* [PATCH net-next 4/8] tg3: Relax 5717 serdes restriction
From: Matt Carlson @ 2010-07-11 19:31 UTC (permalink / raw)
To: davem; +Cc: netdev, andy, mcarlson
tg3 is coded to refuse to attach to 5717 serdes devices. Now that the
hardware is better supported, we can relax this restriction. This patch
also fixes a recently introduced bug which will cause serdes parallel
detection not to work with 5780 class devices.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
---
drivers/net/tg3.c | 5 ++---
1 files changed, 2 insertions(+), 3 deletions(-)
diff --git a/drivers/net/tg3.c b/drivers/net/tg3.c
index efac448..c91049b 100644
--- a/drivers/net/tg3.c
+++ b/drivers/net/tg3.c
@@ -8551,7 +8551,7 @@ static void tg3_timer(unsigned long __opaque)
tg3_setup_phy(tp, 0);
}
} else if ((tp->tg3_flags2 & TG3_FLG2_MII_SERDES) &&
- !(tp->tg3_flags2 & TG3_FLG2_5780_CLASS)) {
+ (tp->tg3_flags2 & TG3_FLG2_5780_CLASS)) {
tg3_serdes_parallel_detect(tp);
}
@@ -13427,8 +13427,7 @@ static int __devinit tg3_get_invariants(struct tg3 *tp)
return err;
if (GET_ASIC_REV(tp->pci_chip_rev_id) == ASIC_REV_5717 &&
- (tp->pci_chip_rev_id != CHIPREV_ID_5717_A0 ||
- (tp->tg3_flags2 & TG3_FLG2_MII_SERDES)))
+ tp->pci_chip_rev_id != CHIPREV_ID_5717_A0)
return -ENOTSUPP;
/* Initialize data/descriptor byte/word swapping. */
--
1.6.4.4
^ permalink raw reply related
* [PATCH net-next 1/8] tg3: Revert RSS indir tbl setup change
From: Matt Carlson @ 2010-07-11 19:31 UTC (permalink / raw)
To: davem; +Cc: netdev, andy, mcarlson
This patch reverts commit 2601d8a0049c8b5d29cd5adb844a305a804e505f. A
spectacular set of coincidences made it look as though the table was
setup incorrectly. The original version was correct.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
---
drivers/net/tg3.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/net/tg3.c b/drivers/net/tg3.c
index aa2d64f..89aa486 100644
--- a/drivers/net/tg3.c
+++ b/drivers/net/tg3.c
@@ -8237,7 +8237,7 @@ static int tg3_reset_hw(struct tg3 *tp, int reset_phy)
for (i = 0; i < TG3_RSS_INDIR_TBL_SIZE; i++) {
int idx = i % sizeof(val);
- ent[idx] = (i % (tp->irq_cnt - 1)) + 1;
+ ent[idx] = i % (tp->irq_cnt - 1);
if (idx == sizeof(val) - 1) {
tw32(reg, val);
reg += 4;
--
1.6.4.4
^ permalink raw reply related
* [PATCH net-next 0/8] tg3 bugfixes
From: Matt Carlson @ 2010-07-11 19:31 UTC (permalink / raw)
To: davem; +Cc: netdev, andy, mcarlson
This patchset fixes a few MSI-X related problems and performs a
few driver enhancements.
^ permalink raw reply
* [PATCH net-next 2/8] tg3: Fix single MSI-X vector coalescing
From: Matt Carlson @ 2010-07-11 19:31 UTC (permalink / raw)
To: davem; +Cc: netdev, andy, mcarlson
The interrupt coalescing setup code used the TG3_FLG2_USING_MSIX flag to
determine whether or not to configure the rx coalescing parameters.
This is incorrect for the single MSI-X vector case. This patch changes
the code to look at the TG3_FLG3_ENABLE_RSS instead.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
---
drivers/net/tg3.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/net/tg3.c b/drivers/net/tg3.c
index 89aa486..57dba79 100644
--- a/drivers/net/tg3.c
+++ b/drivers/net/tg3.c
@@ -7447,7 +7447,7 @@ static void __tg3_set_coalesce(struct tg3 *tp, struct ethtool_coalesce *ec)
tw32(HOSTCC_TXCOAL_MAXF_INT, 0);
}
- if (!(tp->tg3_flags2 & TG3_FLG2_USING_MSIX)) {
+ if (!(tp->tg3_flags3 & TG3_FLG3_ENABLE_RSS)) {
tw32(HOSTCC_RXCOL_TICKS, ec->rx_coalesce_usecs);
tw32(HOSTCC_RXMAX_FRAMES, ec->rx_max_coalesced_frames);
tw32(HOSTCC_RXCOAL_MAXF_INT, ec->rx_max_coalesced_frames_irq);
--
1.6.4.4
^ permalink raw reply related
* [PATCH net-next 3/8] tg3: Fix IPv6 TSO code in tg3_start_xmit_dma_bug()
From: Matt Carlson @ 2010-07-11 19:31 UTC (permalink / raw)
To: davem; +Cc: netdev, andy, mcarlson
The tg3_start_xmit_dma_bug() function was missing code to process IPv6
TSO packets. This patch adds the missing support.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
---
drivers/net/tg3.c | 20 ++++++++++++++------
1 files changed, 14 insertions(+), 6 deletions(-)
diff --git a/drivers/net/tg3.c b/drivers/net/tg3.c
index 57dba79..efac448 100644
--- a/drivers/net/tg3.c
+++ b/drivers/net/tg3.c
@@ -5779,7 +5779,7 @@ static netdev_tx_t tg3_start_xmit_dma_bug(struct sk_buff *skb,
if ((mss = skb_shinfo(skb)->gso_size) != 0) {
struct iphdr *iph;
- u32 tcp_opt_len, ip_tcp_len, hdr_len;
+ u32 tcp_opt_len, hdr_len;
if (skb_header_cloned(skb) &&
pskb_expand_head(skb, 0, 0, GFP_ATOMIC)) {
@@ -5787,10 +5787,21 @@ static netdev_tx_t tg3_start_xmit_dma_bug(struct sk_buff *skb,
goto out_unlock;
}
+ iph = ip_hdr(skb);
tcp_opt_len = tcp_optlen(skb);
- ip_tcp_len = ip_hdrlen(skb) + sizeof(struct tcphdr);
- hdr_len = ip_tcp_len + tcp_opt_len;
+ if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6) {
+ hdr_len = skb_headlen(skb) - ETH_HLEN;
+ } else {
+ u32 ip_tcp_len;
+
+ ip_tcp_len = ip_hdrlen(skb) + sizeof(struct tcphdr);
+ hdr_len = ip_tcp_len + tcp_opt_len;
+
+ iph->check = 0;
+ iph->tot_len = htons(mss + hdr_len);
+ }
+
if (unlikely((ETH_HLEN + hdr_len) > 80) &&
(tp->tg3_flags2 & TG3_FLG2_TSO_BUG))
return tg3_tso_bug(tp, skb);
@@ -5798,9 +5809,6 @@ static netdev_tx_t tg3_start_xmit_dma_bug(struct sk_buff *skb,
base_flags |= (TXD_FLAG_CPU_PRE_DMA |
TXD_FLAG_CPU_POST_DMA);
- iph = ip_hdr(skb);
- iph->check = 0;
- iph->tot_len = htons(mss + hdr_len);
if (tp->tg3_flags2 & TG3_FLG2_HW_TSO) {
tcp_hdr(skb)->check = 0;
base_flags &= ~TXD_FLAG_TCPUDP_CSUM;
--
1.6.4.4
^ permalink raw reply related
* [PATCH net-next 6/8] tg3: Revert PCIe tx glitch fix
From: Matt Carlson @ 2010-07-11 19:31 UTC (permalink / raw)
To: davem; +Cc: netdev, andy, mcarlson
This patch reverts commit 52cdf8526fe24f11d300b75458ddee017f3f4c88,
entitled "tg3: Prevent a PCIe tx glitch". The problem does not have
any visible side-effects and happens too early for the driver to do
anything about it. The proper place for this code is within the
device's bootcode.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
---
drivers/net/tg3.c | 24 ------------------------
drivers/net/tg3.h | 22 ----------------------
2 files changed, 0 insertions(+), 46 deletions(-)
diff --git a/drivers/net/tg3.c b/drivers/net/tg3.c
index e6da16a..f61a4d8 100644
--- a/drivers/net/tg3.c
+++ b/drivers/net/tg3.c
@@ -7072,30 +7072,6 @@ static int tg3_chip_reset(struct tg3 *tp)
tg3_mdio_start(tp);
- if (GET_ASIC_REV(tp->pci_chip_rev_id) == ASIC_REV_57780) {
- u8 phy_addr;
-
- phy_addr = tp->phy_addr;
- tp->phy_addr = TG3_PHY_PCIE_ADDR;
-
- tg3_writephy(tp, TG3_PCIEPHY_BLOCK_ADDR,
- TG3_PCIEPHY_TXB_BLK << TG3_PCIEPHY_BLOCK_SHIFT);
- val = TG3_PCIEPHY_TX0CTRL1_TXOCM | TG3_PCIEPHY_TX0CTRL1_RDCTL |
- TG3_PCIEPHY_TX0CTRL1_TXCMV | TG3_PCIEPHY_TX0CTRL1_TKSEL |
- TG3_PCIEPHY_TX0CTRL1_NB_EN;
- tg3_writephy(tp, TG3_PCIEPHY_TX0CTRL1, val);
- udelay(10);
-
- tg3_writephy(tp, TG3_PCIEPHY_BLOCK_ADDR,
- TG3_PCIEPHY_XGXS_BLK1 << TG3_PCIEPHY_BLOCK_SHIFT);
- val = TG3_PCIEPHY_PWRMGMT4_LOWPWR_EN |
- TG3_PCIEPHY_PWRMGMT4_L1PLLPD_EN;
- tg3_writephy(tp, TG3_PCIEPHY_PWRMGMT4, val);
- udelay(10);
-
- tp->phy_addr = phy_addr;
- }
-
if ((tp->tg3_flags2 & TG3_FLG2_PCI_EXPRESS) &&
tp->pci_chip_rev_id != CHIPREV_ID_5750_A0 &&
GET_ASIC_REV(tp->pci_chip_rev_id) != ASIC_REV_5785 &&
diff --git a/drivers/net/tg3.h b/drivers/net/tg3.h
index be7ab91..0432399 100644
--- a/drivers/net/tg3.h
+++ b/drivers/net/tg3.h
@@ -2032,31 +2032,9 @@
/* Currently this is fixed. */
-#define TG3_PHY_PCIE_ADDR 0x00
#define TG3_PHY_MII_ADDR 0x01
-/*** Tigon3 specific PHY PCIE registers. ***/
-
-#define TG3_PCIEPHY_BLOCK_ADDR 0x1f
-#define TG3_PCIEPHY_XGXS_BLK1 0x0801
-#define TG3_PCIEPHY_TXB_BLK 0x0861
-#define TG3_PCIEPHY_BLOCK_SHIFT 4
-
-/* TG3_PCIEPHY_TXB_BLK */
-#define TG3_PCIEPHY_TX0CTRL1 0x15
-#define TG3_PCIEPHY_TX0CTRL1_TXOCM 0x0003
