When unloading the MANA driver, mana_dealloc_queues() waits for the MANA
hardware to complete any inflight packets and set the pending send count
to zero. But if the hardware has failed, mana_dealloc_queues()
could wait forever.
Fix this by adding a timeout to the wait. Set the timeout to 120 seconds,
which is a somewhat arbitrary value that is more than long enough for
functional hardware to complete any sends.
Cc: [email protected]
Fixes: ca9c54d2d6a5 ("net: mana: Add a driver for Microsoft Azure Network Adapter (MANA)")
Signed-off-by: Souradeep Chakrabarti <[email protected]>
---
V6 -> V7:
* Optimized the while loop for freeing skb.
V5 -> V6:
* Added pcie_flr to reset the pci after timeout.
* Fixed the position of changelog.
* Removed unused variable like cq.
V4 -> V5:
* Added fixes tag
* Changed the usleep_range from static to incremental value.
* Initialized timeout in the begining.
V3 -> V4:
* Removed the unnecessary braces from mana_dealloc_queues().
V2 -> V3:
* Removed the unnecessary braces from mana_dealloc_queues().
V1 -> V2:
* Added net branch
* Removed the typecasting to (struct mana_context*) of void pointer
* Repositioned timeout variable in mana_dealloc_queues()
* Repositioned vf_unload_timeout in mana_context struct, to utilise the
6 bytes hole
---
drivers/net/ethernet/microsoft/mana/mana_en.c | 37 +++++++++++++++++--
1 file changed, 33 insertions(+), 4 deletions(-)
diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c
index a499e460594b..3c5552a176d0 100644
--- a/drivers/net/ethernet/microsoft/mana/mana_en.c
+++ b/drivers/net/ethernet/microsoft/mana/mana_en.c
@@ -8,6 +8,7 @@
#include <linux/ethtool.h>
#include <linux/filter.h>
#include <linux/mm.h>
+#include <linux/pci.h>
#include <net/checksum.h>
#include <net/ip6_checksum.h>
@@ -2345,9 +2346,12 @@ int mana_attach(struct net_device *ndev)
static int mana_dealloc_queues(struct net_device *ndev)
{
struct mana_port_context *apc = netdev_priv(ndev);
+ unsigned long timeout = jiffies + 120 * HZ;
struct gdma_dev *gd = apc->ac->gdma_dev;
struct mana_txq *txq;
+ struct sk_buff *skb;
int i, err;
+ u32 tsleep;
if (apc->port_is_up)
return -EINVAL;
@@ -2363,15 +2367,40 @@ static int mana_dealloc_queues(struct net_device *ndev)
* to false, but it doesn't matter since mana_start_xmit() drops any
* new packets due to apc->port_is_up being false.
*
- * Drain all the in-flight TX packets
+ * Drain all the in-flight TX packets.
+ * A timeout of 120 seconds for all the queues is used.
+ * This will break the while loop when h/w is not responding.
+ * This value of 120 has been decided here considering max
+ * number of queues.
*/
+
for (i = 0; i < apc->num_queues; i++) {
txq = &apc->tx_qp[i].txq;
-
- while (atomic_read(&txq->pending_sends) > 0)
- usleep_range(1000, 2000);
+ tsleep = 1000;
+ while (atomic_read(&txq->pending_sends) > 0 &&
+ time_before(jiffies, timeout)) {
+ usleep_range(tsleep, tsleep + 1000);
+ tsleep <<= 1;
+ }
+ if (atomic_read(&txq->pending_sends)) {
+ err = pcie_flr(to_pci_dev(gd->gdma_context->dev));
+ if (err) {
+ netdev_err(ndev, "flr failed %d with %d pkts pending in txq %u\n",
+ err, atomic_read(&txq->pending_sends),
+ txq->gdma_txq_id);
+ }
+ break;
+ }
}
+ for (i = 0; i < apc->num_queues; i++) {
+ txq = &apc->tx_qp[i].txq;
+ while (skb = skb_dequeue(&txq->pending_skbs)) {
+ mana_unmap_skb(skb, apc);
+ dev_consume_skb_any(skb);
+ }
+ atomic_set(&txq->pending_sends, 0);
+ }
/* We're 100% sure the queues can no longer be woken up, because
* we're sure now mana_poll_tx_cq() can't be running.
*/
--
2.34.1
On Tue, Aug 01, 2023 at 05:29:13AM -0700, Souradeep Chakrabarti wrote:
...
