Re: [PATCH net-next 5/7] cxgb4: add support for setting u32 filters

[Jiri Pirko <jiri@resnulli.us>]

Mon, Sep 12, 2016 at 10:12:38AM CEST, rahul.lakkireddy@chelsio.com wrote:
>Add support for offloading u32 filter onto hardware.  Links are stored
>in a jump table to perform necessary jumps to match TCP/UDP header.
>When inserting rules in the linked bucket, the TCP/UDP match fields
>in the corresponding entry of the jump table are appended to the filter
>rule before insertion.
>
>Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>
>Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
>---
> drivers/net/ethernet/chelsio/cxgb4/Makefile        |   2 +-
> drivers/net/ethernet/chelsio/cxgb4/cxgb4.h         |   3 +
> drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c    |  36 +++
> drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_u32.c  | 343 +++++++++++++++++++++
> drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_u32.h  |  55 ++++
> .../ethernet/chelsio/cxgb4/cxgb4_tc_u32_parse.h    |  12 +
> 6 files changed, 450 insertions(+), 1 deletion(-)
> create mode 100644 drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_u32.c
> create mode 100644 drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_u32.h
>
>diff --git a/drivers/net/ethernet/chelsio/cxgb4/Makefile b/drivers/net/ethernet/chelsio/cxgb4/Makefile
>index da88981..c6b71f6 100644
>--- a/drivers/net/ethernet/chelsio/cxgb4/Makefile
>+++ b/drivers/net/ethernet/chelsio/cxgb4/Makefile
>@@ -4,7 +4,7 @@
> 
> obj-$(CONFIG_CHELSIO_T4) += cxgb4.o
> 
>-cxgb4-objs := cxgb4_main.o l2t.o t4_hw.o sge.o clip_tbl.o cxgb4_ethtool.o cxgb4_uld.o sched.o cxgb4_filter.o
>+cxgb4-objs := cxgb4_main.o l2t.o t4_hw.o sge.o clip_tbl.o cxgb4_ethtool.o cxgb4_uld.o sched.o cxgb4_filter.o cxgb4_tc_u32.o
> cxgb4-$(CONFIG_CHELSIO_T4_DCB) +=  cxgb4_dcb.o
> cxgb4-$(CONFIG_CHELSIO_T4_FCOE) +=  cxgb4_fcoe.o
> cxgb4-$(CONFIG_DEBUG_FS) += cxgb4_debugfs.o
>diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
>index fbd593a..1adb28f 100644
>--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
>+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
>@@ -867,6 +867,9 @@ struct adapter {
> 
> 	spinlock_t stats_lock;
> 	spinlock_t win0_lock ____cacheline_aligned_in_smp;
>+
>+	/* TC u32 offload */
>+	struct cxgb4_tc_u32_table *tc_u32;
> };
> 
> /* Support for "sched-class" command to allow a TX Scheduling Class to be
>diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
>index af07f9d..28396f5 100644
>--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
>+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
>@@ -78,6 +78,7 @@
> #include "clip_tbl.h"
> #include "l2t.h"
> #include "sched.h"
>+#include "cxgb4_tc_u32.h"
> 
> char cxgb4_driver_name[] = KBUILD_MODNAME;
> 
>@@ -3027,6 +3028,33 @@ static int cxgb_set_tx_maxrate(struct net_device *dev, int index, u32 rate)
> 	return err;
> }
> 
>+int cxgb_setup_tc(struct net_device *dev, u32 handle, __be16 proto,
>+		  struct tc_to_netdev *tc)
>+{
>+	struct adapter *adap = netdev2adap(dev);
>+	struct port_info *pi = netdev2pinfo(dev);
>+
>+	if (!(adap->flags & FULL_INIT_DONE)) {
>+		dev_err(adap->pdev_dev,
>+			"Failed to setup tc on port %d. Link Down?\n",
>+			pi->port_id);
