pass smtp.mailfrom=intel.com; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b=hFmjpJGs; arc=none smtp.client-ip=192.198.163.8 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=intel.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b="hFmjpJGs" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1723587782; x=1755123782; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=CcQgzfTQFOOFWMKkzuwilAdxpY11f/Y1xW1+W48wJd8=; b=hFmjpJGs0Ez8cSF4ViN8Q+X22nwFRMfF4qVWhUHUPMlKLdL/vQijcBg2 FD6qpAlkwUJbVZA7Lh64F9hvVWsPBkCCwlkbZ61hLwTEIkprN+2zBY1Pm jppr4L+lutIHDADwBG59uY2fUxeJ7GXg1t2qWTJheGShcL26CvWN0CfdO MZ2+VWODU6By2kQPWsb3QdyItqYMx3NE/Qf7ZxMIyJYfraGyDX9eY1O1q 08Pf20FuVdVXf6NSkyfFxa5yd8ELoOtcfdSmR89A9Bew0CU2vGn2o2c8b tFCbbtb6jFaIrRmln3i07FPHZ5yDxzJif7i2YA3uDs9xMG47UC4ASvv6k A==; X-CSE-ConnectionGUID: zJ12VPp7T0SnvxYZXPc0QA== X-CSE-MsgGUID: w8GQBEetRPeL4AO64I2k9A== X-IronPort-AV: E=McAfee;i="6700,10204,11163"; a="39287122" X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="39287122" Received: from orviesa007.jf.intel.com ([10.64.159.147]) by fmvoesa102.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 13 Aug 2024 15:22:56 -0700 X-CSE-ConnectionGUID: 7tJgwjDkTW2ggQ2hXmFpaw== X-CSE-MsgGUID: T+K0+24PTwaNLkfDZk+aGg== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="59381082" Received: from anguy11-upstream.jf.intel.com ([10.166.9.133]) by orviesa007.jf.intel.com with ESMTP; 13 Aug 2024 15:22:55 -0700 From: Tony Nguyen To: davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com, edumazet@google.com, netdev@vger.kernel.org Cc: Junfeng Guo , anthony.l.nguyen@intel.com, ahmed.zaki@intel.com, madhu.chittim@intel.com, horms@kernel.org, hkelam@marvell.com, Marcin Szycik , Qi Zhang , Rafal Romanowski Subject: [PATCH net-next v2 09/13] virtchnl: support raw packet in protocol header Date: Tue, 13 Aug 2024 15:22:44 -0700 Message-ID: <20240813222249.3708070-10-anthony.l.nguyen@intel.com> X-Mailer: git-send-email 2.42.0 In-Reply-To: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> References: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Patchwork-Delegate: kuba@kernel.org From: Junfeng Guo The patch extends existing virtchnl_proto_hdrs structure to allow VF to pass a pair of buffers as packet data and mask that describe a match pattern of a filter rule. Then the kernel PF driver is requested to parse the pair of buffer and figure out low level hardware metadata (ptype, profile, field vector.. ) to program the expected FDIR or RSS rules. Reviewed-by: Simon Horman Reviewed-by: Marcin Szycik Signed-off-by: Qi Zhang Signed-off-by: Junfeng Guo Signed-off-by: Ahmed Zaki Tested-by: Rafal Romanowski Signed-off-by: Tony Nguyen --- include/linux/avf/virtchnl.h | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-) diff --git a/include/linux/avf/virtchnl.h b/include/linux/avf/virtchnl.h index 8e177b67e82f..4f78a65e33dc 100644 --- a/include/linux/avf/virtchnl.h +++ b/include/linux/avf/virtchnl.h @@ -1121,6 +1121,7 @@ enum virtchnl_vfr_states { }; #define VIRTCHNL_MAX_NUM_PROTO_HDRS 32 +#define VIRTCHNL_MAX_SIZE_RAW_PACKET 1024 #define PROTO_HDR_SHIFT 5 #define PROTO_HDR_FIELD_START(proto_hdr_type) ((proto_hdr_type) << PROTO_HDR_SHIFT) #define PROTO_HDR_FIELD_MASK ((1UL << PROTO_HDR_SHIFT) - 1) @@ -1266,13 +1267,22 @@ struct virtchnl_proto_hdrs { u8 pad[3]; /** * specify where protocol header start from. + * must be 0 when sending a raw packet request. * 0 - from the outer layer * 1 - from the first inner layer * 2 - from the second inner layer * .... **/ int count; /* the proto layers must < VIRTCHNL_MAX_NUM_PROTO_HDRS */ - struct virtchnl_proto_hdr proto_hdr[VIRTCHNL_MAX_NUM_PROTO_HDRS]; + union { + struct virtchnl_proto_hdr + proto_hdr[VIRTCHNL_MAX_NUM_PROTO_HDRS]; + struct { + u16 pkt_len; + u8 spec[VIRTCHNL_MAX_SIZE_RAW_PACKET]; + u8 mask[VIRTCHNL_MAX_SIZE_RAW_PACKET]; + } raw; + }; }; VIRTCHNL_CHECK_STRUCT_LEN(2312, virtchnl_proto_hdrs); From patchwork Tue Aug 13 22:22:45 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Tony Nguyen X-Patchwork-Id: 13762631 X-Patchwork-Delegate: kuba@kernel.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.8]) (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 4D9361ABEDC for ; Tue, 13 Aug 2024 22:23:02 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=192.198.163.8 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587784; cv=none; b=SMv4j8gM/cJbOQS4b8xEdOzE9WllkZ4Xzhy5sfyVpPcXYfCwGNVqm6G/fqYb3S8I+d129pQv8NoAya4nTyx3Mbd5z53HgLdWCtcM51brO/HZuZPzTor0qbDRSPoc/LXvN9ygoyZc+Gq1kYmuyEoqbbcmOXrbh9r6bC9X/ltaT+Y= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587784; c=relaxed/simple; bh=iQOr94O3Zb5G7nLuSiVpB97At12+UMuNm63e7A7kzgo=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=VVKjUGoFav6QlizDceDA+go/AQvOHhXZB+DgTWCk55ec9RfAyury5IxOS4qje/b8M87WcImXR0cnsETrGZpJGcQeIzGXIFVMWP6qkRJhggTSSWHYELZmNCOX3J7TDDOTdcgbQc9FNlB1Lj6c6tmma5YdxeuJdqayd0+hN1AWXVA= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com; spf=pass smtp.mailfrom=intel.com; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b=lblAKkTE; arc=none smtp.client-ip=192.198.163.8 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=intel.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b="lblAKkTE" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1723587782; x=1755123782; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=iQOr94O3Zb5G7nLuSiVpB97At12+UMuNm63e7A7kzgo=; b=lblAKkTEmlsMp6RWMDcDZ2emp27Hi3fpVH+uD1hpUBBR3zRr7BfOTOxc SNP8YpW1UtJYoRDIOvqLJh5f0iMIcPIvHmyWpfVJ05t8+tZXK24gd3yMn JuNOelQeGQxkdRVPR4vJZf1tUYWxnVu/eeMbm+2/3VNQsffUw1esXCeFt Efb5lyuHHUnVrUlQJXRmurkaZ3tzCHGt8MRHXWcMl9wAss9XdWfEWhSC0 qjCzbuMFZiJI71HGpj0PhUfvU3Kt27IayxpEsrZ2hbrFAG+y8zntw9FXt gEGgk2Ud9TQEYMr+sppOqu7uYaqTgSlmw04cf47jInsJG1c/Ipmjl3JGn Q==; X-CSE-ConnectionGUID: kvEKg8EdQuavoSMssuU2Ww== X-CSE-MsgGUID: 9atyxhffTk6bVc/N0qRwdQ== X-IronPort-AV: E=McAfee;i="6700,10204,11163"; a="39287130" X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="39287130" Received: from orviesa007.jf.intel.com ([10.64.159.147]) by fmvoesa102.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 13 Aug 2024 15:22:56 -0700 X-CSE-ConnectionGUID: 3ll8KUjXTOqmWsUxhEq0/Q== X-CSE-MsgGUID: 0eDQKvaVT6WCHijNNBqWjg== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="59381086" Received: from anguy11-upstream.jf.intel.com ([10.166.9.133]) by orviesa007.jf.intel.com with ESMTP; 13 Aug 2024 15:22:55 -0700 From: Tony Nguyen To: davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com, edumazet@google.com, netdev@vger.kernel.org Cc: Junfeng Guo , anthony.l.nguyen@intel.com, ahmed.zaki@intel.com, madhu.chittim@intel.com, horms@kernel.org, hkelam@marvell.com, Marcin Szycik , Rafal Romanowski Subject: [PATCH net-next v2 10/13] ice: add method to disable FDIR SWAP option Date: Tue, 13 Aug 2024 15:22:45 -0700 Message-ID: <20240813222249.3708070-11-anthony.l.nguyen@intel.com> X-Mailer: git-send-email 2.42.0 In-Reply-To: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> References: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Patchwork-Delegate: kuba@kernel.org From: Junfeng Guo The SWAP Flag in the FDIR Programming Descriptor doesn't work properly, it is always set and cannot be unset (hardware bug). Thus, add a method to effectively disable the FDIR SWAP option by setting the FDSWAP instead of FDINSET registers. Reviewed-by: Marcin Szycik Signed-off-by: Junfeng Guo Signed-off-by: Ahmed Zaki Tested-by: Rafal Romanowski Reviewed-by: Simon Horman Signed-off-by: Tony Nguyen --- .../net/ethernet/intel/ice/ice_flex_pipe.c | 51 ++++++++++++++++++- .../net/ethernet/intel/ice/ice_flex_pipe.h | 4 +- drivers/net/ethernet/intel/ice/ice_flow.c | 2 +- 3 files changed, 53 insertions(+), 4 deletions(-) diff --git a/drivers/net/ethernet/intel/ice/ice_flex_pipe.c b/drivers/net/ethernet/intel/ice/ice_flex_pipe.c index 20d5db88c99f..7f182a6f72aa 100644 --- a/drivers/net/ethernet/intel/ice/ice_flex_pipe.c +++ b/drivers/net/ethernet/intel/ice/ice_flex_pipe.c @@ -2981,6 +2981,50 @@ ice_add_prof_attrib(struct ice_prof_map *prof, u8 ptg, u16 ptype, } /** + * ice_disable_fd_swap - set register appropriately to disable FD SWAP + * @hw: pointer to the HW struct + * @prof_id: profile ID + */ +static void +ice_disable_fd_swap(struct ice_hw *hw, u8 prof_id) +{ + u16 swap_val, fvw_num; + unsigned int i; + + swap_val = ICE_SWAP_VALID; + fvw_num = hw->blk[ICE_BLK_FD].es.fvw / ICE_FDIR_REG_SET_SIZE; + + /* Since the SWAP Flag in the Programming Desc doesn't work, + * here add method to disable the SWAP Option via setting + * certain SWAP and INSET register sets. + */ + for (i = 0; i < fvw_num ; i++) { + u32 raw_swap, raw_in; + unsigned int j; + + raw_swap = 0; + raw_in = 0; + + for (j = 0; j < ICE_FDIR_REG_SET_SIZE; j++) { + raw_swap |= (swap_val++) << (j * BITS_PER_BYTE); + raw_in |= ICE_INSET_DFLT << (j * BITS_PER_BYTE); + } + + /* write the FDIR swap register set */ + wr32(hw, GLQF_FDSWAP(prof_id, i), raw_swap); + + ice_debug(hw, ICE_DBG_INIT, "swap wr(%d, %d): 0x%x = 0x%08x\n", + prof_id, i, GLQF_FDSWAP(prof_id, i), raw_swap); + + /* write the FDIR inset register set */ + wr32(hw, GLQF_FDINSET(prof_id, i), raw_in); + + ice_debug(hw, ICE_DBG_INIT, "inset wr(%d, %d): 0x%x = 0x%08x\n", + prof_id, i, GLQF_FDINSET(prof_id, i), raw_in); + } +} + +/* * ice_add_prof - add profile * @hw: pointer to the HW struct * @blk: hardware block @@ -2991,6 +3035,7 @@ ice_add_prof_attrib(struct