Hi Souradeep,
> + for (i = 0; i < apc->num_queues; i++) {
> + txq = &apc->tx_qp[i].txq;
> + while (skb = skb_dequeue(&txq->pending_skbs)) {
W=1 builds with both clang-16 and gcc-12 complain that
they would like an extra set of parentheses around
an assignment used as a truth value.
> + mana_unmap_skb(skb, apc);
> + dev_consume_skb_any(skb);
> + }
> + atomic_set(&txq->pending_sends, 0);
> + }
> /* We're 100% sure the queues can no longer be woken up, because
> * we're sure now mana_poll_tx_cq() can't be running.
> */
> --
> 2.34.1
>
>
On 8/1/2023 5:29 AM, Souradeep Chakrabarti wrote:
> When unloading the MANA driver, mana_dealloc_queues() waits for the MANA
> hardware to complete any inflight packets and set the pending send count
> to zero. But if the hardware has failed, mana_dealloc_queues()
> could wait forever.
>
> Fix this by adding a timeout to the wait. Set the timeout to 120 seconds,
tx timeout in stack defaults to 5 seconds, do you not have that on? What
happens when you start getting resets while unloading?
> which is a somewhat arbitrary value that is more than long enough for
> functional hardware to complete any sends.
I'd say 2 or 5 seconds is probably plenty of time to hang up a driver
unload.
>
> Cc: [email protected]
> Fixes: ca9c54d2d6a5 ("net: mana: Add a driver for Microsoft Azure Network Adapter (MANA)")
>
> Signed-off-by: Souradeep Chakrabarti <[email protected]>
keep s-o-b and other trailers together please, no spaces, it messes up
git and doesn't conform to kernel standards.
> ---
> V6 -> V7:
> * Optimized the while loop for freeing skb.
>
> V5 -> V6:
> * Added pcie_flr to reset the pci after timeout.
> * Fixed the position of changelog.
> * Removed unused variable like cq.
>
> V4 -> V5:
> * Added fixes tag
> * Changed the usleep_range from static to incremental value.
> * Initialized timeout in the begining.
>
> V3 -> V4:
> * Removed the unnecessary braces from mana_dealloc_queues().
>
> V2 -> V3:
> * Removed the unnecessary braces from mana_dealloc_queues().
>
> V1 -> V2:
> * Added net branch
> * Removed the typecasting to (struct mana_context*) of void pointer
> * Repositioned timeout variable in mana_dealloc_queues()
> * Repositioned vf_unload_timeout in mana_context struct, to utilise the
> 6 bytes hole
> ---
> drivers/net/ethernet/microsoft/mana/mana_en.c | 37 +++++++++++++++++--
> 1 file changed, 33 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c
> index a499e460594b..3c5552a176d0 100644
> --- a/drivers/net/ethernet/microsoft/mana/mana_en.c
> +++ b/drivers/net/ethernet/microsoft/mana/mana_en.c
> @@ -8,6 +8,7 @@
> #include <linux/ethtool.h>
> #include <linux/filter.h>
> #include <linux/mm.h>
> +#include <linux/pci.h>
>
> #include <net/checksum.h>
> #include <net/ip6_checksum.h>
> @@ -2345,9 +2346,12 @@ int mana_attach(struct net_device *ndev)
> static int mana_dealloc_queues(struct net_device *ndev)
> {
> struct mana_port_context *apc = netdev_priv(ndev);
> + unsigned long timeout = jiffies + 120 * HZ;
> struct gdma_dev *gd = apc->ac->gdma_dev;
> struct mana_txq *txq;
> + struct sk_buff *skb;
> int i, err;
> + u32 tsleep;
>
> if (apc->port_is_up)
> return -EINVAL;
> @@ -2363,15 +2367,40 @@ static int mana_dealloc_queues(struct net_device *ndev)
> * to false, but it doesn't matter since mana_start_xmit() drops any
> * new packets due to apc->port_is_up being false.
> *
> - * Drain all the in-flight TX packets
> + * Drain all the in-flight TX packets.
> + * A timeout of 120 seconds for all the queues is used.
> + * This will break the while loop when h/w is not responding.
> + * This value of 120 has been decided here considering max
> + * number of queues.