>+		return -EINVAL;
>+	}
>+
>+	if (TC_H_MAJ(handle) == TC_H_MAJ(TC_H_INGRESS) &&
>+	    tc->type == TC_SETUP_CLSU32) {
>+		switch (tc->cls_u32->command) {
>+		case TC_CLSU32_NEW_KNODE:
>+		case TC_CLSU32_REPLACE_KNODE:
>+			return cxgb4_config_knode(dev, proto, tc->cls_u32);
>+		default:
>+			return -EOPNOTSUPP;
>+		}
>+	}
>+
>+	return -EOPNOTSUPP;
>+}
>+
> static const struct net_device_ops cxgb4_netdev_ops = {
> 	.ndo_open             = cxgb_open,
> 	.ndo_stop             = cxgb_close,
>@@ -3050,6 +3078,7 @@ static const struct net_device_ops cxgb4_netdev_ops = {
> 	.ndo_busy_poll        = cxgb_busy_poll,
> #endif
> 	.ndo_set_tx_maxrate   = cxgb_set_tx_maxrate,
>+	.ndo_setup_tc         = cxgb_setup_tc,
> };
> 
> #ifdef CONFIG_PCI_IOV
>@@ -4781,6 +4810,7 @@ static void free_some_resources(struct adapter *adapter)
> 	t4_free_mem(adapter->l2t);
> 	t4_cleanup_sched(adapter);
> 	t4_free_mem(adapter->tids.tid_tab);
>+	cxgb4_cleanup_tc_u32(adapter);
> 	kfree(adapter->sge.egr_map);
> 	kfree(adapter->sge.ingr_map);
> 	kfree(adapter->sge.starving_fl);
>@@ -5213,6 +5243,12 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
> 		dev_warn(&pdev->dev, "could not allocate TID table, "
> 			 "continuing\n");
> 		adapter->params.offload = 0;
>+	} else {
>+		adapter->tc_u32 = cxgb4_init_tc_u32(adapter,
>+						    CXGB4_MAX_LINK_HANDLE);
>+		if (!adapter->tc_u32)
>+			dev_warn(&pdev->dev,
>+				 "could not offload tc u32, continuing\n");
> 	}
> 
> 	if (is_offload(adapter)) {
>diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_u32.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_u32.c
>new file mode 100644
>index 0000000..62c1695
>--- /dev/null
>+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_u32.c
>@@ -0,0 +1,343 @@
>+/*
>+ * This file is part of the Chelsio T4 Ethernet driver for Linux.
>+ *
>+ * Copyright (c) 2016 Chelsio Communications, Inc. All rights reserved.
>+ *
>+ * This software is available to you under a choice of one of two
>+ * licenses.  You may choose to be licensed under the terms of the GNU
>+ * General Public License (GPL) Version 2, available from the file
>+ * COPYING in the main directory of this source tree, or the
>+ * OpenIB.org BSD license below:
>+ *
>+ *     Redistribution and use in source and binary forms, with or
>+ *     without modification, are permitted provided that the following
>+ *     conditions are met:
>+ *
>+ *      - Redistributions of source code must retain the above
>+ *        copyright notice, this list of conditions and the following
>+ *        disclaimer.
>+ *
>+ *      - Redistributions in binary form must reproduce the above
>+ *        copyright notice, this list of conditions and the following
>+ *        disclaimer in the documentation and/or other materials
>+ *        provided with the distribution.
>+ *
>+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
>+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
>+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
>+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
>+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
>+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
>+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
>+ * SOFTWARE.