-#define TG3_PCIEPHY_TX0CTRL1_RDCTL 0x0008
-#define TG3_PCIEPHY_TX0CTRL1_TXCMV 0x0030
-#define TG3_PCIEPHY_TX0CTRL1_TKSEL 0x0040
-#define TG3_PCIEPHY_TX0CTRL1_NB_EN 0x0400
-
-/* TG3_PCIEPHY_XGXS_BLK1 */
-#define TG3_PCIEPHY_PWRMGMT4 0x1a
-#define TG3_PCIEPHY_PWRMGMT4_L1PLLPD_EN 0x0038
-#define TG3_PCIEPHY_PWRMGMT4_LOWPWR_EN 0x4000
-
-
/*** Tigon3 specific PHY MII registers. ***/
#define TG3_BMCR_SPEED1000 0x0040
--
1.6.4.4
^ permalink raw reply related
* [PATCH net-next 5/8] tg3: Report driver version to firmware
From: Matt Carlson @ 2010-07-11 19:31 UTC (permalink / raw)
To: davem; +Cc: netdev, andy, mcarlson
This patch changes the code so that the driver version can be reported
to the firmware in addition to the current use.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
---
drivers/net/tg3.c | 8 ++++++--
drivers/net/tg3.h | 4 +++-
2 files changed, 9 insertions(+), 3 deletions(-)
diff --git a/drivers/net/tg3.c b/drivers/net/tg3.c
index c91049b..e6da16a 100644
--- a/drivers/net/tg3.c
+++ b/drivers/net/tg3.c
@@ -18,6 +18,7 @@
#include <linux/module.h>
#include <linux/moduleparam.h>
+#include <linux/stringify.h>
#include <linux/kernel.h>
#include <linux/types.h>
#include <linux/compiler.h>
@@ -67,7 +68,10 @@
#include "tg3.h"
#define DRV_MODULE_NAME "tg3"
-#define DRV_MODULE_VERSION "3.111"
+#define TG3_MAJ_NUM 3
+#define TG3_MIN_NUM 111
+#define DRV_MODULE_VERSION \
+ __stringify(TG3_MAJ_NUM) "." __stringify(TG3_MIN_NUM)
#define DRV_MODULE_RELDATE "June 5, 2010"
#define TG3_DEF_MAC_MODE 0
@@ -6629,7 +6633,7 @@ static void tg3_ape_driver_state_change(struct tg3 *tp, int kind)
apedata = tg3_ape_read32(tp, TG3_APE_HOST_INIT_COUNT);
tg3_ape_write32(tp, TG3_APE_HOST_INIT_COUNT, ++apedata);
tg3_ape_write32(tp, TG3_APE_HOST_DRIVER_ID,
- APE_HOST_DRIVER_ID_MAGIC);
+ APE_HOST_DRIVER_ID_MAGIC(TG3_MAJ_NUM, TG3_MIN_NUM));
tg3_ape_write32(tp, TG3_APE_HOST_BEHAVIOR,
APE_HOST_BEHAV_NO_PHYLOCK);
diff --git a/drivers/net/tg3.h b/drivers/net/tg3.h
index b72ac52..be7ab91 100644
--- a/drivers/net/tg3.h
+++ b/drivers/net/tg3.h
@@ -2193,7 +2193,9 @@
#define APE_HOST_SEG_LEN_MAGIC 0x0000001c
#define TG3_APE_HOST_INIT_COUNT 0x4208
#define TG3_APE_HOST_DRIVER_ID 0x420c
-#define APE_HOST_DRIVER_ID_MAGIC 0xf0035100
+#define APE_HOST_DRIVER_ID_LINUX 0xf0000000
+#define APE_HOST_DRIVER_ID_MAGIC(maj, min) \
+ (APE_HOST_DRIVER_ID_LINUX | (maj & 0xff) << 16 | (min & 0xff) << 8)
#define TG3_APE_HOST_BEHAVIOR 0x4210
#define APE_HOST_BEHAV_NO_PHYLOCK 0x00000001
#define TG3_APE_HOST_HEARTBEAT_INT_MS 0x4214
--
1.6.4.4
^ permalink raw reply related
* [PATCH net-next 8/8] tg3: Update version to 3.112
From: Matt Carlson @ 2010-07-11 19:31 UTC (permalink / raw)
To: davem; +Cc: netdev, andy, mcarlson
This patch updates the tg3 version to 3.112.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
---
drivers/net/tg3.c | 4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/tg3.c b/drivers/net/tg3.c
index 7116a47..5769e15 100644
--- a/drivers/net/tg3.c
+++ b/drivers/net/tg3.c
@@ -69,10 +69,10 @@
#define DRV_MODULE_NAME "tg3"
#define TG3_MAJ_NUM 3
-#define TG3_MIN_NUM 111
+#define TG3_MIN_NUM 112
#define DRV_MODULE_VERSION \
__stringify(TG3_MAJ_NUM) "." __stringify(TG3_MIN_NUM)
-#define DRV_MODULE_RELDATE "June 5, 2010"
+#define DRV_MODULE_RELDATE "July 11, 2010"
#define TG3_DEF_MAC_MODE 0
#define TG3_DEF_RX_MODE 0
--
1.6.4.4
^ permalink raw reply related
* [PATCH net-next 7/8] tg3: Fix some checkpatch errors
From: Matt Carlson @ 2010-07-11 19:31 UTC (permalink / raw)
To: davem; +Cc: netdev, andy, mcarlson
This patch fixes the following checkpatch errors:
ERROR: do not use assignment in if condition
+ if ((mss = skb_shinfo(skb)->gso_size) != 0) {
ERROR: do not use assignment in if condition
+ if ((mss = skb_shinfo(skb)->gso_size) != 0) {
ERROR: space prohibited after that '!' (ctx:BxW)
+ if (! netif_carrier_ok(tp->dev) &&
^
ERROR: space required after that ',' (ctx:VxV)
+#define GET_REG32_LOOP(base,len) \
^
ERROR: "(foo*)" should be "(foo *)"
+ memcpy(data, ((char*)&val) + b_offset, b_count);
ERROR: do not use assignment in if condition
+ if ((err = tg3_do_mem_test(tp, mem_tbl[i].offset,
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
---
drivers/net/tg3.c | 17 +++++++++--------
1 files changed, 9 insertions(+), 8 deletions(-)
diff --git a/drivers/net/tg3.c b/drivers/net/tg3.c
index f61a4d8..7116a47 100644
--- a/drivers/net/tg3.c
+++ b/drivers/net/tg3.c
@@ -5574,8 +5574,8 @@ static netdev_tx_t tg3_start_xmit(struct sk_buff *skb,
entry = tnapi->tx_prod;
base_flags = 0;
- mss = 0;
- if ((mss = skb_shinfo(skb)->gso_size) != 0) {
+ mss = skb_shinfo(skb)->gso_size;
+ if (mss) {
int tcp_opt_len, ip_tcp_len;
u32 hdrlen;
@@ -5781,7 +5781,8 @@ static netdev_tx_t tg3_start_xmit_dma_bug(struct sk_buff *skb,
if (skb->ip_summed == CHECKSUM_PARTIAL)
base_flags |= TXD_FLAG_TCPUDP_CSUM;
- if ((mss = skb_shinfo(skb)->gso_size) != 0) {
+ mss = skb_shinfo(skb)->gso_size;
+ if (mss) {
struct iphdr *iph;
u32 tcp_opt_len, hdr_len;
@@ -8514,7 +8515,7 @@ static void tg3_timer(unsigned long __opaque)
(mac_stat & MAC_STATUS_LNKSTATE_CHANGED)) {
need_setup = 1;
}
- if (! netif_carrier_ok(tp->dev) &&
+ if (!netif_carrier_ok(tp->dev) &&
(mac_stat & (MAC_STATUS_PCS_SYNCED |
MAC_STATUS_SIGNAL_DET))) {
need_setup = 1;
@@ -9372,7 +9373,7 @@ static void tg3_get_regs(struct net_device *dev,
tg3_full_lock(tp, 0);
#define __GET_REG32(reg) (*(p)++ = tr32(reg))
-#define GET_REG32_LOOP(base,len) \
+#define GET_REG32_LOOP(base, len) \
do { p = (u32 *)(orig_p + (base)); \
for (i = 0; i < len; i += 4) \
__GET_REG32((base) + i); \
@@ -9465,7 +9466,7 @@ static int tg3_get_eeprom(struct net_device *dev, struct ethtool_eeprom *eeprom,
ret = tg3_nvram_read_be32(tp, offset-b_offset, &val);
if (ret)
return ret;
- memcpy(data, ((char*)&val) + b_offset, b_count);
+ memcpy(data, ((char *)&val) + b_offset, b_count);
len -= b_count;
offset += b_count;
eeprom->len += b_count;
@@ -10585,8 +10586,8 @@ static int tg3_test_memory(struct tg3 *tp)
mem_tbl = mem_tbl_570x;
for (i = 0; mem_tbl[i].offset != 0xffffffff; i++) {
- if ((err = tg3_do_mem_test(tp, mem_tbl[i].offset,
- mem_tbl[i].len)) != 0)
+ err = tg3_do_mem_test(tp, mem_tbl[i].offset, mem_tbl[i].len);
+ if (err)
break;
}
--
1.6.4.4
^ permalink raw reply related
* Re: oops in tcp_xmit_retransmit_queue() w/ v2.6.32.15
From: Ilpo Järvinen @ 2010-07-11 19:44 UTC (permalink / raw)
To: Tejun Heo
Cc: Eric Dumazet, David S. Miller, lkml, netdev@vger.kernel.org,
Fehrmann, Henning, Carsten Aulbert
In-Reply-To: <alpine.DEB.2.00.1007112224110.15736@melkinpaasi.cs.helsinki.fi>
[-- Attachment #1: Type: TEXT/PLAIN, Size: 4203 bytes --]
On Sun, 11 Jul 2010, Ilpo Järvinen wrote:
> On Sun, 11 Jul 2010, Ilpo Järvinen wrote:
>
> > On Sun, 11 Jul 2010, Eric Dumazet wrote:
> >
> > > Le dimanche 11 juillet 2010 à 19:46 +0200, Eric Dumazet a écrit :
> > > > Le dimanche 11 juillet 2010 à 19:06 +0200, Eric Dumazet a écrit :
> > > > > Le dimanche 11 juillet 2010 à 19:09 +0300, Ilpo Järvinen a écrit :
> > > > > > On Thu, 8 Jul 2010, Tejun Heo wrote:
> > > > > >
> > > > > > > We've been seeing oops in tcp_xmit_retransmit_queue() w/ 2.6.32.15.
> > > > > > > Please see the attached photoshoot. This is happening on a HPC
> > > > > > > cluster and very interestingly caused by one particular job. How long
> > > > > > > it takes isn't clear yet (at least more than a day) but when it
> > > > > > > happens it happens on a lot of machines in relatively short time.
> > > > > > >
> > > > > > > With a bit of disassemblying, I've found that the oops is happening
> > > > > > > during tcp_for_write_queue_from() because the skb->next points to
> > > > > > > NULL.
> > > > > > >
> > > > > > > void tcp_xmit_retransmit_queue(struct sock *sk)
> > > > > > > {
> > > > > > > ...
