Closed d-m-bailey closed 2 years ago
The problem is not with the gds, it seems. Investigating futher...
The problem is that the verilog is not being updated correctly for the first 5 gpio_defaults_blocks
gpio_defaults_block gpio_defaults_block_29 ( | gpio_defaults_block_1808 gpio_defaults_block_29 (
.VGND(vssd_core), | .VGND(vssd_core),
.VPWR(vccd_core), | .VPWR(vccd_core),
.gpio_defaults({ \gpio_defaults[389] , \gpio_defaults[388] , \gpio_| .gpio_defaults({ \gpio_defaults[389] , \gpio_defaults[388] , \gpio
); | );
gpio_defaults_block \gpio_defaults_block_2[0] ( | gpio_defaults_block \gpio_defaults_block_2[0] (
.VGND(vssd_core), | .VGND(vssd_core),
You can see the instances that we're updated and those that weren't.
Patched scripts/gen_gpio_defaults.py
. Testing now.
Thanks for investigating. Let me know when you have a pull request, or if you need any help.
During
make ship
andmake truck
,mag
data is read fromcaravel_user_project/mag
andcaravel_user_project/caravel/mag
. Older versions ofgpio_default_block_1803
andgpio_default_block_0403
exist incaravel/mag
and appear to override the newer versions incaravel_user_project/mag
.To reproduce, in
caravel_user_project
The output gds should match the
caravel_user_project/mag
data, but I think it'scaravel_user_project/caravel/mag
data.LVS doesn't match for these 2 cells and the mag data is different, so I'm assuming this is the reason.
Will submit a pull request to remove the mag data from
caravel/mag
.There are many of the following messages generated during
make ship
, which indicates that the location of the data is very relevant when there are versions in both directories.