>+ */
>+
>+#include "cxgb4.h"
>+#include "cxgb4_tc_u32_parse.h"
>+#include "cxgb4_tc_u32.h"
>+
>+/* Fill ch_filter_specification with parsed match value/mask pair. */
>+static int fill_match_fields(struct adapter *adap,
>+			     struct ch_filter_specification *fs,
>+			     struct tc_cls_u32_offload *cls,
>+			     const struct cxgb4_match_field *entry,
>+			     bool next_header)
>+{
>+	unsigned int i, j;
>+	int off;
>+	u32 val, mask;
>+	int err;
>+	bool found = false;
>+
>+	for (i = 0; i < cls->knode.sel->nkeys; i++) {
>+		off = cls->knode.sel->keys[i].off;
>+		val = cls->knode.sel->keys[i].val;
>+		mask = cls->knode.sel->keys[i].mask;
>+
>+		if (next_header) {
>+			/* For next headers, parse only keys with offmask */
>+			if (!cls->knode.sel->keys[i].offmask)
>+				continue;
>+		} else {
>+			/* For the remaining, parse only keys without offmask */
>+			if (cls->knode.sel->keys[i].offmask)
>+				continue;
>+		}
>+
>+		found = false;
>+
>+		for (j = 0; entry[j].val; j++) {
>+			if (off == entry[j].off) {
>+				found = true;
>+				err = entry[j].val(fs, val, mask);
>+				if (err)
>+					return err;
>+				break;
>+			}
>+		}
>+
>+		if (!found)
>+			return -EINVAL;
>+	}
>+
>+	return 0;
>+}
>+
>+int cxgb4_config_knode(struct net_device *dev, __be16 protocol,
>+		       struct tc_cls_u32_offload *cls)
>+{
>+	struct adapter *adapter = netdev2adap(dev);
>+	struct cxgb4_tc_u32_table *t;
>+	const struct cxgb4_match_field *start, *link_start = NULL;
>+	struct cxgb4_link *link;
>+	struct ch_filter_specification fs;
>+	struct filter_ctx ctx;
>+	unsigned int filter_id;
>+	u32 uhtid, link_uhtid;
>+	int ret;
>+	bool is_ipv6 = false;
>+
>+	if (!can_tc_u32_offload(dev))
>+		return -EOPNOTSUPP;
>+
>+	if ((protocol != htons(ETH_P_IP)) && (protocol != htons(ETH_P_IPV6)))

You use a lot of unnecessary ()


>+		return -EOPNOTSUPP;
>+
>+	/* Fetch the location to insert the filter. */
>+	filter_id = (cls->knode.handle & 0xFFFFF);

here as well


>+
>+	if (filter_id > adapter->tids.nftids) {
>+		dev_err(adapter->pdev_dev,
>+			"Location %d out of range for insertion. Max: %d\n",
>+			filter_id, adapter->tids.nftids);
>+		return -ERANGE;
>+	}
>+
>+	t = adapter->tc_u32;
>+	uhtid = TC_U32_USERHTID(cls->knode.handle);
>+	link_uhtid = TC_U32_USERHTID(cls->knode.link_handle);
>+
>+	/* Ensure that uhtid is either root u32 (i.e. 0x800)
>+	 * or a a valid linked bucket.
>+	 */
>+	if (uhtid != 0x800 && uhtid >= t->size)
>+		return -EINVAL;
>+
>+	/* Ensure link handle uhtid is sane, if specified. */
>+	if (link_uhtid >= t->size)
>+		return -EINVAL;
>+
>+	memset(&fs, 0, sizeof(fs));
>+
>+	if (protocol == htons(ETH_P_IPV6)) {
>+		start = cxgb4_ipv6_fields;
>+		is_ipv6 = true;
>+	} else {
>+		start = cxgb4_ipv4_fields;
>+		is_ipv6 = false;
>+	}
>+
>+	if (uhtid != 0x800) {
>+		/* Link must exist from root node before insertion. */
>+		if (!t->table[uhtid - 1].link_handle)
>+			return -EINVAL;
>+
>+		/* Link must have a valid supported next header. */
>+		link_start = (&t->table[uhtid - 1])->match_field;

here as well

>+		if (!link_start)
>+			return -EINVAL;
>+	}
>+
>+	/* Parse links and record them for subsequent jumps to valid
>+	 * next headers.
>+	 */
>+	if (link_uhtid) {
>+		const struct cxgb4_next_header *next;
>+		unsigned int i, j;
>+		int off;
>+		u32 val, mask;
>+		bool found = false;
>+
>+		if (t->table[link_uhtid - 1].link_handle) {
>+			dev_err(adapter->pdev_dev,
>+				"Link handle exists for: 0x%x\n",
>+				link_uhtid);
>+			return -EINVAL;
>+		}
>+
>+		next = is_ipv6 ? cxgb4_ipv6_jumps : cxgb4_ipv4_jumps;
>+
>+		/* Try to find matches that allow jumps to next header. */
>+		for (i = 0; next[i].jump; i++) {
>+			if (next[i].offoff != cls->knode.sel->offoff ||
>+			    next[i].shift != cls->knode.sel->offshift ||
>+			    next[i].mask != cls->knode.sel->offmask ||
>+			    next[i].offset != cls->knode.sel->off)
>+				continue;
>+
>+			/* Found a possible candidate.  Find a key that
>+			 * matches the corresponding offset, value, and
>+			 * mask to jump to next header.