> > > > > > > if (tp->retransmit_skb_hint) {
> > > > > > > skb = tp->retransmit_skb_hint;
> > > > > > > last_lost = TCP_SKB_CB(skb)->end_seq;
> > > > > > > if (after(last_lost, tp->retransmit_high))
> > > > > > > last_lost = tp->retransmit_high;
> > > > > > > } else {
> > > > > > > skb = tcp_write_queue_head(sk);
> > > > > > > last_lost = tp->snd_una;
> > > > > > > }
> > > > > > >
> > > > > > > => tcp_for_write_queue_from(skb, sk) {
> > > > > > > __u8 sacked = TCP_SKB_CB(skb)->sacked;
> > > > > > >
> > > > > > > if (skb == tcp_send_head(sk))
> > > > > > > break;
> > > > > > > /* we could do better than to assign each time */
> > > > > > > if (hole == NULL)
> > > > > > >
> > > > > > > This can happen for one of the following reasons,
> > > > > > >
> > > > > > > 1. tp->retransmit_skb_hint is NULL and tcp_write_queue_head() is NULL
> > > > > > > too. ie. tcp_xmit_retransmit_queue() is called on an empty write
> > > > > > > queue for some reason.
> > > > > > >
> > > > > > > 2. tp->retransmit_skb_hint is pointing to a skb which is not on the
> > > > > > > write_queue. ie. somebody forgot to update hint while removing the
> > > > > > > skb from the write queue.
> > > > > >
> > > > > > Once again I've read the unlinkers through, and only thing that could
> > > > > > cause this is tcp_send_synack (others do deal with the hints) but I think
> > > > > > Eric already proposed a patch to that but we never got anywhere due to
> > > > > > some counterargument why it wouldn't take place (too far away for me to
> > > > > > remember, see archives about the discussions). ...But if you want be dead
> > > > > > sure some WARN_ON there might not hurt. Also the purging of the whole
> > > > > > queue was a similar suspect I then came across (but that would only
> > > > > > materialize with sk reuse happening e.g., with nfs which the other guys
> > > > > > weren't using).
> > > > > >
> > > > >
> > > > > Hmm.
> > > > >
> > > > > This sounds familiar to me, but I cannot remember the discussion you
> > > > > mention or the patch.
> > > > >
> > > > > Or maybe it was the TCP transaction thing ? (including data in SYN or
> > > > > SYN-ACK packet)
> >
> > No. That's another thing. ...I've already found it with google today but
> > cannot seem to find it again. I thought I used tcp_make_synack eric but
> > for some reason I only get these transaction fix hits. I'll keep looking.
>
> Right, this one:
>
> http://kerneltrap.org/mailarchive/linux-netdev/2009/10/29/6259073
Hmm, another idea... It might be useful to try to disable
tcp_retrans_try_collapse in tcp_retransmit_skb as a test. I think that it
might be possible that tcp_xmit_retransmit_queue might end up holding
a stale reference either in hole or skb. Kind of shot into the dark still,
no actual theory on how that could happen but that
tcp_xmit_retransmit_queue logic is rather tricky because of returning to
the first hole if such exists so that I couldn't immediately rule out the
possibility.
--
i.
^ permalink raw reply
* [PATCH 00/12] autoconvert trivial BKL users to private mutex
From: Arnd Bergmann @ 2010-07-11 21:18 UTC (permalink / raw)
To: linux-kernel
Cc: devel, Jesper Nilsson, linux-usb, Arnd Bergmann, Corey Minyard,
linux-cris-kernel, Frederic Weisbecker, Greg Kroah-Hartman,
Karsten Keil, Mauro Carvalho Chehab, linux-scsi, linuxppc-dev,
James E.J. Bottomley, linux-mtd, John Kacur, netdev, linux-media,
openipmi-developer, David S. Miller, David Woodhouse
This is a repost of an earlier patch to remove
those users of the big kernel lock that can be
converted to a mutex using a simple script.
The only use of the BKL is in file operations
that are called without any other lock, so the
new mutex is the top-level serialization
and cannot introduce any AB-BA deadlock.
Please apply to the respective maintainer trees
if the patches look good.
Arnd Bergmann (12):
staging: autoconvert trivial BKL users to private mutex
isdn: autoconvert trivial BKL users to private mutex
scsi: autoconvert trivial BKL users to private mutex
media: autoconvert trivial BKL users to private mutex
usb: autoconvert trivial BKL users to private mutex
net: autoconvert trivial BKL users to private mutex
cris: autoconvert trivial BKL users to private mutex
sbus: autoconvert trivial BKL users to private mutex
mtd: autoconvert trivial BKL users to private mutex
mac: autoconvert trivial BKL users to private mutex
ipmi: autoconvert trivial BKL users to private mutex
drivers: autoconvert trivial BKL users to private mutex
arch/cris/arch-v10/drivers/eeprom.c | 2 -
arch/cris/arch-v10/drivers/i2c.c | 2 -
arch/cris/arch-v32/drivers/cryptocop.c | 2 -
arch/cris/arch-v32/drivers/i2c.c | 12 ++++----
drivers/block/paride/pg.c | 7 ++--
drivers/block/paride/pt.c | 19 ++++++------
drivers/char/apm-emulation.c | 11 ++++---
drivers/char/applicom.c | 9 +++--
drivers/char/ds1302.c | 15 +++++----
drivers/char/ds1620.c | 8 ++--
drivers/char/dsp56k.c | 27 +++++++++--------
drivers/char/dtlk.c | 8 ++--
drivers/char/generic_nvram.c | 7 ++--
drivers/char/genrtc.c | 13 ++++----
drivers/char/i8k.c | 7 ++--
drivers/char/ip2/ip2main.c | 8 ++--
drivers/char/ipmi/ipmi_devintf.c | 14 ++++----
drivers/char/ipmi/ipmi_watchdog.c | 8 ++--
drivers/char/lp.c | 15 +++++----
drivers/char/mbcs.c | 8 ++--
drivers/char/mmtimer.c | 7 ++--
drivers/char/mwave/mwavedd.c | 44 ++++++++++++++--------------
drivers/char/nvram.c | 11 ++++---
drivers/char/nwflash.c | 12 ++++----
drivers/char/pcmcia/cm4000_cs.c | 11 ++++---
drivers/char/pcmcia/cm4040_cs.c | 7 ++--
drivers/char/ppdev.c | 8 ++--
drivers/char/rio/rio_linux.c | 7 ++--
drivers/char/snsc.c | 9 +++--
drivers/char/toshiba.c | 9 +++--
drivers/char/viotape.c | 11 ++++---
drivers/char/xilinx_hwicap/xilinx_hwicap.c | 6 ++--
drivers/hwmon/fschmd.c | 6 ++--
drivers/hwmon/w83793.c | 6 ++--
drivers/input/misc/hp_sdc_rtc.c | 7 ++--
drivers/isdn/capi/capi.c | 6 ++--
drivers/isdn/divert/divert_procfs.c | 7 ++--
drivers/isdn/hardware/eicon/divamnt.c | 7 ++--
drivers/isdn/hardware/eicon/divasi.c | 2 -
drivers/isdn/hardware/eicon/divasmain.c | 2 -
drivers/isdn/hysdn/hysdn_procconf.c | 21 +++++++------
drivers/isdn/hysdn/hysdn_proclog.c | 15 +++++----
drivers/isdn/i4l/isdn_common.c | 27 +++++++++--------
drivers/isdn/mISDN/timerdev.c | 7 ++--
drivers/macintosh/adb.c | 10 +++---
drivers/macintosh/smu.c | 6 ++--
drivers/macintosh/via-pmu.c | 11 ++++---
drivers/media/dvb/bt8xx/dst_ca.c | 7 ++--
drivers/media/video/cx88/cx88-blackbird.c | 13 ++++----
drivers/media/video/dabusb.c | 18 ++++++------
drivers/media/video/se401.c | 9 +++--
drivers/media/video/stradis.c | 9 +++--
drivers/media/video/usbvideo/vicam.c | 14 ++++----
drivers/message/fusion/mptctl.c | 15 +++++----
drivers/message/i2o/i2o_config.c | 23 +++++++-------
drivers/misc/phantom.c | 11 ++++---
drivers/mtd/mtdchar.c | 15 +++++----
drivers/net/ppp_generic.c | 19 ++++++------
drivers/net/wan/cosa.c | 10 +++---
drivers/pci/hotplug/cpqphp_sysfs.c | 13 ++++----
drivers/rtc/rtc-m41t80.c | 13 ++++----
drivers/sbus/char/display7seg.c | 8 ++--
drivers/sbus/char/envctrl.c | 2 -
drivers/sbus/char/flash.c | 15 +++++----
drivers/sbus/char/openprom.c | 15 +++++----
drivers/sbus/char/uctrl.c | 7 ++--
drivers/scsi/3w-9xxx.c | 7 ++--
drivers/scsi/3w-sas.c | 7 ++--
drivers/scsi/3w-xxxx.c | 9 ++---
drivers/scsi/aacraid/linit.c | 15 +++++----
drivers/scsi/ch.c | 8 ++--
drivers/scsi/dpt_i2o.c | 18 ++++++------
drivers/scsi/gdth.c | 11 ++++---
drivers/scsi/megaraid.c | 8 ++--
drivers/scsi/megaraid/megaraid_mm.c | 8 ++--
drivers/scsi/megaraid/megaraid_sas.c | 2 -
drivers/scsi/mpt2sas/mpt2sas_ctl.c | 11 ++++---
drivers/scsi/osst.c | 15 +++++----
drivers/scsi/scsi_tgt_if.c | 2 -
drivers/scsi/sg.c | 11 ++++---
drivers/staging/crystalhd/crystalhd_lnx.c | 9 +++--
drivers/staging/dt3155/dt3155_drv.c | 6 ++-
drivers/staging/vme/devices/vme_user.c | 7 ++--
drivers/telephony/ixj.c | 7 ++--
drivers/usb/gadget/printer.c | 7 ++--
drivers/usb/misc/iowarrior.c | 15 +++++----
drivers/usb/misc/rio500.c | 15 +++++----
drivers/usb/misc/usblcd.c | 16 +++++-----
drivers/watchdog/cpwd.c | 15 +++++----
fs/hfsplus/ioctl.c | 11 ++++---
net/wanrouter/wanmain.c | 7 ++--
net/wanrouter/wanproc.c | 7 ++--
92 files changed, 505 insertions(+), 469 deletions(-)
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Corey Minyard <minyard@acm.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: David Woodhouse <David.Woodhouse@intel.com>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Cc: "James E.J. Bottomley" <James.Bottomley@suse.de>
Cc: Jesper Nilsson <jesper.nilsson@axis.com>
Cc: Karsten Keil <isdn@linux-pingi.de>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: netdev@vger.kernel.org
Cc: openipmi-developer@lists.sourceforge.net
Cc: devel@driverdev.osuosl.org
Cc: linux-cris-kernel@axis.com
Cc: linux-media@vger.kernel.org
Cc: linux-mtd@lists.infradead.org
Cc: linuxppc-dev@ozlabs.org
Cc: linux-scsi@vger.kernel.org
Cc: linux-usb@vger.kernel.org
^ permalink raw reply
* [PATCH 02/12] isdn: autoconvert trivial BKL users to private mutex
From: Arnd Bergmann @ 2010-07-11 21:18 UTC (permalink / raw)
To: linux-kernel
Cc: John Kacur, Frederic Weisbecker, Arnd Bergmann, Karsten Keil,
netdev
In-Reply-To: <1278883143-29035-1-git-send-email-arnd@arndb.de>
All these files use the big kernel lock in a trivial
way to serialize their private file operations,
typically resulting from an earlier semi-automatic
pushdown from VFS.
None of these drivers appears to want to lock against
other code, and they all use the BKL as the top-level
lock in their file operations, meaning that there
is no lock-order inversion problem.
Consequently, we can remove the BKL completely,
replacing it with a per-file mutex in every case.
