From: frank.blaschka@de.ibm.com
To: davem@davemloft.net
Cc: netdev@vger.kernel.org, linux-s390@vger.kernel.org
Subject: [patch 1/3] [PATCH] qeth: exploit HW TX checksumming
Date: Wed, 12 May 2010 07:34:45 +0200 [thread overview]
Message-ID: <20100512053528.463770000@de.ibm.com> (raw)
In-Reply-To: 20100512053444.035939000@de.ibm.com
[-- Attachment #1: 601-qeth-hw-tx-checksum.diff --]
[-- Type: text/plain, Size: 6549 bytes --]
From: Frank Blaschka <frank.blaschka@de.ibm.com>
OSA supports HW TX checksumming in layer 3 mode. Enable this
feature and remove software fallback used for TSO. Cleanup
checksum bits to indicate OSA can do checksumming only for
IPv4 TCP and UDP.
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
---
drivers/s390/net/qeth_core.h | 3 -
drivers/s390/net/qeth_l3_main.c | 92 +++++++++++++++++++++++++++++-----------
2 files changed, 69 insertions(+), 26 deletions(-)
diff -urpN linux-2.6/drivers/s390/net/qeth_core.h linux-2.6-patched/drivers/s390/net/qeth_core.h
--- linux-2.6/drivers/s390/net/qeth_core.h 2010-05-11 22:10:12.000000000 +0200
+++ linux-2.6-patched/drivers/s390/net/qeth_core.h 2010-05-11 22:10:34.000000000 +0200
@@ -351,7 +351,7 @@ enum qeth_header_ids {
#define QETH_HDR_EXT_SRC_MAC_ADDR 0x08
#define QETH_HDR_EXT_CSUM_HDR_REQ 0x10
#define QETH_HDR_EXT_CSUM_TRANSP_REQ 0x20
-#define QETH_HDR_EXT_UDP_TSO 0x40 /*bit off for TCP*/
+#define QETH_HDR_EXT_UDP 0x40 /*bit off for TCP*/
static inline int qeth_is_last_sbale(struct qdio_buffer_element *sbale)
{
@@ -630,6 +630,7 @@ struct qeth_card_info {
int unique_id;
struct qeth_card_blkt blkt;
__u32 csum_mask;
+ __u32 tx_csum_mask;
enum qeth_ipa_promisc_modes promisc_mode;
};
diff -urpN linux-2.6/drivers/s390/net/qeth_l3_main.c linux-2.6-patched/drivers/s390/net/qeth_l3_main.c
--- linux-2.6/drivers/s390/net/qeth_l3_main.c 2010-05-11 22:10:12.000000000 +0200
+++ linux-2.6-patched/drivers/s390/net/qeth_l3_main.c 2010-05-11 22:10:34.000000000 +0200
@@ -54,16 +54,16 @@ int qeth_l3_set_large_send(struct qeth_c
if (card->options.large_send == QETH_LARGE_SEND_TSO) {
if (qeth_is_supported(card, IPA_OUTBOUND_TSO)) {
card->dev->features |= NETIF_F_TSO | NETIF_F_SG |
- NETIF_F_HW_CSUM;
+ NETIF_F_IP_CSUM;
} else {
card->dev->features &= ~(NETIF_F_TSO | NETIF_F_SG |
- NETIF_F_HW_CSUM);
+ NETIF_F_IP_CSUM);
card->options.large_send = QETH_LARGE_SEND_NO;
rc = -EOPNOTSUPP;
}
} else {
card->dev->features &= ~(NETIF_F_TSO | NETIF_F_SG |
- NETIF_F_HW_CSUM);
+ NETIF_F_IP_CSUM);
card->options.large_send = QETH_LARGE_SEND_NO;
}
return rc;
@@ -1108,6 +1108,13 @@ static int qeth_l3_default_setassparms_c
card->info.csum_mask = cmd->data.setassparms.data.flags_32bit;
QETH_DBF_TEXT_(TRACE, 3, "csum:%d", card->info.csum_mask);
}
+ if (cmd->data.setassparms.hdr.assist_no == IPA_OUTBOUND_CHECKSUM &&
+ cmd->data.setassparms.hdr.command_code == IPA_CMD_ASS_START) {
+ card->info.tx_csum_mask =
+ cmd->data.setassparms.data.flags_32bit;
+ QETH_DBF_TEXT_(TRACE, 3, "tcsu:%d", card->info.tx_csum_mask);
+ }
+
return 0;
}
@@ -1536,6 +1543,28 @@ static int qeth_l3_start_ipa_checksum(st
return rc;
}
+static int qeth_l3_start_ipa_tx_checksum(struct qeth_card *card)
+{
+ int rc = 0;
+
+ if (!qeth_is_supported(card, IPA_OUTBOUND_CHECKSUM))
+ return rc;
+ rc = qeth_l3_send_simple_setassparms(card, IPA_OUTBOUND_CHECKSUM,
+ IPA_CMD_ASS_START, 0);
+ if (rc)
+ goto err_out;
+ rc = qeth_l3_send_simple_setassparms(card, IPA_OUTBOUND_CHECKSUM,
+ IPA_CMD_ASS_ENABLE, card->info.tx_csum_mask);
+ if (rc)
+ goto err_out;
+ dev_info(&card->gdev->dev, "HW TX Checksumming enabled\n");
+ return rc;
+err_out:
+ dev_warn(&card->gdev->dev, "Enabling HW TX checksumming for %s "
+ "failed, using SW TX checksumming\n", QETH_CARD_IFNAME(card));