>+			 */
>+			for (j = 0; j < cls->knode.sel->nkeys; j++) {
>+				off = cls->knode.sel->keys[j].off;
>+				val = cls->knode.sel->keys[j].val;
>+				mask = cls->knode.sel->keys[j].mask;
>+
>+				if (next[i].match_off == off &&
>+				    next[i].match_val == val &&
>+				    next[i].match_mask == mask) {
>+					found = true;
>+					break;
>+				}
>+			}
>+
>+			if (!found)
>+				continue; /* Try next candidate. */
>+
>+			/* Candidate to jump to next header found.
>+			 * Translate all keys to internal specification
>+			 * and store them in jump table. This spec is copied
>+			 * later to set the actual filters.
>+			 */
>+			ret = fill_match_fields(adapter, &fs, cls,
>+						start, false);
>+			if (ret)
>+				goto out;
>+
>+			link = &t->table[link_uhtid - 1];
>+			link->match_field = next[i].jump;
>+			link->link_handle = cls->knode.handle;
>+			memcpy(&link->fs, &fs, sizeof(fs));
>+			break;
>+		}
>+
>+		/* No candidate found to jump to next header. */
>+		if (!found)
>+			return -EINVAL;
>+
>+		return 0;
>+	}
>+
>+	/* Fill ch_filter_specification match fields to be shipped to hardware.
>+	 * Copy the linked spec (if any) first.  And then update the spec as
>+	 * needed.
>+	 */
>+	if ((uhtid != 0x800) && t->table[uhtid - 1].link_handle) {
>+		/* Copy linked ch_filter_specification */
>+		memcpy(&fs, &t->table[uhtid - 1].fs, sizeof(fs));
>+		ret = fill_match_fields(adapter, &fs, cls,
>+					link_start, true);
>+		if (ret)
>+			goto out;
>+	}
>+
>+	ret = fill_match_fields(adapter, &fs, cls, start, false);
>+	if (ret)
>+		goto out;
>+
>+	/* The filter spec has been completely built from the info
>+	 * provided from u32.  We now set some default fields in the
>+	 * spec for sanity.
>+	 */
>+
>+	/* Match only packets coming from the ingress port where this
>+	 * filter will be created.
>+	 */
>+	fs.val.iport = netdev2pinfo(dev)->port_id;
>+	fs.mask.iport = ~0;
>+
>+	/* Enable filter hit counts. */
>+	fs.hitcnts = 1;
>+
>+	/* Set type of filter - IPv6 or IPv4 */
>+	fs.type = is_ipv6 ? 1 : 0;
>+
>+	init_completion(&ctx.completion);
>+
>+	/* Set the filter */
>+	ret = cxgb4_set_filter(dev, filter_id, &fs, &ctx);
>+	if (ret)
>+		goto out;
>+
>+	/* Wait for reply */
>+	ret = wait_for_completion_timeout(&ctx.completion, 10 * HZ);
>+	if (!ret)
>+		return -ETIMEDOUT;
>+
>+	ret = ctx.result;
>+	if (!ret) {
>+		/* If this is a linked bucket, then set the corresponding
>+		 * entry in the bitmap to mark it as belonging to this linked
>+		 * bucket.