Using a scripted approach means we can avoid
typos.
file=$1
name=$2
if grep -q lock_kernel ${file} ; then
if grep -q 'include.*linux.mutex.h' ${file} ; then
sed -i '/include.*<linux\/smp_lock.h>/d' ${file}
else
sed -i 's/include.*<linux\/smp_lock.h>.*$/include <linux\/mutex.h>/g' ${file}
fi
sed -i ${file} \
-e "/^#include.*linux.mutex.h/,$ {
1,/^\(static\|int\|long\)/ {
/^\(static\|int\|long\)/istatic DEFINE_MUTEX(${name}_mutex);
} }" \
-e "s/\(un\)*lock_kernel\>[ ]*()/mutex_\1lock(\&${name}_mutex)/g" \
-e '/[ ]*cycle_kernel_lock();/d'
else
sed -i -e '/include.*\<smp_lock.h\>/d' ${file} \
-e '/cycle_kernel_lock()/d'
fi
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Karsten Keil <isdn@linux-pingi.de>
Cc: netdev@vger.kernel.org
---
drivers/isdn/capi/capi.c | 6 +++---
drivers/isdn/divert/divert_procfs.c | 7 ++++---
drivers/isdn/hardware/eicon/divamnt.c | 7 ++++---
drivers/isdn/hardware/eicon/divasi.c | 2 --
drivers/isdn/hardware/eicon/divasmain.c | 2 --
drivers/isdn/hysdn/hysdn_procconf.c | 21 +++++++++++----------
drivers/isdn/hysdn/hysdn_proclog.c | 15 ++++++++-------
drivers/isdn/i4l/isdn_common.c | 27 ++++++++++++++-------------
drivers/isdn/mISDN/timerdev.c | 7 ++++---
9 files changed, 48 insertions(+), 46 deletions(-)
diff --git a/drivers/isdn/capi/capi.c b/drivers/isdn/capi/capi.c
index 0cabe31..b0a4a69 100644
--- a/drivers/isdn/capi/capi.c
+++ b/drivers/isdn/capi/capi.c
@@ -20,7 +20,6 @@
#include <linux/signal.h>
#include <linux/mutex.h>
#include <linux/mm.h>
-#include <linux/smp_lock.h>
#include <linux/timer.h>
#include <linux/wait.h>
#include <linux/tty.h>
@@ -50,6 +49,7 @@ MODULE_LICENSE("GPL");
/* -------- driver information -------------------------------------- */
+static DEFINE_MUTEX(capi_mutex);
static struct class *capi_class;
static int capi_major = 68; /* allocated */
@@ -985,9 +985,9 @@ capi_unlocked_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
{
int ret;
- lock_kernel();
+ mutex_lock(&capi_mutex);
ret = capi_ioctl(file, cmd, arg);
- unlock_kernel();
+ mutex_unlock(&capi_mutex);
return ret;
}
diff --git a/drivers/isdn/divert/divert_procfs.c b/drivers/isdn/divert/divert_procfs.c
index c53e241..33ec9e4 100644
--- a/drivers/isdn/divert/divert_procfs.c
+++ b/drivers/isdn/divert/divert_procfs.c
@@ -20,7 +20,7 @@
#include <linux/sched.h>
#include <linux/isdnif.h>
#include <net/net_namespace.h>
-#include <linux/smp_lock.h>
+#include <linux/mutex.h>
#include "isdn_divert.h"
@@ -28,6 +28,7 @@
/* Variables for interface queue */
/*********************************/
ulong if_used = 0; /* number of interface users */
+static DEFINE_MUTEX(isdn_divert_mutex);
static struct divert_info *divert_info_head = NULL; /* head of queue */
static struct divert_info *divert_info_tail = NULL; /* pointer to last entry */
static DEFINE_SPINLOCK(divert_info_lock);/* lock for queue */
@@ -261,9 +262,9 @@ static long isdn_divert_ioctl(struct file *file, uint cmd, ulong arg)
{
long ret;
- lock_kernel();
+ mutex_lock(&isdn_divert_mutex);
ret = isdn_divert_ioctl_unlocked(file, cmd, arg);
- unlock_kernel();
+ mutex_unlock(&isdn_divert_mutex);
return ret;
}
diff --git a/drivers/isdn/hardware/eicon/divamnt.c b/drivers/isdn/hardware/eicon/divamnt.c
index 1e85f74..f1d464f 100644
--- a/drivers/isdn/hardware/eicon/divamnt.c
+++ b/drivers/isdn/hardware/eicon/divamnt.c
@@ -14,7 +14,7 @@
#include <linux/init.h>
#include <linux/kernel.h>
#include <linux/poll.h>
-#include <linux/smp_lock.h>
+#include <linux/mutex.h>
#include <asm/uaccess.h>
#include "platform.h"
@@ -22,6 +22,7 @@
#include "divasync.h"
#include "debug_if.h"
+static DEFINE_MUTEX(maint_mutex);
static char *main_revision = "$Revision: 1.32.6.10 $";
static int major;
@@ -130,7 +131,7 @@ static int maint_open(struct inode *ino, struct file *filep)
{
int ret;
- lock_kernel();
+ mutex_lock(&maint_mutex);
/* only one open is allowed, so we test
it atomically */
if (test_and_set_bit(0, &opened))
@@ -139,7 +140,7 @@ static int maint_open(struct inode *ino, struct file *filep)
filep->private_data = NULL;
ret = nonseekable_open(ino, filep);
}
- unlock_kernel();
+ mutex_unlock(&maint_mutex);
return ret;
}
diff --git a/drivers/isdn/hardware/eicon/divasi.c b/drivers/isdn/hardware/eicon/divasi.c
index f577719..42d3b83 100644
--- a/drivers/isdn/hardware/eicon/divasi.c
+++ b/drivers/isdn/hardware/eicon/divasi.c
@@ -18,7 +18,6 @@
#include <linux/proc_fs.h>
#include <linux/skbuff.h>
#include <linux/seq_file.h>
-#include <linux/smp_lock.h>
#include <asm/uaccess.h>
#include "platform.h"
@@ -402,7 +401,6 @@ static unsigned int um_idi_poll(struct file *file, poll_table * wait)
static int um_idi_open(struct inode *inode, struct file *file)
{
- cycle_kernel_lock();
return (0);
}
diff --git a/drivers/isdn/hardware/eicon/divasmain.c b/drivers/isdn/hardware/eicon/divasmain.c
index fbbcb27..16a874b 100644
--- a/drivers/isdn/hardware/eicon/divasmain.c
+++ b/drivers/isdn/hardware/eicon/divasmain.c
@@ -21,7 +21,6 @@
#include <linux/list.h>
#include <linux/poll.h>
#include <linux/kmod.h>
-#include <linux/smp_lock.h>
#include "platform.h"
#undef ID_MASK
@@ -581,7 +580,6 @@ xdi_copy_from_user(void *os_handle, void *dst, const void __user *src, int lengt
*/
static int divas_open(struct inode *inode, struct file *file)
{
- cycle_kernel_lock();
return (0);
}
diff --git a/drivers/isdn/hysdn/hysdn_procconf.c b/drivers/isdn/hysdn/hysdn_procconf.c
index 8096646..96b3e39 100644
--- a/drivers/isdn/hysdn/hysdn_procconf.c
+++ b/drivers/isdn/hysdn/hysdn_procconf.c
@@ -17,11 +17,12 @@
#include <linux/proc_fs.h>
#include <linux/pci.h>
#include <linux/slab.h>
-#include <linux/smp_lock.h>
+#include <linux/mutex.h>
#include <net/net_namespace.h>
#include "hysdn_defs.h"
+static DEFINE_MUTEX(hysdn_conf_mutex);
static char *hysdn_procconf_revision = "$Revision: 1.8.6.4 $";
#define INFO_OUT_LEN 80 /* length of info line including lf */
@@ -234,7 +235,7 @@ hysdn_conf_open(struct inode *ino, struct file *filep)
char *cp, *tmp;
/* now search the addressed card */
- lock_kernel();
+ mutex_lock(&hysdn_conf_mutex);
card = card_root;
while (card) {
pd = card->procconf;
@@ -243,7 +244,7 @@ hysdn_conf_open(struct inode *ino, struct file *filep)
card = card->next; /* search next entry */
}
if (!card) {
- unlock_kernel();
+ mutex_unlock(&hysdn_conf_mutex);
return (-ENODEV); /* device is unknown/invalid */
}
if (card->debug_flags & (LOG_PROC_OPEN | LOG_PROC_ALL))
@@ -255,7 +256,7 @@ hysdn_conf_open(struct inode *ino, struct file *filep)
/* write only access -> write boot file or conf line */
if (!(cnf = kmalloc(sizeof(struct conf_writedata), GFP_KERNEL))) {
- unlock_kernel();
+ mutex_unlock(&hysdn_conf_mutex);
return (-EFAULT);
}
cnf->card = card;
@@ -267,7 +268,7 @@ hysdn_conf_open(struct inode *ino, struct file *filep)
/* read access -> output card info data */
if (!(tmp = kmalloc(INFO_OUT_LEN * 2 + 2, GFP_KERNEL))) {
- unlock_kernel();
+ mutex_unlock(&hysdn_conf_mutex);
return (-EFAULT); /* out of memory */
}
filep->private_data = tmp; /* start of string */
@@ -301,10 +302,10 @@ hysdn_conf_open(struct inode *ino, struct file *filep)
*cp++ = '\n';
*cp = 0; /* end of string */
} else { /* simultaneous read/write access forbidden ! */
- unlock_kernel();
+ mutex_unlock(&hysdn_conf_mutex);
return (-EPERM); /* no permission this time */
}
- unlock_kernel();
+ mutex_unlock(&hysdn_conf_mutex);
return nonseekable_open(ino, filep);
} /* hysdn_conf_open */
@@ -319,7 +320,7 @@ hysdn_conf_close(struct inode *ino, struct file *filep)
int retval = 0;
struct proc_dir_entry *pd;
- lock_kernel();
+ mutex_lock(&hysdn_conf_mutex);
/* search the addressed card */
card = card_root;
while (card) {
@@ -329,7 +330,7 @@ hysdn_conf_close(struct inode *ino, struct file *filep)
card = card->next; /* search next entry */
}
if (!card) {
- unlock_kernel();
+ mutex_unlock(&hysdn_conf_mutex);
return (-ENODEV); /* device is unknown/invalid */
}
if (card->debug_flags & (LOG_PROC_OPEN | LOG_PROC_ALL))
@@ -352,7 +353,7 @@ hysdn_conf_close(struct inode *ino, struct file *filep)
kfree(filep->private_data); /* release memory */
}
- unlock_kernel();
+ mutex_unlock(&hysdn_conf_mutex);
return (retval);
} /* hysdn_conf_close */
diff --git a/drivers/isdn/hysdn/hysdn_proclog.c b/drivers/isdn/hysdn/hysdn_proclog.c
index e83f6fd..37a9dd3 100644
--- a/drivers/isdn/hysdn/hysdn_proclog.c
+++ b/drivers/isdn/hysdn/hysdn_proclog.c
@@ -15,13 +15,14 @@
#include <linux/proc_fs.h>
