xrt build and install error on aws F1.4

Question

Hello please halpe to check below issue:  
  
I am now upgrading the AWS and XRT on my AWS F1 machine, but afterwards I can not find the fpga devices any more. lspci I can find the device, which I can load afi (hellowrold afi)  
$ lspci  
...  
00:1b.0 Memory controller: Amazon.com, Inc. Device f000  
00:1c.0 Memory controller: Amazon.com, Inc. Device 1041  
00:1d.0 Memory controller: Amazon.com, Inc. Device f000  
00:1e.0 Memory controller: Amazon.com, Inc. Device 1041  
  
$ xbutil scan  
INFO: Found total 0 card(s), 0 are usable  
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~  
System Configuration  
OS name:	Linux  
Release:	5.4.0-1025-aws  
Version:	#25~18.04.1-Ubuntu SMP Fri Sep 11 12:03:04 UTC 2020  
Machine:	x86_64  
Model:		HVM domU  
CPU cores:	16  
Memory:		245642 MB  
Glibc:		2.27  
Distribution:	Ubuntu 18.04.4 LTS  
Now:		Tue Sep 29 04:46:36 2020 GMT  
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~  
XRT Information  
Version:	2.8.0  
Git Hash:	5d0ccb4f90d39202644f07de9648743d986a7e68  
Git Branch:	master  
Build Date:	2020-09-29 04:26:24  
XOCL:		unknown  
XCLMGMT:	unknown  
-----------------------------------------------  
The problem looks like XOCL is not installed.  
So I redo all the installation again and check each steps, I find that when I am soing this step:   
  
sudo apt install --reinstall ./xrt_202020.2.8.0_18.04-amd64-xrt.deb  
  
There are error as below:  
-----------------------------------------------  
Unregistering old XRT Linux kernel module sources 2.8.0 from dkms on Ubuntu/Debian  
  
------------------------------  
Deleting module version: 2.8.0  
completely from the DKMS tree.  
------------------------------  
Done.  
Cleanup is skipped for package upgrade/downgrade/re-install on Ubuntu/Debian  
Unpacking xrt (2.8.0) over (2.8.0) ...  
Setting up xrt (2.8.0) ...  
Unloading old XRT Linux kernel modules  
rmmod: ERROR: Module xocl is not currently loaded  
rmmod: ERROR: Module xclmgmt is not currently loaded  
Invoking DKMS common.postinst for xrt  
Loading new xrt-2.8.0 DKMS files...  
Building for 5.4.0-1025-aws  
Building initial module for 5.4.0-1025-aws  
ERROR: Cannot create report: \[Errno 17] File exists: '/var/crash/xrt.0.crash'  
Error!  Build of xclmgmt.ko failed for: 5.4.0-1025-aws (x86_64)  
Consult the make.log in the build directory  
/var/lib/dkms/xrt/2.8.0/build/ for more information.  
Finished DKMS common.postinst  
Loading new XRT Linux kernel modules  
modprobe: FATAL: Module xclmgmt not found in directory /lib/modules/5.4.0-1025-aws  
modprobe: FATAL: Module xocl not found in directory /lib/modules/5.4.0-1025-aws  
****************************************************************  
* DKMS failed to install XRT drivers.  
* Please check if kernel development headers are installed for OS variant used.  
*   
* Check build logs in /var/lib/dkms/xrt/2.8.0  
****************************************************************  
Installing MSD / MPD daemons  
  
The aws-fpga branch I am using is 1.4.15;  
The xrt version: I have been used xrt2.6.0 and 2.8.0 (xrt master).

Accepted Answer

Hello,  
  
It seems like the Ubuntu AWS optimized kernel and module extras do not have the fpga-mgr module that is required by this XRT.  
Can you try using a different kernel?  
  
For eg:  
  
sudo apt update  
sudo apt install linux-virtual  
sudo apt install linux-modules-extra-$(uname -r)  
sudo apt purge linux*aws  
sudo reboot  
  
This should get the XRT to work. Let me know if it doesn't work for you.  
  
-Deep

Answer

Hi Deep,  
My current aws kernel version is: 5.4.0-1025-aws  
Do you mean this kernel is not supported by XRT?  
  
for your suggested step: sudo apt purge linux*aws   
it actually removes the current kernel -- there will be warning that remove kernel may cause system problem so I would need your confirm.

Answer

Thanks, after removing the old kernel which is 5.4.0-1025-aws, reboot, the kernel version now is 4.15.0-118-generic.   
  
XRT need to be rebuild and reinstalled then.  
Afterwards the xbutil scan will get the correct result.

xrt build and install error on aws F1.4

XRT Information Version: 2.8.0 Git Hash: 5d0ccb4f90d39202644f07de9648743d986a7e68 Git Branch: master Build Date: 2020-09-29 04:26:24 XOCL: unknown XCLMGMT: unknown

There are error as below:

Deleting module version: 2.8.0 completely from the DKMS tree.

Relevant content

XRT Information
Version: 2.8.0
Git Hash: 5d0ccb4f90d39202644f07de9648743d986a7e68
Git Branch: master
Build Date: 2020-09-29 04:26:24
XOCL: unknown
XCLMGMT: unknown

Deleting module version: 2.8.0
completely from the DKMS tree.