<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40"> <head> <meta http-equiv="Content-Type" content="text/html; charset=utf-8"> <meta name="Generator" content="Microsoft Word 15 (filtered medium)"> <style></style> </head> <body lang="EN-US" link="blue" vlink="purple" style="word-wrap:break-word"> <div class="WordSection1"> <p class="MsoNormal"><span style="font-size:11.0pt">Confirm, it’s a bug, IIUC was introduced by reporting function update.<br> AFAIK, we do not test with non-DevX environment anymore, so missed this.<o:p></o:p></span></p> <p class="MsoNormal"><span style="font-size:11.0pt">Fix should be provided.<o:p></o:p></span></p> <p class="MsoNormal"><span style="font-size:11.0pt"><o:p> </o:p></span></p> <p class="MsoNormal"><span style="font-size:11.0pt">With best regards,<o:p></o:p></span></p> <p class="MsoNormal"><span style="font-size:11.0pt">Slava<o:p></o:p></span></p> <p class="MsoNormal"><span style="font-size:11.0pt"><o:p> </o:p></span></p> <div style="border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt"> <div> <div style="border:none;border-top:solid #E1E1E1 1.0pt;padding:3.0pt 0cm 0cm 0cm"> <p class="MsoNormal"><b><span style="font-size:11.0pt;font-family:"Calibri",sans-serif">From:</span></b><span style="font-size:11.0pt;font-family:"Calibri",sans-serif"> Asaf Penso <asafp@nvidia.com> <br> <b>Sent:</b> Sunday, December 22, 2024 9:39 AM<br> <b>To:</b> igootorov@gmail.com; Slava Ovsiienko <viacheslavo@nvidia.com><br> <b>Cc:</b> Laurent Hardy <laurent.hardy@6wind.com>; Olivier Matz <olivier.matz@6wind.com>; Didier Pallard <didier.pallard@6wind.com>; Jean-Mickael Guerin <jmg@6wind.com>; Edwin Brossette <edwin.brossette@6wind.com>; dev@dpdk.org<br> <b>Subject:</b> RE: net/mlx5: wrong Rx/Tx descriptor limits when DevX is off<o:p></o:p></span></p> </div> </div> <p class="MsoNormal"><o:p> </o:p></p> <p class="MsoNormal"><span style="font-size:11.0pt">Hello Igor and Slava,<o:p></o:p></span></p> <p class="MsoNormal"><span style="font-size:11.0pt">Can you please check out this issue?<o:p></o:p></span></p> <p class="MsoNormal"><span style="font-size:11.0pt"><o:p> </o:p></span></p> <p class="MsoNormal"><span style="font-size:11.0pt">Regards,<o:p></o:p></span></p> <p class="MsoNormal"><span style="font-size:11.0pt">Asaf Penso<o:p></o:p></span></p> <p class="MsoNormal"><span style="font-size:11.0pt"><o:p> </o:p></span></p> <div style="border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt"> <div> <div style="border:none;border-top:solid #E1E1E1 1.0pt;padding:3.0pt 0cm 0cm 0cm"> <p class="MsoNormal"><b><span style="font-size:11.0pt;font-family:"Calibri",sans-serif">From:</span></b><span style="font-size:11.0pt;font-family:"Calibri",sans-serif"> Edwin Brossette <<a href="mailto:edwin.brossette@6wind.com">edwin.brossette@6wind.com</a>> <br> <b>Sent:</b> Friday, 20 December 2024 19:06<br> <b>To:</b> <a href="mailto:dev@dpdk.org">dev@dpdk.org</a><br> <b>Cc:</b> Laurent Hardy <<a href="mailto:laurent.hardy@6wind.com">laurent.hardy@6wind.com</a>>; Olivier Matz <<a href="mailto:olivier.matz@6wind.com">olivier.matz@6wind.com</a>>; Didier Pallard <<a href="mailto:didier.pallard@6wind.com">didier.pallard@6wind.com</a>>; Jean-Mickael Guerin <<a href="mailto:jmg@6wind.com">jmg@6wind.com</a>><br> <b>Subject:</b> net/mlx5: wrong Rx/Tx descriptor limits when DevX is off<o:p></o:p></span></p> </div> </div> <p class="MsoNormal"><o:p> </o:p></p> <div> <p class="MsoNormal" style="margin-bottom:12.0pt">Hello,<br> <br> I have run into a regression following an update to stable dpdk-24.11 with a number of my Mellanox cx4/5/6 nics. This regression occurs with all nics in my lab which have DevX disabled: using mstconfig utility, I can see the flag UCTX_EN is not set.<br> <br> Mainly, the issue is that the ports cannot be started, with the following error logs in the journal:<o:p></o:p></p> <div style="margin-left:30.0pt"> <p class="MsoNormal">Set nb_rxd=1 (asked=512) for port=0<br> Set nb_txd=1 (asked=512) for port=0<br> starting port 0<br> Initializing port 0 [7c:fe:90:65:e6:54]<br> port 0: ntfp1 (mlx5_pci)<br> nb_rxq=2 nb_txq=2<br> rxq0=c9 rxq1=c25<br> txq0=c9 txq1=c25<br> port 0: rx_scatter=0 tx_scatter=0 max_rx_frame=1526<br> mlx5_net: port 0 number of descriptors requested for Tx queue 0 must be higher than MLX5_TX_COMP_THRESH, using 33 instead of 1<br> mlx5_net: port 0 increased number of descriptors in Tx queue 0 to the next power of two (64)<br> mlx5_net: port 0 number of descriptors requested for Tx queue 1 must be higher than MLX5_TX_COMP_THRESH, using 33 instead of 1<br> mlx5_net: port 0 increased number of descriptors in Tx queue 1 to the next power of two (64)<br> mlx5_net: Port 0 Rx queue 0 CQ creation failure.