[PATCH v3 net] qed/qede: Fix scheduling while atomic

[PATCH v3 net] qed/qede: Fix scheduling while atomic

[Manish Chopra]

Bonding module collects the statistics while holding
the spinlock, beneath that qede->qed driver statistics
flow gets scheduled out due to usleep_range() used in PTT
acquire logic which results into below bug and traces -

[ 3673.988874] Hardware name: HPE ProLiant DL365 Gen10 Plus/ProLiant DL365 Gen10 Plus, BIOS A42 10/29/2021
[ 3673.988878] Call Trace:
[ 3673.988891]  dump_stack_lvl+0x34/0x44
[ 3673.988908]  __schedule_bug.cold+0x47/0x53
[ 3673.988918]  __schedule+0x3fb/0x560
[ 3673.988929]  schedule+0x43/0xb0
[ 3673.988932]  schedule_hrtimeout_range_clock+0xbf/0x1b0
[ 3673.988937]  ? __hrtimer_init+0xc0/0xc0
[ 3673.988950]  usleep_range+0x5e/0x80
[ 3673.988955]  qed_ptt_acquire+0x2b/0xd0 [qed]
[ 3673.988981]  _qed_get_vport_stats+0x141/0x240 [qed]
[ 3673.989001]  qed_get_vport_stats+0x18/0x80 [qed]
[ 3673.989016]  qede_fill_by_demand_stats+0x37/0x400 [qede]
[ 3673.989028]  qede_get_stats64+0x19/0xe0 [qede]
[ 3673.989034]  dev_get_stats+0x5c/0xc0
[ 3673.989045]  netstat_show.constprop.0+0x52/0xb0
[ 3673.989055]  dev_attr_show+0x19/0x40
[ 3673.989065]  sysfs_kf_seq_show+0x9b/0xf0
[ 3673.989076]  seq_read_iter+0x120/0x4b0
[ 3673.989087]  new_sync_read+0x118/0x1a0
[ 3673.989095]  vfs_read+0xf3/0x180
[ 3673.989099]  ksys_read+0x5f/0xe0
[ 3673.989102]  do_syscall_64+0x3b/0x90
[ 3673.989109]  entry_SYSCALL_64_after_hwframe+0x44/0xae
[ 3673.989115] RIP: 0033:0x7f8467d0b082
[ 3673.989119] Code: c0 e9 b2 fe ff ff 50 48 8d 3d ca 05 08 00 e8 35 e7 01 00 0f 1f 44 00 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 0f 05 <48> 3d 00 f0 ff ff 77 56 c3 0f 1f 44 00 00 48 83 ec 28 48 89 54 24
[ 3673.989121] RSP: 002b:00007ffffb21fd08 EFLAGS: 00000246 ORIG_RAX: 0000000000000000
[ 3673.989127] RAX: ffffffffffffffda RBX: 000000000100eca0 RCX: 00007f8467d0b082
[ 3673.989128] RDX: 00000000000003ff RSI: 00007ffffb21fdc0 RDI: 0000000000000003
[ 3673.989130] RBP: 00007f8467b96028 R08: 0000000000000010 R09: 00007ffffb21ec00
[ 3673.989132] R10: 00007ffffb27b170 R11: 0000000000000246 R12: 00000000000000f0
[ 3673.989134] R13: 0000000000000003 R14: 00007f8467b92000 R15: 0000000000045a05
[ 3673.989139] CPU: 30 PID: 285188 Comm: read_all Kdump: loaded Tainted: G        W  OE

Fix this by having caller (QEDE driver flows) to provide the context
whether it could be in atomic context flow or not when getting the
vport stats from QED driver. QED driver based on the context provided
decide to schedule out or not when acquiring the PTT BAR window.

Fixes: 133fac0eedc3 ("qede: Add basic ethtool support")
Cc: Sudarsana Kalluru <skalluru@marvell.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Manish Chopra <manishc@marvell.com>
Signed-off-by: Ariel Elior <aelior@marvell.com>
Signed-off-by: Alok Prasad <palok@marvell.com>
---
v1->v2:
 - Fixed checkpatch and kdoc warnings.
v2->v3:
 - Moving the changelog after tags.
---
 drivers/net/ethernet/qlogic/qed/qed_dev_api.h | 12 +++++++-
 drivers/net/ethernet/qlogic/qed/qed_hw.c      | 28 +++++++++++++++----
 drivers/net/ethernet/qlogic/qed/qed_l2.c      | 11 ++++----
 drivers/net/ethernet/qlogic/qed/qed_l2.h      |  3 +-
 drivers/net/ethernet/qlogic/qed/qed_main.c    |  4 +--
 drivers/net/ethernet/qlogic/qede/qede.h       |  2 +-
 .../net/ethernet/qlogic/qede/qede_ethtool.c   |  2 +-
 drivers/net/ethernet/qlogic/qede/qede_main.c  |  6 ++--
 include/linux/qed/qed_eth_if.h                |  2 +-
 9 files changed, 50 insertions(+), 20 deletions(-)

diff --git a/drivers/net/ethernet/qlogic/qed/qed_dev_api.h b/drivers/net/ethernet/qlogic/qed/qed_dev_api.h
index f8682356d0cf..379169076aa4 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_dev_api.h
+++ b/drivers/net/ethernet/qlogic/qed/qed_dev_api.h
@@ -182,7 +182,7 @@ int qed_hw_prepare(struct qed_dev *cdev,
 void qed_hw_remove(struct qed_dev *cdev);
 
 /**
- * qed_ptt_acquire(): Allocate a PTT window.
+ * qed_ptt_acquire(): Allocate a PTT window in sleepable context.
  *
  * @p_hwfn: HW device data.
  *
@@ -193,6 +193,16 @@ void qed_hw_remove(struct qed_dev *cdev);
  */
 struct qed_ptt *qed_ptt_acquire(struct qed_hwfn *p_hwfn);
 
+/**
+ *  _qed_ptt_acquire(): Allocate a PTT window based on the context.
+ *
+ *  @p_hwfn: HW device data.
+ *  @is_atomic: context (sleepable or unsleepable) based on which ptt is acquired.
+ *
+ *  Return: struct qed_ptt.
+ */
+struct qed_ptt *_qed_ptt_acquire(struct qed_hwfn *p_hwfn, bool is_atomic);
+
 /**
  * qed_ptt_release(): Release PTT Window.
  *
diff --git a/drivers/net/ethernet/qlogic/qed/qed_hw.c b/drivers/net/ethernet/qlogic/qed/qed_hw.c
index 554f30b0cfd5..4e8bfa0194e7 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_hw.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_hw.c
@@ -23,7 +23,10 @@
 #include "qed_reg_addr.h"
 #include "qed_sriov.h"
 
-#define QED_BAR_ACQUIRE_TIMEOUT 1000
+#define QED_BAR_ACQUIRE_TIMEOUT_USLEEP_CNT	1000
+#define QED_BAR_ACQUIRE_TIMEOUT_USLEEP		1000
+#define QED_BAR_ACQUIRE_TIMEOUT_UDELAY_CNT	100000
+#define QED_BAR_ACQUIRE_TIMEOUT_UDELAY		10
 
 /* Invalid values */
 #define QED_BAR_INVALID_OFFSET          (cpu_to_le32(-1))
@@ -83,13 +86,18 @@ void qed_ptt_pool_free(struct qed_hwfn *p_hwfn)
 	p_hwfn->p_ptt_pool = NULL;
 }
 
-struct qed_ptt *qed_ptt_acquire(struct qed_hwfn *p_hwfn)
+struct qed_ptt *_qed_ptt_acquire(struct qed_hwfn *p_hwfn, bool is_atomic)
 {
 	struct qed_ptt *p_ptt;
-	unsigned int i;
+	unsigned int i, count;
+
+	if (is_atomic)
+		count = QED_BAR_ACQUIRE_TIMEOUT_UDELAY_CNT;
+	else
+		count = QED_BAR_ACQUIRE_TIMEOUT_USLEEP_CNT;
 
 	/* Take the free PTT from the list */
-	for (i = 0; i < QED_BAR_ACQUIRE_TIMEOUT; i++) {
+	for (i = 0; i < count; i++) {
 		spin_lock_bh(&p_hwfn->p_ptt_pool->lock);
 
 		if (!list_empty(&p_hwfn->p_ptt_pool->free_list)) {
@@ -105,13 +113,23 @@ struct qed_ptt *qed_ptt_acquire(struct qed_hwfn *p_hwfn)
 		}
 
 		spin_unlock_bh(&p_hwfn->p_ptt_pool->lock);
-		usleep_range(1000, 2000);
+
+		if (is_atomic)
+			udelay(QED_BAR_ACQUIRE_TIMEOUT_UDELAY);
+		else
+			usleep_range(QED_BAR_ACQUIRE_TIMEOUT_USLEEP,
+				     QED_BAR_ACQUIRE_TIMEOUT_USLEEP * 2);
 	}
 
 	DP_NOTICE(p_hwfn, "PTT acquire timeout - failed to allocate PTT\n");
 	return NULL;
 }
 
+struct qed_ptt *qed_ptt_acquire(struct qed_hwfn *p_hwfn)
+{
+	return _qed_ptt_acquire(p_hwfn, false);
+}
+
 void qed_ptt_release(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt)
 {
 	spin_lock_bh(&p_hwfn->p_ptt_pool->lock);
diff --git a/drivers/net/ethernet/qlogic/qed/qed_l2.c b/drivers/net/ethernet/qlogic/qed/qed_l2.c
index 2edd6bf64a3c..46d8d35dc7ac 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_l2.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_l2.c
@@ -1863,7 +1863,7 @@ static void __qed_get_vport_stats(struct qed_hwfn *p_hwfn,
 }
 
 static void _qed_get_vport_stats(struct qed_dev *cdev,
-				 struct qed_eth_stats *stats)
+				 struct qed_eth_stats *stats, bool is_atomic)
 {
 	u8 fw_vport = 0;
 	int i;
@@ -1872,7 +1872,7 @@ static void _qed_get_vport_stats(struct qed_dev *cdev,
 
 	for_each_hwfn(cdev, i) {
 		struct qed_hwfn *p_hwfn = &cdev->hwfns[i];
-		struct qed_ptt *p_ptt = IS_PF(cdev) ? qed_ptt_acquire(p_hwfn)
+		struct qed_ptt *p_ptt = IS_PF(cdev) ? _qed_ptt_acquire(p_hwfn, is_atomic)
 						    :  NULL;
 		bool b_get_port_stats;
 
@@ -1899,7 +1899,8 @@ static void _qed_get_vport_stats(struct qed_dev *cdev,
 	}
 }
 
-void qed_get_vport_stats(struct qed_dev *cdev, struct qed_eth_stats *stats)
+void qed_get_vport_stats(struct qed_dev *cdev, struct qed_eth_stats *stats,
+			 bool is_atomic)
 {
 	u32 i;
 
@@ -1908,7 +1909,7 @@ void qed_get_vport_stats(struct qed_dev *cdev, struct qed_eth_stats *stats)
 		return;
 	}
 
-	_qed_get_vport_stats(cdev, stats);
+	_qed_get_vport_stats(cdev, stats, is_atomic);
 
 	if (!cdev->reset_stats)
 		return;
@@ -1960,7 +1961,7 @@ void qed_reset_vport_stats(struct qed_dev *cdev)
 	if (!cdev->reset_stats) {
 		DP_INFO(cdev, "Reset stats not allocated\n");
 	} else {
-		_qed_get_vport_stats(cdev, cdev->reset_stats);
+		_qed_get_vport_stats(cdev, cdev->reset_stats, false);
 		cdev->reset_stats->common.link_change_count = 0;
 	}
 }
diff --git a/drivers/net/ethernet/qlogic/qed/qed_l2.h b/drivers/net/ethernet/qlogic/qed/qed_l2.h
index a538cf478c14..2bb93c50a2e4 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_l2.h
+++ b/drivers/net/ethernet/qlogic/qed/qed_l2.h
@@ -249,7 +249,8 @@ qed_sp_eth_rx_queues_update(struct qed_hwfn *p_hwfn,
 			    enum spq_mode comp_mode,
 			    struct qed_spq_comp_cb *p_comp_data);
 
-void qed_get_vport_stats(struct qed_dev *cdev, struct qed_eth_stats *stats);
+void qed_get_vport_stats(struct qed_dev *cdev,
+			 struct qed_eth_stats *stats, bool is_atomic);
 
 void qed_reset_vport_stats(struct qed_dev *cdev);
 
diff --git a/drivers/net/ethernet/qlogic/qed/qed_main.c b/drivers/net/ethernet/qlogic/qed/qed_main.c
index c91898be7c03..307856c4ed22 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_main.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_main.c
@@ -3101,7 +3101,7 @@ void qed_get_protocol_stats(struct qed_dev *cdev,
 
 	switch (type) {
 	case QED_MCP_LAN_STATS:
-		qed_get_vport_stats(cdev, &eth_stats);
+		qed_get_vport_stats(cdev, &eth_stats, false);
 		stats->lan_stats.ucast_rx_pkts =
 					eth_stats.common.rx_ucast_pkts;
 		stats->lan_stats.ucast_tx_pkts =
@@ -3161,7 +3161,7 @@ qed_fill_generic_tlv_data(struct qed_dev *cdev, struct qed_mfw_tlv_generic *tlv)
 		}
 	}
 
-	qed_get_vport_stats(cdev, &stats);
+	qed_get_vport_stats(cdev, &stats, false);
 	p_common = &stats.common;
 	tlv->rx_frames = p_common->rx_ucast_pkts + p_common->rx_mcast_pkts +
 			 p_common->rx_bcast_pkts;
diff --git a/drivers/net/ethernet/qlogic/qede/qede.h b/drivers/net/ethernet/qlogic/qede/qede.h
index f90dcfe9ee68..312b1c2484fe 100644
--- a/drivers/net/ethernet/qlogic/qede/qede.h
+++ b/drivers/net/ethernet/qlogic/qede/qede.h
@@ -569,7 +569,7 @@ void qede_set_udp_tunnels(struct qede_dev *edev);
 void qede_reload(struct qede_dev *edev,
 		 struct qede_reload_args *args, bool is_locked);
 int qede_change_mtu(struct net_device *dev, int new_mtu);
-void qede_fill_by_demand_stats(struct qede_dev *edev);
+void qede_fill_by_demand_stats(struct qede_dev *edev, bool is_atomic);
 void __qede_lock(struct qede_dev *edev);
 void __qede_unlock(struct qede_dev *edev);
 bool qede_has_rx_work(struct qede_rx_queue *rxq);
diff --git a/drivers/net/ethernet/qlogic/qede/qede_ethtool.c b/drivers/net/ethernet/qlogic/qede/qede_ethtool.c
index 8034d812d5a0..7e40e35d990c 100644
--- a/drivers/net/ethernet/qlogic/qede/qede_ethtool.c
+++ b/drivers/net/ethernet/qlogic/qede/qede_ethtool.c
@@ -408,7 +408,7 @@ static void qede_get_ethtool_stats(struct net_device *dev,
 	struct qede_fastpath *fp;
 	int i;
 
-	qede_fill_by_demand_stats(edev);
+	qede_fill_by_demand_stats(edev, false);
 
 	/* Need to protect the access to the fastpath array */
 	__qede_lock(edev);
diff --git a/drivers/net/ethernet/qlogic/qede/qede_main.c b/drivers/net/ethernet/qlogic/qede/qede_main.c
index 953f304b8588..6c4187e5faa5 100644
--- a/drivers/net/ethernet/qlogic/qede/qede_main.c
+++ b/drivers/net/ethernet/qlogic/qede/qede_main.c
@@ -301,12 +301,12 @@ module_exit(qede_cleanup);
 static int qede_open(struct net_device *ndev);
 static int qede_close(struct net_device *ndev);
 
-void qede_fill_by_demand_stats(struct qede_dev *edev)
+void qede_fill_by_demand_stats(struct qede_dev *edev, bool is_atomic)
 {
 	struct qede_stats_common *p_common = &edev->stats.common;
 	struct qed_eth_stats stats;
 
-	edev->ops->get_vport_stats(edev->cdev, &stats);
+	edev->ops->get_vport_stats(edev->cdev, &stats, is_atomic);
 
 	p_common->no_buff_discards = stats.common.no_buff_discards;
 	p_common->packet_too_big_discard = stats.common.packet_too_big_discard;
@@ -413,7 +413,7 @@ static void qede_get_stats64(struct net_device *dev,
 	struct qede_dev *edev = netdev_priv(dev);
 	struct qede_stats_common *p_common;
 
-	qede_fill_by_demand_stats(edev);
+	qede_fill_by_demand_stats(edev, true);
 	p_common = &edev->stats.common;
 
 	stats->rx_packets = p_common->rx_ucast_pkts + p_common->rx_mcast_pkts +
diff --git a/include/linux/qed/qed_eth_if.h b/include/linux/qed/qed_eth_if.h
index e1bf3219b4e6..f2893b6b4cb3 100644
--- a/include/linux/qed/qed_eth_if.h
+++ b/include/linux/qed/qed_eth_if.h
@@ -319,7 +319,7 @@ struct qed_eth_ops {
 				  struct eth_slow_path_rx_cqe *cqe);
 
 	void (*get_vport_stats)(struct qed_dev *cdev,
-				struct qed_eth_stats *stats);
+				struct qed_eth_stats *stats, bool is_atomic);
 
 	int (*tunn_config)(struct qed_dev *cdev,
 			   struct qed_tunn_params *params);
-- 
2.27.0

Re: [PATCH v3 net] qed/qede: Fix scheduling while atomic

[Stephen Hemminger]

On Fri, 28 Apr 2023 09:13:37 -0700
Manish Chopra <manishc@marvell.com> wrote:

> -		usleep_range(1000, 2000);
> +
> +		if (is_atomic)
> +			udelay(QED_BAR_ACQUIRE_TIMEOUT_UDELAY);
> +		else
> +			usleep_range(QED_BAR_ACQUIRE_TIMEOUT_USLEEP,
> +				     QED_BAR_ACQUIRE_TIMEOUT_USLEEP * 2);
>  	}

This is a variant of the conditional locking which is an ugly design pattern.
It makes static checking tools break and
a source of more bugs.

Better to fix the infrastructure or caller to not spin, or have two different
functions.

Re: [PATCH v3 net] qed/qede: Fix scheduling while atomic

[Jakub Kicinski]

On Fri, 28 Apr 2023 10:26:51 -0700 Stephen Hemminger wrote:
> On Fri, 28 Apr 2023 09:13:37 -0700
> Manish Chopra <manishc@marvell.com> wrote:
> 
> > -		usleep_range(1000, 2000);
> > +
> > +		if (is_atomic)
> > +			udelay(QED_BAR_ACQUIRE_TIMEOUT_UDELAY);
> > +		else
> > +			usleep_range(QED_BAR_ACQUIRE_TIMEOUT_USLEEP,
> > +				     QED_BAR_ACQUIRE_TIMEOUT_USLEEP * 2);
> >  	}  
> 
> This is a variant of the conditional locking which is an ugly design pattern.
> It makes static checking tools break and
> a source of more bugs.
> 
> Better to fix the infrastructure or caller to not spin, or have two different
> functions.

FWIW the most common way to solve this issue is using a delayed work
which reads out the stats periodically from a non-atomic context, and
return a stashed copy from get_stat64.

RE: [EXT] Re: [PATCH v3 net] qed/qede: Fix scheduling while atomic

[Manish Chopra]

Hi Jakub,

> -----Original Message-----
> From: Jakub Kicinski <kuba@kernel.org>
> Sent: Wednesday, May 3, 2023 8:33 AM
> To: Manish Chopra <manishc@marvell.com>
> Cc: Stephen Hemminger <stephen@networkplumber.org>;
> netdev@vger.kernel.org; Ariel Elior <aelior@marvell.com>; Alok Prasad
> <palok@marvell.com>; Sudarsana Reddy Kalluru <skalluru@marvell.com>;
> David S . Miller <davem@davemloft.net>
> Subject: [EXT] Re: [PATCH v3 net] qed/qede: Fix scheduling while atomic
> 
> External Email
> 
> ----------------------------------------------------------------------
> On Fri, 28 Apr 2023 10:26:51 -0700 Stephen Hemminger wrote:
> > On Fri, 28 Apr 2023 09:13:37 -0700
> > Manish Chopra <manishc@marvell.com> wrote:
> >
> > > -		usleep_range(1000, 2000);
> > > +
> > > +		if (is_atomic)
> > > +			udelay(QED_BAR_ACQUIRE_TIMEOUT_UDELAY);
> > > +		else
> > > +			usleep_range(QED_BAR_ACQUIRE_TIMEOUT_USLEEP,
> > > +				     QED_BAR_ACQUIRE_TIMEOUT_USLEEP *
> 2);
> > >  	}
> >
> > This is a variant of the conditional locking which is an ugly design pattern.
> > It makes static checking tools break and a source of more bugs.
> >
> > Better to fix the infrastructure or caller to not spin, or have two
> > different functions.
> 
> FWIW the most common way to solve this issue is using a delayed work which
> reads out the stats periodically from a non-atomic context, and return a
> stashed copy from get_stat64.

Thanks for the suggestion, I will send v4 soon with required changes.

Regards,
Manish