ice_prof_map *prof, u8 ptg, u16 ptype, * @es: extraction sequence (length of array is determined by the block) * @masks: mask for extraction sequence * @symm: symmetric setting for RSS profiles + * @fd_swap: enable/disable FDIR paired src/dst fields swap option * * This function registers a profile, which matches a set of PTYPES with a * particular extraction sequence. While the hardware profile is allocated @@ -3000,7 +3045,7 @@ ice_add_prof_attrib(struct ice_prof_map *prof, u8 ptg, u16 ptype, int ice_add_prof(struct ice_hw *hw, enum ice_block blk, u64 id, u8 ptypes[], const struct ice_ptype_attributes *attr, u16 attr_cnt, - struct ice_fv_word *es, u16 *masks, bool symm) + struct ice_fv_word *es, u16 *masks, bool symm, bool fd_swap) { u32 bytes = DIV_ROUND_UP(ICE_FLOW_PTYPE_MAX, BITS_PER_BYTE); DECLARE_BITMAP(ptgs_used, ICE_XLT1_CNT); @@ -3020,7 +3065,7 @@ ice_add_prof(struct ice_hw *hw, enum ice_block blk, u64 id, u8 ptypes[], status = ice_alloc_prof_id(hw, blk, &prof_id); if (status) goto err_ice_add_prof; - if (blk == ICE_BLK_FD) { + if (blk == ICE_BLK_FD && fd_swap) { /* For Flow Director block, the extraction sequence may * need to be altered in the case where there are paired * fields that have no match. This is necessary because @@ -3031,6 +3076,8 @@ ice_add_prof(struct ice_hw *hw, enum ice_block blk, u64 id, u8 ptypes[], status = ice_update_fd_swap(hw, prof_id, es); if (status) goto err_ice_add_prof; + } else if (blk == ICE_BLK_FD) { + ice_disable_fd_swap(hw, prof_id); } status = ice_update_prof_masking(hw, blk, prof_id, masks); if (status) diff --git a/drivers/net/ethernet/intel/ice/ice_flex_pipe.h b/drivers/net/ethernet/intel/ice/ice_flex_pipe.h index b39d7cdc381f..7c66652dadd6 100644 --- a/drivers/net/ethernet/intel/ice/ice_flex_pipe.h +++ b/drivers/net/ethernet/intel/ice/ice_flex_pipe.h @@ -6,6 +6,8 @@ #include "ice_type.h" +#define ICE_FDIR_REG_SET_SIZE 4 + int ice_acquire_change_lock(struct ice_hw *hw, enum ice_aq_res_access_type access); void ice_release_change_lock(struct ice_hw *hw); @@ -42,7 +44,7 @@ bool ice_hw_ptype_ena(struct ice_hw *hw, u16 ptype); int ice_add_prof(struct ice_hw *hw, enum ice_block blk, u64 id, u8 ptypes[], const struct ice_ptype_attributes *attr, u16 attr_cnt, - struct ice_fv_word *es, u16 *masks, bool symm); + struct ice_fv_word *es, u16 *masks, bool symm, bool fd_swap); struct ice_prof_map * ice_search_prof_id(struct ice_hw *hw, enum ice_block blk, u64 id); int diff --git a/drivers/net/ethernet/intel/ice/ice_flow.c b/drivers/net/ethernet/intel/ice/ice_flow.c index fc2b58f56279..79106503194b 100644 --- a/drivers/net/ethernet/intel/ice/ice_flow.c +++ b/drivers/net/ethernet/intel/ice/ice_flow.c @@ -1400,7 +1400,7 @@ ice_flow_add_prof_sync(struct ice_hw *hw, enum ice_block blk, /* Add a HW profile for this flow profile */ status = ice_add_prof(hw, blk, prof_id, (u8 *)params->ptypes, params->attr, params->attr_cnt, params->es, - params->mask, symm); + params->mask, symm, true); if (status) { ice_debug(hw, ICE_DBG_FLOW, "Error adding a HW flow profile\n"); goto out; From patchwork Tue Aug 13 22:22:46 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Tony Nguyen X-Patchwork-Id: 13762633 X-Patchwork-Delegate: kuba@kernel.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.8]) (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 A4E7D1AC439 for ; Tue, 13 Aug 2024 22:23:03 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=192.198.163.8 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587786; cv=none; b=UfYuTqBP0Cw3MFir4njVOPGLBM8+X0XdQorsuudjisfRbaNu2pG3MRgQfFMU/SDwUeh88MG6UlxOTBqWX+/04PXjKJBSbP13qKjvxqG/+AYYYvfBeIH59JeSrJVQ0a9PnjIjH0YVC00o0ID7RXgkUEdhUIYSj5KDs1Msw+6v6fg= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587786; c=relaxed/simple; bh=599h/P+gzQrmg6YW3ZcuY8O4lLqVh8bHMMfrSUEWrzw=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=ciSCDDlzZJT8V3ugfWdo7WTlvqIiBhSSJ8mJnNps9JbmIUeBwkzNH3vYvhGqPM3RLcXoMalfdl+15KnO0zZbHNh1Gz47ILxlXrf3k2wPmz8nli1yX21U9PicWHCofmirqauLJhtqFQ8ehMQl68787/+VBBhArPDQ+SzrzEbNe7U= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com; spf=pass smtp.mailfrom=intel.com; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b=WOqKnBkD; arc=none smtp.client-ip=192.198.163.8 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=intel.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b="WOqKnBkD" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1723587784; x=1755123784; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=599h/P+gzQrmg6YW3ZcuY8O4lLqVh8bHMMfrSUEWrzw=; b=WOqKnBkD8ebCm/GSvA+VA1FmwyFnZN0XkLWpXmvPC6o0GrshJ1mU+Bz8 2CGi46iQDyuXEPtwH4kAR+mtxOVCGU+L8j01zvTfuKFyyHM9upRi2oY8J sPfq5JxgZm9NjcvKjPBqeLTscrR7tvNedCdchTasxEZfoc1hlr9R/OtEM 4pMkXrJbplxTf6D+KlcJtlILK8sWPcLFyBwwO1TmzaQB6hXOWEvLnrHX5 1eZ+bY/VgDz2IiHu5SFQIMBfd8VpabhG1tUP0obELpSDUzD0iS84si8U5 zaGYXpFOAy1g/m57F7V8Kvlb0hkkKzcFCi22uDftlG9vWbZuy4X4hzHNo g==; X-CSE-ConnectionGUID: 6ACG4ugNRtKqlHXiihsHRA== X-CSE-MsgGUID: fmQuW/w8SVSBoxPYdo5uJg== X-IronPort-AV: E=McAfee;i="6700,10204,11163"; a="39287135" X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="39287135" Received: from orviesa007.jf.intel.com ([10.64.159.147]) by fmvoesa102.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 13 Aug 2024 15:22:56 -0700 X-CSE-ConnectionGUID: 6XQDZO8/TielY6qNOq5FeQ== X-CSE-MsgGUID: h05d55LGSIawWDHg6717Dw== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="59381089" Received: from anguy11-upstream.jf.intel.com ([10.166.9.133]) by orviesa007.jf.intel.com with ESMTP; 13 Aug 2024 15:22:55 -0700 From: Tony Nguyen To: davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com, edumazet@google.com, netdev@vger.kernel.org Cc: Junfeng Guo , anthony.l.nguyen@intel.com, ahmed.zaki@intel.com, madhu.chittim@intel.com, horms@kernel.org, hkelam@marvell.com, Marcin Szycik , Rafal Romanowski Subject: [PATCH net-next v2 11/13] ice: enable FDIR filters from raw binary patterns for VFs Date: Tue, 13 Aug 2024 15:22:46 -0700 Message-ID: <20240813222249.3708070-12-anthony.l.nguyen@intel.com> X-Mailer: git-send-email 2.42.0 In-Reply-To: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> References: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Patchwork-Delegate: kuba@kernel.org From: Junfeng Guo Enable VFs to create FDIR filters from raw binary patterns. The corresponding processes for raw flow are added in the Parse / Create / Destroy stages. Reviewed-by: Marcin Szycik Signed-off-by: Junfeng Guo Co-developed-by: Ahmed Zaki Signed-off-by: Ahmed Zaki Tested-by: Rafal Romanowski Reviewed-by: Simon Horman Signed-off-by: Tony Nguyen --- .../net/ethernet/intel/ice/ice_flex_pipe.c | 48 +++ .../net/ethernet/intel/ice/ice_flex_pipe.h | 3 + drivers/net/ethernet/intel/ice/ice_flow.c | 107 +++++ drivers/net/ethernet/intel/ice/ice_flow.h | 5 + drivers/net/ethernet/intel/ice/ice_vf_lib.h | 8 + .../ethernet/intel/ice/ice_virtchnl_fdir.c | 403 +++++++++++++++++- 6 files changed, 566 insertions(+), 8 deletions(-) diff --git a/drivers/net/ethernet/intel/ice/ice_flex_pipe.c b/drivers/net/ethernet/intel/ice/ice_flex_pipe.c index 7f182a6f72aa..ed95072ca6e3 100644 --- a/drivers/net/ethernet/intel/ice/ice_flex_pipe.c +++ b/drivers/net/ethernet/intel/ice/ice_flex_pipe.c @@ -4145,6 +4145,54 @@ ice_add_prof_id_flow(struct ice_hw *hw, enum ice_block blk, u16 vsi, u64 hdl) return status; } +/** + * ice_flow_assoc_fdir_prof - add an FDIR profile for main/ctrl VSI + * @hw: pointer to the HW struct + * @blk: HW block + * @dest_vsi: dest VSI + * @fdir_vsi: fdir programming VSI + * @hdl: profile handle + * + * Update the hardware tables to enable the FDIR profile indicated by @hdl for + * the VSI specified by @dest_vsi. On success, the flow will be enabled. + * + * Return: 0 on success or negative errno on failure. + */ +int +ice_flow_assoc_fdir_prof(struct ice_hw *hw, enum ice_block blk, + u16 dest_vsi, u16 fdir_vsi, u64 hdl) +{ + u16 vsi_num; + int status; + + if (blk != ICE_BLK_FD) + return -EINVAL; + + vsi_num = ice_get_hw_vsi_num(hw, dest_vsi); + status = ice_add_prof_id_flow(hw, blk, vsi_num, hdl); + if (status) { + ice_debug(hw, ICE_DBG_FLOW, "Adding HW profile failed for main VSI flow entry: %d\n", + status); + return status; + } + + vsi_num = ice_get_hw_vsi_num(hw, fdir_vsi); + status = ice_add_prof_id_flow(hw, blk, vsi_num, hdl); + if (status) { + ice_debug(hw, ICE_DBG_FLOW, "Adding HW profile failed for ctrl VSI flow entry: %d\n", + status); + goto err; + } + + return 0; + +err: + vsi_num = ice_get_hw_vsi_num(hw, dest_vsi); + ice_rem_prof_id_flow(hw, blk, vsi_num, hdl); + + return status; +} + /** * ice_rem_prof_from_list - remove a profile from list * @hw: pointer to the HW struct diff --git a/drivers/net/ethernet/intel/ice/ice_flex_pipe.h b/drivers/net/ethernet/intel/ice/ice_flex_pipe.h index 7c66652dadd6..90b9b0993122 100644 --- a/drivers/net/ethernet/intel/ice/ice_flex_pipe.h +++ b/drivers/net/ethernet/intel/ice/ice_flex_pipe.h @@ -51,6 +51,9 @@ int ice_add_prof_id_flow(struct ice_hw *hw, enum ice_block blk, u16 vsi, u64 hdl); int ice_rem_prof_id_flow(struct ice_hw *hw, enum ice_block blk, u16 vsi, u64 hdl); +int +ice_flow_assoc_fdir_prof(struct ice_hw *hw, enum ice_block blk, + u16 dest_vsi, u16 fdir_vsi, u64 hdl); enum ice_ddp_state ice_init_pkg(struct ice_hw *hw, u8 *buff, u32 len); enum ice_ddp_state ice_copy_and_init_pkg(struct ice_hw *hw, const u8 *buf, u32 len); diff --git a/drivers/net/ethernet/intel/ice/ice_flow.c