| *** tpb has joined #litex | 00:00 | |
| *** futarisIRCcloud has quit IRC | 02:03 | |
| *** futarisIRCcloud has joined #litex | 02:10 | |
| *** nrossi has joined #litex | 03:35 | |
| *** futarisIRCcloud has quit IRC | 04:13 | |
| nrossi | so I am trying to get the linux-on-litex-vexriscv going a on versa ecp5. However I am running into some trouble with both the prebuilt binaries and any self built binaries where the kernel seems to hang. It looks like it is hanging in the kernel on a BUG_ON in setup_bootmem where it checks if mem_size is 0. | 06:35 |
|---|---|---|
| daveshah | nrossi: can you post the full startup log? | 06:47 |
| nrossi | daveshah: https://gist.github.com/nathanrossi/9135ac0889efc7256c402a3b4e318ada | 07:10 |
| tpb | Title: litex-linux-vexriscv-log-prebuilt-binaries · GitHub (at gist.github.com) | 07:10 |
| nrossi | oh sorry, that is with my built kernel. Though it looks exactly the same with the kernel image from the repo, just different size in bytes | 07:12 |
| daveshah | nrossi: just to double check, is your litex up to date? | 07:27 |
| nrossi | daveshah: i did just update it, I noticed an issue with the CSR mapping stuff. However it looks with a rebuilt bitstream and emulator.bin it boots. Just checking if it was something I had done wrong previously though | 07:29 |
| nrossi | daveshah: I think i traced the issue... TFTP.... it looks like the images are not copying correctly... | 07:33 |
| daveshah | nrossi: I've seen this too | 07:37 |
| daveshah | Maybe try a different tftp server | 07:37 |
| daveshah | I've seen the dnsmasq tftp server recommended here, although haven't tried it myself | 07:38 |
| nrossi | hmmm, tftppy and atftpd seem to have issues. I did not know dnsmasq had a tftp server will have a look at that | 07:44 |
| nrossi | daveshah: no luck with dnsmasq, this makes me think its just packet loss on the network and the tftp client in the bios code does not handle that well? | 07:47 |
| Dolu | Just one thing, how the address system work in litex ? because in the .py, i can see that the ethernet mac is mapped at 0x30000000, which isn't an IO region of the CPU | 08:59 |
| Dolu | For me it is suspicious, and can be the source of the ethernet issue, unless there is some kind of address MSB trickery + software accessing the ethernet via 0xB | 09:02 |
| xobs | Dolu: I have absolutely no context here, so take this with a grain of salt, but I seem to recall in litex, the MSB indicates whether an access is cached or not. | 09:17 |
| Dolu | It would make sense | 09:17 |
| Dolu | Right that's it | 09:26 |
| Dolu | so that's fine, that's not the issue here | 09:26 |
| *** futarisIRCcloud has joined #litex | 09:33 | |
| _florent__ | nrossi: can you try to load the images over serial? (it's a bit slow, a few minutes, but then we'll see if it's related to the TFTP or the bitstreams/images) | 10:14 |
| _florent__ | nrossi: https://github.com/enjoy-digital/linux-on-litex-vexriscv#load-the-linux-images-over-serial | 10:15 |
| tpb | Title: GitHub - enjoy-digital/linux-on-litex-vexriscv: Experiments with Linux on LiteX-VexRiscv (at github.com) | 10:15 |
| _florent__ | btw, the prevuilt bitstreams/images are now there: https://github.com/enjoy-digital/linux-on-litex-vexriscv-prebuilt | 10:15 |
| tpb | Title: GitHub - enjoy-digital/linux-on-litex-vexriscv-prebuilt: Prebuilt bitstreams / linux images for litex-on-litex-vexriscv repository (at github.com) | 10:15 |
| _florent__ | and how to use them: https://github.com/enjoy-digital/linux-on-litex-vexriscv#pre-built-bitstreamslinux-images | 10:15 |
| tpb | Title: GitHub - enjoy-digital/linux-on-litex-vexriscv: Experiments with Linux on LiteX-VexRiscv (at github.com) | 10:15 |
| *** spacekookie has quit IRC | 11:04 | |
| nrossi | _florent__: serial loading works and kernel boots (with the prebuilt binaries you linked above) | 11:05 |
| *** spacekookie has joined #litex | 11:05 | |
| *** futarisIRCcloud has quit IRC | 11:43 | |
| _florent__ | nrossi: thanks, i'll do more tests with tftp | 12:18 |
| nrossi | _florent__: I added some instrumentation to the bios code to crc the transferred files... it looks like they transfer fine, but now it seems to hang just before booting (and part of the uart printed text is missing) | 12:20 |
| _florent__ | nrossi: when doing tftp, are you using the exact same images you were using with serial? | 12:22 |
| nrossi | _florent__: yep, the same prebuilt ones from the repo you linked | 12:23 |
| _florent__ | nrossi: i just tested with the prebuilt bitstream and images, it works on my machine | 12:58 |
| _florent__ | nrossi: are you also using the provided bitstream? | 12:58 |
| nrossi | _florent__: just tested it and it works (tried a few boots). However I did have issues earlier, so not sure if it works consistently | 13:02 |
| nrossi | _florent__: although i did notice that with the most recent litex the build now fails with csr already mapped errors | 13:03 |
| _florent__ | nrossi: ok thanks, sorry a commit was not pushed, it's done now and should build correctly (you also need to update litedram) | 13:59 |
| _florent__ | daveshah: just for info, i fixed the reported memory size for the versa | 13:59 |
| daveshah | _florent__: thanks! | 14:45 |
| _florent__ | https://twitter.com/enjoy_digital/status/1126897441201119237 :) | 17:20 |
| somlo | Congrats!!! This is a huge step forward for those of us who care about actually being in control of our own computers! | 17:37 |
| somlo | now, on to 64bit Linux, and maybe Fedora ;) | 17:38 |
| _florent__ | this is also a good regression test for LiteX/LiteDRAM :) | 17:50 |
| somlo | should writing a 32'h0 to 0xe0000000 cause a reset, or should that only happen for specific bit patterns, e.g. 32'h01)? | 18:08 |
| somlo | right now any write reboots (tested on vexriscv), but is that the desired behavior? | 18:09 |
| somlo | ^any value, I mean | 18:09 |
| somlo | I found the answer here: https://github.com/enjoy-digital/litex/blob/master/litex/soc/integration/soc_core.py#L152 | 18:40 |
| tpb | Title: litex/soc_core.py at master · enjoy-digital/litex · GitHub (at github.com) | 18:40 |
| somlo | it's *any* write to the register, regardless of the value... | 18:41 |
| *** futarisIRCcloud has joined #litex | 21:55 | |
Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!