* [dpdk-dev] [PATCH] net/i40e: fix mbuf fast free wrong check
@ 2021-04-07 6:37 Murphy Yang
2021-04-08 8:57 ` Zhang, Qi Z
0 siblings, 1 reply; 2+ messages in thread
From: Murphy Yang @ 2021-04-07 6:37 UTC (permalink / raw)
To: dev
Cc: qiming.yang, jia.guo, beilei.xing, qi.z.zhang, stevex.yang,
robinx.zhang, Murphy Yang
If no offload flags are set, then 0 == 0 & <anything> is true.
This means that if DEV_TX_OFFLOAD_MBUF_FAST_FREE is not
set (and no other flags are set) then this check will still
enable `tx_simple_allowed`.
This patch adds the 'offloads' flag not equal zero to avoid this
scenario.
Bugzilla ID: 665
Fixes: 399421100e08 ("net/i40e: fix missing mbuf fast free offload")
Signed-off-by: Murphy Yang <murphyx.yang@intel.com>
---
drivers/net/i40e/i40e_rxtx.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/i40e/i40e_rxtx.c b/drivers/net/i40e/i40e_rxtx.c
index b3d7765e3b..1450b228ca 100644
--- a/drivers/net/i40e/i40e_rxtx.c
+++ b/drivers/net/i40e/i40e_rxtx.c
@@ -3345,7 +3345,7 @@ i40e_set_tx_function_flag(struct rte_eth_dev *dev, struct i40e_tx_queue *txq)
/* Use a simple Tx queue if possible (only fast free is allowed) */
ad->tx_simple_allowed =
- (txq->offloads ==
+ (txq->offloads != 0 && txq->offloads ==
(txq->offloads & DEV_TX_OFFLOAD_MBUF_FAST_FREE) &&
txq->tx_rs_thresh >= RTE_PMD_I40E_TX_MAX_BURST);
ad->tx_vec_allowed = (ad->tx_simple_allowed &&
--
2.17.1
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [dpdk-dev] [PATCH] net/i40e: fix mbuf fast free wrong check
2021-04-07 6:37 [dpdk-dev] [PATCH] net/i40e: fix mbuf fast free wrong check Murphy Yang
@ 2021-04-08 8:57 ` Zhang, Qi Z
0 siblings, 0 replies; 2+ messages in thread
From: Zhang, Qi Z @ 2021-04-08 8:57 UTC (permalink / raw)
To: Yang, MurphyX, dev
Cc: Yang, Qiming, Guo, Jia, Xing, Beilei, Yang, SteveX, Zhang, RobinX
> -----Original Message-----
> From: Yang, MurphyX <murphyx.yang@intel.com>
> Sent: Wednesday, April 7, 2021 2:37 PM
> To: dev@dpdk.org
> Cc: Yang, Qiming <qiming.yang@intel.com>; Guo, Jia <jia.guo@intel.com>;
> Xing, Beilei <beilei.xing@intel.com>; Zhang, Qi Z <qi.z.zhang@intel.com>; Yang,
> SteveX <stevex.yang@intel.com>; Zhang, RobinX <robinx.zhang@intel.com>;
> Yang, MurphyX <murphyx.yang@intel.com>
> Subject: [PATCH] net/i40e: fix mbuf fast free wrong check
>
> If no offload flags are set, then 0 == 0 & <anything> is true.
> This means that if DEV_TX_OFFLOAD_MBUF_FAST_FREE is not set (and no
> other flags are set) then this check will still enable `tx_simple_allowed`.
>
> This patch adds the 'offloads' flag not equal zero to avoid this scenario.
>
> Bugzilla ID: 665
> Fixes: 399421100e08 ("net/i40e: fix missing mbuf fast free offload")
>
> Signed-off-by: Murphy Yang <murphyx.yang@intel.com>
> ---
> drivers/net/i40e/i40e_rxtx.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/net/i40e/i40e_rxtx.c b/drivers/net/i40e/i40e_rxtx.c index
> b3d7765e3b..1450b228ca 100644
> --- a/drivers/net/i40e/i40e_rxtx.c
> +++ b/drivers/net/i40e/i40e_rxtx.c
> @@ -3345,7 +3345,7 @@ i40e_set_tx_function_flag(struct rte_eth_dev *dev,
> struct i40e_tx_queue *txq)
>
> /* Use a simple Tx queue if possible (only fast free is allowed) */
> ad->tx_simple_allowed =
> - (txq->offloads ==
> + (txq->offloads != 0 && txq->offloads ==
> (txq->offloads & DEV_TX_OFFLOAD_MBUF_FAST_FREE) &&
I don't think this is correct.
Why simple_tx can't be selected if no DEV_TX_OFFLOAD_MBUF_FAST_FREE is required?
Actually, the original code means, the max Tx offload that simple_tx support is DEV_TX_OFFLOAD_MBUF_FAST_FREE, you can't have more than it, but cannot have it.
> txq->tx_rs_thresh >= RTE_PMD_I40E_TX_MAX_BURST);
> ad->tx_vec_allowed = (ad->tx_simple_allowed &&
> --
> 2.17.1
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2021-04-08 8:57 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-04-07 6:37 [dpdk-dev] [PATCH] net/i40e: fix mbuf fast free wrong check Murphy Yang
2021-04-08 8:57 ` Zhang, Qi Z
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).