b/drivers/net/ethernet/intel/ice/ice_flow.c index 79106503194b..d97b751052f2 100644 --- a/drivers/net/ethernet/intel/ice/ice_flow.c +++ b/drivers/net/ethernet/intel/ice/ice_flow.c @@ -409,6 +409,29 @@ static const u32 ice_ptypes_gtpc_tid[] = { }; /* Packet types for GTPU */ +static const struct ice_ptype_attributes ice_attr_gtpu_session[] = { + { ICE_MAC_IPV4_GTPU_IPV4_FRAG, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV4_GTPU_IPV4_PAY, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV4_GTPU_IPV4_UDP_PAY, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV4_GTPU_IPV4_TCP, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV4_GTPU_IPV4_ICMP, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV6_GTPU_IPV4_FRAG, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV6_GTPU_IPV4_PAY, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV6_GTPU_IPV4_UDP_PAY, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV6_GTPU_IPV4_TCP, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV6_GTPU_IPV4_ICMP, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV4_GTPU_IPV6_FRAG, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV4_GTPU_IPV6_PAY, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV4_GTPU_IPV6_UDP_PAY, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV4_GTPU_IPV6_TCP, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV4_GTPU_IPV6_ICMPV6, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV6_GTPU_IPV6_FRAG, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV6_GTPU_IPV6_PAY, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV6_GTPU_IPV6_UDP_PAY, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV6_GTPU_IPV6_TCP, ICE_PTYPE_ATTR_GTP_SESSION }, + { ICE_MAC_IPV6_GTPU_IPV6_ICMPV6, ICE_PTYPE_ATTR_GTP_SESSION }, +}; + static const struct ice_ptype_attributes ice_attr_gtpu_eh[] = { { ICE_MAC_IPV4_GTPU_IPV4_FRAG, ICE_PTYPE_ATTR_GTP_PDU_EH }, { ICE_MAC_IPV4_GTPU_IPV4_PAY, ICE_PTYPE_ATTR_GTP_PDU_EH }, @@ -1523,6 +1546,90 @@ ice_flow_disassoc_prof(struct ice_hw *hw, enum ice_block blk, return status; } +#define FLAG_GTP_EH_PDU_LINK BIT_ULL(13) +#define FLAG_GTP_EH_PDU BIT_ULL(14) + +#define HI_BYTE_IN_WORD GENMASK(15, 8) +#define LO_BYTE_IN_WORD GENMASK(7, 0) + +#define FLAG_GTPU_MSK \ + (FLAG_GTP_EH_PDU | FLAG_GTP_EH_PDU_LINK) +#define FLAG_GTPU_UP \ + (FLAG_GTP_EH_PDU | FLAG_GTP_EH_PDU_LINK) +#define FLAG_GTPU_DW FLAG_GTP_EH_PDU + +/** + * ice_flow_set_parser_prof - Set flow profile based on the parsed profile info + * @hw: pointer to the HW struct + * @dest_vsi: dest VSI + * @fdir_vsi: fdir programming VSI + * @prof: stores parsed profile info from raw flow + * @blk: classification blk + * + * Return: 0 on success or negative errno on failure. + */ +int +ice_flow_set_parser_prof(struct ice_hw *hw, u16 dest_vsi, u16 fdir_vsi, + struct ice_parser_profile *prof, enum ice_block blk) +{ + u64 id = find_first_bit(prof->ptypes, ICE_FLOW_PTYPE_MAX); + struct ice_flow_prof_params *params __free(kfree); + u8 fv_words = hw->blk[blk].es.fvw; + int status; + int i, idx; + + params = kzalloc(sizeof(*params), GFP_KERNEL); + if (!params) + return -ENOMEM; + + for (i = 0; i < ICE_MAX_FV_WORDS; i++) { + params->es[i].prot_id = ICE_PROT_INVALID; + params->es[i].off = ICE_FV_OFFSET_INVAL; + } + + for (i = 0; i < prof->fv_num; i++) { + if (hw->blk[blk].es.reverse) + idx = fv_words - i - 1; + else + idx = i; + params->es[idx].prot_id = prof->fv[i].proto_id; + params->es[idx].off = prof->fv[i].offset; + params->mask[idx] = (((prof->fv[i].msk) << BITS_PER_BYTE) & + HI_BYTE_IN_WORD) | + (((prof->fv[i].msk) >> BITS_PER_BYTE) & + LO_BYTE_IN_WORD); + } + + switch (prof->flags) { + case FLAG_GTPU_DW: + params->attr = ice_attr_gtpu_down; + params->attr_cnt = ARRAY_SIZE(ice_attr_gtpu_down); + break; + case FLAG_GTPU_UP: + params->attr = ice_attr_gtpu_up; + params->attr_cnt = ARRAY_SIZE(ice_attr_gtpu_up); + break; + default: + if (prof->flags_msk & FLAG_GTPU_MSK) { + params->attr = ice_attr_gtpu_session; + params->attr_cnt = ARRAY_SIZE(ice_attr_gtpu_session); + } + break; + } + + status = ice_add_prof(hw, blk, id, (u8 *)prof->ptypes, + params->attr, params->attr_cnt, + params->es, params->mask, false, false); + if (status) + return status; + + status = ice_flow_assoc_fdir_prof(hw, blk, dest_vsi, fdir_vsi, id); + if (status) + ice_rem_prof(hw, blk, id); + + return status; +} + /** * ice_flow_add_prof - Add a flow profile for packet segments and matched fields * @hw: pointer to the HW struct diff --git a/drivers/net/ethernet/intel/ice/ice_flow.h b/drivers/net/ethernet/intel/ice/ice_flow.h index 2fd2e0cb483d..6cb7bb879c98 100644 --- a/drivers/net/ethernet/intel/ice/ice_flow.h +++ b/drivers/net/ethernet/intel/ice/ice_flow.h @@ -5,6 +5,7 @@ #define _ICE_FLOW_H_ #include "ice_flex_type.h" +#include "ice_parser.h" #define ICE_FLOW_ENTRY_HANDLE_INVAL 0 #define ICE_FLOW_FLD_OFF_INVAL 0xffff @@ -326,6 +327,7 @@ enum ice_rss_cfg_hdr_type { ICE_RSS_ANY_HEADERS }; +struct ice_vsi; struct ice_rss_hash_cfg { u32 addl_hdrs; /* protocol header fields */ u64 hash_flds; /* hash bit field (ICE_FLOW_HASH_*) to configure */ @@ -445,6 +447,9 @@ ice_flow_add_prof(struct ice_hw *hw, enum ice_block blk, enum ice_flow_dir dir, bool symm, struct ice_flow_prof **prof); int ice_flow_rem_prof(struct ice_hw *hw, enum ice_block blk, u64 prof_id); int +ice_flow_set_parser_prof(struct ice_hw *hw, u16 dest_vsi, u16 fdir_vsi, + struct ice_parser_profile *prof, enum ice_block blk); +int ice_flow_add_entry(struct ice_hw *hw, enum ice_block blk, u64 prof_id, u64 entry_id, u16 vsi, enum ice_flow_priority prio, void *data, u64 *entry_h); diff --git a/drivers/net/ethernet/intel/ice/ice_vf_lib.h b/drivers/net/ethernet/intel/ice/ice_vf_lib.h index fec16919ec19..be4266899690 100644 --- a/drivers/net/ethernet/intel/ice/ice_vf_lib.h +++ b/drivers/net/ethernet/intel/ice/ice_vf_lib.h @@ -12,6 +12,7 @@ #include #include #include "ice_type.h" +#include "ice_flow.h" #include "ice_virtchnl_fdir.h" #include "ice_vsi_vlan_ops.h" @@ -52,6 +53,12 @@ struct ice_mdd_vf_events { u16 last_printed; }; +/* Structure to store fdir fv entry */ +struct ice_fdir_prof_info { + struct ice_parser_profile prof; + u64 fdir_active_cnt; +}; + /* VF operations */ struct ice_vf_ops { enum ice_disq_rst_src reset_type; @@ -91,6 +98,7 @@ struct ice_vf { u16 lan_vsi_idx; /* index into PF struct */ u16 ctrl_vsi_idx; struct ice_vf_fdir fdir; + struct ice_fdir_prof_info fdir_prof_info[ICE_MAX_PTGS]; /* first vector index of this VF in the PF space */ int first_vector_idx; struct ice_sw *vf_sw_id; /* switch ID the VF VSIs connect to */ diff --git a/drivers/net/ethernet/intel/ice/ice_virtchnl_fdir.c b/drivers/net/ethernet/intel/ice/ice_virtchnl_fdir.c index b4feb0927687..14e3f0f89c78 100644 --- a/drivers/net/ethernet/intel/ice/ice_virtchnl_fdir.c +++ b/drivers/net/ethernet/intel/ice/ice_virtchnl_fdir.c @@ -26,6 +26,15 @@ enum ice_fdir_tunnel_type { ICE_FDIR_TUNNEL_TYPE_NONE = 0, ICE_FDIR_TUNNEL_TYPE_GTPU, ICE_FDIR_TUNNEL_TYPE_GTPU_EH, + ICE_FDIR_TUNNEL_TYPE_ECPRI, + ICE_FDIR_TUNNEL_TYPE_GTPU_INNER, + ICE_FDIR_TUNNEL_TYPE_GTPU_EH_INNER, + ICE_FDIR_TUNNEL_TYPE_GRE, + ICE_FDIR_TUNNEL_TYPE_GTPOGRE, + ICE_FDIR_TUNNEL_TYPE_GTPOGRE_INNER, + ICE_FDIR_TUNNEL_TYPE_GRE_INNER, + ICE_FDIR_TUNNEL_TYPE_L2TPV2, + ICE_FDIR_TUNNEL_TYPE_L2TPV2_INNER, }; struct virtchnl_fdir_fltr_conf { @@ -33,6 +42,11 @@ struct virtchnl_fdir_fltr_conf { enum ice_fdir_tunnel_type ttype; u64 inset_flag; u32 flow_id; + + struct ice_parser_profile *prof; + bool parser_ena; + u8 *pkt_buf; + u8 pkt_len; }; struct virtchnl_fdir_inset_map { @@ -786,6 +800,107 @@ ice_vc_fdir_config_input_set(struct ice_vf *vf, struct virtchnl_fdir_add *fltr, return ret; } +/** + * ice_vc_fdir_is_raw_flow - check if FDIR flow is raw (binary) + * @proto: virtchnl protocol headers + * + * Check if the FDIR rule is raw flow (protocol agnostic flow) or not. Note + * that common FDIR rule must have non-zero proto->count. Thus, we choose the + * tunnel_level and count of proto as the indicators. If both tunnel_level and + * count of proto are zero, this FDIR rule will be regarded as raw flow. + * + * Returns: true if headers describe raw flow, false otherwise. + */ +static bool +ice_vc_fdir_is_raw_flow(struct virtchnl_proto_hdrs *proto) +{ + return (proto->tunnel_level == 0 && proto->count == 0); +} + +/** + * ice_vc_fdir_parse_raw - parse a virtchnl raw FDIR rule + * @vf: pointer to the VF info + * @proto: virtchnl protocol headers + * @conf: FDIR configuration for each filter + * + * Parse the virtual channel filter's raw flow and store it in @conf + * + * Return: 0 on success or negative errno on failure. + */ +static int +ice_vc_fdir_parse_raw(struct ice_vf *vf, + struct virtchnl_proto_hdrs *proto, + struct virtchnl_fdir_fltr_conf *conf) +{ + u8 *pkt_buf, *msk_buf __free(kfree); + struct ice_parser_result rslt; + struct ice_pf *pf = vf->pf; + struct ice_parser *psr; + int status = -ENOMEM; + struct ice_hw *hw; + u16 udp_port = 0; + + pkt_buf = kzalloc(proto->raw.pkt_len, GFP_KERNEL); + msk_buf = kzalloc(proto->raw.pkt_len, GFP_KERNEL); + if (!pkt_buf || !msk_buf) + goto err_mem_alloc; + + memcpy(pkt_buf, proto->raw.spec, proto->raw.pkt_len); + memcpy(msk_buf, proto->raw.mask, proto->raw.pkt_len); + + hw = &pf->hw; + + /* Get raw profile info via Parser Lib */ + psr = ice_parser_create(hw); + if (IS_ERR(psr)) { + status = PTR_ERR(psr); + goto