> */
> +
> for (i = 0; i < apc->num_queues; i++) {
> txq = &apc->tx_qp[i].txq;
> -
> - while (atomic_read(&txq->pending_sends) > 0)
> - usleep_range(1000, 2000);
> + tsleep = 1000;
> + while (atomic_read(&txq->pending_sends) > 0 &&
> + time_before(jiffies, timeout)) {
> + usleep_range(tsleep, tsleep + 1000);
> + tsleep <<= 1;
> + }
> + if (atomic_read(&txq->pending_sends)) {
> + err = pcie_flr(to_pci_dev(gd->gdma_context->dev));
> + if (err) {
> + netdev_err(ndev, "flr failed %d with %d pkts pending in txq %u\n",
> + err, atomic_read(&txq->pending_sends),
> + txq->gdma_txq_id);
> + }
> + break;
> + }
> }
>
> + for (i = 0; i < apc->num_queues; i++) {
> + txq = &apc->tx_qp[i].txq;
> + while (skb = skb_dequeue(&txq->pending_skbs)) {
> + mana_unmap_skb(skb, apc);
> + dev_consume_skb_any(skb);
dev_kfree_skb_any() would be more correct here since this is an error
path and the transmit is presumed dropped, isn't it?
> + }
> + atomic_set(&txq->pending_sends, 0);
> + }
> /* We're 100% sure the queues can no longer be woken up, because
> * we're sure now mana_poll_tx_cq() can't be running.
> */
>-----Original Message-----
>From: Jesse Brandeburg <[email protected]>
>Sent: Tuesday, August 1, 2023 11:34 PM
>To: Souradeep Chakrabarti <[email protected]>; KY Srinivasan
><[email protected]>; Haiyang Zhang <[email protected]>;
>[email protected]; Dexuan Cui <[email protected]>;
>[email protected]; [email protected]; [email protected];
>[email protected]; Long Li <[email protected]>; Ajay Sharma
><[email protected]>; [email protected]; [email protected];
>[email protected]; vkuznets <[email protected]>;
>[email protected]; [email protected]; [email protected];
>[email protected]; [email protected]
>Cc: Souradeep Chakrabarti <[email protected]>;
>[email protected]
>Subject: [EXTERNAL] Re: [PATCH V7 net] net: mana: Fix MANA VF unload when
>hardware is
>
>On 8/1/2023 5:29 AM, Souradeep Chakrabarti wrote:
>> When unloading the MANA driver, mana_dealloc_queues() waits for the
>> MANA hardware to complete any inflight packets and set the pending
>> send count to zero. But if the hardware has failed,
>> mana_dealloc_queues() could wait forever.
>>
>> Fix this by adding a timeout to the wait. Set the timeout to 120
>> seconds,
>
>tx timeout in stack defaults to 5 seconds, do you not have that on? What
>happens when you start getting resets while unloading?
>
>> which is a somewhat arbitrary value that is more than long enough for
>> functional hardware to complete any sends.
>
>I'd say 2 or 5 seconds is probably plenty of time to hang up a driver unload.
>
Thank you for the comment. This was already discussed in V4.
I am just sharing the summary here:
This waiting time is usually much shorter than 120 sec.
The long wait only happens in rare and unexpected NIC HW non-responding cases.
At that point, we don't actually care if the pending packets are sent or not.
But if we free the queues too soon, and the HW is slow for unexpected reasons,
a delayed completion notice will DMA into the freed memory and cause corruption.
That's why we have a longer waiting time.
>>
>> Cc: [email protected]
>> Fixes: ca9c54d2d6a5 ("net: mana: Add a driver for Microsoft Azure
>> Network Adapter (MANA)")
>>
>> Signed-off-by: Souradeep Chakrabarti
>> <[email protected]>
>
>keep s-o-b and other trailers together please, no spaces, it messes up git and
>doesn't conform to kernel standards.
>
>
>> ---
>> V6 -> V7:
>> * Optimized the while loop for freeing skb.
>>
>> V5 -> V6:
>> * Added pcie_flr to reset the pci after timeout.
>> * Fixed the position of changelog.
>> * Removed unused variable like cq.
>>
>> V4 -> V5:
>> * Added fixes tag
>> * Changed the usleep_range from static to incremental value.
>> * Initialized timeout in the begining.
>>
>> V3 -> V4:
>> * Removed the unnecessary braces from mana_dealloc_queues().
>>
>> V2 -> V3:
>> * Removed the unnecessary braces from mana_dealloc_queues().