>+		 */
>+		if ((uhtid != 0x800) && t->table[uhtid - 1].link_handle)
>+			set_bit(filter_id, (&t->table[uhtid - 1])->tid_map);
>+	}
>+
>+out:
>+	return ret;
>+}
>+
>+void cxgb4_cleanup_tc_u32(struct adapter *adap)
>+{
>+	struct cxgb4_tc_u32_table *t;
>+	unsigned int i;
>+
>+	if (!adap->tc_u32)
>+		return;
>+
>+	/* Free up all allocated memory. */
>+	t = adap->tc_u32;
>+	for (i = 0; i < t->size; i++) {
>+		struct cxgb4_link *link = &t->table[i];
>+
>+		t4_free_mem(link->tid_map);
>+	}
>+	t4_free_mem(adap->tc_u32);
>+}
>+
>+struct cxgb4_tc_u32_table *cxgb4_init_tc_u32(struct adapter *adap,
>+					     unsigned int size)
>+{
>+	struct cxgb4_tc_u32_table *t;
>+	unsigned int i;
>+
>+	if (!size)
>+		return NULL;
>+
>+	t = t4_alloc_mem(sizeof(*t) +
>+			 (size * sizeof(struct cxgb4_link)));
>+	if (!t)
>+		return NULL;
>+
>+	t->size = size;
>+
>+	for (i = 0; i < t->size; i++) {
>+		struct cxgb4_link *link = &t->table[i];
>+		unsigned int bmap_size;
>+		unsigned int max_tids;
>+
>+		max_tids = adap->tids.nftids;
>+		bmap_size = BITS_TO_LONGS(max_tids);
>+		link->tid_map = t4_alloc_mem(sizeof(unsigned long) * bmap_size);
>+		if (!link->tid_map)
>+			goto out_no_mem;
>+		bitmap_zero(link->tid_map, max_tids);
>+	}
>+
>+	return t;
>+
>+out_no_mem:
>+	for (i = 0; i < t->size; i++) {
>+		struct cxgb4_link *link = &t->table[i];
>+
>+		if (link->tid_map)
>+			t4_free_mem(link->tid_map);
>+	}
>+
>+	if (t)
>+		t4_free_mem(t);
>+
>+	return NULL;
>+}
>diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_u32.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_u32.h
>new file mode 100644
>index 0000000..46575843
>--- /dev/null
>+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_u32.h
>@@ -0,0 +1,55 @@
>+/*
>+ * This file is part of the Chelsio T4 Ethernet driver for Linux.
>+ *
>+ * Copyright (c) 2016 Chelsio Communications, Inc. All rights reserved.
>+ *
>+ * This software is available to you under a choice of one of two
>+ * licenses.  You may choose to be licensed under the terms of the GNU
>+ * General Public License (GPL) Version 2, available from the file
>+ * COPYING in the main directory of this source tree, or the
>+ * OpenIB.org BSD license below:
>+ *
>+ *     Redistribution and use in source and binary forms, with or
>+ *     without modification, are permitted provided that the following
>+ *     conditions are met:
>+ *
>+ *      - Redistributions of source code must retain the above
>+ *        copyright notice, this list of conditions and the following
>+ *        disclaimer.
>+ *
>+ *      - Redistributions in binary form must reproduce the above
>+ *        copyright notice, this list of conditions and the following
>+ *        disclaimer in the documentation and/or other materials
>+ *        provided with the distribution.
>+ *
>+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
>+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
>+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
>+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
>+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
>+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
>+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
>+ * SOFTWARE.
>+ */
>+
>+#ifndef __CXGB4_TC_U32_H
>+#define __CXGB4_TC_U32_H
>+
>+#include <net/pkt_cls.h>
>+
>+#define CXGB4_MAX_LINK_HANDLE 32
>+
>+static inline bool can_tc_u32_offload(struct net_device *dev)
>+{
>+	struct adapter *adap = netdev2adap(dev);
>+
>+	return (dev->features & NETIF_F_HW_TC) && adap->tc_u32 ? true : false;
>+}
>+
>+int cxgb4_config_knode(struct net_device *dev, __be16 protocol,
>+		       struct tc_cls_u32_offload *cls);
>+
>+void cxgb4_cleanup_tc_u32(struct adapter *adapter);
>+struct cxgb4_tc_u32_table *cxgb4_init_tc_u32(struct adapter *adap,
>+					     unsigned int size);
>+#endif /* __CXGB4_TC_U32_H */
>diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_u32_parse.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_u32_parse.h
>index 261aa4a..de321bf 100644
>--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_u32_parse.h
>+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_u32_parse.h
>@@ -279,4 +279,16 @@ static const struct cxgb4_next_header cxgb4_ipv6_jumps[] = {
> 	  .jump = cxgb4_udp_fields },
> 	{ .jump = NULL }
> };
>+
>+struct cxgb4_link {
>+	const struct cxgb4_match_field *match_field;  /* Next header */
>+	struct ch_filter_specification fs; /* Match spec associated with link */
>+	u32 link_handle;         /* Knode handle associated with the link */
>+	unsigned long *tid_map;  /* Bitmap for filter tids */
>+};
>+
>+struct cxgb4_tc_u32_table {
>+	unsigned int size;          /* number of entries in table */
>+	struct cxgb4_link table[0]; /* Jump table */
>+};
> #endif /* __CXGB4_TC_U32_PARSE_H */
>-- 
>2.5.3
>