[v2] app/testpmd: fix segment number check
Checks
Commit Message
From: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
The --txpkts command line parameter was silently ignored due to
application was unable to check the Tx queue ring sizes for non
configured ports [1].
The "set txpkts <len0[,len1]*>" was also rejected if there
was some stopped or /unconfigured port.
This provides the following:
- If fails to get ring size from the port, this can be because port is
not initialized yet, ignore the check and just be sure segment size
won't cause an out of bound access. The port descriptor check will
be done during Tx setup.
- The capability to send single packet is supposed to be very basic
and always supported, the setting segment number to 1 is always
allowed, no check performed
- At the moment of Tx queue setup the descriptor number is checked
against configured segment number
Bugzilla ID: 584
Fixes: 8dae835d88b7 ("app/testpmd: remove restriction on Tx segments set")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
---
Cc: Andrew Boyer <aboyer@pensando.io>
v2:
* Become more flexible for the '--txpkts' command line, if not able to
get the descriptor size from port, ignore the check.
('nb_txd' check was proposed before, this will require '--txd'
parameter, but also enforces a specific order on the parameters,
instead going with the option to flex the checks for parameter.)
---
app/test-pmd/cmdline.c | 4 ++++
app/test-pmd/config.c | 32 ++++++++++++++++++++++++--------
2 files changed, 28 insertions(+), 8 deletions(-)
Comments
Hi
> -----Original Message-----
> From: Yigit, Ferruh <ferruh.yigit@intel.com>
> Sent: Saturday, April 24, 2021 00:10
> To: Viacheslav Ovsiienko <viacheslavo@nvidia.com>; Li, Xiaoyun
> <xiaoyun.li@intel.com>; Wei Hu (Xavier) <xavier.huwei@huawei.com>;
> Chengchang Tang <tangchengchang@huawei.com>
> Cc: Yigit, Ferruh <ferruh.yigit@intel.com>; dev@dpdk.org; stable@dpdk.org;
> Andrew Boyer <aboyer@pensando.io>
> Subject: [PATCH v2] app/testpmd: fix segment number check
>
> From: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
>
> The --txpkts command line parameter was silently ignored due to application
> was unable to check the Tx queue ring sizes for non configured ports [1].
Remove this [1] or mark the following items as [1] [2] [3].
>
> The "set txpkts <len0[,len1]*>" was also rejected if there was some stopped or
> /unconfigured port.
>
> This provides the following:
>
> - If fails to get ring size from the port, this can be because port is
> not initialized yet, ignore the check and just be sure segment size
> won't cause an out of bound access. The port descriptor check will
> be done during Tx setup.
>
> - The capability to send single packet is supposed to be very basic
> and always supported, the setting segment number to 1 is always
> allowed, no check performed
>
> - At the moment of Tx queue setup the descriptor number is checked
> against configured segment number
>
> Bugzilla ID: 584
> Fixes: 8dae835d88b7 ("app/testpmd: remove restriction on Tx segments set")
> Cc: stable@dpdk.org
>
> Signed-off-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
> Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
> ---
> Cc: Andrew Boyer <aboyer@pensando.io>
>
> v2:
> * Become more flexible for the '--txpkts' command line, if not able to
> get the descriptor size from port, ignore the check.
>
> ('nb_txd' check was proposed before, this will require '--txd'
> parameter, but also enforces a specific order on the parameters,
> instead going with the option to flex the checks for parameter.)
> ---
> app/test-pmd/cmdline.c | 4 ++++
> app/test-pmd/config.c | 32 ++++++++++++++++++++++++--------
> 2 files changed, 28 insertions(+), 8 deletions(-)
Except the one comment above for commit log,
Acked-by: Xiaoyun Li <xiaoyun.li@intel.com>
On 4/26/2021 12:23 PM, Li, Xiaoyun wrote:
> Hi
>
>> -----Original Message-----
>> From: Yigit, Ferruh <ferruh.yigit@intel.com>
>> Sent: Saturday, April 24, 2021 00:10
>> To: Viacheslav Ovsiienko <viacheslavo@nvidia.com>; Li, Xiaoyun
>> <xiaoyun.li@intel.com>; Wei Hu (Xavier) <xavier.huwei@huawei.com>;
>> Chengchang Tang <tangchengchang@huawei.com>
>> Cc: Yigit, Ferruh <ferruh.yigit@intel.com>; dev@dpdk.org; stable@dpdk.org;
>> Andrew Boyer <aboyer@pensando.io>
>> Subject: [PATCH v2] app/testpmd: fix segment number check
>>
>> From: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
>>
>> The --txpkts command line parameter was silently ignored due to application
>> was unable to check the Tx queue ring sizes for non configured ports [1].
>
> Remove this [1] or mark the following items as [1] [2] [3].
>
>>
>> The "set txpkts <len0[,len1]*>" was also rejected if there was some stopped or
>> /unconfigured port.
>>
>> This provides the following:
>>
>> - If fails to get ring size from the port, this can be because port is
>> not initialized yet, ignore the check and just be sure segment size
>> won't cause an out of bound access. The port descriptor check will
>> be done during Tx setup.
>>
>> - The capability to send single packet is supposed to be very basic
>> and always supported, the setting segment number to 1 is always
>> allowed, no check performed
>>
>> - At the moment of Tx queue setup the descriptor number is checked
>> against configured segment number
>>
>> Bugzilla ID: 584
>> Fixes: 8dae835d88b7 ("app/testpmd: remove restriction on Tx segments set")
>> Cc: stable@dpdk.org
>>
>> Signed-off-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
>> Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
>> ---
>> Cc: Andrew Boyer <aboyer@pensando.io>
>>
>> v2:
>> * Become more flexible for the '--txpkts' command line, if not able to
>> get the descriptor size from port, ignore the check.
>>
>> ('nb_txd' check was proposed before, this will require '--txd'
>> parameter, but also enforces a specific order on the parameters,
>> instead going with the option to flex the checks for parameter.)
>> ---
>> app/test-pmd/cmdline.c | 4 ++++
>> app/test-pmd/config.c | 32 ++++++++++++++++++++++++--------
>> 2 files changed, 28 insertions(+), 8 deletions(-)
>
> Except the one comment above for commit log,
> Acked-by: Xiaoyun Li <xiaoyun.li@intel.com>
>
Applied to dpdk-next-net/main, thanks.
Above missing reference in the commit log removed while merging.
@@ -2910,6 +2910,10 @@ cmd_setup_rxtx_queue_parsed(
if (!numa_support || socket_id == NUMA_NO_CONFIG)
socket_id = port->socket_id;
+ if (port->nb_tx_desc[res->qid] < tx_pkt_nb_segs) {
+ printf("Failed to setup TX queue: not enough descriptors\n");
+ return;
+ }
ret = rte_eth_tx_queue_setup(res->portid,
res->qid,
port->nb_tx_desc[res->qid],
@@ -3697,13 +3697,15 @@ nb_segs_is_invalid(unsigned int nb_segs)
RTE_ETH_FOREACH_DEV(port_id) {
for (queue_id = 0; queue_id < nb_txq; queue_id++) {
ret = get_tx_ring_size(port_id, queue_id, &ring_size);
-
- if (ret)
- return true;
-
+ if (ret) {
+ /* Port may not be initialized yet, can't say
+ * the port is invalid in this stage.
+ */
+ continue;
+ }
if (ring_size < nb_segs) {
- printf("nb segments per TX packets=%u >= "
- "TX queue(%u) ring_size=%u - ignored\n",
+ printf("nb segments per TX packets=%u >= TX "
+ "queue(%u) ring_size=%u - txpkts ignored\n",
nb_segs, queue_id, ring_size);
return true;
}
@@ -3719,12 +3721,26 @@ set_tx_pkt_segments(unsigned int *seg_lengths, unsigned int nb_segs)
uint16_t tx_pkt_len;
unsigned int i;
- if (nb_segs_is_invalid(nb_segs))
+ /*
+ * For single segment settings failed check is ignored.
+ * It is a very basic capability to send the single segment
+ * packets, suppose it is always supported.
+ */
+ if (nb_segs > 1 && nb_segs_is_invalid(nb_segs)) {
+ printf("Tx segment size(%u) is not supported - txpkts ignored\n",
+ nb_segs);
return;
+ }
+
+ if (nb_segs > RTE_MAX_SEGS_PER_PKT) {
+ printf("Tx segment size(%u) is bigger than max number of segment(%u)\n",
+ nb_segs, RTE_MAX_SEGS_PER_PKT);
+ return;
+ }
/*
* Check that each segment length is greater or equal than
- * the mbuf data sise.
+ * the mbuf data size.
* Check also that the total packet length is greater or equal than the
* size of an empty UDP/IP packet (sizeof(struct rte_ether_hdr) +
* 20 + 8).