03-30-2016 04:53 PM
I'm finding that about once per day, Vivado will reset all PS settings by itself.
I'll be working deep down in a heiarchical block for a couple hours, making various changes, and
then suddenly my design spews a bunch of errors because all my PS peripherals have been eliminated.
I've been forced to manually re-enter all the peripheral, clock, DDS timing etc etc. What a hassel.
Sorry I can't (yet) provide a reproducible procedure, but I was wondering if anyone else was experiencing this.
03-30-2016 05:07 PM
Doh! Forgot to say:
Vivado 2015.4, Windows 10, 64-bit. Vivado project created initially with Microzed board file.
03-30-2016 05:15 PM
Looks like it is related to home-built packaged IP.
I made some changed to some IP I wrote, and changed the version number from 1.0 to 1.1 and repackaged the IP
and closed the custom-IP-project.
Then I opend the target project that uses this IP and it let me know that a new version was available and I should
upgrade. This is what I expected, and I did so.
Boom! All my PS settings are gone.
03-31-2016 07:35 AM
So I'm using presets to restore PS settings after they are trashed by Vivado, but I still must re-wire
all my peripherals each time this happens.
I'd be grateful for any guidance on troubleshooting this issue.
04-07-2016 05:03 AM
07-01-2016 10:46 AM
Still seeing this phenomenon on 2015.4. It is definitely related to loading the block design when there is user IP
that has undergone a revision change.
08-04-2016 08:48 AM
"Good morning all, thank you for coming to our quarterly engineering update meeting."
"Before we begin, David has an update on the problem with Vivado losing his PS settings.
David what can you tell us?"
"Well, Bob, it's still doing it."
"OK. Thanks David. Now on to our scheduled agenda..."
08-05-2016 09:57 AM
Hi, Sorry to hear that- data corruption is never good.
I support this IP and can say I have not heard of this before.
Unfortunately, this sounds rare, and so to get it fixed, we'll need a testcase/methodology to reproduce. It doesn't have to be reproducible on the first try, but the number of attempts to observer the corruption must be bounded.
It sounds like it may be related to hierarchical BD + upgrading custom IP, but I'm not sure. If you can invest in a testcase, feel free to send to me and I'll forward to development.