Problems building 'Hello World CL VHDL Example'

Question

I wanted to build this example on my EC2 instance (which was prepared from the AWS FGA Developer AL2 ami):
* https://github.com/aws/aws-fpga/tree/master/hdk/cl/examples/cl_hello_world_vhdl

I was following steps from this page:
* https://github.com/aws/aws-fpga/blob/master/hdk/README.md#endtoend

But I bumped into issues in the build output looking like this:
```
ERROR: [Common 17-141] Failed to write file content of top_sp.xbdc in zip archive.
...
...
Abnormal program termination (11)
Please check '/home/ec2-user/src/project_data/aws-fpga/hdk/cl/examples/cl_hello_world_vhdl/build/scripts/hs_err_pid11287.log' for details
```

If I take a look at the suggested `hs_err_pid11287.log` file, it contains this:
```
Stack:
/lib64/libc.so.6(+0x33c90) [0x7ffa2444bc90]
/opt/Xilinx/Vivado/2021.1/lib/lnx64.o/libtcmalloc.so.4(tcmalloc::ThreadCache::ReleaseToCentralCache(tcmalloc::ThreadCache::FreeList*, unsigned long, int)+0xf3) [0x7ffa263a8973]
```

I noticed there were a few other forum threads with similar errors:
* https://forums.aws.amazon.com/thread.jspa?threadID=256609
* https://forums.aws.amazon.com/thread.jspa?threadID=261665&tstart=0

I saw a mention of trying to fiddle with the value of a `resource_sharing` parameter, but the thread didn't mention where to tweak such a setting. Hunting around, I thought perhaps I needed to edit the `$HDK_DIR/common/shell_v04261818/build/scripts/strategy_TIMING.tcl` file, then tweak a line in there to change from `-resource_sharing off` to `-resource_sharing auto`, and then try re-build with `./aws_build_dcp_from_cl.sh -strategy TIMING`.

But nope, got a similar error as above, though this time, the 'Failed to write' error related to a `top_sp.edf` file rather than the `top_sp.xbdc` file mentioned earlier.

Wondering what else can I try?

Gurce · Accepted Answer

Ok, gave this another try tonight, and this time when I ran `./aws_build_dcp_from_cl.sh -foreground`, I got this hint:

```
ERROR: [Common 17-1294] Unable to create directory [/home/ec2-user/src/project_data/aws-fpga/hdk/cl/examples/cl_hello_world_vhdl/build/scripts/.Xil/Vivado-5384-ip-10-161-10-222.ap-southeast-2.compute.internal].
```
This surprised me, so I tried manually doing a `mkdir`of this directory and got:

```
mkdir: cannot create directory ‘/home/ec2-user/src/project_data/aws-fpga/hdk/cl/examples/cl_hello_world_vhdl/build/scripts/.Xil/Vivado-5384-ip-10-161-10-222.ap-southeast-2.compute.internal’: No space left on device
```

Oh my! So it turns out my `/dev/xvdb` EBS storage (used for `/home/ec2-user/src/project_data/` where all the examples are) was only 5GB in size and was already pretty full after compiling one example in the past, and then trying to compile this example now.

So looking back, this was my fault for only giving it the default 5GB while preparing my VM. I've managed to manually expand the volume to be 10GB now, but if I had my way again, I'd have initially configured this drive to be 50GB, to give me plenty of leg-room in future.

Anyways, I've run `./aws_build_dcp_from_cl.sh -foreground` again now, and it has worked successfully, so thanks for your assistance along the way and for the reminder to get back to you on this, working fine now :)

AWS-User-7905434 · Answer

Dear customer,
just checking in to see if you need any further help with this issue or if it is resolved. Thanks

AWS-User-7905434 · Answer

Hi, Thank you so much for the additional information. I was able to successfully generate DCP and AFIs for cl_hello_world_vhdl using FPGA Developer AMI (AL2) versions 1.11.0 and 1.11.3. Just to make sure, are you seeing the error during dcp generation or is it after dcp gen and in the afi generation step?

In addition, can you please try using the latest AL2 dev AMI version 1.11.3 and see if it works.

Thanks

In addition, can you please try using the latest AL2 dev AMI version 1.11.3 and see if it works.

Thanks

AWS-User-7905434 · Answer

Dear customer,

Thank you so much for contacting aws and your interest in F1. I tried to re-create the error you mentioned; but I was able to successfully generate dcp for cl_hello_world_vhdl. Can you please provide the following information 1.) dev-AMI version you used 2.) instance type 3.) also can you please confirm if you pointed $CL_DIR to cl_hello_world_vhdl and also ran hdk_setup.sh

Thanks

Problems building 'Hello World CL VHDL Example'

Relevant content