+ return rc;
+}
+
static int qeth_l3_start_ipa_tso(struct qeth_card *card)
{
int rc;
@@ -1578,6 +1607,7 @@ static int qeth_l3_start_ipassists(struc
qeth_l3_start_ipa_ipv6(card); /* go on*/
qeth_l3_start_ipa_broadcast(card); /* go on*/
qeth_l3_start_ipa_checksum(card); /* go on*/
+ qeth_l3_start_ipa_tx_checksum(card);
qeth_l3_start_ipa_tso(card); /* go on*/
return 0;
}
@@ -2817,6 +2847,21 @@ static void qeth_l3_fill_header(struct q
}
}
+static inline void qeth_l3_hdr_csum(struct qeth_card *card,
+ struct qeth_hdr *hdr, struct sk_buff *skb)
+{
+ struct iphdr *iph = ip_hdr(skb);
+
+ /* tcph->check contains already the pseudo hdr checksum
+ * so just set the header flags
+ */
+ if (iph->protocol == IPPROTO_UDP)
+ hdr->hdr.l3.ext_flags |= QETH_HDR_EXT_UDP;
+ hdr->hdr.l3.ext_flags |= QETH_HDR_EXT_CSUM_TRANSP_REQ;
+ if (card->options.performance_stats)
+ card->perf_stats.tx_csum++;
+}
+
static void qeth_tso_fill_header(struct qeth_card *card,
struct qeth_hdr *qhdr, struct sk_buff *skb)
{
@@ -2852,21 +2897,6 @@ static void qeth_tso_fill_header(struct
}
}
-static void qeth_tx_csum(struct sk_buff *skb)
-{
- __wsum csum;
- int offset;
-
- skb_set_transport_header(skb, skb->csum_start - skb_headroom(skb));
- offset = skb->csum_start - skb_headroom(skb);
- BUG_ON(offset >= skb_headlen(skb));
- csum = skb_checksum(skb, offset, skb->len - offset, 0);
-
- offset += skb->csum_offset;
- BUG_ON(offset + sizeof(__sum16) > skb_headlen(skb));
- *(__sum16 *)(skb->data + offset) = csum_fold(csum);
-}
-
static inline int qeth_l3_tso_elements(struct sk_buff *skb)
{
unsigned long tcpd = (unsigned long)tcp_hdr(skb) +
@@ -2923,12 +2953,6 @@ static int qeth_l3_hard_start_xmit(struc
if (skb_is_gso(skb))
large_send = card->options.large_send;
- else
- if (skb->ip_summed == CHECKSUM_PARTIAL) {
- qeth_tx_csum(skb);
- if (card->options.performance_stats)
- card->perf_stats.tx_csum++;
- }
if ((card->info.type == QETH_CARD_TYPE_IQD) && (!large_send) &&
(skb_shinfo(skb)->nr_frags == 0)) {
@@ -3007,6 +3031,9 @@ static int qeth_l3_hard_start_xmit(struc
cast_type);
hdr->hdr.l3.length = new_skb->len - data_offset;
}
+
+ if (skb->ip_summed == CHECKSUM_PARTIAL)
+ qeth_l3_hdr_csum(card, hdr, new_skb);
}
elems = qeth_get_elements_no(card, (void *)hdr, new_skb,
@@ -3132,10 +3159,25 @@ static int qeth_l3_ethtool_set_tso(struc
return rc;
}
+static int qeth_l3_ethtool_set_tx_csum(struct net_device *dev, u32 data)
+{
+ struct qeth_card *card = dev->ml_priv;
+
+ if (data) {
+ if (qeth_is_supported(card, IPA_OUTBOUND_CHECKSUM))
+ dev->features |= NETIF_F_IP_CSUM;
+ else
+ return -EPERM;
+ } else
+ dev->features &= ~NETIF_F_IP_CSUM;
+
+ return 0;
+}
+
static const struct ethtool_ops qeth_l3_ethtool_ops = {
.get_link = ethtool_op_get_link,
.get_tx_csum = ethtool_op_get_tx_csum,
- .set_tx_csum = ethtool_op_set_tx_hw_csum,
+ .set_tx_csum = qeth_l3_ethtool_set_tx_csum,
.get_rx_csum = qeth_l3_ethtool_get_rx_csum,
.set_rx_csum = qeth_l3_ethtool_set_rx_csum,
.get_sg = ethtool_op_get_sg,
next prev parent reply other threads:[~2010-05-12 5:34 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-05-12 5:34 [patch 0/3] s390: qeth patches for 2.6.35 frank.blaschka
2010-05-12 5:34 ` frank.blaschka [this message]
2010-05-12 5:34 ` [patch 2/3] [PATCH] qeth: new message if OLM limit is reached frank.blaschka
2010-05-12 5:34 ` [patch 3/3] [PATCH] qeth: synchronize configuration interface frank.blaschka
2010-05-16 7:50 ` [patch 0/3] s390: qeth patches for 2.6.35 David Miller
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20100512053528.463770000@de.ibm.com \
--to=frank.blaschka@de.ibm.com \
--cc=davem@davemloft.net \
--cc=linux-s390@vger.kernel.org \
--cc=netdev@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.