00:04 <sf-slack2> <timo.callahan> @ric96 -- quick and maybe not useful answer, but yes, I've booted Linux on Litex gateware from prjxray/symbiflow. Which instructions are you using?

00:06 <sf-slack2> <timo.callahan> I pulled recent changes on symbiflow-arch-defs under my additions for 100T, and now I see this error. Is there something I need to clean or reset?

00:06 <sf-slack2> <timo.callahan> `100%] Built target file_xc_xc7_archs_artix7_100t_tiles_slicel_slicel.pb_type.xml [100%] Built target file_xc_xc7_archs_artix7_100t_tiles_slicel_slicel.model.xml [100%] Generating slicel.tile.xml`

00:06 <sf-slack2> <timo.callahan> Traceback (most recent call last): File "/home/tcal/2nd-tcal-x/symbiflow-arch-defs/xc/common/utils/prjxray_physical_tile_import.py", line 195, in <module> main() File "/home/tcal/2nd-tcal-x/symbiflow-arch-defs/xc/common/utils/prjxray_physical_tile_import.py", line 191, in main import_physical_tile(args) File

00:06 <sf-slack2> "/home/tcal/2nd-tcal-x/symbiflow-arch-defs/xc/common/utils/prjxray_physical_tile_import.py", line 142, in import_physical_tile add_pinlocations(tile_name, sub_tile_xml, fc_xml, pin_assignments, ports) File "/home/tcal/2nd-tcal-x/symbiflow-arch-defs/utils/lib/pb_type_xml.py", line 55, in add_pinlocations for side in pin_assignments['pin_directions'][tile_name][pin]: KeyError: 'SLICEL'

00:06 <sf-slack2> xc/xc7/archs/artix7_100t/tiles/slicel/CMakeFiles/file_xc_xc7_archs_artix7_100t_tiles_slicel_slicel.tile.xml.dir/build.make:95: recipe for target 'xc/xc7/archs/artix7_100t/tiles/slicel/slicel.tile.xml' failed

00:10 <litghost> acomodi's change to how SLICEL/M are imported was merged

00:11 <litghost> I suspect you need to apply some of the changes from that PR to your PR

00:32 <sf-slack2> <timo.callahan> Maybe I forgot to do a `make all_conda` after I pulled the changes. Let me give that a quick try....

00:33 <litghost> timo.callahan: https://github.com/SymbiFlow/symbiflow-arch-defs/pull/1484 is the PR I was thinking of

00:33 <tpb> Title: xc7: avoid splitting CLBs by acomodi · Pull Request #1484 · SymbiFlow/symbiflow-arch-defs · GitHub (at github.com)

00:34 <litghost> I believe https://github.com/SymbiFlow/symbiflow-arch-defs/pull/1484/files#diff-496a133e3b8ce0856128264ee020c3ffR5 is the type of changes you need to make

00:34 <tpb> Title: xc7: avoid splitting CLBs by acomodi · Pull Request #1484 · SymbiFlow/symbiflow-arch-defs · GitHub (at github.com)

00:42 <sf-slack2> <timo.callahan> Ah yes yes, stuff in my uncommitted code that needs updates. Got it.

01:20 <mithro> ric96: Are you @sahajsarup on twitter?

03:23 <ric96> mithro: yes

04:22 <ric96> sf-slack2: the litex soc is from https://github.com/SymbiFlow/symbiflow-examples and i use the vexriscv prebuilt images

04:22 <tpb> Title: GitHub - SymbiFlow/symbiflow-examples (at github.com)

08:59 <sf-slack2> <acomodi> ric96: This may be an issue with the Linux images. I am currently using this ones https://we.tl/t-e8Uq147RAD (The problem is that I have generated them quite a while ago and I do not remember their origin)

09:05 <ric96> sf-slack2: any idea on the commit id? I can build

09:39 <sf-slack2> <acomodi> ric96: I was able to regenerate working Linux images for the litex-linux example from https://github.com/timvideos/litex-buildenv (commit 17e230ec8af0838060d12e3d810a448825d13fd2) and following the Linux wiki (https://github.com/timvideos/litex-buildenv/wiki/Linux)