#include <linux/sched.h>
#include <linux/slab.h>
-#include <linux/smp_lock.h>
+#include <linux/mutex.h>
#include "hysdn_defs.h"
/* the proc subdir for the interface is defined in the procconf module */
extern struct proc_dir_entry *hysdn_proc_entry;
+static DEFINE_MUTEX(hysdn_log_mutex);
static void put_log_buffer(hysdn_card * card, char *cp);
/*************************************************/
@@ -251,7 +252,7 @@ hysdn_log_open(struct inode *ino, struct file *filep)
struct procdata *pd = NULL;
unsigned long flags;
- lock_kernel();
+ mutex_lock(&hysdn_log_mutex);
card = card_root;
while (card) {
pd = card->proclog;
@@ -260,7 +261,7 @@ hysdn_log_open(struct inode *ino, struct file *filep)
card = card->next; /* search next entry */
}
if (!card) {
- unlock_kernel();
+ mutex_unlock(&hysdn_log_mutex);
return (-ENODEV); /* device is unknown/invalid */
}
filep->private_data = card; /* remember our own card */
@@ -278,10 +279,10 @@ hysdn_log_open(struct inode *ino, struct file *filep)
filep->private_data = &pd->log_head;
spin_unlock_irqrestore(&card->hysdn_lock, flags);
} else { /* simultaneous read/write access forbidden ! */
- unlock_kernel();
+ mutex_unlock(&hysdn_log_mutex);
return (-EPERM); /* no permission this time */
}
- unlock_kernel();
+ mutex_unlock(&hysdn_log_mutex);
return nonseekable_open(ino, filep);
} /* hysdn_log_open */
@@ -300,7 +301,7 @@ hysdn_log_close(struct inode *ino, struct file *filep)
hysdn_card *card;
int retval = 0;
- lock_kernel();
+ mutex_lock(&hysdn_log_mutex);
if ((filep->f_mode & (FMODE_READ | FMODE_WRITE)) == FMODE_WRITE) {
/* write only access -> write debug level written */
retval = 0; /* success */
@@ -339,7 +340,7 @@ hysdn_log_close(struct inode *ino, struct file *filep)
kfree(inf);
}
} /* read access */
- unlock_kernel();
+ mutex_unlock(&hysdn_log_mutex);
return (retval);
} /* hysdn_log_close */
diff --git a/drivers/isdn/i4l/isdn_common.c b/drivers/isdn/i4l/isdn_common.c
index a44cdb4..15632bd 100644
--- a/drivers/isdn/i4l/isdn_common.c
+++ b/drivers/isdn/i4l/isdn_common.c
@@ -17,7 +17,7 @@
#include <linux/slab.h>
#include <linux/vmalloc.h>
#include <linux/isdn.h>
-#include <linux/smp_lock.h>
+#include <linux/mutex.h>
#include "isdn_common.h"
#include "isdn_tty.h"
#include "isdn_net.h"
@@ -42,6 +42,7 @@ MODULE_LICENSE("GPL");
isdn_dev *dev;
+static DEFINE_MUTEX(isdn_mutex);
static char *isdn_revision = "$Revision: 1.1.2.3 $";
extern char *isdn_net_revision;
@@ -1070,7 +1071,7 @@ isdn_read(struct file *file, char __user *buf, size_t count, loff_t * off)
int retval;
char *p;
- lock_kernel();
+ mutex_lock(&isdn_mutex);
if (minor == ISDN_MINOR_STATUS) {
if (!file->private_data) {
if (file->f_flags & O_NONBLOCK) {
@@ -1163,7 +1164,7 @@ isdn_read(struct file *file, char __user *buf, size_t count, loff_t * off)
#endif
retval = -ENODEV;
out:
- unlock_kernel();
+ mutex_unlock(&isdn_mutex);
return retval;
}
@@ -1180,7 +1181,7 @@ isdn_write(struct file *file, const char __user *buf, size_t count, loff_t * off
if (!dev->drivers)
return -ENODEV;
- lock_kernel();
+ mutex_lock(&isdn_mutex);
if (minor <= ISDN_MINOR_BMAX) {
printk(KERN_WARNING "isdn_write minor %d obsolete!\n", minor);
drvidx = isdn_minor2drv(minor);
@@ -1225,7 +1226,7 @@ isdn_write(struct file *file, const char __user *buf, size_t count, loff_t * off
#endif
retval = -ENODEV;
out:
- unlock_kernel();
+ mutex_unlock(&isdn_mutex);
return retval;
}
@@ -1236,7 +1237,7 @@ isdn_poll(struct file *file, poll_table * wait)
unsigned int minor = iminor(file->f_path.dentry->d_inode);
int drvidx = isdn_minor2drv(minor - ISDN_MINOR_CTRL);
- lock_kernel();
+ mutex_lock(&isdn_mutex);
if (minor == ISDN_MINOR_STATUS) {
poll_wait(file, &(dev->info_waitq), wait);
/* mask = POLLOUT | POLLWRNORM; */
@@ -1266,7 +1267,7 @@ isdn_poll(struct file *file, poll_table * wait)
#endif
mask = POLLERR;
out:
- unlock_kernel();
+ mutex_unlock(&isdn_mutex);
return mask;
}
@@ -1727,9 +1728,9 @@ isdn_unlocked_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
{
int ret;
- lock_kernel();
+ mutex_lock(&isdn_mutex);
ret = isdn_ioctl(file, cmd, arg);
- unlock_kernel();
+ mutex_unlock(&isdn_mutex);
return ret;
}
@@ -1745,7 +1746,7 @@ isdn_open(struct inode *ino, struct file *filep)
int chidx;
int retval = -ENODEV;
- lock_kernel();
+ mutex_lock(&isdn_mutex);
if (minor == ISDN_MINOR_STATUS) {
infostruct *p;
@@ -1796,7 +1797,7 @@ isdn_open(struct inode *ino, struct file *filep)
#endif
out:
nonseekable_open(ino, filep);
- unlock_kernel();
+ mutex_unlock(&isdn_mutex);
return retval;
}
@@ -1805,7 +1806,7 @@ isdn_close(struct inode *ino, struct file *filep)
{
uint minor = iminor(ino);
- lock_kernel();
+ mutex_lock(&isdn_mutex);
if (minor == ISDN_MINOR_STATUS) {
infostruct *p = dev->infochain;
infostruct *q = NULL;
@@ -1839,7 +1840,7 @@ isdn_close(struct inode *ino, struct file *filep)
#endif
out:
- unlock_kernel();
+ mutex_unlock(&isdn_mutex);
return 0;
}
diff --git a/drivers/isdn/mISDN/timerdev.c b/drivers/isdn/mISDN/timerdev.c
index 81048b8..de43c8c 100644
--- a/drivers/isdn/mISDN/timerdev.c
+++ b/drivers/isdn/mISDN/timerdev.c
@@ -24,9 +24,10 @@
#include <linux/miscdevice.h>
#include <linux/module.h>
#include <linux/mISDNif.h>
-#include <linux/smp_lock.h>
+#include <linux/mutex.h>
#include "core.h"
+static DEFINE_MUTEX(mISDN_mutex);
static u_int *debug;
@@ -224,7 +225,7 @@ mISDN_ioctl(struct file *filep, unsigned int cmd, unsigned long arg)
if (*debug & DEBUG_TIMER)
printk(KERN_DEBUG "%s(%p, %x, %lx)\n", __func__,
filep, cmd, arg);
- lock_kernel();
+ mutex_lock(&mISDN_mutex);
switch (cmd) {
case IMADDTIMER:
if (get_user(tout, (int __user *)arg)) {
@@ -256,7 +257,7 @@ mISDN_ioctl(struct file *filep, unsigned int cmd, unsigned long arg)
default:
ret = -EINVAL;
}
- unlock_kernel();
+ mutex_unlock(&mISDN_mutex);
return ret;
}
--
1.7.1
^ permalink raw reply related
* [PATCH 06/12] net: autoconvert trivial BKL users to private mutex
From: Arnd Bergmann @ 2010-07-11 21:18 UTC (permalink / raw)
To: linux-kernel
Cc: John Kacur, Frederic Weisbecker, Arnd Bergmann, David S. Miller,
netdev
In-Reply-To: <1278883143-29035-1-git-send-email-arnd@arndb.de>
All these files use the big kernel lock in a trivial
way to serialize their private file operations,
typically resulting from an earlier semi-automatic
pushdown from VFS.
None of these drivers appears to want to lock against
other code, and they all use the BKL as the top-level
lock in their file operations, meaning that there
is no lock-order inversion problem.
Consequently, we can remove the BKL completely,
replacing it with a per-file mutex in every case.
Using a scripted approach means we can avoid
typos.
file=$1
name=$2
if grep -q lock_kernel ${file} ; then
if grep -q 'include.*linux.mutex.h' ${file} ; then
sed -i '/include.*<linux\/smp_lock.h>/d' ${file}
else
sed -i 's/include.*<linux\/smp_lock.h>.*$/include <linux\/mutex.h>/g' ${file}
fi
sed -i ${file} \
-e "/^#include.*linux.mutex.h/,$ {
1,/^\(static\|int\|long\)/ {
/^\(static\|int\|long\)/istatic DEFINE_MUTEX(${name}_mutex);
} }" \
-e "s/\(un\)*lock_kernel\>[ ]*()/mutex_\1lock(\&${name}_mutex)/g" \
-e '/[ ]*cycle_kernel_lock();/d'
else
sed -i -e '/include.*\<smp_lock.h\>/d' ${file} \
-e '/cycle_kernel_lock()/d'
fi
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: netdev@vger.kernel.org
---
drivers/net/ppp_generic.c | 19 +++++++++----------
drivers/net/wan/cosa.c | 10 +++++-----
net/wanrouter/wanmain.c | 7 ++++---
net/wanrouter/wanproc.c | 7 ++++---
4 files changed, 22 insertions(+), 21 deletions(-)
diff --git a/drivers/net/ppp_generic.c b/drivers/net/ppp_generic.c
index 1b2c291..a72a335 100644
--- a/drivers/net/ppp_generic.c
+++ b/drivers/net/ppp_generic.c
@@ -40,7 +40,6 @@
#include <linux/if_arp.h>
#include <linux/ip.h>
#include <linux/tcp.h>
-#include <linux/smp_lock.h>
#include <linux/spinlock.h>
#include <linux/rwsem.h>
#include <linux/stddef.h>
@@ -181,6 +180,7 @@ struct channel {
* channel.downl.
*/
+static DEFINE_MUTEX(ppp_mutex);
static atomic_t ppp_unit_count = ATOMIC_INIT(0);
static atomic_t channel_count = ATOMIC_INIT(0);
@@ -363,7 +363,6 @@ static const int npindex_to_ethertype[NUM_NP] = {
*/
static int ppp_open(struct inode *inode, struct file *file)
{
- cycle_kernel_lock();
/*
* This could (should?) be enforced by the permissions on /dev/ppp.
*/
@@ -588,7 +587,7 @@ static long ppp_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
* this fd and reopening /dev/ppp.