<br> mlx5_net: port 0 Rx queue allocation failed: Cannot allocate memory<br> rte_eth_dev_start(port 0) failed, error=-12<br> Failed to start port 0, set link down<br> Failed to start port 0<o:p></o:p></p> </div> <p class="MsoNormal" style="margin-bottom:12.0pt"><br> Looking more precisely into the problem, it appears that the number of Rx and Tx descriptors configured for my queues is 1. This happens because mlx5_dev_infos_get() return a limit of 1 for both Rx and Tx, which is unexpected. I identified this patch to be responsible for the regression:<br> <br> 4c3d7961d9002: net/mlx5: fix reported Rx/Tx descriptor limits<br> <a href="https://git.dpdk.org/dpdk/commit/?id=4c3d7961d9002bb715a8ee76bcf464d633316d4c" target="_blank">https://git.dpdk.org/dpdk/commit/?id=4c3d7961d9002bb715a8ee76bcf464d633316d4c</a><br> <br> After doing some debugging, I noticed that hca_attr.log_max_wq_sz is never configured. This should be done in mlx5_devx_cmd_query_hca_attr() which is called in this bit of code:<br> <br> <a href="https://git.dpdk.org/dpdk/tree/drivers/common/mlx5/mlx5_common.c#n681" target="_blank">https://git.dpdk.org/dpdk/tree/drivers/common/mlx5/mlx5_common.c#n681</a><o:p></o:p></p> <div style="margin-left:30.0pt"> <p class="MsoNormal">/*<br> * When CTX is created by Verbs, query HCA attribute is unsupported.<br> * When CTX is imported, we cannot know if it is created by DevX or<br> * Verbs. So, we use query HCA attribute function to check it.<br> */<br> if (cdev->config.devx || cdev->config.device_fd != MLX5_ARG_UNSET) {<o:p></o:p></p> </div> <div style="margin-left:30.0pt"> <div style="margin-left:30.0pt"> <p class="MsoNormal">/* Query HCA attributes. */<br> ret = mlx5_devx_cmd_query_hca_attr(cdev->ctx, &cdev->config.hca_attr);<br> if (ret) {<o:p></o:p></p> </div> </div> <div style="margin-left:30.0pt"> <div style="margin-left:30.0pt"> <div style="margin-left:30.0pt"> <p class="MsoNormal">DRV_LOG(ERR, "Unable to read HCA caps in DevX mode.");<br> rte_errno = ENOTSUP;<br> goto error;<o:p></o:p></p> </div> </div> </div> <div style="margin-left:30.0pt"> <div style="margin-left:30.0pt"> <p class="MsoNormal">}<br> cdev->config.devx = 1;<o:p></o:p></p> </div> </div> <div style="margin-left:30.0pt"> <p class="MsoNormal">}<br> DRV_LOG(DEBUG, "DevX is %ssupported.", cdev->config.devx ? "" : "NOT ");<o:p></o:p></p> </div> <p class="MsoNormal"><o:p> </o:p></p> <div> <p class="MsoNormal">I deduced that following the above patch, the correct value for maximum Rx and Tx descriptors will only be set if DevX is enabled (see the if condition on cdev->config.devx). If it is disabled, then maximum Rx and Tx descriptors will be 1, which will make the ports fail to start. Perhaps we should keep the previous default value (65535) if config.devx == 0 (DevX off)? This could be done like this, for example:<br> <br> diff --git a/drivers/net/mlx5/mlx5_ethdev.c b/drivers/net/mlx5/mlx5_ethdev.c<br> index 7708a0b80883..8ba3eb4a32de 100644<br> --- a/drivers/net/mlx5/mlx5_ethdev.c<br> +++ b/drivers/net/mlx5/mlx5_ethdev.c<br> @@ -359,10 +359,12 @@ mlx5_dev_infos_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *info)<br> info->flow_type_rss_offloads = ~MLX5_RSS_HF_MASK;<br> mlx5_set_default_params(dev, info);<br> mlx5_set_txlimit_params(dev, info);<br> - info->rx_desc_lim.nb_max =<br> - 1 << priv->sh->cdev->config.hca_attr.log_max_wq_sz;<br> - info->tx_desc_lim.nb_max =<br> - 1 << priv->sh->cdev->config.hca_attr.log_max_wq_sz;<br> + if (priv->sh->cdev->config.devx) {<br> + info->rx_desc_lim.nb_max =<br> + 1 << priv->sh->cdev->config.hca_attr.log_max_wq_sz;<br> + info->tx_desc_lim.nb_max =<br> + 1 << priv->sh->cdev->config.hca_attr.log_max_wq_sz;<br> + }<br> if (priv->sh->cdev->config.hca_attr.mem_rq_rmp &&<br> priv->obj_ops.rxq_obj_new == devx_obj_ops.rxq_obj_new)<br> info->dev_capa |= RTE_ETH_DEV_CAPA_RXQ_SHARE;<br> <br> Thanks in advance for your help.<br> <br> Regards,<br> Edwin Brossette.<o:p></o:p></p> <div> <p class="MsoNormal" style="margin-bottom:12.0pt"><o:p> </o:p></p> </div> </div> </div> </div> </div> </div> </body> </html>