err_mem_alloc; + } + + ice_parser_dvm_set(psr, ice_is_dvm_ena(hw)); + + if (ice_get_open_tunnel_port(hw, &udp_port, TNL_VXLAN)) + ice_parser_vxlan_tunnel_set(psr, udp_port, true); + + status = ice_parser_run(psr, pkt_buf, proto->raw.pkt_len, &rslt); + if (status) + goto err_parser_destroy; + + if (hw->debug_mask & ICE_DBG_PARSER) + ice_parser_result_dump(hw, &rslt); + + conf->prof = kzalloc(sizeof(*conf->prof), GFP_KERNEL); + if (!conf->prof) { + status = -ENOMEM; + goto err_parser_destroy; + } + + status = ice_parser_profile_init(&rslt, pkt_buf, msk_buf, + proto->raw.pkt_len, ICE_BLK_FD, + conf->prof); + if (status) + goto err_parser_profile_init; + + if (hw->debug_mask & ICE_DBG_PARSER) + ice_parser_profile_dump(hw, conf->prof); + + /* Store raw flow info into @conf */ + conf->pkt_len = proto->raw.pkt_len; + conf->pkt_buf = pkt_buf; + conf->parser_ena = true; + + ice_parser_destroy(psr); + return 0; + +err_parser_profile_init: + kfree(conf->prof); +err_parser_destroy: + ice_parser_destroy(psr); +err_mem_alloc: + kfree(pkt_buf); + return status; +} + /** * ice_vc_fdir_parse_pattern * @vf: pointer to the VF info @@ -813,6 +928,10 @@ ice_vc_fdir_parse_pattern(struct ice_vf *vf, struct virtchnl_fdir_add *fltr, return -EINVAL; } + /* For raw FDIR filters created by the parser */ + if (ice_vc_fdir_is_raw_flow(proto)) + return ice_vc_fdir_parse_raw(vf, proto, conf); + for (i = 0; i < proto->count; i++) { struct virtchnl_proto_hdr *hdr = &proto->proto_hdr[i]; struct ip_esp_hdr *esph; @@ -1101,8 +1220,10 @@ ice_vc_validate_fdir_fltr(struct ice_vf *vf, struct virtchnl_fdir_add *fltr, struct virtchnl_proto_hdrs *proto = &fltr->rule_cfg.proto_hdrs; int ret; - if (!ice_vc_validate_pattern(vf, proto)) - return -EINVAL; + /* For raw FDIR filters created by the parser */ + if (!ice_vc_fdir_is_raw_flow(proto)) + if (!ice_vc_validate_pattern(vf, proto)) + return -EINVAL; ret = ice_vc_fdir_parse_pattern(vf, fltr, conf); if (ret) @@ -1295,11 +1416,15 @@ static int ice_vc_fdir_write_fltr(struct ice_vf *vf, return -ENOMEM; ice_fdir_get_prgm_desc(hw, input, &desc, add); - ret = ice_fdir_get_gen_prgm_pkt(hw, input, pkt, false, is_tun); - if (ret) { - dev_dbg(dev, "Gen training pkt for VF %d ptype %d failed\n", - vf->vf_id, input->flow_type); - goto err_free_pkt; + if (conf->parser_ena) { + memcpy(pkt, conf->pkt_buf, conf->pkt_len); + } else { + ret = ice_fdir_get_gen_prgm_pkt(hw, input, pkt, false, is_tun); + if (ret) { + dev_dbg(dev, "Gen training pkt for VF %d ptype %d failed\n", + vf->vf_id, input->flow_type); + goto err_free_pkt; + } } ret = ice_prgm_fdir_fltr(ctrl_vsi, &desc, pkt); @@ -1521,6 +1646,16 @@ ice_vf_verify_rx_desc(struct ice_vf *vf, struct ice_vf_fdir_ctx *ctx, return ret; } +static int ice_fdir_is_tunnel(enum ice_fdir_tunnel_type ttype) +{ + return (ttype == ICE_FDIR_TUNNEL_TYPE_GRE_INNER || + ttype == ICE_FDIR_TUNNEL_TYPE_GTPU_INNER || + ttype == ICE_FDIR_TUNNEL_TYPE_GTPU_EH_INNER || + ttype == ICE_FDIR_TUNNEL_TYPE_GTPOGRE_INNER || + ttype == ICE_FDIR_TUNNEL_TYPE_ECPRI || + ttype == ICE_FDIR_TUNNEL_TYPE_L2TPV2_INNER); +} + /** * ice_vc_add_fdir_fltr_post * @vf: pointer to the VF structure @@ -1781,6 +1916,158 @@ static void ice_vc_fdir_clear_irq_ctx(struct ice_vf *vf) spin_unlock_irqrestore(&vf->fdir.ctx_lock, flags); } +/** + * ice_vc_parser_fv_check_diff - check two parsed FDIR profile fv context + * @fv_a: struct of parsed FDIR profile field vector + * @fv_b: struct of parsed FDIR profile field vector + * + * Check if the two parsed FDIR profile field vector context are different, + * including proto_id, offset and mask. + * + * Return: true on different, false on otherwise. + */ +static bool ice_vc_parser_fv_check_diff(struct ice_parser_fv *fv_a, + struct ice_parser_fv *fv_b) +{ + return (fv_a->proto_id != fv_b->proto_id || + fv_a->offset != fv_b->offset || + fv_a->msk != fv_b->msk); +} + +/** + * ice_vc_parser_fv_save - save parsed FDIR profile fv context + * @fv: struct of parsed FDIR profile field vector + * @fv_src: parsed FDIR profile field vector context to save + * + * Save the parsed FDIR profile field vector context, including proto_id, + * offset and mask. + * + * Return: Void. + */ +static void ice_vc_parser_fv_save(struct ice_parser_fv *fv, + struct ice_parser_fv *fv_src) +{ + fv->proto_id = fv_src->proto_id; + fv->offset = fv_src->offset; + fv->msk = fv_src->msk; + fv->spec = 0; +} + +/** + * ice_vc_add_fdir_raw - add a raw FDIR filter for VF + * @vf: pointer to the VF info + * @conf: FDIR configuration for each filter + * @v_ret: the final VIRTCHNL code + * @stat: pointer to the VIRTCHNL_OP_ADD_FDIR_FILTER + * @len: length of the stat + * + * Return: 0 on success or negative errno on failure. + */ +static int +ice_vc_add_fdir_raw(struct ice_vf *vf, + struct virtchnl_fdir_fltr_conf *conf, + enum virtchnl_status_code *v_ret, + struct virtchnl_fdir_add *stat, int len) +{ + struct ice_vsi *vf_vsi, *ctrl_vsi; + struct ice_fdir_prof_info *pi; + struct ice_pf *pf = vf->pf; + int ret, ptg, id, i; + struct device *dev; + struct ice_hw *hw; + bool fv_found; + + dev = ice_pf_to_dev(pf); + hw = &pf->hw; + *v_ret = VIRTCHNL_STATUS_ERR_PARAM; + stat->status = VIRTCHNL_FDIR_FAILURE_RULE_NORESOURCE; + + id = find_first_bit(conf->prof->ptypes, ICE_FLOW_PTYPE_MAX); + ptg = hw->blk[ICE_BLK_FD].xlt1.t[id]; + + vf_vsi = ice_get_vf_vsi(vf); + if (!vf_vsi) { + dev_err(dev, "Can not get FDIR vf_vsi for VF %d\n", vf->vf_id); + return -ENODEV; + } + + ctrl_vsi = pf->vsi[vf->ctrl_vsi_idx]; + if (!ctrl_vsi) { + dev_err(dev, "Can not get FDIR ctrl_vsi for VF %d\n", + vf->vf_id); + return -ENODEV; + } + + fv_found = false; + + /* Check if profile info already exists, then update the counter */ + pi = &vf->fdir_prof_info[ptg]; + if (pi->fdir_active_cnt != 0) { + for (i = 0; i < ICE_MAX_FV_WORDS; i++) + if (ice_vc_parser_fv_check_diff(&pi->prof.fv[i], + &conf->prof->fv[i])) + break; + if (i == ICE_MAX_FV_WORDS) { + fv_found = true; + pi->fdir_active_cnt++; + } + } + + /* HW profile setting is only required for the first time */ + if (!fv_found) { + ret = ice_flow_set_parser_prof(hw, vf_vsi->idx, + ctrl_vsi->idx, conf->prof, + ICE_BLK_FD); + + if (ret) { + *v_ret = VIRTCHNL_STATUS_ERR_NO_MEMORY; + dev_dbg(dev, "VF %d: insert hw prof failed\n", + vf->vf_id); + return ret; + } + } + + ret = ice_vc_fdir_insert_entry(vf, conf, &conf->flow_id); + if (ret) { + *v_ret = VIRTCHNL_STATUS_ERR_NO_MEMORY; + dev_dbg(dev, "VF %d: insert FDIR list failed\n", + vf->vf_id); + return ret; + } + + ret = ice_vc_fdir_set_irq_ctx(vf, conf, + VIRTCHNL_OP_ADD_FDIR_FILTER); + if (ret) { + dev_dbg(dev, "VF %d: set FDIR context failed\n", + vf->vf_id); + goto err_rem_entry; + } + + ret = ice_vc_fdir_write_fltr(vf, conf, true, false); + if (ret) { + dev_err(dev, "VF %d: adding FDIR raw flow rule failed, ret:%d\n", + vf->vf_id, ret); + goto err_clr_irq; + } + + /* Save parsed profile fv info of the FDIR rule for the first time */ + if (!fv_found) { + for (i = 0; i < conf->prof->fv_num; i++) + ice_vc_parser_fv_save(&pi->prof.fv[i], + &conf->prof->fv[i]); + pi->prof.fv_num = conf->prof->fv_num; + pi->fdir_active_cnt = 1; + } + + return 0; + +err_clr_irq: + ice_vc_fdir_clear_irq_ctx(vf); +err_rem_entry: + ice_vc_fdir_remove_entry(vf, conf, conf->flow_id); + return ret; +} + /** * ice_vc_add_fdir_fltr - add a FDIR filter for VF by the msg buffer * @vf: pointer to the VF info @@ -1846,7 +2133,7 @@ int ice_vc_add_fdir_fltr(struct ice_vf *vf, u8 *msg) len = sizeof(*stat); ret = ice_vc_validate_fdir_fltr(vf, fltr, conf); if (ret) { - v_ret = VIRTCHNL_STATUS_SUCCESS; + v_ret = VIRTCHNL_STATUS_ERR_PARAM; stat->status = VIRTCHNL_FDIR_FAILURE_RULE_INVALID; dev_dbg(dev, "Invalid FDIR filter from VF %d\n", vf->vf_id); goto err_free_conf; @@ -1861,6 +2148,15 @@ int ice_vc_add_fdir_fltr(struct ice_vf *vf, u8 *msg) goto exit; } + /* For raw FDIR filters created by the parser */ + if (conf->parser_ena) { + ret = ice_vc_add_fdir_raw(vf, conf, &v_ret, stat, len); + if (ret) + goto err_free_conf; + goto exit; + } + + is_tun = ice_fdir_is_tunnel(conf->ttype); ret = ice_vc_fdir_config_input_set(vf, fltr, conf, is_tun); if (ret) { v_ret = VIRTCHNL_STATUS_SUCCESS; @@ -1921,6 +2217,78 @@ int ice_vc_add_fdir_fltr(struct ice_vf *vf, u8 *msg) return ret; } +/** + * ice_vc_del_fdir_raw - delete a raw FDIR filter for VF + * @vf: pointer to the VF info + * @conf: FDIR configuration for each filter + * @v_ret: the final VIRTCHNL code + * @stat: pointer to the VIRTCHNL_OP_DEL_FDIR_FILTER + * @len: length of the stat + * + * Return: 0 on success or negative errno on failure. + */ +static int +ice_vc_del_fdir_raw(struct ice_vf *vf, + struct virtchnl_fdir_fltr_conf *conf, + enum virtchnl_status_code *v_ret, + struct virtchnl_fdir_del *stat, int len) +{ + struct ice_vsi *vf_vsi, *ctrl_vsi; + enum ice_block blk = ICE_BLK_FD; + struct ice_fdir_prof_info *pi; + struct ice_pf *pf = vf->pf; + struct device *dev; + struct ice_hw *hw; + unsigned long id; + u16 vsi_num; + int ptg; + int ret; + + dev = ice_pf_to_dev(pf); + hw = &pf->hw; + *v_ret = VIRTCHNL_STATUS_ERR_PARAM; + stat->status = VIRTCHNL_FDIR_FAILURE_RULE_NORESOURCE; + + id = find_first_bit(conf->prof->ptypes, ICE_FLOW_PTYPE_MAX); + ptg = hw->blk[ICE_BLK_FD].xlt1.t[id]; + + ret = ice_vc_fdir_write_fltr(vf, conf, false, false); + if (ret) { + dev_err(dev, "VF %u: deleting FDIR raw flow rule failed: %d\n", + vf->vf_id, ret); + return ret; + } + + vf_vsi = ice_get_vf_vsi(vf); + if (!vf_vsi) { + dev_err(dev, "Can not get FDIR vf_vsi for VF %u\n", vf->vf_id); + return -ENODEV; + } + + ctrl_vsi = pf->vsi[vf->ctrl_vsi_idx]; + if (!ctrl_vsi) { + dev_err(dev, "Can not get FDIR ctrl_vsi for VF %u\n", + vf->vf_id); + return -ENODEV; + } + + pi = &vf->fdir_prof_info[ptg]; + if (pi->fdir_active_cnt != 0) { + pi->fdir_active_cnt--; + /* Remove the profile id flow if no active FDIR rule left */ + if (!pi->fdir_active_cnt) { + vsi_num = ice_get_hw_vsi_num(hw, ctrl_vsi->idx); + ice_rem_prof_id_flow(hw, blk, vsi_num, id); + + vsi_num = ice_get_hw_vsi_num(hw, vf_vsi->idx); + ice_rem_prof_id_flow(hw, blk, vsi_num, id); + } + } + + conf->parser_ena = false; + return 0; +} + /** * ice_vc_del_fdir_fltr - delete a FDIR filter for VF by the msg buffer * @vf: pointer to the VF info @@ -1933,7 +2301,10 @@ int ice_vc_del_fdir_fltr(struct ice_vf *vf, u8 *msg) struct virtchnl_fdir_del *fltr = (struct virtchnl_fdir_del *)msg; struct virtchnl_fdir_del *stat = NULL; struct virtchnl_fdir_fltr_conf *conf; + struct ice_vf_fdir *fdir = &vf->fdir; enum virtchnl_status_code v_ret; + struct ice_fdir_fltr *input; + enum ice_fltr_ptype flow; struct device *dev; struct ice_pf *pf; int is_tun = 0; @@ -1983,6 +2354,15 @@ int ice_vc_del_fdir_fltr(struct ice_vf *vf, u8 *msg) goto err_exit; } + /* For raw FDIR filters created by the parser */ + if (conf->parser_ena) { + ret = ice_vc_del_fdir_raw(vf, conf, &v_ret, stat, len); + if (ret) + goto err_del_tmr; + goto exit; + } + + is_tun = ice_fdir_is_tunnel(conf->ttype); ret = ice_vc_fdir_write_fltr(vf, conf, false, is_tun); if (ret) { v_ret = VIRTCHNL_STATUS_SUCCESS; @@ -1992,6 +2372,13 @@ int ice_vc_del_fdir_fltr(struct ice_vf *vf, u8 *msg) goto err_del_tmr; } + /* Remove unused profiles to avoid unexpected behaviors */ + input = &conf->input; + flow = input->flow_type; + if (fdir->fdir_fltr_cnt[flow][is_tun] == 1) + ice_vc_fdir_rem_prof(vf, flow, is_tun); + +exit: kfree(stat); return ret; From patchwork Tue Aug 13 22:22:47 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Tony Nguyen X-Patchwork-Id: 13762632 X-Patchwork-Delegate: kuba@kernel.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.8]) (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 EFCF21AC43E for ; Tue, 13 Aug 2024 22:23:03 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=192.198.163.8 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587785; cv=none; b=fEWKaZy+n5DTekNFu0TZRz4CNzrCPX1/PaxAK0S8yxkHIKcAe+bbDnWgvCTzSJzRoMNA4yWxFhKqYjF2hQjJ/WZefo5VyNIn9cdJAqMKuZo++6fafqt8Clpuz93Sf0lc5OdxqO4Am1jDOrmRCzWfHQ9NEv7+xzMyZPabPafYhnk= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587785; c=relaxed/simple; bh=tbYV53tO1ExEVxJSZAHIvE3/sh+rRMVQRouoky16wH8=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=sHsQ4XMMATpWi/B9H3/i3CAtWfRXo5ymnsJgvlmiQG2jO3oJKwtL/vW5ENK36DEHERND4zXR/7gQ7/0yroksCOe2vY/FKYNK43LcVEd+CGCFQPr4gGIFnAnVmBbTaWscgEX8v5xdqndcElIJRGkQzMFnkIBbz4MPlAM1ywM5BoE= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com; spf=pass smtp.mailfrom=intel.com; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b=K4dU8eWZ; arc=none smtp.client-ip=192.198.163.8 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=intel.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b="K4dU8eWZ" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1723587784; x=1755123784; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=tbYV53tO1ExEVxJSZAHIvE3/sh+rRMVQRouoky16wH8=; b=K4dU8eWZTsCG12OFsSa4sopnVmYdZWcEGCS6evUutp5wAivudMn3wDeK 1yTCRJFQKPyTolQUCx6YHiP+neW5oyd7x+LrsFQ/5s6HG5w6TVKGTBbLZ dbBZD+Y9EMzwcJUi8V8TsDkVdXL9n0p3WOYlp+7ovOuM5k0rqGfC5RTy0 +DK/9YyJG6AQHACHhPeyfPY5drlcLjuSqgm5YiXjzKWJFmc8FgfI6wzxc 6B+hfYzQCA894ZdviPQoKVLb7ly0h5MDj0XGRBa11BxDzYJwRmkYg4e7U pLFWCRc30lPqyO/3cMnGJxcNai4VLz2Oi6WtQk0T4eVBqXXU1lsIKbkJw w==; X-CSE-ConnectionGUID: C+x+E7ARQDuhrnXBwaCUcg== X-CSE-MsgGUID: IYWLpSbTSC6T5FdntYtjZg== X-IronPort-AV: E=McAfee;i="6700,10204,11163"; a="39287141" X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="39287141" Received: from orviesa007.jf.intel.com ([10.64.159.147]) by fmvoesa102.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 13 Aug 2024 15:22:56 -0700 X-CSE-ConnectionGUID: iRu2wgxuThSZpMUyG5HbgQ== X-CSE-MsgGUID: ZP3agfc3SR+waSQaBN5ONw== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="59381092" Received: from anguy11-upstream.jf.intel.com ([10.166.9.133]) by orviesa007.jf.intel.com with ESMTP; 13 Aug 2024 15:22:55 -0700 From: Tony Nguyen To: davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com, edumazet@google.com, netdev@vger.kernel.org Cc: Ahmed Zaki , anthony.l.nguyen@intel.com, madhu.chittim@intel.com, horms@kernel.org, hkelam@marvell.com, Sridhar Samudrala , Marcin Szycik , Rafal Romanowski Subject: [PATCH net-next v2 12/13] iavf: refactor add/del FDIR filters Date: Tue, 13 Aug 2024 15:22:47 -0700 Message-ID: <20240813222249.3708070-13-anthony.l.nguyen@intel.com> X-Mailer: git-send-email 2.42.0 In-Reply-To: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> References: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Patchwork-Delegate: kuba@kernel.org From: Ahmed Zaki In preparation for a second type of FDIR filters that can be added by tc-u32, move the add/del of the FDIR logic to be entirely contained in iavf_fdir.c. The iavf_find_fdir_fltr_by_loc() is renamed to iavf_find_fdir_fltr() to be more agnostic to the filter ID parameter (for now @loc, which is relevant only to current FDIR filters added via ethtool). The FDIR filter deletion is moved from iavf_del_fdir_ethtool() in ethtool.c to iavf_fdir_del_fltr(). While at it, fix a minor bug where the "fltr" is accessed out of the fdir_fltr_lock spinlock protection. Reviewed-by: Sridhar Samudrala Reviewed-by: Marcin Szycik Signed-off-by: Ahmed Zaki Tested-by: Rafal Romanowski Reviewed-by: Simon Horman Signed-off-by: Tony Nguyen --- drivers/net/ethernet/intel/iavf/iavf.h | 5 ++ .../net/ethernet/intel/iavf/iavf_ethtool.c | 56 ++------------- drivers/net/ethernet/intel/iavf/iavf_fdir.c | 72 +++++++++++++++++-- drivers/net/ethernet/intel/iavf/iavf_fdir.h | 7 +- 4 files changed, 83 insertions(+), 57 deletions(-) diff --git a/drivers/net/ethernet/intel/iavf/iavf.h b/drivers/net/ethernet/intel/iavf/iavf.h index 23a6557fc3db..dfed22baebf8 100644 --- a/drivers/net/ethernet/intel/iavf/iavf.h +++ b/drivers/net/ethernet/intel/iavf/iavf.h @@ -444,6 +444,11 @@ struct iavf_adapter { spinlock_t adv_rss_lock; /* protect the RSS management list */ }; +/* Must be called with fdir_fltr_lock lock held */ +static inline bool iavf_fdir_max_reached(struct iavf_adapter *adapter) +{ + return adapter->fdir_active_fltr >= IAVF_MAX_FDIR_FILTERS; +} /* Ethtool Private Flags */ diff --git a/drivers/net/ethernet/intel/iavf/iavf_ethtool.c b/drivers/net/ethernet/intel/iavf/iavf_ethtool.c index 52273f7eab2c..7ab445eeee18 100644 --- a/drivers/net/ethernet/intel/iavf/iavf_ethtool.c +++ b/drivers/net/ethernet/intel/iavf/iavf_ethtool.c @@ -927,7 +927,7 @@ iavf_get_ethtool_fdir_entry(struct iavf_adapter *adapter, spin_lock_bh(&adapter->fdir_fltr_lock); - rule = iavf_find_fdir_fltr_by_loc(adapter, fsp->location); + rule = iavf_find_fdir_fltr(adapter, fsp->location); if (!rule) { ret = -EINVAL; goto release_lock; @@ -1263,15 +1263,7 @@ static int iavf_add_fdir_ethtool(struct iavf_adapter *adapter, struct ethtool_rx return -EINVAL; spin_lock_bh(&adapter->fdir_fltr_lock); - if (adapter->fdir_active_fltr >= IAVF_MAX_FDIR_FILTERS) { - spin_unlock_bh(&adapter->fdir_fltr_lock); - dev_err(&adapter->pdev->dev, - "Unable to add Flow Director filter because VF reached the limit of max allowed filters (%u)\n", - IAVF_MAX_FDIR_FILTERS); - return -ENOSPC; - } - - if (iavf_find_fdir_fltr_by_loc(adapter, fsp->location)) { + if (iavf_find_fdir_fltr(adapter, fsp->location)) { dev_err(&adapter->pdev->dev, "Failed to add Flow Director filter, it already exists\n"); spin_unlock_bh(&adapter->fdir_fltr_lock); return -EEXIST; @@ -1291,23 +1283,10 @@ static int iavf_add_fdir_ethtool(struct iavf_adapter *adapter, struct ethtool_rx } err = iavf_add_fdir_fltr_info(adapter, fsp, fltr); - if (err) - goto ret; - - spin_lock_bh(&adapter->fdir_fltr_lock); - iavf_fdir_list_add_fltr(adapter, fltr); - adapter->fdir_active_fltr++; - - if (adapter->link_up) - fltr->state = IAVF_FDIR_FLTR_ADD_REQUEST; - else - fltr->state = IAVF_FDIR_FLTR_INACTIVE; - spin_unlock_bh(&adapter->fdir_fltr_lock); + if (!err) + err = iavf_fdir_add_fltr(adapter, fltr); - if (adapter->link_up) - iavf_schedule_aq_request(adapter, IAVF_FLAG_AQ_ADD_FDIR_FILTER); -ret: - if (err && fltr) + if (err) kfree(fltr); mutex_unlock(&adapter->crit_lock); @@ -1324,34 +1303,11 @@ static int iavf_add_fdir_ethtool(struct iavf_adapter *adapter, struct ethtool_rx static int iavf_del_fdir_ethtool(struct iavf_adapter *adapter, struct ethtool_rxnfc *cmd) { struct ethtool_rx_flow_spec *fsp = (struct ethtool_rx_flow_spec *)&cmd->fs; - struct iavf_fdir_fltr *fltr = NULL; - int err = 0; if (!