*/
err = -EINVAL;
- lock_kernel();
+ mutex_lock(&ppp_mutex);
if (pf->kind == INTERFACE) {
ppp = PF_TO_PPP(pf);
if (file == ppp->owner)
@@ -600,7 +599,7 @@ static long ppp_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
} else
printk(KERN_DEBUG "PPPIOCDETACH file->f_count=%ld\n",
atomic_long_read(&file->f_count));
- unlock_kernel();
+ mutex_unlock(&ppp_mutex);
return err;
}
@@ -608,7 +607,7 @@ static long ppp_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
struct channel *pch;
struct ppp_channel *chan;
- lock_kernel();
+ mutex_lock(&ppp_mutex);
pch = PF_TO_CHANNEL(pf);
switch (cmd) {
@@ -630,7 +629,7 @@ static long ppp_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
err = chan->ops->ioctl(chan, cmd, arg);
up_read(&pch->chan_sem);
}
- unlock_kernel();
+ mutex_unlock(&ppp_mutex);
return err;
}
@@ -640,7 +639,7 @@ static long ppp_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
return -EINVAL;
}
- lock_kernel();
+ mutex_lock(&ppp_mutex);
ppp = PF_TO_PPP(pf);
switch (cmd) {
case PPPIOCSMRU:
@@ -788,7 +787,7 @@ static long ppp_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
default:
err = -ENOTTY;
}
- unlock_kernel();
+ mutex_unlock(&ppp_mutex);
return err;
}
@@ -801,7 +800,7 @@ static int ppp_unattached_ioctl(struct net *net, struct ppp_file *pf,
struct ppp_net *pn;
int __user *p = (int __user *)arg;
- lock_kernel();
+ mutex_lock(&ppp_mutex);
switch (cmd) {
case PPPIOCNEWUNIT:
/* Create a new ppp unit */
@@ -852,7 +851,7 @@ static int ppp_unattached_ioctl(struct net *net, struct ppp_file *pf,
default:
err = -ENOTTY;
}
- unlock_kernel();
+ mutex_unlock(&ppp_mutex);
return err;
}
diff --git a/drivers/net/wan/cosa.c b/drivers/net/wan/cosa.c
index f0bd70f..04c6cd4 100644
--- a/drivers/net/wan/cosa.c
+++ b/drivers/net/wan/cosa.c
@@ -89,7 +89,6 @@
#include <linux/spinlock.h>
#include <linux/mutex.h>
#include <linux/device.h>
-#include <linux/smp_lock.h>
#include <asm/io.h>
#include <asm/dma.h>
#include <asm/byteorder.h>
@@ -174,6 +173,7 @@ struct cosa_data {
* Character device major number. 117 was allocated for us.
* The value of 0 means to allocate a first free one.
*/
+static DEFINE_MUTEX(cosa_chardev_mutex);
static int cosa_major = 117;
/*
@@ -944,7 +944,7 @@ static int cosa_open(struct inode *inode, struct file *file)
int n;
int ret = 0;
- lock_kernel();
+ mutex_lock(&cosa_chardev_mutex);
if ((n=iminor(file->f_path.dentry->d_inode)>>CARD_MINOR_BITS)
>= nr_cards) {
ret = -ENODEV;
@@ -976,7 +976,7 @@ static int cosa_open(struct inode *inode, struct file *file)
chan->rx_done = chrdev_rx_done;
spin_unlock_irqrestore(&cosa->lock, flags);
out:
- unlock_kernel();
+ mutex_unlock(&cosa_chardev_mutex);
return ret;
}
@@ -1212,10 +1212,10 @@ static long cosa_chardev_ioctl(struct file *file, unsigned int cmd,
struct cosa_data *cosa;
long ret;
- lock_kernel();
+ mutex_lock(&cosa_chardev_mutex);
cosa = channel->cosa;
ret = cosa_ioctl_common(cosa, channel, cmd, arg);
- unlock_kernel();
+ mutex_unlock(&cosa_chardev_mutex);
return ret;
}
diff --git a/net/wanrouter/wanmain.c b/net/wanrouter/wanmain.c
index 258daa8..2bf2340 100644
--- a/net/wanrouter/wanmain.c
+++ b/net/wanrouter/wanmain.c
@@ -48,7 +48,7 @@
#include <linux/kernel.h>
#include <linux/module.h> /* support for loadable modules */
#include <linux/slab.h> /* kmalloc(), kfree() */
-#include <linux/smp_lock.h>
+#include <linux/mutex.h>
#include <linux/mm.h>
#include <linux/string.h> /* inline mem*, str* functions */
@@ -71,6 +71,7 @@
* WAN device IOCTL handlers
*/
+static DEFINE_MUTEX(wanrouter_mutex);
static int wanrouter_device_setup(struct wan_device *wandev,
wandev_conf_t __user *u_conf);
static int wanrouter_device_stat(struct wan_device *wandev,
@@ -376,7 +377,7 @@ long wanrouter_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
if (wandev->magic != ROUTER_MAGIC)
return -EINVAL;
- lock_kernel();
+ mutex_lock(&wanrouter_mutex);
switch (cmd) {
case ROUTER_SETUP:
err = wanrouter_device_setup(wandev, data);
@@ -408,7 +409,7 @@ long wanrouter_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
err = wandev->ioctl(wandev, cmd, arg);
else err = -EINVAL;
}
- unlock_kernel();
+ mutex_unlock(&wanrouter_mutex);
return err;
}
diff --git a/net/wanrouter/wanproc.c b/net/wanrouter/wanproc.c
index c44d96b..11f25c7 100644
--- a/net/wanrouter/wanproc.c
+++ b/net/wanrouter/wanproc.c
@@ -27,7 +27,7 @@
#include <linux/module.h>
#include <linux/wanrouter.h> /* WAN router API definitions */
#include <linux/seq_file.h>
-#include <linux/smp_lock.h>
+#include <linux/mutex.h>
#include <net/net_namespace.h>
#include <asm/io.h>
@@ -66,6 +66,7 @@
* /proc/net/router
*/
+static DEFINE_MUTEX(config_mutex);
static struct proc_dir_entry *proc_router;
/* Strings */
@@ -85,7 +86,7 @@ static void *r_start(struct seq_file *m, loff_t *pos)
struct wan_device *wandev;
loff_t l = *pos;
- lock_kernel();
+ mutex_lock(&config_mutex);
if (!l--)
return SEQ_START_TOKEN;
for (wandev = wanrouter_router_devlist; l-- && wandev;
@@ -104,7 +105,7 @@ static void *r_next(struct seq_file *m, void *v, loff_t *pos)
static void r_stop(struct seq_file *m, void *v)
__releases(kernel_lock)
{
- unlock_kernel();
+ mutex_unlock(&config_mutex);
}
static int config_show(struct seq_file *m, void *v)
--
1.7.1
^ permalink raw reply related
* [PATCH] r8169: incorrect identifier for a 8168dp
From: Francois Romieu @ 2010-07-11 21:48 UTC (permalink / raw)
To: netdev; +Cc: David Miller, Hayes
Merge error.
See CFG_METHOD_8 (0x3c800000 + 0x00300000) since version 8.002.00
of Realtek's driver.
Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Cc: Hayes <hayeswang@realtek.com>
---
drivers/net/r8169.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/net/r8169.c b/drivers/net/r8169.c
index 96b6cfb..cdc6a5c 100644
--- a/drivers/net/r8169.c
+++ b/drivers/net/r8169.c
@@ -1316,7 +1316,7 @@ static void rtl8169_get_mac_version(struct rtl8169_private *tp,
{ 0x7c800000, 0x28000000, RTL_GIGA_MAC_VER_26 },
/* 8168C family. */
- { 0x7cf00000, 0x3ca00000, RTL_GIGA_MAC_VER_24 },
+ { 0x7cf00000, 0x3cb00000, RTL_GIGA_MAC_VER_24 },
{ 0x7cf00000, 0x3c900000, RTL_GIGA_MAC_VER_23 },
{ 0x7cf00000, 0x3c800000, RTL_GIGA_MAC_VER_18 },
{ 0x7c800000, 0x3c800000, RTL_GIGA_MAC_VER_24 },
--
1.7.1.1
^ permalink raw reply related
* [PATCH net-next 1/5] cxgb4: move the choice of interrupt type before net_device registration
From: Dimitris Michailidis @ 2010-07-11 22:01 UTC (permalink / raw)
To: netdev; +Cc: Dimitris Michailidis
In-Reply-To: <1278885679-3738-1-git-send-email-dm@chelsio.com>
We need to settle on the kind of interrupts we'll be using, a choice that
also impacts the number of queues, before registering and making visible
the net_devices. Move the relevant code up a bit.
Signed-off-by: Dimitris Michailidis <dm@chelsio.com>
---
drivers/net/cxgb4/cxgb4_main.c | 13 +++++++------
1 files changed, 7 insertions(+), 6 deletions(-)
diff --git a/drivers/net/cxgb4/cxgb4_main.c b/drivers/net/cxgb4/cxgb4_main.c
index 2619997..743dc6f 100644
--- a/drivers/net/cxgb4/cxgb4_main.c
+++ b/drivers/net/cxgb4/cxgb4_main.c
@@ -3507,6 +3507,12 @@ static int __devinit init_one(struct pci_dev *pdev,
adapter->params.offload = 0;
}
+ /* See what interrupts we'll be using */
+ if (msi > 1 && enable_msix(adapter) == 0)
+ adapter->flags |= USING_MSIX;
+ else if (msi > 0 && pci_enable_msi(pdev) == 0)
+ adapter->flags |= USING_MSI;
+
/*
* The card is now ready to go. If any errors occur during device
* registration we do not fail the whole card but rather proceed only
@@ -3542,12 +3548,6 @@ static int __devinit init_one(struct pci_dev *pdev,
setup_debugfs(adapter);
}
- /* See what interrupts we'll be using */
- if (msi > 1 && enable_msix(adapter) == 0)
- adapter->flags |= USING_MSIX;
- else if (msi > 0 && pci_enable_msi(pdev) == 0)
- adapter->flags |= USING_MSI;
-
if (is_offload(adapter))
attach_ulds(adapter);
@@ -3571,6 +3571,7 @@ sriov:
free_netdev(adapter->port[i]);
if (adapter->flags & FW_OK)
t4_fw_bye(adapter, 0);
+ disable_msi(adapter);
out_unmap_bar:
iounmap(adapter->regs);
out_free_adapter:
--
1.5.4
^ permalink raw reply related
* [PATCH net-next 0/5] cxgb4 updates
From: Dimitris Michailidis @ 2010-07-11 22:01 UTC (permalink / raw)
To: netdev
Following is a series of 5 cxgb4 patches, all for net-next. Mainly they
implement the get_rxnfc, get_rxfh_indir, and set_rxfh_indir ethtool ops,
plus a couple of fixes.
^ permalink raw reply
* [PATCH net-next 2/5] cxgb4: avoid duplicating some resource freeing code
From: Dimitris Michailidis @ 2010-07-11 22:01 UTC (permalink / raw)
To: netdev; +Cc: Dimitris Michailidis
In-Reply-To: <1278885679-3738-2-git-send-email-dm@chelsio.com>
Currently there are two copies of some resource freeing code, turn it into
a function and call it.
