https://open-source-silicon.dev logo
Join Slack
Powered by
Hi, we got `routes over obstructions` during `27-e...
# openlane
j
Hi, we got 
routes over obstructions
during 
27-ext2spice
Copy code
[INFO]: Running Magic Spice Export from LEF...
[ERROR]: There are illegal overlaps (e.g., routes over obstructions) in your design.
[ERROR]: See /scratch/users/jakeke/caravel_fast_ann_fieldious/openlane/user_project_wrapper/runs/user_project_wrapper/logs/finishing/27-ext2spice.feedback.txt for more.
In the feedback.txt, there are many metal 2 with obsm2 and metal 3 with obsm3 overlaps like shown below
Copy code
box 4124 4000 4152 4012
feedback add "Illegal overlap between obsm2 and metal2 (types do not connect)" medium
box 5228 4000 5256 4012
feedback add "Illegal overlap between obsm2 and metal2 (types do not connect)" medium
box 4000 4104 4048 4164
feedback add "Illegal overlap between obsm3 and metal3 (types do not connect)" medium
When I opened the locations in magic, many are under the met4 and met5 power ring, but there is no met2 in those places. Some boxes are not even in the wrapper
m
I may be mistaken, but I believe the coordinates here are in database units. Divide by 200 to get user-units. Or copy-paste 
box 4124 4000 4152 4012
into the magic console.
j
i did just copy-pasted directly in the magic console
m
Does that show an error in the layout?
j
i had error display enabled in the magic screenshot but i do not see anything in the layout
m
How about 
zoom [box 4124 4000 4152 4012]
?
Sorry if I'm not being much help.
j
the zoom command returns 
zoom factor must be between 0 and 20
. Thank you Mitch. I just really need more hints to debug these 
route over obstruction
issues. I did not have those issues with a previous version user_proj_example layout
m
j
I do not understand how this is related to my routes over obstruction error in ext2spice
@Tim Edwards, could you please help us locate the problem nets in magic? We do not see anything based on the boxes in the report.
m
Use the command "feedback find" followed by "zoom box" to see the area of the error in question.
t
@Jake Ke: More specifically, you would need to do this: 
snap internal ; box values 4124 4000 4152 4012 ; zoom box
. The 
snap internal
is significant; it determines whether magic interprets bare units given to it as internal units or lambda units (a multiple of internal units). You can also do 
box values 41.24um 40um 41.52um 40.12um
. They really should not be writing out raw data from magic like that, particularly as there is a nice built-in script that formats it into microns.
👍 2
5 Views
Open in Slack
PreviousNext
Powered by