* [PATCH net v4 1/6] bnx2x: properly initialize statistic counters
2013-08-05 21:35 [PATCH net v4 0/6] bnx2x: fixes Dmitry Kravkov
@ 2013-08-05 21:35 ` Dmitry Kravkov
2013-08-12 20:08 ` Benjamin Poirier
2013-08-05 21:35 ` [PATCH net v4 2/6] bnx2x: protect different statistics flows Dmitry Kravkov
` (4 subsequent siblings)
5 siblings, 1 reply; 10+ messages in thread
From: Dmitry Kravkov @ 2013-08-05 21:35 UTC (permalink / raw)
To: davem, netdev; +Cc: eilong, Dmitry Kravkov, Ariel Elior
This prevent second statistics query be sent before first one is complete.
This is required since two outstanding queries may cause FW assert.
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c | 7 ++++++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
index 98366ab..a22ad61 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
@@ -1589,12 +1589,17 @@ void bnx2x_memset_stats(struct bnx2x *bp)
void bnx2x_stats_init(struct bnx2x *bp)
{
- int /*abs*/port = BP_PORT(bp);
+ struct stats_counter *counters = &bp->fw_stats_data->storm_counters;
int mb_idx = BP_FW_MB_IDX(bp);
+ int port = BP_PORT(bp);
bp->stats_pending = 0;
bp->executer_idx = 0;
bp->stats_counter = 0;
+ counters->xstats_counter = cpu_to_le16(0xFFFF);
+ counters->tstats_counter = cpu_to_le16(0xFFFF);
+ counters->ustats_counter = cpu_to_le16(0xFFFF);
+ counters->cstats_counter = cpu_to_le16(0xFFFF);
/* port and func stats for management */
if (!BP_NOMCP(bp)) {
--
1.8.1.4
^ permalink raw reply related [flat|nested] 10+ messages in thread* Re: [PATCH net v4 1/6] bnx2x: properly initialize statistic counters
2013-08-05 21:35 ` [PATCH net v4 1/6] bnx2x: properly initialize statistic counters Dmitry Kravkov
@ 2013-08-12 20:08 ` Benjamin Poirier
2013-08-12 22:48 ` Dmitry Kravkov
0 siblings, 1 reply; 10+ messages in thread
From: Benjamin Poirier @ 2013-08-12 20:08 UTC (permalink / raw)
To: Dmitry Kravkov; +Cc: davem, netdev, eilong, Ariel Elior
On 2013/08/06 00:35, Dmitry Kravkov wrote:
> This prevent second statistics query be sent before first one is complete.
> This is required since two outstanding queries may cause FW assert.
>
> Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
> Signed-off-by: Ariel Elior <ariele@broadcom.com>
> Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
> ---
> drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c | 7 ++++++-
> 1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
> index 98366ab..a22ad61 100644
> --- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
> +++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
> @@ -1589,12 +1589,17 @@ void bnx2x_memset_stats(struct bnx2x *bp)
>
> void bnx2x_stats_init(struct bnx2x *bp)
> {
> - int /*abs*/port = BP_PORT(bp);
> + struct stats_counter *counters = &bp->fw_stats_data->storm_counters;
> int mb_idx = BP_FW_MB_IDX(bp);
> + int port = BP_PORT(bp);
>
> bp->stats_pending = 0;
> bp->executer_idx = 0;
> bp->stats_counter = 0;
> + counters->xstats_counter = cpu_to_le16(0xFFFF);
> + counters->tstats_counter = cpu_to_le16(0xFFFF);
> + counters->ustats_counter = cpu_to_le16(0xFFFF);
> + counters->cstats_counter = cpu_to_le16(0xFFFF);
I think the following code executed a little later becomes redundant:
static void bnx2x_prep_fw_stats_req(struct bnx2x *bp)
{
[...]
/* prepare to the first stats ramrod (will be completed with
* the counters equal to zero) - init counters to somethig different.