(adapter->flags & IAVF_FLAG_FDIR_ENABLED)) return -EOPNOTSUPP; - spin_lock_bh(&adapter->fdir_fltr_lock); - fltr = iavf_find_fdir_fltr_by_loc(adapter, fsp->location); - if (fltr) { - if (fltr->state == IAVF_FDIR_FLTR_ACTIVE) { - fltr->state = IAVF_FDIR_FLTR_DEL_REQUEST; - } else if (fltr->state == IAVF_FDIR_FLTR_INACTIVE) { - list_del(&fltr->list); - kfree(fltr); - adapter->fdir_active_fltr--; - fltr = NULL; - } else { - err = -EBUSY; - } - } else if (adapter->fdir_active_fltr) { - err = -EINVAL; - } - spin_unlock_bh(&adapter->fdir_fltr_lock); - - if (fltr && fltr->state == IAVF_FDIR_FLTR_DEL_REQUEST) - iavf_schedule_aq_request(adapter, IAVF_FLAG_AQ_DEL_FDIR_FILTER); - - return err; + return iavf_fdir_del_fltr(adapter, fsp->location); } /** diff --git a/drivers/net/ethernet/intel/iavf/iavf_fdir.c b/drivers/net/ethernet/intel/iavf/iavf_fdir.c index 2d47b0b4640e..1e1daf71dfa0 100644 --- a/drivers/net/ethernet/intel/iavf/iavf_fdir.c +++ b/drivers/net/ethernet/intel/iavf/iavf_fdir.c @@ -815,13 +815,14 @@ bool iavf_fdir_is_dup_fltr(struct iavf_adapter *adapter, struct iavf_fdir_fltr * } /** - * iavf_find_fdir_fltr_by_loc - find filter with location + * iavf_find_fdir_fltr - find FDIR filter * @adapter: pointer to the VF adapter structure * @loc: location to find. * - * Returns pointer to Flow Director filter if found or null + * Returns: pointer to Flow Director filter if found or NULL. Lock must be held. */ -struct iavf_fdir_fltr *iavf_find_fdir_fltr_by_loc(struct iavf_adapter *adapter, u32 loc) +struct iavf_fdir_fltr *iavf_find_fdir_fltr(struct iavf_adapter *adapter, + u32 loc) { struct iavf_fdir_fltr *rule; @@ -833,14 +834,26 @@ struct iavf_fdir_fltr *iavf_find_fdir_fltr_by_loc(struct iavf_adapter *adapter, } /** - * iavf_fdir_list_add_fltr - add a new node to the flow director filter list + * iavf_fdir_add_fltr - add a new node to the flow director filter list * @adapter: pointer to the VF adapter structure * @fltr: filter node to add to structure + * + * Return: 0 on success or negative errno on failure. */ -void iavf_fdir_list_add_fltr(struct iavf_adapter *adapter, struct iavf_fdir_fltr *fltr) +int iavf_fdir_add_fltr(struct iavf_adapter *adapter, + struct iavf_fdir_fltr *fltr) { struct iavf_fdir_fltr *rule, *parent = NULL; + spin_lock_bh(&adapter->fdir_fltr_lock); + if (iavf_fdir_max_reached(adapter)) { + spin_unlock_bh(&adapter->fdir_fltr_lock); + dev_err(&adapter->pdev->dev, + "Unable to add Flow Director filter (limit (%u) reached)\n", + IAVF_MAX_FDIR_FILTERS); + return -ENOSPC; + } + list_for_each_entry(rule, &adapter->fdir_list_head, list) { if (rule->loc >= fltr->loc) break; @@ -851,4 +864,53 @@ void iavf_fdir_list_add_fltr(struct iavf_adapter *adapter, struct iavf_fdir_fltr list_add(&fltr->list, &parent->list); else list_add(&fltr->list, &adapter->fdir_list_head); + adapter->fdir_active_fltr++; + + if (adapter->link_up) + fltr->state = IAVF_FDIR_FLTR_ADD_REQUEST; + else + fltr->state = IAVF_FDIR_FLTR_INACTIVE; + spin_unlock_bh(&adapter->fdir_fltr_lock); + + if (adapter->link_up) + iavf_schedule_aq_request(adapter, IAVF_FLAG_AQ_ADD_FDIR_FILTER); + + return 0; +} + +/** + * iavf_fdir_del_fltr - delete a flow director filter from the list + * @adapter: pointer to the VF adapter structure + * @loc: location to delete. + * + * Return: 0 on success or negative errno on failure. + */ +int iavf_fdir_del_fltr(struct iavf_adapter *adapter, u32 loc) +{ + struct iavf_fdir_fltr *fltr = NULL; + int err = 0; + + spin_lock_bh(&adapter->fdir_fltr_lock); + fltr = iavf_find_fdir_fltr(adapter, loc); + + if (fltr) { + if (fltr->state == IAVF_FDIR_FLTR_ACTIVE) { + fltr->state = IAVF_FDIR_FLTR_DEL_REQUEST; + } else if (fltr->state == IAVF_FDIR_FLTR_INACTIVE) { + list_del(&fltr->list); + kfree(fltr); + adapter->fdir_active_fltr--; + fltr = NULL; + } else { + err = -EBUSY; + } + } else if (adapter->fdir_active_fltr) { + err = -EINVAL; + } + + if (fltr && fltr->state == IAVF_FDIR_FLTR_DEL_REQUEST) + iavf_schedule_aq_request(adapter, IAVF_FLAG_AQ_DEL_FDIR_FILTER); + + spin_unlock_bh(&adapter->fdir_fltr_lock); + return err; } diff --git a/drivers/net/ethernet/intel/iavf/iavf_fdir.h b/drivers/net/ethernet/intel/iavf/iavf_fdir.h index d31bd923ba8c..5c85eb25fa2a 100644 --- a/drivers/net/ethernet/intel/iavf/iavf_fdir.h +++ b/drivers/net/ethernet/intel/iavf/iavf_fdir.h @@ -128,6 +128,9 @@ int iavf_validate_fdir_fltr_masks(struct iavf_adapter *adapter, int iavf_fill_fdir_add_msg(struct iavf_adapter *adapter, struct iavf_fdir_fltr *fltr); void iavf_print_fdir_fltr(struct iavf_adapter *adapter, struct iavf_fdir_fltr *fltr); bool iavf_fdir_is_dup_fltr(struct iavf_adapter *adapter, struct iavf_fdir_fltr *fltr); -void iavf_fdir_list_add_fltr(struct iavf_adapter *adapter, struct iavf_fdir_fltr *fltr); -struct iavf_fdir_fltr *iavf_find_fdir_fltr_by_loc(struct iavf_adapter *adapter, u32 loc); +int iavf_fdir_add_fltr(struct iavf_adapter *adapter, + struct iavf_fdir_fltr *fltr); +int iavf_fdir_del_fltr(struct iavf_adapter *adapter, u32 loc); +struct iavf_fdir_fltr *iavf_find_fdir_fltr(struct iavf_adapter *adapter, + u32 loc); #endif /* _IAVF_FDIR_H_ */ From patchwork Tue Aug 13 22:22:48 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Tony Nguyen X-Patchwork-Id: 13762634 X-Patchwork-Delegate: kuba@kernel.org Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.8]) (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 225E31AC444 for ; Tue, 13 Aug 2024 22:23:04 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=192.198.163.8 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587786; cv=none; b=Vz02hO60+Y3bKreDNiS4cvlSFSGs5UdhPHdHGToN90qvi9luBd0t3s/h09sHMD40wBJcRVdhOOZARLsAUGVM/AHoLHJU0GwjZ1BNIQX/3srbko3d2zIWJrcfcMYx+b2L+/QLEVXy6oNCezgMWcWhy5XwEgnfhTbxGw9ZdpF5vLI= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1723587786; c=relaxed/simple; bh=A3nKOK+yVWLQoe+70gUUapS8qaPgPZfS1DlF5nh4dgw=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=orUhQp0UqNav+MaDzL+NKcE5UiHP/b6PDUU1qHaN9uxGA4H1yAHVH3RuOJUnMR3SHdsv1G7IrsSBK4SuMmvL2vH+Js28xvTlSOAFlY0tmUqSl6qc+PdR+Lq8rMFXq2lIO3bycJE0wGElzximVOyCcmkUxxtOQI9Q4/RxuSQ9s5g= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com; spf=pass smtp.mailfrom=intel.com; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b=ZnM5WMTD; arc=none smtp.client-ip=192.198.163.8 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=intel.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=intel.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.b="ZnM5WMTD" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1723587784; x=1755123784; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=A3nKOK+yVWLQoe+70gUUapS8qaPgPZfS1DlF5nh4dgw=; b=ZnM5WMTDoonZhtXiW4ZH2N22r/eFbrZN+IQWGDEN0uDG6iqPjevjQdMg HaMbXF6AYO7gOrDvjD/7/BfGthuDKnEptj51lYOq8xNlfzQWbUN76oWcn wWBoQSC18XfiQAoOB5Ra441FahsGFPkiJzJ+M0vP+PRsGm6r21z9bSuQH bHfTlsCKZp6B1+q6W1AbcgexSrl9ie7kAak/5Ja9zBLxVhmIFu967Oxd+ B7Cf5ije3woW9TilFFbxSCHWWZUUK0DoViYC3aGUjC16uDq5pJaHx/1Jh c7pM6WhnrS25I8wNlG9088wdOgXcSZiwHJPvO56iRWMaAMaOZS9HlZ6Fm Q==; X-CSE-ConnectionGUID: L+QzJvbER6qrJqNK+bXs6w== X-CSE-MsgGUID: JEZbZNuzQkOJs8jr25wCNw== X-IronPort-AV: E=McAfee;i="6700,10204,11163"; a="39287145" X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="39287145" Received: from orviesa007.jf.intel.com ([10.64.159.147]) by fmvoesa102.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 13 Aug 2024 15:22:56 -0700 X-CSE-ConnectionGUID: Kxo0O2nRSkazYlJNXG5KDg== X-CSE-MsgGUID: WNkxs+TsRxiyaZsHwkD7RQ== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.09,287,1716274800"; d="scan'208";a="59381095" Received: from anguy11-upstream.jf.intel.com ([10.166.9.133]) by orviesa007.jf.intel.com with ESMTP; 13 Aug 2024 15:22:55 -0700 From: Tony Nguyen To: davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com, edumazet@google.com, netdev@vger.kernel.org Cc: Ahmed Zaki , anthony.l.nguyen@intel.com, madhu.chittim@intel.com, horms@kernel.org, hkelam@marvell.com, Sridhar Samudrala , Marcin Szycik , Rafal Romanowski Subject: [PATCH net-next v2 13/13] iavf: add support for offloading tc U32 cls filters Date: Tue, 13 Aug 2024 15:22:48 -0700 Message-ID: <20240813222249.3708070-14-anthony.l.nguyen@intel.com> X-Mailer: git-send-email 2.42.0 In-Reply-To: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> References: <20240813222249.3708070-1-anthony.l.nguyen@intel.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Patchwork-Delegate: kuba@kernel.org From: Ahmed Zaki Add support for offloading cls U32 filters. Only "skbedit queue_mapping" and "drop" actions are supported. Also, only "ip" and "802_3" tc protocols are allowed. The PF must advertise the VIRTCHNL_VF_OFFLOAD_TC_U32 capability flag. Since the filters will be enabled via the FD stage at the PF, a new type of FDIR filters is added and the existing list and state machine are used. The new filters can be used to configure flow directors based on raw (binary) pattern in the rx packet. Examples: 0. # tc qdisc add dev enp175s0v0 ingress 1. Redirect UDP from src IP 192.168.2.1 to queue 12: # tc filter add dev protocol ip ingress u32 \ match u32 0x45000000 0xff000000 at 0 \ match u32 0x00110000 0x00ff0000 at 8 \ match u32 0xC0A80201 0xffffffff at 12 \ match u32 0x00000000 0x00000000 at 24 \ action skbedit queue_mapping 12 skip_sw 2. Drop all ICMP: # tc filter add dev protocol ip ingress u32 \ match u32 0x45000000 0xff000000 at 0 \ match u32 0x00010000 0x00ff0000 at 8 \ match u32 0x00000000 0x00000000 at 24 \ action drop skip_sw 3. Redirect ICMP traffic from MAC 3c:fd:fe:a5:47:e0 to queue 7 (note proto: 802_3): # tc filter add dev protocol 802_3 ingress u32 \ match u32 0x00003CFD 0x0000ffff at 4 \ match u32 0xFEA547E0 0xffffffff at 8 \ match u32 0x08004500 0xffffff00 at 12 \ match u32 0x00000001 0x000000ff at 20 \ match u32 0x0000 0x0000 at 40 \ action skbedit queue_mapping 7 skip_sw Notes on matches: 1 - All intermediate fields that are needed to parse the correct PTYPE must be provided (in e.g. 3: Ethernet Type 0x0800 in MAC, IP version and IP length: 0x45 and protocol: 0x01 (ICMP)). 