Merge to upstream

[MarekPikula]

Do you consider merging this version upstream so that there is no need for custom build in future?

This would require also making it Windows-compatible (mainly WinSock-compatible for jtag-tcp). I'll create separate issue for this though.

[MarekPikula]

Also keeping it aligned with upstream would be beneficial to prevent issues like #6 from happening, which were fixed in upstream already some time ago.

[Dolu1990]

Do you consider merging this version upstream so that there is no need for custom build in future?

I personnaly have no interrest in that, but i understand that some people have.

This would require also making it Windows-compatible (mainly WinSock-compatible for jtag-tcp). I'll create separate issue for this though.

Yes that's one of the issues :/ jtag-tcp is quite usefull to connect with simulations, so we should not drop it.

Also keeping it aligned with upstream would be beneficial to prevent issues like #6 from happening, which were fixed in upstream already some time ago.

Yes that's right, already had issues with update of GCC being more picky ^^

So, i'm all for what you propose, the issue is just man power and my personnal willing to do it myself, as i never use it on windows. If you are interrested into it, i would be happy to review pullrequest and merge them, and at the end try to get this fork upstream.

9

[MarekPikula]

Good to hear that. I'm adding it to my backlog for next week.

[MarekPikula]

Note that for some time I won't have a Windows machine at hand, so I won't be able to work on #9.

[adamfeuer]

+1 on this.

@MarekPikula Did you ever make progress on this? I'd like to help if I can.

[Dolu1990]

@MarekPikula @adamfeuer I got the official RISC-V debug spec implemented on VexRiscv via : https://github.com/SpinalHDL/VexRiscv/blob/0979f8ba805d23be7989ffff44cb2a383f7fb61d/src/main/scala/vexriscv/TestsWorkspace.scala#L149

• CsrPluginConfig(withPrivilegedDebug=true)

( Efinix provide the funding :) )

I will have to do some area overhead mesurement to soo how much is lost exactly.

[Dolu1990]

@MarekPikula @adamfeuer

Amagad, on Artix 7, the whole CPU (in a small config), with the debug infrastructure + jtag tap : VexRiscv style debug -> 938 LUT 788 FF RISC-V official style debug -> 1297 LUT 1235 FF

So, not a small overhead to get the official style debug :(

[Dolu1990]

So, i think it would make sense, even if VexRiscv implement the official RISC-V debug spec too, to merge the lightweight VexRiscv non standard debug implemention in upstream openocd.

[adamfeuer]

@Dolu1990 That's fantastic about getting the standard RISC-V debug spec implemented. Personally that would be the way I would prefer, despite the lots of extra LUTs/FFs. Has this change already been merged to the master/main branch?

Re: getting the lightweight JTAG for VexRiscv merged to upstream, does it provide the same functionality?

I guess either way I'd be willing to try to get it merged to upstream. I'll look at it today and see if I can update the current code to the current OpenOCD head, I guess that would be the first step.

[Dolu1990]

Has this change already been merged to the master/main branch?

Not yet merged, but soon will be merged in dev for sure (~2 weeks).

getting the lightweight JTAG for VexRiscv merged to upstream, does it provide the same functionality?

For regular baremetal stuff, yes, that's mostly the same, but things as debug linux kernel / mmu handeling could be way better with the official RISC-V openocd implementation.

I guess either way I'd be willing to try to get it merged to upstream. I'll look at it today and see if I can update the current code to the current OpenOCD head, I guess that would be the first step.

See https://github.com/SpinalHDL/openocd_riscv/pull/24

And as far as i know, that should be all. let's me know if you have any question / issue :)

[adamfeuer]

Thanks @Dolu1990. When you say this will be merged to dev, does that mean once this PR gets merged to the dnltz:WIP/dnltz/fix-upstream branch, @dnltz wants to submit a PR to merge his branch to OpenOCD?

[MarekPikula]

@adamfeuer Hi, unfortunately I didn't have much of a chance to work on it since the 2 years or so.

[Dolu1990]

@adamfeuer

When you say this will be merged to dev, does that mean once this PR gets merged to the dnltz:WIP/dnltz/fix-upstream branch, @dnltz wants to submit a PR to merge his branch to OpenOCD?

Ahh that's two diferent things, i was talking about the official RISC-V debug implemention in VexRiscv to be merge in the VexRiscv dev branch in 2 weeks.

The dnltz thing is about update this openocd fork with openocd upstream (VexRiscv style debug support)

[Dolu1990]

@adamfeuer Note, that this openocd fork also add a jtag driver over TCP, with a very simple protocol : Each byte from openocd to the jtag client (can be a sim) -> on byte : jtag.tms #= (byte & 1) != 0; jtag.tdi #= (byte & 2) != 0; jtag.tck #= (byte & 8) != 0; if ((buffer & 4) != 0) { tcpSend(jtag.tdo) }

And that's all. Here is a client implementation for SpinalSim : https://github.com/SpinalHDL/SpinalHDL/blob/dev/lib/src/main/scala/spinal/lib/com/jtag/sim/JtagTcp.scala#L32