*/
memset(&bp->fw_stats_data->storm_counters, 0xff,
sizeof(struct stats_counter));
^ permalink raw reply [flat|nested] 10+ messages in thread* RE: [PATCH net v4 1/6] bnx2x: properly initialize statistic counters
2013-08-12 20:08 ` Benjamin Poirier
@ 2013-08-12 22:48 ` Dmitry Kravkov
0 siblings, 0 replies; 10+ messages in thread
From: Dmitry Kravkov @ 2013-08-12 22:48 UTC (permalink / raw)
To: Benjamin Poirier; +Cc: davem, netdev, Eilon Greenstein, Ariel Elior
> -----Original Message-----
> From: Benjamin Poirier [mailto:benjamin.poirier@gmail.com]
> Sent: Monday, August 12, 2013 11:08 PM
> To: Dmitry Kravkov
> Cc: davem@davemloft.net; netdev@vger.kernel.org; Eilon Greenstein;
> Ariel Elior
> Subject: Re: [PATCH net v4 1/6] bnx2x: properly initialize statistic counters
>
> On 2013/08/06 00:35, Dmitry Kravkov wrote:
> > This prevent second statistics query be sent before first one is complete.
> > This is required since two outstanding queries may cause FW assert.
> >
> > Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
> > Signed-off-by: Ariel Elior <ariele@broadcom.com>
> > Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
> > ---
> > drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c | 7 ++++++-
> > 1 file changed, 6 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
> > b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
> > index 98366ab..a22ad61 100644
> > --- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
> > +++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
> > @@ -1589,12 +1589,17 @@ void bnx2x_memset_stats(struct bnx2x *bp)
> >
> > void bnx2x_stats_init(struct bnx2x *bp) {
> > - int /*abs*/port = BP_PORT(bp);
> > + struct stats_counter *counters = &bp->fw_stats_data-
> >storm_counters;
> > int mb_idx = BP_FW_MB_IDX(bp);
> > + int port = BP_PORT(bp);
> >
> > bp->stats_pending = 0;
> > bp->executer_idx = 0;
> > bp->stats_counter = 0;
> > + counters->xstats_counter = cpu_to_le16(0xFFFF);
> > + counters->tstats_counter = cpu_to_le16(0xFFFF);
> > + counters->ustats_counter = cpu_to_le16(0xFFFF);
> > + counters->cstats_counter = cpu_to_le16(0xFFFF);
>
> I think the following code executed a little later becomes redundant:
>
> static void bnx2x_prep_fw_stats_req(struct bnx2x *bp) { [...]
> /* prepare to the first stats ramrod (will be completed with
> * the counters equal to zero) - init counters to somethig different.
> */
> memset(&bp->fw_stats_data->storm_counters, 0xff,
> sizeof(struct stats_counter));
You absolutely right. Thanks!
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH net v4 2/6] bnx2x: protect different statistics flows
2013-08-05 21:35 [PATCH net v4 0/6] bnx2x: fixes Dmitry Kravkov
2013-08-05 21:35 ` [PATCH net v4 1/6] bnx2x: properly initialize statistic counters Dmitry Kravkov
@ 2013-08-05 21:35 ` Dmitry Kravkov
2013-08-05 23:26 ` David Miller
2013-08-05 21:35 ` [PATCH net v4 3/6] bnx2x: update fairness parameters following DCB negotiation Dmitry Kravkov
` (3 subsequent siblings)
5 siblings, 1 reply; 10+ messages in thread
From: Dmitry Kravkov @ 2013-08-05 21:35 UTC (permalink / raw)
To: davem, netdev; +Cc: eilong, Dmitry Kravkov, Ariel Elior
Add locking to protect different statistics flows from
running simultaneously.
This in order to serialize statistics requests sent to FW,
otherwise two outstanding queries may cause FW assert.
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x.h | 3 +
drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c | 88 +++++++++++++++++++----
2 files changed, 79 insertions(+), 12 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index d80e34b..c7ffff3 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -1830,6 +1830,9 @@ struct bnx2x {
int fp_array_size;
u32 dump_preset_idx;
+ bool stats_started;
+ unsigned long stats_flags;
+#define BNX2X_STATS_UPDATE_BIT 0
};
/* Tx queues may be less or equal to Rx queues */
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
index a22ad61..bf7f9a7 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_stats.c
@@ -221,7 +221,8 @@ static int bnx2x_stats_comp(struct bnx2x *bp)
* Statistics service functions
*/
-static void bnx2x_stats_pmf_update(struct bnx2x *bp)
+/* should be called under bnx2x_lock_stat_transition */
+static void __bnx2x_stats_pmf_update(struct bnx2x *bp)
{
struct dmae_command *dmae;
u32 opcode;
@@ -518,7 +519,8 @@ static void bnx2x_func_stats_init(struct bnx2x *bp)
*stats_comp = 0;
}
-static void bnx2x_stats_start(struct bnx2x *bp)
+/* should be called under bnx2x_lock_stat_transition */
+static void __bnx2x_stats_start(struct bnx2x *bp)
{
/* vfs travel through here as part of the statistics FSM, but no action
* is required
@@ -534,13 +536,58 @@ static void bnx2x_stats_start(struct bnx2x *bp)
bnx2x_hw_stats_post(bp);
bnx2x_storm_stats_post(bp);
+
+ bp->stats_started = true;
+}
+
+static bool bnx2x_trylock_stat_transition(struct bnx2x *bp)
+{
+ return !test_and_set_bit(BNX2X_STATS_UPDATE_BIT, &bp->stats_flags);
+}
+
+static void bnx2x_lock_stat_transition(struct bnx2x *bp)
+{
+ int cnt = 10; /* 10-20mSec */
+
+ while (test_and_set_bit(BNX2X_STATS_UPDATE_BIT, &bp->stats_flags)) {
+ if (!cnt) {
+ BNX2X_ERR("timeout waiting for stats lock\n");
+ bnx2x_panic();
+ break;
+ }
+ cnt--;
+ usleep_range(1000, 2000);
+ }
+}
+
+static void bnx2x_unlock_stat_transition(struct bnx2x *bp)
+{
+ smp_mb__before_clear_bit();
+ clear_bit(BNX2X_STATS_UPDATE_BIT, &bp->stats_flags);
+ smp_mb__after_clear_bit();
+}
+
+static void bnx2x_stats_start(struct bnx2x *bp)
+{
+ bnx2x_lock_stat_transition(bp);
+ __bnx2x_stats_start(bp);
+ bnx2x_unlock_stat_transition(bp);
}
static void bnx2x_stats_pmf_start(struct bnx2x *bp)
{
+ bnx2x_lock_stat_transition(bp);
bnx2x_stats_comp(bp);
- bnx2x_stats_pmf_update(bp);
- bnx2x_stats_start(bp);
+ __bnx2x_stats_pmf_update(bp);
+ __bnx2x_stats_start(bp);
+ bnx2x_unlock_stat_transition(bp);
+}
+
+static void bnx2x_stats_pmf_update(struct bnx2x *bp)
+{
+ bnx2x_lock_stat_transition(bp);
+ __bnx2x_stats_pmf_update(bp);
+ bnx2x_unlock_stat_transition(bp);
}
static void bnx2x_stats_restart(struct bnx2x *bp)
@@ -550,8 +597,10 @@ static void bnx2x_stats_restart(struct bnx2x *bp)
*/
if (IS_VF(bp))
return;
+ bnx2x_lock_stat_transition(bp);
bnx2x_stats_comp(bp);
- bnx2x_stats_start(bp);
+ __bnx2x_stats_start(bp);
+ bnx2x_unlock_stat_transition(bp);
}
static void bnx2x_bmac_stats_update(struct bnx2x *bp)
@@ -888,9 +937,7 @@ static int bnx2x_storm_stats_validate_counters(struct bnx2x *bp)
/* Make sure we use the value of the counter
* used for sending the last stats ramrod.
*/
- spin_lock_bh(&bp->stats_lock);
cur_stats_counter = bp->stats_counter - 1;
- spin_unlock_bh(&bp->stats_lock);
/* are storm stats valid? */
if (le16_to_cpu(counters->xstats_counter) != cur_stats_counter) {
@@ -1227,12 +1274,18 @@ static void bnx2x_stats_update(struct bnx2x *bp)
{
u32 *stats_comp = bnx2x_sp(bp, stats_comp);
- if (bnx2x_edebug_stats_stopped(bp))
+ /* we run update from timer context, so give up
+ * if somebody in the middle of transition
+ */
+ if (!bnx2x_trylock_stat_transition(bp))
return;
+ if (bnx2x_edebug_stats_stopped(bp) || !bp->stats_started)
+ goto out;
+
if (IS_PF(bp)) {
if (*stats_comp != DMAE_COMP_VAL)
- return;
+ goto out;
if (bp->port.pmf)
bnx2x_hw_stats_update(bp);
@@ -1242,7 +1295,7 @@ static void bnx2x_stats_update(struct bnx2x *bp)
BNX2X_ERR("storm stats were not updated for 3 times\n");
bnx2x_panic();
}
- return;
+ goto out;
}
} else {
/* vf doesn't collect HW statistics, and doesn't get completions
@@ -1256,7 +1309,7 @@ static void bnx2x_stats_update(struct bnx2x *bp)
/* vf is done */
if (IS_VF(bp))
- return;
+ goto out;
if (netif_msg_timer(bp)) {
struct bnx2x_eth_stats *estats = &bp->eth_stats;
@@ -1267,6 +1320,9 @@ static void bnx2x_stats_update(struct bnx2x *bp)
bnx2x_hw_stats_post(bp);
bnx2x_storm_stats_post(bp);
+
+out:
+ bnx2x_unlock_stat_transition(bp);
}
static void bnx2x_port_stats_stop(struct bnx2x *bp)
@@ -1332,6 +1388,10 @@ static void bnx2x_stats_stop(struct bnx2x *bp)
{
int update = 0;
+ bnx2x_lock_stat_transition(bp);
+
+ bp->stats_started = false;
+
bnx2x_stats_comp(bp);
if (bp->port.pmf)
@@ -1348,6 +1408,8 @@ static void bnx2x_stats_stop(struct bnx2x *bp)
bnx2x_hw_stats_post(bp);
bnx2x_stats_comp(bp);
}
+
+ bnx2x_unlock_stat_transition(bp);
}
static void bnx2x_stats_do_nothing(struct bnx2x *bp)
@@ -1376,15 +1438,17 @@ static const struct {
void bnx2x_stats_handle(struct bnx2x *bp, enum bnx2x_stats_event event)
{
enum bnx2x_stats_state state;
+ void (*action)(struct bnx2x *bp);
if (unlikely(bp->panic))
return;
spin_lock_bh(&bp->stats_lock);
state = bp->stats_state;
bp->stats_state = bnx2x_stats_stm[state][event].next_state;
+ action = bnx2x_stats_stm[state][event].action;
spin_unlock_bh(&bp->stats_lock);
- bnx2x_stats_stm[state][event].action(bp);
+ action(bp);
if ((event != STATS_EVENT_UPDATE) || netif_msg_timer(bp))
DP(BNX2X_MSG_STATS, "state %d -> event %d -> state %d\n",
--
1.8.1.4
^ permalink raw reply related [flat|nested] 10+ messages in thread* Re: [PATCH net v4 2/6] bnx2x: protect different statistics flows
2013-08-05 21:35 ` [PATCH net v4 2/6] bnx2x: protect different statistics flows Dmitry Kravkov
@ 2013-08-05 23:26 ` David Miller
0 siblings, 0 replies; 10+ messages in thread
From: David Miller @ 2013-08-05 23:26 UTC (permalink / raw)
To: dmitry; +Cc: netdev, eilong, ariele
From: "Dmitry Kravkov" <dmitry@broadcom.com>
Date: Tue, 6 Aug 2013 00:35:40 +0300
> Add locking to protect different statistics flows from
> running simultaneously.
> This in order to serialize statistics requests sent to FW,
> otherwise two outstanding queries may cause FW assert.
>
> Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
> Signed-off-by: Ariel Elior <ariele@broadcom.com>
> Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Please do not reinvent the wheel, use a standard semaphore
and down_timeout().
Anytime you feel that you have to implement your own fancy locking
primitives, you are doing it wrong.
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH net v4 3/6] bnx2x: update fairness parameters following DCB negotiation
2013-08-05 21:35 [PATCH net v4 0/6] bnx2x: fixes Dmitry Kravkov
2013-08-05 21:35 ` [PATCH net v4 1/6] bnx2x: properly initialize statistic counters Dmitry Kravkov
2013-08-05 21:35 ` [PATCH net v4 2/6] bnx2x: protect different statistics flows Dmitry Kravkov
@ 2013-08-05 21:35 ` Dmitry Kravkov
2013-08-05 21:35 ` [PATCH net v4 4/6] bnx2x: fix memory leak in VF Dmitry Kravkov
` (2 subsequent siblings)
5 siblings, 0 replies; 10+ messages in thread
From: Dmitry Kravkov @ 2013-08-05 21:35 UTC (permalink / raw)
To: davem, netdev; +Cc: eilong, Dmitry Kravkov, Ariel Elior
ETS can be enabled as a result of DCB negotiation, then
fairness must be recalculated after each negotiation.
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x.h | 2 ++
drivers/net/ethernet/broadcom/bnx2x/bnx2x_dcb.c | 4 ++++
drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c | 30 ++++++++++++++----------
3 files changed, 24 insertions(+), 12 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index c7ffff3..062480a 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -2454,4 +2454,6 @@ enum bnx2x_pci_bus_speed {
BNX2X_PCI_LINK_SPEED_5000 = 5000,
BNX2X_PCI_LINK_SPEED_8000 = 8000
};
+
+void bnx2x_set_local_cmng(struct bnx2x *bp);
#endif /* bnx2x.h */
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_dcb.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_dcb.c
index 0c94df4..f9122f2 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_dcb.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_dcb.c
@@ -753,6 +753,10 @@ void bnx2x_dcbx_set_params(struct bnx2x *bp, u32 state)
bnx2x_pfc_set_pfc(bp);
bnx2x_dcbx_update_ets_params(bp);
+
+ /* ets may affect cmng configuration: reinit it in hw */
+ bnx2x_set_local_cmng(bp);
+
bnx2x_dcbx_resume_hw_tx(bp);
return;
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index e06186c..90a84af 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -2476,7 +2476,7 @@ static void bnx2x_cmng_fns_init(struct bnx2x *bp, u8 read_cfg, u8 cmng_type)
input.port_rate = bp->link_vars.line_speed;
- if (cmng_type == CMNG_FNS_MINMAX) {
+ if (cmng_type == CMNG_FNS_MINMAX && input.port_rate) {
int vn;
/* read mf conf from shmem */
@@ -2533,6 +2533,21 @@ static void storm_memset_cmng(struct bnx2x *bp,
}
}
+/* init cmng mode in HW according to local configuration */
+void bnx2x_set_local_cmng(struct bnx2x *bp)
+{
+ int cmng_fns = bnx2x_get_cmng_fns_mode(bp);
+
+ if (cmng_fns != CMNG_FNS_NONE) {
+ bnx2x_cmng_fns_init(bp, false, cmng_fns);
+ storm_memset_cmng(bp, &bp->cmng, BP_PORT(bp));
+ } else {
+ /* rate shaping and fairness are disabled */
+ DP(NETIF_MSG_IFUP,
+ "single function mode without fairness\n");
+ }
+}
+
/* This function is called upon link interrupt */
static void bnx2x_link_attn(struct bnx2x *bp)
{
@@ -2568,17 +2583,8 @@ static void bnx2x_link_attn(struct bnx2x *bp)
bnx2x_stats_handle(bp, STATS_EVENT_LINK_UP);
}
- if (bp->link_vars.link_up && bp->link_vars.line_speed) {
- int cmng_fns = bnx2x_get_cmng_fns_mode(bp);
-
- if (cmng_fns != CMNG_FNS_NONE) {
- bnx2x_cmng_fns_init(bp, false, cmng_fns);
- storm_memset_cmng(bp, &bp->cmng, BP_PORT(bp));
- } else
- /* rate shaping and fairness are disabled */
- DP(NETIF_MSG_IFUP,
- "single function mode without fairness\n");
- }
+ if (bp->link_vars.link_up && bp->link_vars.line_speed)
+ bnx2x_set_local_cmng(bp);
__bnx2x_link_report(bp);
--
1.8.1.4
^ permalink raw reply related [flat|nested] 10+ messages in thread* [PATCH net v4 4/6] bnx2x: fix memory leak in VF
2013-08-05 21:35 [PATCH net v4 0/6] bnx2x: fixes Dmitry Kravkov
` (2 preceding siblings ...)
2013-08-05 21:35 ` [PATCH net v4 3/6] bnx2x: update fairness parameters following DCB negotiation Dmitry Kravkov
@ 2013-08-05 21:35 ` Dmitry Kravkov
2013-08-05 21:35 ` [PATCH net v4 5/6] bnx2x: fix PTE write access error Dmitry Kravkov
2013-08-05 21:35 ` [PATCH net v4 6/6] bnx2x: prevent crash in shutdown flow with CNIC Dmitry Kravkov
5 siblings, 0 replies; 10+ messages in thread
From: Dmitry Kravkov @ 2013-08-05 21:35 UTC (permalink / raw)
To: davem, netdev; +Cc: eilong, Ariel Elior, Dmitry Kravkov
From: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c
index 95861ef..44104fb 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c
@@ -3463,7 +3463,7 @@ int bnx2x_vf_pci_alloc(struct bnx2x *bp)
alloc_mem_err:
BNX2X_PCI_FREE(bp->vf2pf_mbox, bp->vf2pf_mbox_mapping,
sizeof(struct bnx2x_vf_mbx_msg));
- BNX2X_PCI_FREE(bp->vf2pf_mbox, bp->vf2pf_mbox_mapping,
+ BNX2X_PCI_FREE(bp->vf2pf_mbox, bp->pf2vf_bulletin_mapping,
sizeof(union pf_vf_bulletin));
return -ENOMEM;
}
--
1.8.1.4
^ permalink raw reply related [flat|nested] 10+ messages in thread* [PATCH net v4 5/6] bnx2x: fix PTE write access error
2013-08-05 21:35 [PATCH net v4 0/6] bnx2x: fixes Dmitry Kravkov
` (3 preceding siblings ...)
2013-08-05 21:35 ` [PATCH net v4 4/6] bnx2x: fix memory leak in VF Dmitry Kravkov
@ 2013-08-05 21:35 ` Dmitry Kravkov
2013-08-05 21:35 ` [PATCH net v4 6/6] bnx2x: prevent crash in shutdown flow with CNIC Dmitry Kravkov
5 siblings, 0 replies; 10+ messages in thread
From: Dmitry Kravkov @ 2013-08-05 21:35 UTC (permalink / raw)
To: davem, netdev; +Cc: eilong, Barak Witkowsky, Dmitry Kravkov, Ariel Elior
From: Barak Witkowsky <barak@broadcom.com>
PTE write access error might occur in MF_ALLOWED mode when IOMMU
is active. The patch adds rmmod HSI indicating to MFW to stop
running queries which might trigger this failure.