2 - The last match must provide an offset that guarantees all required headers are accounted for, even if the last header is not matched. For example, in #2, the last match is 4 bytes at offset 24 starting from IP header, so the total is 14 (MAC) + 24 + 4 = 42, which is the sum of MAC+IP+ICMP headers. Reviewed-by: Sridhar Samudrala Reviewed-by: Marcin Szycik Signed-off-by: Ahmed Zaki Tested-by: Rafal Romanowski Reviewed-by: Simon Horman Signed-off-by: Tony Nguyen --- drivers/net/ethernet/intel/iavf/iavf.h | 27 ++- .../net/ethernet/intel/iavf/iavf_ethtool.c | 9 +- drivers/net/ethernet/intel/iavf/iavf_fdir.c | 29 +++- drivers/net/ethernet/intel/iavf/iavf_fdir.h | 10 +- drivers/net/ethernet/intel/iavf/iavf_main.c | 160 +++++++++++++++++- .../net/ethernet/intel/iavf/iavf_virtchnl.c | 25 ++- drivers/net/ethernet/intel/ice/ice_virtchnl.c | 4 + include/linux/avf/virtchnl.h | 1 + 8 files changed, 238 insertions(+), 27 deletions(-) diff --git a/drivers/net/ethernet/intel/iavf/iavf.h b/drivers/net/ethernet/intel/iavf/iavf.h index dfed22baebf8..48cd1d06761c 100644 --- a/drivers/net/ethernet/intel/iavf/iavf.h +++ b/drivers/net/ethernet/intel/iavf/iavf.h @@ -33,6 +33,7 @@ #include #include #include +#include #include "iavf_type.h" #include @@ -393,6 +394,8 @@ struct iavf_adapter { VIRTCHNL_VF_OFFLOAD_VLAN_V2) #define CRC_OFFLOAD_ALLOWED(_a) ((_a)->vf_res->vf_cap_flags & \ VIRTCHNL_VF_OFFLOAD_CRC) +#define TC_U32_SUPPORT(_a) ((_a)->vf_res->vf_cap_flags & \ + VIRTCHNL_VF_OFFLOAD_TC_U32) #define VLAN_V2_FILTERING_ALLOWED(_a) \ (VLAN_V2_ALLOWED((_a)) && \ ((_a)->vlan_v2_caps.filtering.filtering_support.outer || \ @@ -437,6 +440,7 @@ struct iavf_adapter { #define IAVF_MAX_FDIR_FILTERS 128 /* max allowed Flow Director filters */ u16 fdir_active_fltr; + u16 raw_fdir_active_fltr; struct list_head fdir_list_head; spinlock_t fdir_fltr_lock; /* protect the Flow Director filter list */ @@ -447,7 +451,28 @@ struct iavf_adapter { /* Must be called with fdir_fltr_lock lock held */ static inline bool iavf_fdir_max_reached(struct iavf_adapter *adapter) { - return adapter->fdir_active_fltr >= IAVF_MAX_FDIR_FILTERS; + return adapter->fdir_active_fltr + adapter->raw_fdir_active_fltr >= + IAVF_MAX_FDIR_FILTERS; +} + +static inline void +iavf_inc_fdir_active_fltr(struct iavf_adapter *adapter, + struct iavf_fdir_fltr *fltr) +{ + if (iavf_is_raw_fdir(fltr)) + adapter->raw_fdir_active_fltr++; + else + adapter->fdir_active_fltr++; +} + +static inline void +iavf_dec_fdir_active_fltr(struct iavf_adapter *adapter, + struct iavf_fdir_fltr *fltr) +{ + if (iavf_is_raw_fdir(fltr)) + adapter->raw_fdir_active_fltr--; + else + adapter->fdir_active_fltr--; } /* Ethtool Private Flags */ diff --git a/drivers/net/ethernet/intel/iavf/iavf_ethtool.c b/drivers/net/ethernet/intel/iavf/iavf_ethtool.c index 7ab445eeee18..74a1e9fe1821 100644 --- a/drivers/net/ethernet/intel/iavf/iavf_ethtool.c +++ b/drivers/net/ethernet/intel/iavf/iavf_ethtool.c @@ -927,7 +927,7 @@ iavf_get_ethtool_fdir_entry(struct iavf_adapter *adapter, spin_lock_bh(&adapter->fdir_fltr_lock); - rule = iavf_find_fdir_fltr(adapter, fsp->location); + rule = iavf_find_fdir_fltr(adapter, false, fsp->location); if (!rule) { ret = -EINVAL; goto release_lock; @@ -1072,6 +1072,9 @@ iavf_get_fdir_fltr_ids(struct iavf_adapter *adapter, struct ethtool_rxnfc *cmd, spin_lock_bh(&adapter->fdir_fltr_lock); list_for_each_entry(fltr, &adapter->fdir_list_head, list) { + if (iavf_is_raw_fdir(fltr)) + continue; + if (cnt == cmd->rule_cnt) { val = -EMSGSIZE; goto release_lock; @@ -1263,7 +1266,7 @@ static int iavf_add_fdir_ethtool(struct iavf_adapter *adapter, struct ethtool_rx return -EINVAL; spin_lock_bh(&adapter->fdir_fltr_lock); - if (iavf_find_fdir_fltr(adapter, fsp->location)) { + if (iavf_find_fdir_fltr(adapter, false, fsp->location)) { dev_err(&adapter->pdev->dev, "Failed to add Flow Director filter, it already exists\n"); spin_unlock_bh(&adapter->fdir_fltr_lock); return -EEXIST; @@ -1307,7 +1310,7 @@ static int iavf_del_fdir_ethtool(struct iavf_adapter *adapter, struct ethtool_rx if (!(adapter->flags & IAVF_FLAG_FDIR_ENABLED)) return -EOPNOTSUPP; - return iavf_fdir_del_fltr(adapter, fsp->location); + return iavf_fdir_del_fltr(adapter, false, fsp->location); } /** diff --git a/drivers/net/ethernet/intel/iavf/iavf_fdir.c b/drivers/net/ethernet/intel/iavf/iavf_fdir.c index 1e1daf71dfa0..a1b3b44cc14a 100644 --- a/drivers/net/ethernet/intel/iavf/iavf_fdir.c +++ b/drivers/net/ethernet/intel/iavf/iavf_fdir.c @@ -796,6 +796,9 @@ bool iavf_fdir_is_dup_fltr(struct iavf_adapter *adapter, struct iavf_fdir_fltr * spin_lock_bh(&adapter->fdir_fltr_lock); list_for_each_entry(tmp, &adapter->fdir_list_head, list) { + if (iavf_is_raw_fdir(fltr)) + continue; + if (tmp->flow_type != fltr->flow_type) continue; @@ -817,18 +820,21 @@ bool iavf_fdir_is_dup_fltr(struct iavf_adapter *adapter, struct iavf_fdir_fltr * /** * iavf_find_fdir_fltr - find FDIR filter * @adapter: pointer to the VF adapter structure - * @loc: location to find. + * @is_raw: filter type, is raw (tc u32) or not (ethtool) + * @data: data to ID the filter, type dependent * * Returns: pointer to Flow Director filter if found or NULL. Lock must be held. */ struct iavf_fdir_fltr *iavf_find_fdir_fltr(struct iavf_adapter *adapter, - u32 loc) + bool is_raw, u32 data) { struct iavf_fdir_fltr *rule; - list_for_each_entry(rule, &adapter->fdir_list_head, list) - if (rule->loc == loc) + list_for_each_entry(rule, &adapter->fdir_list_head, list) { + if ((is_raw && rule->cls_u32_handle == data) || + (!is_raw && rule->loc == data)) return rule; + } return NULL; } @@ -855,6 +861,9 @@ int iavf_fdir_add_fltr(struct iavf_adapter *adapter, } list_for_each_entry(rule, &adapter->fdir_list_head, list) { + if (iavf_is_raw_fdir(fltr)) + break; + if (rule->loc >= fltr->loc) break; parent = rule; @@ -864,7 +873,8 @@ int iavf_fdir_add_fltr(struct iavf_adapter *adapter, list_add(&fltr->list, &parent->list); else list_add(&fltr->list, &adapter->fdir_list_head); - adapter->fdir_active_fltr++; + + iavf_inc_fdir_active_fltr(adapter, fltr); if (adapter->link_up) fltr->state = IAVF_FDIR_FLTR_ADD_REQUEST; @@ -881,25 +891,26 @@ int iavf_fdir_add_fltr(struct iavf_adapter *adapter, /** * iavf_fdir_del_fltr - delete a flow director filter from the list * @adapter: pointer to the VF adapter structure - * @loc: location to delete. + * @is_raw: filter type, is raw (tc u32) or not (ethtool) + * @data: data to ID the filter, type dependent * * Return: 0 on success or negative errno on failure. */ -int iavf_fdir_del_fltr(struct iavf_adapter *adapter, u32 loc) +int iavf_fdir_del_fltr(struct iavf_adapter *adapter, bool is_raw, u32 data) { struct iavf_fdir_fltr *fltr = NULL; int err = 0; spin_lock_bh(&adapter->fdir_fltr_lock); - fltr = iavf_find_fdir_fltr(adapter, loc); + fltr = iavf_find_fdir_fltr(adapter, is_raw, data); if (fltr) { if (fltr->state == IAVF_FDIR_FLTR_ACTIVE) { fltr->state = IAVF_FDIR_FLTR_DEL_REQUEST; } else if (fltr->state == IAVF_FDIR_FLTR_INACTIVE) { list_del(&fltr->list); + iavf_dec_fdir_active_fltr(adapter, fltr); kfree(fltr); - adapter->fdir_active_fltr--; fltr = NULL; } else { err = -EBUSY; diff --git a/drivers/net/ethernet/intel/iavf/iavf_fdir.h b/drivers/net/ethernet/intel/iavf/iavf_fdir.h index 5c85eb25fa2a..e84a5351162f 100644 --- a/drivers/net/ethernet/intel/iavf/iavf_fdir.h +++ b/drivers/net/ethernet/intel/iavf/iavf_fdir.h @@ -117,12 +117,18 @@ struct iavf_fdir_fltr { u32 flow_id; + u32 cls_u32_handle; /* for FDIR added via tc u32 */ u32 loc; /* Rule location inside the flow table */ u32 q_index; struct virtchnl_fdir_add vc_add_msg; }; +static inline bool iavf_is_raw_fdir(struct iavf_fdir_fltr *fltr) +{ + return !fltr->vc_add_msg.rule_cfg.proto_hdrs.count; +} + int iavf_validate_fdir_fltr_masks(struct iavf_adapter *adapter, struct iavf_fdir_fltr *fltr); int iavf_fill_fdir_add_msg(struct iavf_adapter *adapter, struct iavf_fdir_fltr *fltr); @@ -130,7 +136,7 @@ void iavf_print_fdir_fltr(struct iavf_adapter *adapter, struct iavf_fdir_fltr *f bool iavf_fdir_is_dup_fltr(struct iavf_adapter *adapter, struct iavf_fdir_fltr *fltr); int iavf_fdir_add_fltr(struct iavf_adapter *adapter, struct iavf_fdir_fltr *fltr); -int iavf_fdir_del_fltr(struct iavf_adapter *adapter, u32 loc); +int iavf_fdir_del_fltr(struct iavf_adapter *adapter, bool is_raw, u32 data); struct iavf_fdir_fltr *iavf_find_fdir_fltr(struct iavf_adapter *adapter, - u32 loc); + bool is_raw, u32 data); #endif /* _IAVF_FDIR_H_ */ diff --git a/drivers/net/ethernet/intel/iavf/iavf_main.c b/drivers/net/ethernet/intel/iavf/iavf_main.c index ff11bafb3b4f..f782402cd789 100644 --- a/drivers/net/ethernet/intel/iavf/iavf_main.c +++ b/drivers/net/ethernet/intel/iavf/iavf_main.c @@ -4013,7 +4013,7 @@ static int iavf_delete_clsflower(struct iavf_adapter *adapter, /** * iavf_setup_tc_cls_flower - flower classifier offloads - * @adapter: board private structure + * @adapter: pointer to iavf adapter structure * @cls_flower: pointer to flow_cls_offload struct with flow info */ static int iavf_setup_tc_cls_flower(struct iavf_adapter *adapter, @@ -4031,6 +4031,154 @@ static int iavf_setup_tc_cls_flower(struct iavf_adapter *adapter, } } +/** + * iavf_add_cls_u32 - Add U32 classifier