09:39 <tpb> Title: GitHub - timvideos/litex-buildenv: An environment for building LiteX based FPGA designs. Makes it easy to get everything you need! (at github.com)

09:42 <ric96> sf-slack2: acomodi: sure, I am using from their prebuilt repo, and these images aren't booting for me https://github.com/litex-hub/linux-on-litex-vexriscv-prebuilt

09:42 <tpb> Title: GitHub - litex-hub/linux-on-litex-vexriscv-prebuilt: Prebuilt bitstreams / linux images for litex-on-litex-vexriscv repository (at github.com)

09:53 <sf-slack2> <acomodi> ric96: yeah, I have tried some of those commits, but they don't seem to work for me as well. I guess that they are incompatible for the current litex-linux design in symbiflow-examples

10:15 <sf-slack2> <acomodi> ric96: FYI, I have opened a PR in symbiflow-examples to add the working Linux binaries: https://github.com/SymbiFlow/symbiflow-examples/pull/8

10:15 <tpb> Title: linux: add pre-built compatible images for the litex-linux example by acomodi · Pull Request #8 · SymbiFlow/symbiflow-examples · GitHub (at github.com)

10:27 <ric96> sf-slack2: acomodi awesome!

15:41 <sf-slack2> <timo.callahan> Hmm, in my 100T work, I must have missed something...or, I copied something from 50T when I should have copied from 200T. I get this error: `Type: Architecture file File: /tmp/really-really-really-really-really-really-really-really-really-really-really-really-really-long-path/conda/conda-bld/vtr_1589474563511/work/vpr/src/util/vpr_utils.cpp Line: 2318 Message: *Unable to find port CLBLM_L_COUT (on block

15:41 <sf-slack2> BLK-TL-CLBLM_L).*` Surrounding printouts show both `BLK-TL-SLICEL/M` and `BLK-TL-CLBLL/M_L/R` under "Resource usage...", but only `BLK-TL-SLICEL/M` under "Device Utilization". I will try a 35T and compare the logs.

15:58 <ric96> sf-slack2: @acomodi hmmm i'm still stuck at

15:58 <ric96> This is using the we.tl images and symbiflow-examples bit

15:58 <ric96> --============= Liftoff! ===============--

15:58 <ric96> ```

15:58 <ric96> **

15:58 <ric96> ```

15:59 <sf-slack2> <timo.callahan> @ric96 which approach are you using for loading the Linux images? lxterm --images, tftp (netboot), or flashing?

16:02 <ric96> sf-slack2: timo.callahan: lxterm, serial

16:02 <ric96> Images load fine, or so it seems

16:05 <sf-slack2> <acomodi> Can you please print the images.json file?

16:06 <sf-slack2> <timo.callahan> I think there have been some changes at the offsets of the different .... ha yes what @acomodi asks. The offsets in images.json must match the gateware version.

16:06 <sf-slack2> <acomodi> It may be a wrong memory offsets

16:06 <sf-slack2> <acomodi> @timo.callahan Yep, exactly

16:45 <sf-slack2> <timo.callahan> I think I found my 100T issue. From the 35T VPR log: Device Utilization: 0.00 (target 1.00) Physical Tile BLK-TL-CLBLL_L: Block Utilization: 0.02 Logical Block: BLK-TL-SLICEL Physical Tile BLK-TL-CLBLL_R: Block Utilization: 0.02 Logical Block: BLK-TL-SLICEL Physical Tile BLK-TL-CLBLM_L: Block Utilization: 0.02 Logical Block: BLK-TL-SLICEL Block

16:46 <sf-slack2> Utilization: 0.00 Logical Block: BLK-TL-SLICEM Physical Tile BLK-TL-CLBLM_R: Block Utilization: 0.02 Logical Block: BLK-TL-SLICEL Block Utilization: 0.00 Logical Block: BLK-TL-SLICEM From my 100T log: Device Utilization: 0.00 (target 1.00) Physical Tile BLK-TL-CLBLL_L: Block Utilization: 0.00 Logical Block: BLK-TL-SLICEL Physical Tile BLK-TL-CLBLL_R: Block Utilization:

16:46 <sf-slack2> 0.00 Logical Block: BLK-TL-SLICEL Physical Tile BLK-TL-CLBLM_L: Block Utilization: 0.00 Logical Block: BLK-TL-SLICEM Physical Tile BLK-TL-CLBLM_R: Block Utilization: 0.00 Logical Block: BLK-TL-SLICEM So I'm missing the SLICELs in my CLBLMs.

16:49 <sf-slack2> <acomodi> @timo.callahan https://github.com/SymbiFlow/symbiflow-arch-defs/blob/master/xc/xc7/archs/artix7/tiles/clblm_l/CMakeLists.txt This is an example of a CLBLM tile definition. The SLICEM site defines, after the `/` a set of equivalent sites for it, in this case, the SLICEL

16:49 <tpb> Title: symbiflow-arch-defs/CMakeLists.txt at master · SymbiFlow/symbiflow-arch-defs · GitHub (at github.com)

16:50 <sf-slack2> <timo.callahan> Yes @acomodi exactly, I had hand-edited that incorrectly.

16:54 <sf-slack2> <timo.callahan> ...thanks! fingers crossed it works this time

17:07 <sf-slack2> <acomodi> ric96: FYI, the images.json should look like this http://snippi.com/s/e139yip

17:07 <tpb> Title: Snippi (at snippi.com)

17:08 <ric96> @acomodi thanks, mine is mich different, hopefully this fixes it. I'll check it in a bit.

17:12 <_whitenotifier-c> [symbiflow-arch-defs] acomodi opened issue #1496: Changes in miniconda version generate conda env issues - https://git.io/Jfgq5

17:15 <andrewb1999> acomodi: Building the pynqz1 buttons test using the branch you sent yesterday is crashing for me

17:15 <andrewb1999> acomodi: AssertionError: ('BRAM_INT_INTERFACE_L_X6Y0', 'BRAM_INT_INTERFACE_L', 'INT_INTERFACE_LOGIC_OUTS_L_B0', 189)

17:16 <andrewb1999> When building channels.db

17:16 <andrewb1999> Any ideas on what's going wrong here?

17:18 <sf-slack2> <acomodi> andrewb1999: yes, I think that the solution would be to decouple the two architecture definitions in two separate root dirs (zynq010 and zynq020). This because the generic channels.db is shared both for the z010 and z020, hence, when building the z020, the errors you see are produced

17:18 <sf-slack2> <acomodi> It should be similar to what has been done for artix7 and artix_200t

17:19 <andrewb1999> acomodi: Okay I will look into doing that

19:28 <ric96> @acomodi: the address change in image.json worked! it boots!

21:04 <sf-slack2> <timo.callahan> Q: Does PrjXray support "BUFR"?

21:06 <litghost> I believe so

21:06 <litghost> I added fuzzers for it

21:06 <litghost> I don't know if symbiflow-arch-defs integrated it

21:07 <litghost> Now that acomodi's heterogeneous VPR feature is landed, in theory you should be able to just ask the tile import script to import it

21:09 <sf-slack2> <timo.callahan> I tried using it in the counter test, and it complained "Failed to find matching architecture model for 'BUFR'". I wondered if I need to use arty-full instead of arty-swbut. But I'll see if I can do as you describe, add it in the right place.

21:10 <sf-slack2> <timo.callahan> I had previously tried BUFG, which doesn't work with arty-swbut but does work with arty-full in my observation.

21:30 <litghost> arty-swbut is an ROI based graph, which has no clock elements

21:30 <litghost> arty-full is the full graph, which supports clock elements

21:30 <litghost> To be clear BUFG and BUFR are different black boxes

21:30 <litghost> BUFG support is well tested

21:31 <litghost> BUFR support is not

22:50 <sf-slack2> <timo.callahan> Thanks @litghost!