UPGRADE YOUR BROWSER

We have detected your current browser version is not the latest one. Xilinx.com uses the latest web technologies to bring you the best online experience possible. Please upgrade to a Xilinx.com supported browser:Chrome, Firefox, Internet Explorer 11, Safari. Thank you!

cancel
Showing results for 
Search instead for 
Did you mean: 
Highlighted
Visitor regang
Visitor
323 Views
Registered: ‎06-26-2019

NVMe Host Accelerator (NVMeHA) usage on ZCU106 ref board

Jump to solution

Greetings Tech people,

Aim:

I'm trying to use the above-mentioned IP to speed up access to NVMe SSDs.

Setup:

1 PL PCIe port connected to 4 PCIe SSDs via a Microsemi PCIe switch. 

So far:

I've tried the Linux NVMe driver. I can access all 4 SSDs and I'm getting around 160MB/s write speed. I've downloaded all the 2018.3 reference h/w and all s/w for this IP. I've applied the patches to my 2018.3 kernel and have the NVMeHA driver compiling and running. The problem is that the driver crashes. I'm currently trying to debug this issue. I've added some extra prints in the driver. The log is attached as well as my petalinux generated pl.dtsi and my hardware design.

 

If anyone has any ideas please let me know.

Thanks for your time.

1 Solution

Accepted Solutions
Visitor regang
Visitor
175 Views
Registered: ‎06-26-2019

Re: NVMe Host Accelerator (NVMeHA) usage on ZCU106 ref board

Jump to solution

I switched to using 2019.1 version of HA /64-bit interface and performed 32 bytes read from CQ fifo. This resolved the issue for me. 

0 Kudos
4 Replies
Visitor regang
Visitor
278 Views
Registered: ‎06-26-2019

Re: NVMe Host Accelerator (NVMeHA) usage on ZCU106 ref board

Jump to solution

After some digging:

In xpci-ha.c, function nvme_handle_cqe()

My code:

Capture.JPG

 

exception occurs:

dev=nvme0
[ 21.816986] xnvmeha_sw_cq_handler, dev->queues[qid]=ffffffc87bf36a80
[ 21.823323] xnvmeha_read_cqe, 913
[ 21.826622] xnvmeha_read_cqe, 919
[ 21.829922] nvme_handle_cqe, 900*nvmeq->tags=ffffffc877696c00 cqe->command_id=54 nvmeq->qid =2
[ 21.838515] nvme_handle_cqe, 903req=0
[ 21.842166] Unable to handle kernel NULL pointer dereference at virtual address 00000148

 

This happens since in line 902 blk_mq_tag_to_rq(*nvmeq->tags, cqe->command_id); returns 0. I don't know my way around the blk modules so if anyone has some ideas please let me know.

 

Thanks.

0 Kudos
197 Views
Registered: ‎05-07-2019

Re: NVMe Host Accelerator (NVMeHA) usage on ZCU106 ref board

Jump to solution

where are you  ? i met the same question.can you give me your email or QQ?

0 Kudos
Visitor regang
Visitor
176 Views
Registered: ‎06-26-2019

Re: NVMe Host Accelerator (NVMeHA) usage on ZCU106 ref board

Jump to solution

I switched to using 2019.1 version of HA /64-bit interface and performed 32 bytes read from CQ fifo. This resolved the issue for me. 

0 Kudos
165 Views
Registered: ‎05-07-2019

Re: NVMe Host Accelerator (NVMeHA) usage on ZCU106 ref board

Jump to solution

thank you !

i still have some questions.

can you give me a printscreen about configration of NVMEHA IP?

do you update the software VIVADO  to 2019.1 ? the version  of driver  about NVMEHA to 2019.1 ?

 

0 Kudos