offloads + * @adapter: pointer to iavf adapter structure + * @cls_u32: pointer to tc_cls_u32_offload struct with flow info + * + * Return: 0 on success or negative errno on failure. + */ +static int iavf_add_cls_u32(struct iavf_adapter *adapter, + struct tc_cls_u32_offload *cls_u32) +{ + struct netlink_ext_ack *extack = cls_u32->common.extack; + struct virtchnl_fdir_rule *rule_cfg; + struct virtchnl_filter_action *vact; + struct virtchnl_proto_hdrs *hdrs; + struct ethhdr *spec_h, *mask_h; + const struct tc_action *act; + struct iavf_fdir_fltr *fltr; + struct tcf_exts *exts; + unsigned int q_index; + int i, status = 0; + int off_base = 0; + + if (cls_u32->knode.link_handle) { + NL_SET_ERR_MSG_MOD(extack, "Linking not supported"); + return -EOPNOTSUPP; + } + + fltr = kzalloc(sizeof(*fltr), GFP_KERNEL); + if (!fltr) + return -ENOMEM; + + rule_cfg = &fltr->vc_add_msg.rule_cfg; + hdrs = &rule_cfg->proto_hdrs; + hdrs->count = 0; + + /* The parser lib at the PF expects the packet starting with MAC hdr */ + switch (ntohs(cls_u32->common.protocol)) { + case ETH_P_802_3: + break; + case ETH_P_IP: + spec_h = (struct ethhdr *)hdrs->raw.spec; + mask_h = (struct ethhdr *)hdrs->raw.mask; + spec_h->h_proto = htons(ETH_P_IP); + mask_h->h_proto = htons(0xFFFF); + off_base += ETH_HLEN; + break; + default: + NL_SET_ERR_MSG_MOD(extack, "Only 802_3 and ip filter protocols are supported"); + status = -EOPNOTSUPP; + goto free_alloc; + } + + for (i = 0; i < cls_u32->knode.sel->nkeys; i++) { + __be32 val, mask; + int off; + + off = off_base + cls_u32->knode.sel->keys[i].off; + val = cls_u32->knode.sel->keys[i].val; + mask = cls_u32->knode.sel->keys[i].mask; + + if (off >= sizeof(hdrs->raw.spec)) { + NL_SET_ERR_MSG_MOD(extack, "Input exceeds maximum allowed."); + status = -EINVAL; + goto free_alloc; + } + + memcpy(&hdrs->raw.spec[off], &val, sizeof(val)); + memcpy(&hdrs->raw.mask[off], &mask, sizeof(mask)); + hdrs->raw.pkt_len = off + sizeof(val); + } + + /* Only one action is allowed */ + rule_cfg->action_set.count = 1; + vact = &rule_cfg->action_set.actions[0]; + exts = cls_u32->knode.exts; + + tcf_exts_for_each_action(i, act, exts) { + /* FDIR queue */ + if (is_tcf_skbedit_rx_queue_mapping(act)) { + q_index = tcf_skbedit_rx_queue_mapping(act); + if (q_index >= adapter->num_active_queues) { + status = -EINVAL; + goto free_alloc; + } + + vact->type = VIRTCHNL_ACTION_QUEUE; + vact->act_conf.queue.index = q_index; + break; + } + + /* Drop */ + if (is_tcf_gact_shot(act)) { + vact->type = VIRTCHNL_ACTION_DROP; + break; + } + + /* Unsupported action */ + NL_SET_ERR_MSG_MOD(extack, "Unsupported action."); + status = -EOPNOTSUPP; + goto free_alloc; + } + + fltr->vc_add_msg.vsi_id = adapter->vsi.id; + fltr->cls_u32_handle = cls_u32->knode.handle; + return iavf_fdir_add_fltr(adapter, fltr); + +free_alloc: + kfree(fltr); + return status; +} + +/** + * iavf_del_cls_u32 - Delete U32 classifier offloads + * @adapter: pointer to iavf adapter structure + * @cls_u32: pointer to tc_cls_u32_offload struct with flow info + * + * Return: 0 on success or negative errno on failure. + */ +static int iavf_del_cls_u32(struct iavf_adapter *adapter, + struct tc_cls_u32_offload *cls_u32) +{ + return iavf_fdir_del_fltr(adapter, true, cls_u32->knode.handle); +} + +/** + * iavf_setup_tc_cls_u32 - U32 filter offloads + * @adapter: pointer to iavf adapter structure + * @cls_u32: pointer to tc_cls_u32_offload struct with flow info + * + * Return: 0 on success or negative errno on failure. + */ +static int iavf_setup_tc_cls_u32(struct iavf_adapter *adapter, + struct tc_cls_u32_offload *cls_u32) +{ + if (!TC_U32_SUPPORT(adapter) || !FDIR_FLTR_SUPPORT(adapter)) + return -EOPNOTSUPP; + + switch (cls_u32->command) { + case TC_CLSU32_NEW_KNODE: + case TC_CLSU32_REPLACE_KNODE: + return iavf_add_cls_u32(adapter, cls_u32); + case TC_CLSU32_DELETE_KNODE: + return iavf_del_cls_u32(adapter, cls_u32); + default: + return -EOPNOTSUPP; + } +} + /** * iavf_setup_tc_block_cb - block callback for tc * @type: type of offload @@ -4050,6 +4198,8 @@ static int iavf_setup_tc_block_cb(enum tc_setup_type type, void *type_data, switch (type) { case TC_SETUP_CLSFLOWER: return iavf_setup_tc_cls_flower(cb_priv, type_data); + case TC_SETUP_CLSU32: + return iavf_setup_tc_cls_u32(cb_priv, type_data); default: return -EOPNOTSUPP; } @@ -4332,8 +4482,8 @@ static void iavf_disable_fdir(struct iavf_adapter *adapter) fdir->state == IAVF_FDIR_FLTR_INACTIVE) { /* Delete filters not registered in PF */ list_del(&fdir->list); + iavf_dec_fdir_active_fltr(adapter, fdir); kfree(fdir); - adapter->fdir_active_fltr--; } else if (fdir->state == IAVF_FDIR_FLTR_ADD_PENDING || fdir->state == IAVF_FDIR_FLTR_DIS_REQUEST || fdir->state == IAVF_FDIR_FLTR_ACTIVE) { @@ -4843,9 +4993,11 @@ int iavf_process_config(struct iavf_adapter *adapter) /* get HW VLAN features that can be toggled */ hw_vlan_features = iavf_get_netdev_vlan_hw_features(adapter); - /* Enable cloud filter if ADQ is supported */ - if (vfres->vf_cap_flags & VIRTCHNL_VF_OFFLOAD_ADQ) + /* Enable HW TC offload if ADQ or tc U32 is supported */ + if (vfres->vf_cap_flags & VIRTCHNL_VF_OFFLOAD_ADQ || + TC_U32_SUPPORT(adapter)) hw_features |= NETIF_F_HW_TC; + if (vfres->vf_cap_flags & VIRTCHNL_VF_OFFLOAD_USO) hw_features |= NETIF_F_GSO_UDP_L4; diff --git a/drivers/net/ethernet/intel/iavf/iavf_virtchnl.c b/drivers/net/ethernet/intel/iavf/iavf_virtchnl.c index 1e543f6a7c30..7e810b65380c 100644 --- a/drivers/net/ethernet/intel/iavf/iavf_virtchnl.c +++ b/drivers/net/ethernet/intel/iavf/iavf_virtchnl.c @@ -142,6 +142,7 @@ int iavf_send_vf_config_msg(struct iavf_adapter *adapter) VIRTCHNL_VF_OFFLOAD_WB_ON_ITR | VIRTCHNL_VF_OFFLOAD_RSS_PCTYPE_V2 | VIRTCHNL_VF_OFFLOAD_ENCAP | + VIRTCHNL_VF_OFFLOAD_TC_U32 | VIRTCHNL_VF_OFFLOAD_VLAN_V2 | VIRTCHNL_VF_OFFLOAD_CRC | VIRTCHNL_VF_OFFLOAD_ENCAP_CSUM | @@ -1961,8 +1962,8 @@ static void iavf_activate_fdir_filters(struct iavf_adapter *adapter) * list on PF is already cleared after a reset */ list_del(&f->list); + iavf_dec_fdir_active_fltr(adapter, f); kfree(f); - adapter->fdir_active_fltr--; } } spin_unlock_bh(&adapter->fdir_fltr_lock); @@ -2135,8 +2136,8 @@ void iavf_virtchnl_completion(struct iavf_adapter *adapter, dev_err(&adapter->pdev->dev, "%s\n", msg); list_del(&fdir->list); + iavf_dec_fdir_active_fltr(adapter, fdir); kfree(fdir); - adapter->fdir_active_fltr--; } } spin_unlock_bh(&adapter->fdir_fltr_lock); @@ -2451,8 +2452,12 @@ void iavf_virtchnl_completion(struct iavf_adapter *adapter, list) { if (fdir->state == IAVF_FDIR_FLTR_ADD_PENDING) { if (add_fltr->status == VIRTCHNL_FDIR_SUCCESS) { - dev_info(&adapter->pdev->dev, "Flow Director filter with location %u is added\n", - fdir->loc); + if (!iavf_is_raw_fdir(fdir)) + dev_info(&adapter->pdev->dev, "Flow Director filter with location %u is added\n", + fdir->loc); + else + dev_info(&adapter->pdev->dev, "Flow Director filter (raw) for TC handle %x is added\n", + TC_U32_USERHTID(fdir->cls_u32_handle)); fdir->state = IAVF_FDIR_FLTR_ACTIVE; fdir->flow_id = add_fltr->flow_id; } else { @@ -2460,8 +2465,8 @@ void iavf_virtchnl_completion(struct iavf_adapter *adapter, add_fltr->status); iavf_print_fdir_fltr(adapter, fdir); list_del(&fdir->list); + iavf_dec_fdir_active_fltr(adapter, fdir); kfree(fdir); - adapter->fdir_active_fltr--; } } } @@ -2479,11 +2484,15 @@ void iavf_virtchnl_completion(struct iavf_adapter *adapter, if (del_fltr->status == VIRTCHNL_FDIR_SUCCESS || del_fltr->status == VIRTCHNL_FDIR_FAILURE_RULE_NONEXIST) { - dev_info(&adapter->pdev->dev, "Flow Director filter with location %u is deleted\n", - fdir->loc); + if (!iavf_is_raw_fdir(fdir)) + dev_info(&adapter->pdev->dev, "Flow Director filter with location %u is deleted\n", + fdir->loc); + else + dev_info(&adapter->pdev->dev, "Flow Director filter (raw) for TC handle %x is deleted\n", + TC_U32_USERHTID(fdir->cls_u32_handle)); list_del(&fdir->list); + iavf_dec_fdir_active_fltr(adapter, fdir); kfree(fdir); - adapter->fdir_active_fltr--; } else { fdir->state = IAVF_FDIR_FLTR_ACTIVE; dev_info(&adapter->pdev->dev, "Failed to delete Flow Director filter with status: %d\n", diff --git a/drivers/net/ethernet/intel/ice/ice_virtchnl.c b/drivers/net/ethernet/intel/ice/ice_virtchnl.c index 1c6ce0c4ed4e..59f62306b9cb 100644 --- a/drivers/net/ethernet/intel/ice/ice_virtchnl.c +++ b/drivers/net/ethernet/intel/ice/ice_virtchnl.c @@ -461,6 +461,10 @@ static int ice_vc_get_vf_res_msg(struct ice_vf *vf, u8 *msg) if (vf->driver_caps & VIRTCHNL_VF_OFFLOAD_FDIR_PF) vfres->vf_cap_flags |= VIRTCHNL_VF_OFFLOAD_FDIR_PF; + if (vf->driver_caps & VIRTCHNL_VF_OFFLOAD_TC_U32 && + vfres->vf_cap_flags & VIRTCHNL_VF_OFFLOAD_FDIR_PF) + vfres->vf_cap_flags |= VIRTCHNL_VF_OFFLOAD_TC_U32; + if (vf->driver_caps & VIRTCHNL_VF_OFFLOAD_RSS_PCTYPE_V2) vfres->vf_cap_flags |= VIRTCHNL_VF_OFFLOAD_RSS_PCTYPE_V2; diff --git a/include/linux/avf/virtchnl.h b/include/linux/avf/virtchnl.h index 4f78a65e33dc..f41395264dca 100644 --- a/include/linux/avf/virtchnl.h +++ b/include/linux/avf/virtchnl.h @@ -247,6 +247,7 @@ VIRTCHNL_CHECK_STRUCT_LEN(16, virtchnl_vsi_resource); /* used to negotiate communicating link speeds in Mbps */ #define VIRTCHNL_VF_CAP_ADV_LINK_SPEED BIT(7) #define VIRTCHNL_VF_OFFLOAD_CRC BIT(10) +#define VIRTCHNL_VF_OFFLOAD_TC_U32 BIT(11) #define VIRTCHNL_VF_OFFLOAD_VLAN_V2 BIT(15) #define VIRTCHNL_VF_OFFLOAD_VLAN BIT(16) #define VIRTCHNL_VF_OFFLOAD_RX_POLLING BIT(17)