>>
>> V1 -> V2:
>> * Added net branch
>> * Removed the typecasting to (struct mana_context*) of void pointer
>> * Repositioned timeout variable in mana_dealloc_queues()
>> * Repositioned vf_unload_timeout in mana_context struct, to utilise
>> the
>> 6 bytes hole
>> ---
>> drivers/net/ethernet/microsoft/mana/mana_en.c | 37
>> +++++++++++++++++--
>> 1 file changed, 33 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c
>> b/drivers/net/ethernet/microsoft/mana/mana_en.c
>> index a499e460594b..3c5552a176d0 100644
>> --- a/drivers/net/ethernet/microsoft/mana/mana_en.c
>> +++ b/drivers/net/ethernet/microsoft/mana/mana_en.c
>> @@ -8,6 +8,7 @@
>> #include <linux/ethtool.h>
>> #include <linux/filter.h>
>> #include <linux/mm.h>
>> +#include <linux/pci.h>
>>
>> #include <net/checksum.h>
>> #include <net/ip6_checksum.h>
>> @@ -2345,9 +2346,12 @@ int mana_attach(struct net_device *ndev)
>> static int mana_dealloc_queues(struct net_device *ndev) {
>> struct mana_port_context *apc = netdev_priv(ndev);
>> + unsigned long timeout = jiffies + 120 * HZ;
>> struct gdma_dev *gd = apc->ac->gdma_dev;
>> struct mana_txq *txq;
>> + struct sk_buff *skb;
>> int i, err;
>> + u32 tsleep;
>>
>> if (apc->port_is_up)
>> return -EINVAL;
>> @@ -2363,15 +2367,40 @@ static int mana_dealloc_queues(struct
>net_device *ndev)
>> * to false, but it doesn't matter since mana_start_xmit() drops any
>> * new packets due to apc->port_is_up being false.
>> *
>> - * Drain all the in-flight TX packets
>> + * Drain all the in-flight TX packets.
>> + * A timeout of 120 seconds for all the queues is used.
>> + * This will break the while loop when h/w is not responding.
>> + * This value of 120 has been decided here considering max
>> + * number of queues.
>> */
>> +
>> for (i = 0; i < apc->num_queues; i++) {
>> txq = &apc->tx_qp[i].txq;
>> -
>> - while (atomic_read(&txq->pending_sends) > 0)
>> - usleep_range(1000, 2000);
>> + tsleep = 1000;
>> + while (atomic_read(&txq->pending_sends) > 0 &&
>> + time_before(jiffies, timeout)) {
>> + usleep_range(tsleep, tsleep + 1000);
>> + tsleep <<= 1;
>> + }
>> + if (atomic_read(&txq->pending_sends)) {
>> + err = pcie_flr(to_pci_dev(gd->gdma_context->dev));
>> + if (err) {
>> + netdev_err(ndev, "flr failed %d with %d pkts
>pending in txq %u\n",
>> + err, atomic_read(&txq-
>>pending_sends),
>> + txq->gdma_txq_id);
>> + }
>> + break;
>> + }
>> }
>>
>> + for (i = 0; i < apc->num_queues; i++) {
>> + txq = &apc->tx_qp[i].txq;
>> + while (skb = skb_dequeue(&txq->pending_skbs)) {
>> + mana_unmap_skb(skb, apc);
>> + dev_consume_skb_any(skb);
>
>dev_kfree_skb_any() would be more correct here since this is an error path and
>the transmit is presumed dropped, isn't it?
Yes, dev_kfree_skb_any() will be a better approach in this scenario. Will change it in next version.
>
>
>> + }
>> + atomic_set(&txq->pending_sends, 0);
>> + }
>> /* We're 100% sure the queues can no longer be woken up, because
>> * we're sure now mana_poll_tx_cq() can't be running.
>> */
>-----Original Message-----
>From: Simon Horman <[email protected]>
>Sent: Tuesday, August 1, 2023 9:01 PM
>To: Souradeep Chakrabarti <[email protected]>
>Cc: KY Srinivasan <[email protected]>; Haiyang Zhang
><[email protected]>; [email protected]; Dexuan Cui
><[email protected]>; [email protected]; [email protected];
>[email protected]; [email protected]; Long Li <[email protected]>; Ajay
>Sharma <[email protected]>; [email protected];
>[email protected]; [email protected]; vkuznets
><[email protected]>; [email protected]; [email protected];
>[email protected]; [email protected]; linux-
>[email protected]; Souradeep Chakrabarti
><[email protected]>; [email protected]
>Subject: [EXTERNAL] Re: [PATCH V7 net] net: mana: Fix MANA VF unload when
>hardware is
>
>On Tue, Aug 01, 2023 at 05:29:13AM -0700, Souradeep Chakrabarti wrote:
>
>...
>
>Hi Souradeep,
>
>
>> + for (i = 0; i < apc->num_queues; i++) {
>> + txq = &apc->tx_qp[i].txq;
>> + while (skb = skb_dequeue(&txq->pending_skbs)) {
>
>W=1 builds with both clang-16 and gcc-12 complain that they would like an
>extra set of parentheses around an assignment used as a truth value.
Thanks for letting me know. I will fix it in next version.
>
>> + mana_unmap_skb(skb, apc);
>> + dev_consume_skb_any(skb);
>> + }
>> + atomic_set(&txq->pending_sends, 0);
>> + }
>> /* We're 100% sure the queues can no longer be woken up, because
>> * we're sure now mana_poll_tx_cq() can't be running.
>> */
>> --
>> 2.34.1
>>
>>
Hi Souradeep,
kernel test robot noticed the following build warnings:
[auto build test WARNING on net/main]
url: https://github.com/intel-lab-lkp/linux/commits/Souradeep-Chakrabarti/net-mana-Fix-MANA-VF-unload-when-hardware-is/20230801-203141
base: net/main
patch link: https://lore.kernel.org/r/1690892953-25201-1-git-send-email-schakrabarti%40linux.microsoft.com
patch subject: [PATCH V7 net] net: mana: Fix MANA VF unload when hardware is
config: x86_64-allyesconfig (https://download.01.org/0day-ci/archive/20230802/[email protected]/config)
compiler: gcc-12 (Debian 12.2.0-14) 12.2.0
reproduce: (https://download.01.org/0day-ci/archive/20230802/[email protected]/reproduce)
If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/
All warnings (new ones prefixed by >>):
drivers/net/ethernet/microsoft/mana/mana_en.c: In function 'mana_dealloc_queues':
>> drivers/net/ethernet/microsoft/mana/mana_en.c:2398:24: warning: suggest parentheses around assignment used as truth value [-Wparentheses]
2398 | while (skb = skb_dequeue(&txq->pending_skbs)) {
| ^~~
vim +2398 drivers/net/ethernet/microsoft/mana/mana_en.c
2345
2346 static int mana_dealloc_queues(struct net_device *ndev)
2347 {
2348 struct mana_port_context *apc = netdev_priv(ndev);
2349 unsigned long timeout = jiffies + 120 * HZ;
2350 struct gdma_dev *gd = apc->ac->gdma_dev;
2351 struct mana_txq *txq;
2352 struct sk_buff *skb;
2353 int i, err;
2354 u32 tsleep;
2355
2356 if (apc->port_is_up)
2357 return -EINVAL;
2358
2359 mana_chn_setxdp(apc, NULL);
2360
2361 if (gd->gdma_context->is_pf)
2362 mana_pf_deregister_filter(apc);
2363
2364 /* No packet can be transmitted now since apc->port_is_up is false.
2365 * There is still a tiny chance that mana_poll_tx_cq() can re-enable
2366 * a txq because it may not timely see apc->port_is_up being cleared
2367 * to false, but it doesn't matter since mana_start_xmit() drops any
2368 * new packets due to apc->port_is_up being false.
2369 *
2370 * Drain all the in-flight TX packets.
2371 * A timeout of 120 seconds for all the queues is used.
2372 * This will break the while loop when h/w is not responding.
2373 * This value of 120 has been decided here considering max
2374 * number of queues.
2375 */
2376
2377 for (i = 0; i < apc->num_queues; i++) {
2378 txq = &apc->tx_qp[i].txq;
2379 tsleep = 1000;
2380 while (atomic_read(&txq->pending_sends) > 0 &&
2381 time_before(jiffies, timeout)) {
2382 usleep_range(tsleep, tsleep + 1000);
2383 tsleep <<= 1;
2384 }
2385 if (atomic_read(&txq->pending_sends)) {
2386 err = pcie_flr(to_pci_dev(gd->gdma_context->dev));
2387 if (err) {
2388 netdev_err(ndev, "flr failed %d with %d pkts pending in txq %u\n",
2389 err, atomic_read(&txq->pending_sends),
2390 txq->gdma_txq_id);
2391 }
2392 break;
2393 }
2394 }
2395
2396 for (i = 0; i < apc->num_queues; i++) {
2397 txq = &apc->tx_qp[i].txq;
> 2398 while (skb = skb_dequeue(&txq->pending_skbs)) {
2399 mana_unmap_skb(skb, apc);
2400 dev_consume_skb_any(skb);
2401 }
2402 atomic_set(&txq->pending_sends, 0);
2403 }
2404 /* We're 100% sure the queues can no longer be woken up, because
2405 * we're sure now mana_poll_tx_cq() can't be running.
2406 */
2407
2408 apc->rss_state = TRI_STATE_FALSE;
2409 err = mana_config_rss(apc, TRI_STATE_FALSE, false, false);
2410 if (err) {
2411 netdev_err(ndev, "Failed to disable vPort: %d\n", err);
2412 return err;
2413 }
2414
2415 mana_destroy_vport(apc);
2416
2417 return 0;
2418 }
2419
--
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki