<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=Windows-1252">
<style type="text/css" style="display:none;"> P {margin-top:0;margin-bottom:0;} </style>
</head>
<body dir="ltr">
<div class="elementToProof" style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
Hi Yangming, </div>
<div class="elementToProof" style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
<br>
</div>
<div class="elementToProof" style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
How did you configure the queue pairs differently between the primary and secondary processes?</div>
<div class="elementToProof" style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
<br>
</div>
<div class="elementToProof" style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
The application must call rte_cryptodev_queue_pair_setup() with a unique qp_id for each process. This implies that each process should receive distinct queue pair configurations at runtime. From what I can tell, it’s likely that the secondary process is still
using the same queue pair as the primary process.</div>
<div class="elementToProof" style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
<br>
</div>
<div class="elementToProof" style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
Regards</div>
<div class="elementToProof" style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
<br>
</div>
<div class="elementToProof" style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
Kai </div>
<div class="elementToProof" style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
<br>
</div>
<div class="elementToProof" style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
<br>
</div>
<div id="appendonsend" style="color: inherit;"></div>
<div><br>
</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
<br>
</div>
<hr style="display: inline-block; width: 98%;">
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
<b>From:</b> Moses Young <mosesyyoung@gmail.com><br>
<b>Sent:</b> Monday, May 12, 2025 11:10<br>
<b>To:</b> Ji, Kai <kai.ji@intel.com>; Yang Ming <ming.1.yang@nokia-sbell.com>; dev@dpdk.org <dev@dpdk.org><br>
<b>Subject:</b> Re: [PATCH v2 2/2] crypto/ipsec_mb: fix QP release in secondary </div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
<br>
</div>
<p style="margin-top: 0px; margin-bottom: 0px;">On 5/7/2025 11:25 PM, Ji, Kai wrote:</p>
<blockquote>
<div style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
Hi Yangming, </div>
<div style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
<br>
</div>
<div style="line-height: 19px; font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
PID check is implemented here: </div>
<div style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
<a href="https://github.com/DPDK/dpdk/blob/75f179ebe347b6098cf3af26d3d3b7168fe3fe24/drivers/crypto/ipsec_mb/ipsec_mb_ops.c#L376" id="OWAce7322dc-19fb-9a60-b9da-0cc56219bc49" class="x_OWAAutoLink x_moz-txt-link-freetext" data-auth="NotApplicable">https://github.com/DPDK/dpdk/blob/75f179ebe347b6098cf3af26d3d3b7168fe3fe24/drivers/crypto/ipsec_mb/ipsec_mb_ops.c#L376</a></div>
<div style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
<br>
</div>
<div style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
Can you share the steps to re-produce the error ?</div>
<div style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
<br>
</div>
<div style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
Regards</div>
<div style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
<br>
</div>
<div style="font-family: "IntelOne Text"; font-size: 10pt; color: rgb(0, 0, 0);">
Kai </div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
<br>
</div>
<hr style="display: inline-block; width: 98%;">
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
<b>From:</b> Yang Ming<br>
<b>Sent:</b> Thursday, April 24, 2025 15:26<br>
<b>To:</b> <a href="mailto:dev@dpdk.org" id="OWAb157fb33-c271-4671-52d4-55ae7fc02603" class="x_moz-txt-link-abbreviated OWAAutoLink">dev@dpdk.org</a><br>
<b>Subject:</b> Re: [PATCH v2 2/2] crypto/ipsec_mb: fix QP release in secondary</div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
<br>
</div>
<div style="font-size: 11pt;">Hi,<br>
<br>
On 2025/4/7 13:25, Yang Ming wrote:<br>
> From: myang <a href="mailto:ming.1.yang@nokia-sbell.com" id="OWA3f9efb88-ef9c-2a54-dceb-f658b1698fba" class="x_moz-txt-link-rfc2396E OWAAutoLink">
<ming.1.yang@nokia-sbell.com></a><br>
><br>
> When a secondary process tries to release a queue pair (QP) that<br>
> does not belong to it, error logs occur:<br>
> CRYPTODEV: ipsec_mb_ipc_request() line 373: Unable to release<br>
> qp_id=0<br>
> EAL: Message data is too long<br>
> EAL: Fail to handle message: ipsec_mb_mp_msg<br>
> EAL: Fail to recv reply for request /tmp/dpdk/l2hi/mp_socket:<br>
> ipsec_mb_mp_msg<br>
><br>
> This patch ensures that a secondary process only frees a QP if<br>
> it actually owns it, preventing conflicts and resolving the<br>
> issue.<br>
><br>
> Signed-off-by: myang <a href="mailto:ming.1.yang@nokia-sbell.com" id="OWAb2264f81-2e4b-4c96-ab14-a31b1a26ffdf" class="x_moz-txt-link-rfc2396E OWAAutoLink">
<ming.1.yang@nokia-sbell.com></a><br>
> ---<br>
> drivers/crypto/ipsec_mb/ipsec_mb_ops.c | 7 +++++--<br>
> 1 file changed, 5 insertions(+), 2 deletions(-)<br>
><br>
> diff --git a/drivers/crypto/ipsec_mb/ipsec_mb_ops.c b/drivers/crypto/ipsec_mb/ipsec_mb_ops.c<br>
> index 910efb1a97..50ee140ccd 100644<br>
> --- a/drivers/crypto/ipsec_mb/ipsec_mb_ops.c<br>
> +++ b/drivers/crypto/ipsec_mb/ipsec_mb_ops.c<br>
> @@ -138,6 +138,7 @@ int<br>
> ipsec_mb_qp_release(struct rte_cryptodev *dev, uint16_t qp_id)<br>
> {<br>
> struct ipsec_mb_qp *qp = dev->data->queue_pairs[qp_id];<br>
> + uint16_t process_id = (uint16_t)getpid();<br>
> <br>
> if (!qp)<br>
> return 0;<br>
> @@ -152,8 +153,10 @@ ipsec_mb_qp_release(struct rte_cryptodev *dev, uint16_t qp_id)<br>
> rte_free(qp);<br>
> dev->data->queue_pairs[qp_id] = NULL;<br>
> } else { /* secondary process */<br>
> - return ipsec_mb_secondary_qp_op(dev->data->dev_id, qp_id,<br>
> - NULL, 0, RTE_IPSEC_MB_MP_REQ_QP_FREE);<br>
> + if (qp->qp_used_by_pid == process_id)<br>
> + return ipsec_mb_secondary_qp_op(dev->data->dev_id,<br>
> + qp_id, NULL, 0,<br>
> + RTE_IPSEC_MB_MP_REQ_QP_FREE);<br>
> }<br>
> return 0;<br>
> }<br>
<br>
Hi Experts,<br>
<br>
Is there any chance to review and accept this patch?<br>
<br>
Brs,<br>
<br>
Yang Ming<br>
<br>
</div>
</blockquote>
<div>Hi,<br>
<br>
David. Thanks for your feedback. I will Cc maintainers in next version.<br>
<br>
Kai, Thanks for your feedback. Here's our test steps after applying the previous patch called "eal: prevent socket closure before MP sync" in this serious:<br>
1. Start the primary process: Run the DPDK primary process with the IPsec MB crypto device.<br>
2. Launch the secondary process: Start a DPDK secondary process using the same device parameters.<br>
3. Exit the secondary normally.<br>
<br>
On secondary exit, error logs show below as my commit log's description:<br>
CRYPTODEV: ipsec_mb_ipc_request() line 373: Unable to release qp_id=0<br>
EAL: Message data is too long<br>
EAL: Fail to handle message: ipsec_mb_mp_msg<br>
EAL: Fail to recv reply for request /tmp/dpdk/l2hi/mp_socket: ipsec_mb_mp_msg<br>
<br>
This message corresponds exactly to the PID check in the code you referenced:<br>
if (qp->qp_used_by_pid != req_param->process_id) {<br>
CDEV_LOG_ERR("Unable to release qp_id=%d", qp_id);<br>
goto out;<br>
}<br>
<br>
In our view, this log indicates an abnormal condition: a secondary process should be able to release its own QPs without triggering an error during normal shutdown.<br>
<br>
Thanks for your help.<br>
<br>
Best,<br>
Yang Ming</div>
</body>
</html>