Signed-off-by: Dimitris Michailidis <dm@chelsio.com>
---
drivers/net/cxgb4/cxgb4_main.c | 42 +++++++++++++++++++++++----------------
1 files changed, 25 insertions(+), 17 deletions(-)
diff --git a/drivers/net/cxgb4/cxgb4_main.c b/drivers/net/cxgb4/cxgb4_main.c
index 743dc6f..653bb54 100644
--- a/drivers/net/cxgb4/cxgb4_main.c
+++ b/drivers/net/cxgb4/cxgb4_main.c
@@ -3364,6 +3364,29 @@ static void __devinit print_port_info(struct adapter *adap)
}
}
+/*
+ * Free the following resources:
+ * - memory used for tables
+ * - MSI/MSI-X
+ * - net devices
+ * - resources FW is holding for us
+ */
+static void free_some_resources(struct adapter *adapter)
+{
+ unsigned int i;
+
+ t4_free_mem(adapter->l2t);
+ t4_free_mem(adapter->tids.tid_tab);
+ disable_msi(adapter);
+
+ for_each_port(adapter, i)
+ if (adapter->port[i])
+ free_netdev(adapter->port[i]);
+
+ if (adapter->flags & FW_OK)
+ t4_fw_bye(adapter, 0);
+}
+
#define VLAN_FEAT (NETIF_F_SG | NETIF_F_IP_CSUM | NETIF_F_TSO | NETIF_F_TSO6 |\
NETIF_F_IPV6_CSUM | NETIF_F_HIGHDMA)
@@ -3564,14 +3587,7 @@ sriov:
return 0;
out_free_dev:
- t4_free_mem(adapter->tids.tid_tab);
- t4_free_mem(adapter->l2t);
- for_each_port(adapter, i)
- if (adapter->port[i])
- free_netdev(adapter->port[i]);
- if (adapter->flags & FW_OK)
- t4_fw_bye(adapter, 0);
- disable_msi(adapter);
+ free_some_resources(adapter);
out_unmap_bar:
iounmap(adapter->regs);
out_free_adapter:
@@ -3606,16 +3622,8 @@ static void __devexit remove_one(struct pci_dev *pdev)
if (adapter->flags & FULL_INIT_DONE)
cxgb_down(adapter);
- t4_free_mem(adapter->l2t);
- t4_free_mem(adapter->tids.tid_tab);
- disable_msi(adapter);
-
- for_each_port(adapter, i)
- if (adapter->port[i])
- free_netdev(adapter->port[i]);
- if (adapter->flags & FW_OK)
- t4_fw_bye(adapter, 0);
+ free_some_resources(adapter);
iounmap(adapter->regs);
kfree(adapter);
pci_disable_pcie_error_reporting(pdev);
--
1.5.4
^ permalink raw reply related
* [PATCH net-next 4/5] cxgb4: implement the ETHTOOL_GRXFH command
From: Dimitris Michailidis @ 2010-07-11 22:01 UTC (permalink / raw)
To: netdev; +Cc: Dimitris Michailidis
In-Reply-To: <1278885679-3738-4-git-send-email-dm@chelsio.com>
Signed-off-by: Dimitris Michailidis <dm@chelsio.com>
---
drivers/net/cxgb4/cxgb4.h | 1 +
drivers/net/cxgb4/cxgb4_main.c | 54 +++++++++++++++++++++++++++++++++++++++-
drivers/net/cxgb4/t4_hw.c | 11 ++++++++
3 files changed, 65 insertions(+), 1 deletions(-)
diff --git a/drivers/net/cxgb4/cxgb4.h b/drivers/net/cxgb4/cxgb4.h
index a614eb5..4769c1c 100644
--- a/drivers/net/cxgb4/cxgb4.h
+++ b/drivers/net/cxgb4/cxgb4.h
@@ -294,6 +294,7 @@ struct port_info {
u8 rx_offload; /* CSO, etc */
u8 nqsets; /* # of qsets */
u8 first_qset; /* index of first qset */
+ u8 rss_mode;
struct link_config link_cfg;
u16 *rss;
};
diff --git a/drivers/net/cxgb4/cxgb4_main.c b/drivers/net/cxgb4/cxgb4_main.c
index 61d4313..110843c 100644
--- a/drivers/net/cxgb4/cxgb4_main.c
+++ b/drivers/net/cxgb4/cxgb4_main.c
@@ -1851,9 +1851,61 @@ static int set_rss_table(struct net_device *dev,
static int get_rxnfc(struct net_device *dev, struct ethtool_rxnfc *info,
void *rules)
{
+ const struct port_info *pi = netdev_priv(dev);
+
switch (info->cmd) {
+ case ETHTOOL_GRXFH: {
+ unsigned int v = pi->rss_mode;
+
+ info->data = 0;
+ switch (info->flow_type) {
+ case TCP_V4_FLOW:
+ if (v & FW_RSS_VI_CONFIG_CMD_IP4FOURTUPEN)
+ info->data = RXH_IP_SRC | RXH_IP_DST |
+ RXH_L4_B_0_1 | RXH_L4_B_2_3;
+ else if (v & FW_RSS_VI_CONFIG_CMD_IP4TWOTUPEN)
+ info->data = RXH_IP_SRC | RXH_IP_DST;
+ break;
+ case UDP_V4_FLOW:
+ if ((v & FW_RSS_VI_CONFIG_CMD_IP4FOURTUPEN) &&
+ (v & FW_RSS_VI_CONFIG_CMD_UDPEN))
+ info->data = RXH_IP_SRC | RXH_IP_DST |
+ RXH_L4_B_0_1 | RXH_L4_B_2_3;
+ else if (v & FW_RSS_VI_CONFIG_CMD_IP4TWOTUPEN)
+ info->data = RXH_IP_SRC | RXH_IP_DST;
+ break;
+ case SCTP_V4_FLOW:
+ case AH_ESP_V4_FLOW:
+ case IPV4_FLOW:
+ if (v & FW_RSS_VI_CONFIG_CMD_IP4TWOTUPEN)
+ info->data = RXH_IP_SRC | RXH_IP_DST;
+ break;
+ case TCP_V6_FLOW:
+ if (v & FW_RSS_VI_CONFIG_CMD_IP6FOURTUPEN)
+ info->data = RXH_IP_SRC | RXH_IP_DST |
+ RXH_L4_B_0_1 | RXH_L4_B_2_3;
+ else if (v & FW_RSS_VI_CONFIG_CMD_IP6TWOTUPEN)
+ info->data = RXH_IP_SRC | RXH_IP_DST;
+ break;
+ case UDP_V6_FLOW:
+ if ((v & FW_RSS_VI_CONFIG_CMD_IP6FOURTUPEN) &&
+ (v & FW_RSS_VI_CONFIG_CMD_UDPEN))
+ info->data = RXH_IP_SRC | RXH_IP_DST |
+ RXH_L4_B_0_1 | RXH_L4_B_2_3;
+ else if (v & FW_RSS_VI_CONFIG_CMD_IP6TWOTUPEN)
+ info->data = RXH_IP_SRC | RXH_IP_DST;
+ break;
+ case SCTP_V6_FLOW:
+ case AH_ESP_V6_FLOW:
+ case IPV6_FLOW:
+ if (v & FW_RSS_VI_CONFIG_CMD_IP6TWOTUPEN)
+ info->data = RXH_IP_SRC | RXH_IP_DST;
+ break;
+ }
+ return 0;
+ }
case ETHTOOL_GRXRINGS:
- info->data = netdev2pinfo(dev)->nqsets;
+ info->data = pi->nqsets;
return 0;
}
return -EOPNOTSUPP;
diff --git a/drivers/net/cxgb4/t4_hw.c b/drivers/net/cxgb4/t4_hw.c
index 3e63d14..ab46797 100644
--- a/drivers/net/cxgb4/t4_hw.c
+++ b/drivers/net/cxgb4/t4_hw.c
@@ -3133,8 +3133,10 @@ int __devinit t4_port_init(struct adapter *adap, int mbox, int pf, int vf)
u8 addr[6];
int ret, i, j = 0;
struct fw_port_cmd c;
+ struct fw_rss_vi_config_cmd rvc;
memset(&c, 0, sizeof(c));
+ memset(&rvc, 0, sizeof(rvc));
for_each_port(adap, i) {
unsigned int rss_size;
@@ -3171,6 +3173,15 @@ int __devinit t4_port_init(struct adapter *adap, int mbox, int pf, int vf)
p->port_type = FW_PORT_CMD_PTYPE_GET(ret);
p->mod_type = FW_PORT_MOD_TYPE_NA;
+ rvc.op_to_viid = htonl(FW_CMD_OP(FW_RSS_VI_CONFIG_CMD) |
+ FW_CMD_REQUEST | FW_CMD_READ |
+ FW_RSS_VI_CONFIG_CMD_VIID(p->viid));
+ rvc.retval_len16 = htonl(FW_LEN16(rvc));
+ ret = t4_wr_mbox(adap, mbox, &rvc, sizeof(rvc), &rvc);
+ if (ret)
+ return ret;
+ p->rss_mode = ntohl(rvc.u.basicvirtual.defaultq_to_udpen);
+
init_link_config(&p->link_cfg, ntohs(c.u.info.pcap));
j++;
}
--
1.5.4
^ permalink raw reply related
* [PATCH net-next 3/5] cxgb4: add user manipulation of the RSS table
From: Dimitris Michailidis @ 2010-07-11 22:01 UTC (permalink / raw)
To: netdev; +Cc: Dimitris Michailidis
In-Reply-To: <1278885679-3738-3-git-send-email-dm@chelsio.com>
Implement the get_rxnfc, get_rxfh_indir, and set_rxfh_indir ethtool
methods for user manipulation of the RSS table. Besides the methods
themselves the rest of the changes here store, initialize, and write
the table contents.
Signed-off-by: Dimitris Michailidis <dm@chelsio.com>
---
drivers/net/cxgb4/cxgb4.h | 1 +
drivers/net/cxgb4/cxgb4_main.c | 113 ++++++++++++++++++++++++++++++++++------
2 files changed, 98 insertions(+), 16 deletions(-)
diff --git a/drivers/net/cxgb4/cxgb4.h b/drivers/net/cxgb4/cxgb4.h
index 62804bb..a614eb5 100644
--- a/drivers/net/cxgb4/cxgb4.h
+++ b/drivers/net/cxgb4/cxgb4.h
@@ -295,6 +295,7 @@ struct port_info {
u8 nqsets; /* # of qsets */
u8 first_qset; /* index of first qset */
struct link_config link_cfg;
+ u16 *rss;
};
/* port_info.rx_offload flags */
diff --git a/drivers/net/cxgb4/cxgb4_main.c b/drivers/net/cxgb4/cxgb4_main.c
index 653bb54..61d4313 100644
--- a/drivers/net/cxgb4/cxgb4_main.c
+++ b/drivers/net/cxgb4/cxgb4_main.c
@@ -597,30 +597,47 @@ static void free_msix_queue_irqs(struct adapter *adap)
}
/**
+ * write_rss - write the RSS table for a given port
+ * @pi: the port
+ * @queues: array of queue indices for RSS
+ *
+ * Sets up the portion of the HW RSS table for the port's VI to distribute
+ * packets to the Rx queues in @queues.
+ */
+static int write_rss(const struct port_info *pi, const u16 *queues)
+{
+ u16 *rss;
+ int i, err;
+ const struct sge_eth_rxq *q = &pi->adapter->sge.ethrxq[pi->first_qset];
+
+ rss = kmalloc(pi->rss_size * sizeof(u16), GFP_KERNEL);
+ if (!rss)
+ return -ENOMEM;
+
+ /* map the queue indices to queue ids */
+ for (i = 0; i < pi->rss_size; i++, queues++)
+ rss[i] = q[*queues].rspq.abs_id;
+
+ err = t4_config_rss_range(pi->adapter, 0, pi->viid, 0, pi->rss_size,
+ rss, pi->rss_size);
+ kfree(rss);
+ return err;
+}
+
+/**
* setup_rss - configure RSS
* @adap: the adapter
*
- * Sets up RSS to distribute packets to multiple receive queues. We
- * configure the RSS CPU lookup table to distribute to the number of HW
- * receive queues, and the response queue lookup table to narrow that
- * down to the response queues actually configured for each port.
- * We always configure the RSS mapping for all ports since the mapping
- * table has plenty of entries.
+ * Sets up RSS for each port.
*/
static int setup_rss(struct adapter *adap)
{
- int i, j, err;
- u16 rss[MAX_ETH_QSETS];
+ int i, err;
for_each_port(adap, i) {
const struct port_info *pi = adap2pinfo(adap, i);
- const struct sge_eth_rxq *q = &adap->sge.ethrxq[pi->first_qset];
-
- for (j = 0; j < pi->nqsets; j++)
- rss[j] = q[j].rspq.abs_id;
- err = t4_config_rss_range(adap, 0, pi->viid, 0, pi->rss_size,
- rss, pi->nqsets);
+ err = write_rss(pi, pi->rss);
if (err)
return err;
}
@@ -1802,6 +1819,46 @@ static int set_flags(struct net_device *dev, u32 flags)
return ethtool_op_set_flags(dev, flags, ETH_FLAG_RXHASH);
}
+static int get_rss_table(struct net_device *dev, struct ethtool_rxfh_indir *p)
+{
+ const struct port_info *pi = netdev_priv(dev);
+ unsigned int n = min_t(unsigned int, p->size, pi->rss_size);
+
+ p->size = pi->rss_size;
+ while (n--)
+ p->ring_index[n] = pi->rss[n];
+ return 0;
+}
+
+static int set_rss_table(struct net_device *dev,
+ const struct ethtool_rxfh_indir *p)
+{
+ unsigned int i;
+ struct port_info *pi = netdev_priv(dev);
+
+ if (p->size != pi->rss_size)
+ return -EINVAL;
+ for (i = 0; i < p->size; i++)
+ if (p->ring_index[i] >= pi->nqsets)
+ return -EINVAL;
+ for (i = 0; i < p->size; i++)
+ pi->rss[i] = p->ring_index[i];
+ if (pi->adapter->flags & FULL_INIT_DONE)
+ return write_rss(pi, pi->rss);
+ return 0;
+}
+
+static int get_rxnfc(struct net_device *dev, struct ethtool_rxnfc *info,
+ void *rules)
+{
+ switch (info->cmd) {
+ case ETHTOOL_GRXRINGS:
+ info->data = netdev2pinfo(dev)->nqsets;
+ return 0;
+ }
+ return -EOPNOTSUPP;
+}
+
static struct ethtool_ops cxgb_ethtool_ops = {
.get_settings = get_settings,
.set_settings = set_settings,
@@ -1833,6 +1890,9 @@ static struct ethtool_ops cxgb_ethtool_ops = {
.set_wol = set_wol,
.set_tso = set_tso,
.set_flags = set_flags,
+ .get_rxnfc = get_rxnfc,
+ .get_rxfh_indir = get_rss_table,
+ .set_rxfh_indir = set_rss_table,
.flash_device = set_flash,
};
@@ -3318,6 +3378,22 @@ static int __devinit enable_msix(struct adapter *adap)
#undef EXTRA_VECS
+static int __devinit init_rss(struct adapter *adap)
+{
+ unsigned int i, j;
+
+ for_each_port(adap, i) {
+ struct port_info *pi = adap2pinfo(adap, i);
+
+ pi->rss = kcalloc(pi->rss_size, sizeof(u16), GFP_KERNEL);
+ if (!pi->rss)
+ return -ENOMEM;
+ for (j = 0; j < pi->rss_size; j++)
+ pi->rss[j] = j % pi->nqsets;
+ }
+ return 0;
+}
+
static void __devinit print_port_info(struct adapter *adap)
{
static const char *base[] = {
@@ -3380,9 +3456,10 @@ static void free_some_resources(struct adapter *adapter)
disable_msi(adapter);
for_each_port(adapter, i)
- if (adapter->port[i])
+ if (adapter->port[i]) {
+ kfree(adap2pinfo(adapter, i)->rss);
free_netdev(adapter->port[i]);
-
+ }
if (adapter->flags & FW_OK)
t4_fw_bye(adapter, 0);
}
@@ -3536,6 +3613,10 @@ static int __devinit init_one(struct pci_dev *pdev,
else if (msi > 0 && pci_enable_msi(pdev) == 0)
adapter->flags |= USING_MSI;
+ err = init_rss(adapter);
+ if (err)
+ goto out_free_dev;
+
/*
* The card is now ready to go. If any errors occur during device
* registration we do not fail the whole card but rather proceed only
--
1.5.4
^ permalink raw reply related
* [PATCH net-next 5/5] cxgb4: exclude registers with read side effects from register dumps
From: Dimitris Michailidis @ 2010-07-11 22:01 UTC (permalink / raw)
To: netdev; +Cc: Dimitris Michailidis
In-Reply-To: <1278885679-3738-5-git-send-email-dm@chelsio.com>
A few registers have side effects on reads, eg FIFO pointers that
auto advance or mailboxes which change ownership when read. They are
unsafe to read so exclude them from ethtool register dumps.
Signed-off-by: Dimitris Michailidis <dm@chelsio.com>
---
drivers/net/cxgb4/cxgb4_main.c | 20 +++++++++++---------
1 files changed, 11 insertions(+), 9 deletions(-)
diff --git a/drivers/net/cxgb4/cxgb4_main.c b/drivers/net/cxgb4/cxgb4_main.c
index 110843c..d3a486b 100644
--- a/drivers/net/cxgb4/cxgb4_main.c
+++ b/drivers/net/cxgb4/cxgb4_main.c
@@ -1128,7 +1128,9 @@ static void get_regs(struct net_device *dev, struct ethtool_regs *regs,
0xdfc0, 0xdfe0,
0xe000, 0xea7c,
0xf000, 0x11190,
- 0x19040, 0x19124,
+ 0x19040, 0x1906c,
+ 0x19078, 0x19080,
+ 0x1908c, 0x19124,
0x19150, 0x191b0,
0x191d0, 0x191e8,
0x19238, 0x1924c,
@@ -1141,49 +1143,49 @@ static void get_regs(struct net_device *dev, struct ethtool_regs *regs,
0x1a190, 0x1a1c4,
0x1a1fc, 0x1a1fc,
0x1e040, 0x1e04c,
- 0x1e240, 0x1e28c,
+ 0x1e284, 0x1e28c,
0x1e2c0, 0x1e2c0,
0x1e2e0, 0x1e2e0,
0x1e300, 0x1e384,
0x1e3c0, 0x1e3c8,
0x1e440, 0x1e44c,
- 0x1e640, 0x1e68c,
+ 0x1e684, 0x1e68c,
0x1e6c0, 0x1e6c0,
0x1e6e0, 0x1e6e0,
0x1e700, 0x1e784,
0x1e7c0, 0x1e7c8,
0x1e840, 0x1e84c,
- 0x1ea40, 0x1ea8c,
+ 0x1ea84, 0x1ea8c,
0x1eac0, 0x1eac0,
0x1eae0, 0x1eae0,
0x1eb00, 0x1eb84,
0x1ebc0, 0x1ebc8,
0x1ec40, 0x1ec4c,
- 0x1ee40, 0x1ee8c,
+ 0x1ee84, 0x1ee8c,
0x1eec0, 0x1eec0,
0x1eee0, 0x1eee0,
0x1ef00, 0x1ef84,
0x1efc0, 0x1efc8,
0x1f040, 0x1f04c,
- 0x1f240, 0x1f28c,
+ 0x1f284, 0x1f28c,
0x1f2c0, 0x1f2c0,
0x1f2e0, 0x1f2e0,
0x1f300, 0x1f384,
0x1f3c0, 0x1f3c8,
0x1f440, 0x1f44c,
- 0x1f640, 0x1f68c,
+ 0x1f684, 0x1f68c,
0x1f6c0, 0x1f6c0,
0x1f6e0, 0x1f6e0,
0x1f700, 0x1f784,
0x1f7c0, 0x1f7c8,
0x1f840, 0x1f84c,
- 0x1fa40, 0x1fa8c,
+ 0x1fa84, 0x1fa8c,
0x1fac0, 0x1fac0,
0x1fae0, 0x1fae0,
0x1fb00, 0x1fb84,
0x1fbc0, 0x1fbc8,
0x1fc40, 0x1fc4c,
- 0x1fe40, 0x1fe8c,
+ 0x1fe84, 0x1fe8c,
0x1fec0, 0x1fec0,
0x1fee0, 0x1fee0,
0x1ff00, 0x1ff84,
--
1.5.4
^ permalink raw reply related
* Re: [PATCH net-next 5/5] cxgb4: exclude registers with read side effects from register dumps
From: Ben Hutchings @ 2010-07-11 22:17 UTC (permalink / raw)
To: Dimitris Michailidis; +Cc: netdev
In-Reply-To: <1278885679-3738-6-git-send-email-dm@chelsio.com>
On Sun, 2010-07-11 at 15:01 -0700, Dimitris Michailidis wrote:
> A few registers have side effects on reads, eg FIFO pointers that
> auto advance or mailboxes which change ownership when read. They are
> unsafe to read so exclude them from ethtool register dumps.
[...]
If you change the set of registers in the dump, you need to increment
the version number.
Ben.
--
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply
* Re: [PATCH] bnx2x: add support for receive hashing
From: David Miller @ 2010-07-11 22:34 UTC (permalink / raw)
To: vladz; +Cc: therbert, netdev
In-Reply-To: <8628FE4E7912BF47A96AE7DD7BAC0AADDDE64704DC@SJEXCHCCR02.corp.ad.broadcom.com>
From: "Vladislav Zolotarov" <vladz@broadcom.com>
Date: Sun, 11 Jul 2010 03:02:17 -0700
>> We intend to use the card provided RSS hash to optimize GSO
>> flow comparisons at some point.
>
> Could u explain what did u mean here, pls.? GSO is a Tx flow while
> RSS hash is generated for the ingress traffic.
I of course meant GRO, not GSO.
^ permalink raw reply
* Re: [PATCH net-next 5/5] cxgb4: exclude registers with read side effects from register dumps
From: David Miller @ 2010-07-12 0:03 UTC (permalink / raw)
To: bhutchings; +Cc: dm, netdev
In-Reply-To: <1278886626.6139.87.camel@localhost>
From: Ben Hutchings <bhutchings@solarflare.com>
Date: Sun, 11 Jul 2010 23:17:06 +0100
> On Sun, 2010-07-11 at 15:01 -0700, Dimitris Michailidis wrote:
>> A few registers have side effects on reads, eg FIFO pointers that
>> auto advance or mailboxes which change ownership when read. They are
>> unsafe to read so exclude them from ethtool register dumps.
> [...]
>
> If you change the set of registers in the dump, you need to increment
> the version number.
Agreed.
^ permalink raw reply
* Re: [PATCH net-next 0/8] tg3 bugfixes
From: David Miller @ 2010-07-12 0:07 UTC (permalink / raw)
To: mcarlson; +Cc: netdev, andy
In-Reply-To: <1278876707-21390-1-git-send-email-mcarlson@broadcom.com>
From: "Matt Carlson" <mcarlson@broadcom.com>
Date: Sun, 11 Jul 2010 12:31:39 -0700
> This patchset fixes a few MSI-X related problems and performs a
> few driver enhancements.
All applied, thanks Matt.
^ permalink raw reply
* Re: [PATCH net-next 0/5] cxgb4 updates
From: David Miller @ 2010-07-12 0:08 UTC (permalink / raw)
To: dm; +Cc: netdev
In-Reply-To: <1278885679-3738-1-git-send-email-dm@chelsio.com>
From: Dimitris Michailidis <dm@chelsio.com>
Date: Sun, 11 Jul 2010 15:01:14 -0700
>
> Following is a series of 5 cxgb4 patches, all for net-next. Mainly they
> implement the get_rxnfc, get_rxfh_indir, and set_rxfh_indir ethtool ops,
> plus a couple of fixes.
Applied patches 1-4, please fixup patch 5 to bump the register dump
version number.
Thanks.
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox