DPDK patches and discussions
 help / color / mirror / Atom feed
* [PATCH] net/mana: use mana_local_data for tracking usage data for primary process
@ 2025-02-07 23:21 longli
  2025-02-08  0:21 ` Stephen Hemminger
  0 siblings, 1 reply; 3+ messages in thread
From: longli @ 2025-02-07 23:21 UTC (permalink / raw)
  To: Ferruh Yigit, Andrew Rybchenko, Wei Hu; +Cc: dev, Long Li, stable

From: Long Li <longli@microsoft.com>

The driver uses mana_shared_data for tracking usage count for primary
process. This is not correct as the mana_shared_data is allocated
by the primary and is meant to track usage of secondary process by the
primary process. And it creates a race condition when the device is
removed because the counter is no longer available if this shared
memory is being freed.

Move the usage count tracking to mana_local_data and fix the race
condition in mana_pci_remove().

Fixes: 517ed6e2d590 ("net/mana: add basic driver with build environment")
Cc: stable@dpdk.org
Signed-off-by: Long Li <longli@microsoft.com>
---
 drivers/net/mana/mana.c | 76 ++++++++++++++++++++++++++---------------
 1 file changed, 48 insertions(+), 28 deletions(-)

diff --git a/drivers/net/mana/mana.c b/drivers/net/mana/mana.c
index c37c4e3444..da4a54144f 100644
--- a/drivers/net/mana/mana.c
+++ b/drivers/net/mana/mana.c
@@ -1167,8 +1167,12 @@ mana_init_shared_data(void)
 	rte_spinlock_lock(&mana_shared_data_lock);
 
 	/* Skip if shared data is already initialized */
-	if (mana_shared_data)
+	if (mana_shared_data) {
+		DRV_LOG(INFO, "shared data is already initialized");
 		goto exit;
+	}
+
+	memset(&mana_local_data, 0, sizeof(mana_local_data));
 
 	if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
 		mana_shared_mz = rte_memzone_reserve(MZ_MANA_SHARED_DATA,
@@ -1192,7 +1196,6 @@ mana_init_shared_data(void)
 		}
 
 		mana_shared_data = secondary_mz->addr;
-		memset(&mana_local_data, 0, sizeof(mana_local_data));
 	}
 
 exit:
@@ -1213,11 +1216,11 @@ mana_init_once(void)
 	if (ret)
 		return ret;
 
-	rte_spinlock_lock(&mana_shared_data->lock);
+	rte_spinlock_lock(&mana_shared_data_lock);
 
 	switch (rte_eal_process_type()) {
 	case RTE_PROC_PRIMARY:
-		if (mana_shared_data->init_done)
+		if (mana_local_data.init_done)
 			break;
 
 		ret = mana_mp_init_primary();
@@ -1225,7 +1228,7 @@ mana_init_once(void)
 			break;
 		DRV_LOG(ERR, "MP INIT PRIMARY");
 
-		mana_shared_data->init_done = 1;
+		mana_local_data.init_done = 1;
 		break;
 
 	case RTE_PROC_SECONDARY:
@@ -1248,7 +1251,7 @@ mana_init_once(void)
 		break;
 	}
 
-	rte_spinlock_unlock(&mana_shared_data->lock);
+	rte_spinlock_unlock(&mana_shared_data_lock);
 
 	return ret;
 }
@@ -1319,11 +1322,6 @@ mana_probe_port(struct ibv_device *ibdev, struct ibv_device_attr_ex *dev_attr,
 		eth_dev->tx_pkt_burst = mana_tx_burst;
 		eth_dev->rx_pkt_burst = mana_rx_burst;
 
-		rte_spinlock_lock(&mana_shared_data->lock);
-		mana_shared_data->secondary_cnt++;
-		mana_local_data.secondary_cnt++;
-		rte_spinlock_unlock(&mana_shared_data->lock);
-
 		rte_eth_copy_pci_info(eth_dev, pci_dev);
 		rte_eth_dev_probing_finish(eth_dev);
 
@@ -1406,10 +1404,6 @@ mana_probe_port(struct ibv_device *ibdev, struct ibv_device_attr_ex *dev_attr,
 		goto failed;
 	}
 
-	rte_spinlock_lock(&mana_shared_data->lock);
-	mana_shared_data->primary_cnt++;
-	rte_spinlock_unlock(&mana_shared_data->lock);
-
 	eth_dev->device = &pci_dev->device;
 
 	DRV_LOG(INFO, "device %s at port %u", name, eth_dev->data->port_id);
@@ -1552,13 +1546,42 @@ mana_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
 		count = mana_pci_probe_mac(pci_dev, NULL);
 	}
 
+	/* If no device is found, clean up resources if this is the last one */
 	if (!count) {
-		rte_memzone_free(mana_shared_mz);
-		mana_shared_mz = NULL;
-		ret = -ENODEV;
+		rte_spinlock_lock(&mana_shared_data_lock);
+		if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
+			if (!mana_local_data.primary_cnt) {
+				mana_mp_uninit_primary();
+				rte_memzone_free(mana_shared_mz);
+				mana_shared_mz = NULL;
+				mana_shared_data = NULL;
+			}
+		} else {
+			if (!mana_local_data.secondary_cnt) {
+				mana_mp_uninit_secondary();
+				mana_shared_data = NULL;
+			}
+		}
+		rte_spinlock_unlock(&mana_shared_data_lock);
+		return -ENODEV;
 	}
 
-	return ret;
+	/* At least one eth_dev is probed, init shared data */
+	if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
+		rte_spinlock_lock(&mana_shared_data_lock);
+		mana_local_data.primary_cnt++;
+		rte_spinlock_unlock(&mana_shared_data_lock);
+	} else {
+		rte_spinlock_lock(&mana_shared_data_lock);
+		mana_local_data.secondary_cnt++;
+		rte_spinlock_unlock(&mana_shared_data_lock);
+
+		rte_spinlock_lock(&mana_shared_data->lock);
+		mana_shared_data->secondary_cnt++;
+		rte_spinlock_unlock(&mana_shared_data->lock);
+	}
+
+	return 0;
 }
 
 static int
@@ -1576,22 +1599,18 @@ mana_pci_remove(struct rte_pci_device *pci_dev)
 	if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
 		rte_spinlock_lock(&mana_shared_data_lock);
 
-		rte_spinlock_lock(&mana_shared_data->lock);
+		RTE_VERIFY(mana_local_data.primary_cnt > 0);
+		mana_local_data.primary_cnt--;
 
-		RTE_VERIFY(mana_shared_data->primary_cnt > 0);
-		mana_shared_data->primary_cnt--;
-		if (!mana_shared_data->primary_cnt) {
+		if (!mana_local_data.primary_cnt) {
 			DRV_LOG(DEBUG, "mp uninit primary");
 			mana_mp_uninit_primary();
-		}
-
-		rte_spinlock_unlock(&mana_shared_data->lock);
 
-		/* Also free the shared memory if this is the last */
-		if (!mana_shared_data->primary_cnt) {
+			/* Also free the shared memory if this is the last */
 			DRV_LOG(DEBUG, "free shared memezone data");
 			rte_memzone_free(mana_shared_mz);
 			mana_shared_mz = NULL;
+			mana_shared_data = NULL;
 		}
 
 		rte_spinlock_unlock(&mana_shared_data_lock);
@@ -1608,6 +1627,7 @@ mana_pci_remove(struct rte_pci_device *pci_dev)
 		if (!mana_local_data.secondary_cnt) {
 			DRV_LOG(DEBUG, "mp uninit secondary");
 			mana_mp_uninit_secondary();
+			mana_shared_data = NULL;
 		}
 
 		rte_spinlock_unlock(&mana_shared_data_lock);
-- 
2.34.1


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH] net/mana: use mana_local_data for tracking usage data for primary process
  2025-02-07 23:21 [PATCH] net/mana: use mana_local_data for tracking usage data for primary process longli
@ 2025-02-08  0:21 ` Stephen Hemminger
  2025-02-08  1:40   ` [EXTERNAL] " Long Li
  0 siblings, 1 reply; 3+ messages in thread
From: Stephen Hemminger @ 2025-02-08  0:21 UTC (permalink / raw)
  To: longli; +Cc: Ferruh Yigit, Andrew Rybchenko, Wei Hu, dev, Long Li, stable

On Fri,  7 Feb 2025 15:21:45 -0800
longli@linuxonhyperv.com wrote:

> +	/* At least one eth_dev is probed, init shared data */
> +	if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
> +		rte_spinlock_lock(&mana_shared_data_lock);
> +		mana_local_data.primary_cnt++;
> +		rte_spinlock_unlock(&mana_shared_data_lock);
> +	} else {
> +		rte_spinlock_lock(&mana_shared_data_lock);
> +		mana_local_data.secondary_cnt++;
> +		rte_spinlock_unlock(&mana_shared_data_lock);
> +
> +		rte_spinlock_lock(&mana_shared_data->lock);
> +		mana_shared_data->secondary_cnt++;
> +		rte_spinlock_unlock(&mana_shared_data->lock);

If all you are doing is wanting a MP safe counter, use atomic operations
instead of the overhead of a spin lock.

^ permalink raw reply	[flat|nested] 3+ messages in thread

* RE: [EXTERNAL] Re: [PATCH] net/mana: use mana_local_data for tracking usage data for primary process
  2025-02-08  0:21 ` Stephen Hemminger
@ 2025-02-08  1:40   ` Long Li
  0 siblings, 0 replies; 3+ messages in thread
From: Long Li @ 2025-02-08  1:40 UTC (permalink / raw)
  To: Stephen Hemminger, longli
  Cc: Ferruh Yigit, Andrew Rybchenko, Wei Hu, dev, stable

> On Fri,  7 Feb 2025 15:21:45 -0800
> longli@linuxonhyperv.com wrote:
> 
> > +	/* At least one eth_dev is probed, init shared data */
> > +	if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
> > +		rte_spinlock_lock(&mana_shared_data_lock);
> > +		mana_local_data.primary_cnt++;
> > +		rte_spinlock_unlock(&mana_shared_data_lock);
> > +	} else {
> > +		rte_spinlock_lock(&mana_shared_data_lock);
> > +		mana_local_data.secondary_cnt++;
> > +		rte_spinlock_unlock(&mana_shared_data_lock);
> > +
> > +		rte_spinlock_lock(&mana_shared_data->lock);
> > +		mana_shared_data->secondary_cnt++;
> > +		rte_spinlock_unlock(&mana_shared_data->lock);
> 
> If all you are doing is wanting a MP safe counter, use atomic operations instead of
> the overhead of a spin lock.

mana_shared_data_lock is also used to protect init_done, mana_shared_mz and mana_shared_data.  That's why I'm reusing it for primary_cnt and secondary_cnt as those values are initialized in locked context at the beginning for the local process. I think this will make the code look clean. 

I can remove mana_shared_data->lock and use atomic for mana_shared_data->secondary_cnt.

Will send v2.

Thanks,

Long

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2025-02-08  1:40 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2025-02-07 23:21 [PATCH] net/mana: use mana_local_data for tracking usage data for primary process longli
2025-02-08  0:21 ` Stephen Hemminger
2025-02-08  1:40   ` [EXTERNAL] " Long Li

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).