Signed-off-by: Barak Witkowsky <barak@broadcom.com>
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x.h | 1 +
drivers/net/ethernet/broadcom/bnx2x/bnx2x_hsi.h | 5 +++++
drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c | 9 +++++++++
3 files changed, 15 insertions(+)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index 062480a..b82793e 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -1502,6 +1502,7 @@ struct bnx2x {
#define BC_SUPPORTS_DCBX_MSG_NON_PMF (1 << 21)
#define IS_VF_FLAG (1 << 22)
#define INTERRUPTS_ENABLED_FLAG (1 << 23)
+#define BC_SUPPORTS_RMMOD_CMD (1 << 24)
#define BP_NOMCP(bp) ((bp)->flags & NO_MCP_FLAG)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_hsi.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_hsi.h
index 5018e52..32767f6 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_hsi.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_hsi.h
@@ -1300,6 +1300,9 @@ struct drv_func_mb {
#define DRV_MSG_CODE_EEE_RESULTS_ACK 0xda000000
+ #define DRV_MSG_CODE_RMMOD 0xdb000000
+ #define REQ_BC_VER_4_RMMOD_CMD 0x0007080f
+
#define DRV_MSG_CODE_SET_MF_BW 0xe0000000
#define REQ_BC_VER_4_SET_MF_BW 0x00060202
#define DRV_MSG_CODE_SET_MF_BW_ACK 0xe1000000
@@ -1372,6 +1375,8 @@ struct drv_func_mb {
#define FW_MSG_CODE_EEE_RESULS_ACK 0xda100000
+ #define FW_MSG_CODE_RMMOD_ACK 0xdb100000
+
#define FW_MSG_CODE_SET_MF_BW_SENT 0xe0000000
#define FW_MSG_CODE_SET_MF_BW_DONE 0xe1000000
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index 90a84af..5c22081 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -10368,6 +10368,10 @@ static void bnx2x_get_common_hwinfo(struct bnx2x *bp)
bp->flags |= (val >= REQ_BC_VER_4_DCBX_ADMIN_MSG_NON_PMF) ?
BC_SUPPORTS_DCBX_MSG_NON_PMF : 0;
+
+ bp->flags |= (val >= REQ_BC_VER_4_RMMOD_CMD) ?
+ BC_SUPPORTS_RMMOD_CMD : 0;
+
boot_mode = SHMEM_RD(bp,
dev_info.port_feature_config[BP_PORT(bp)].mba_config) &
PORT_FEATURE_MBA_BOOT_AGENT_TYPE_MASK;
@@ -12823,6 +12827,11 @@ static void __bnx2x_remove(struct pci_dev *pdev,
bnx2x_dcbnl_update_applist(bp, true);
#endif
+ if (IS_PF(bp) &&
+ !BP_NOMCP(bp) &&
+ (bp->flags & BC_SUPPORTS_RMMOD_CMD))
+ bnx2x_fw_command(bp, DRV_MSG_CODE_RMMOD, 0);
+
/* Close the interface - either directly or implicitly */
if (remove_netdev) {
unregister_netdev(dev);
--
1.8.1.4
^ permalink raw reply related [flat|nested] 10+ messages in thread* [PATCH net v4 6/6] bnx2x: prevent crash in shutdown flow with CNIC
2013-08-05 21:35 [PATCH net v4 0/6] bnx2x: fixes Dmitry Kravkov
` (4 preceding siblings ...)
2013-08-05 21:35 ` [PATCH net v4 5/6] bnx2x: fix PTE write access error Dmitry Kravkov
@ 2013-08-05 21:35 ` Dmitry Kravkov
5 siblings, 0 replies; 10+ messages in thread
From: Dmitry Kravkov @ 2013-08-05 21:35 UTC (permalink / raw)
To: davem, netdev; +Cc: eilong, Yuval Mintz, Dmitry Kravkov, Ariel Elior
From: Yuval Mintz <yuvalmin@broadcom.com>
There might be a crash as during shutdown flow CNIC might try
to access resources already freed by bnx2x.
Change bnx2x_close() into dev_close() in __bnx2x_remove (shutdown flow)
to guarantee CNIC is notified of the device's change of status.
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index 5c22081..abafb30 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -12837,8 +12837,7 @@ static void __bnx2x_remove(struct pci_dev *pdev,
unregister_netdev(dev);
} else {
rtnl_lock();
- if (netif_running(dev))
- bnx2x_close(dev);
+ dev_close(dev);
rtnl_unlock();
}
--
1.8.1.4
^ permalink raw reply related [flat|nested] 10+ messages in thread