Jtag support

[bugadani]

No description provided.

[korken89]

Looking forward to this! 🚀 You go!

[bugadani]

I think in places it's really obvious that I'm working with a bitbanging implementation on the other end, so I can pretty easily affort 1-bit sequences in the API - for my case. Not sure if that is acceptable for more sophisticated implementations, but regardless of that, this PR needs some debugging work before I can start thinking about that :)

This PR is largely based on the CMSIS-DAP library that the RPi debugprobe sources pull in (which then don't end up using JTAG...), and they have largely been mechanically transcribed. As that wasn't particularly easy, I expect the code to have issues, although right now my setup seems broken and I can't even make sense of the logic analyzer trace I captured yesterday.

My next step is to get an ESP32-based implementation working, so that I can attach a debugger without fear of tearing off test points from the RPi debugprobe, hopefully that will make it easier to find out what goes wrong and where.

[bugadani]

Fixed a few things, probe-rs can now output at least something:

 INFO attach: probe_rs::probe::cmsisdap: Using protocol JTAG
 INFO probe_rs::probe::cmsisdap: JTAG DR scan chain detected as 64 bits long
 INFO probe_rs::probe::cmsisdap: JTAG IR scan chain detected as 9 bits long
 INFO probe_rs::probe::common: Found IDCODE: 0x4BA00477 (ARM Ltd)
 INFO probe_rs::probe::common: Found IDCODE: 0x16468041 (STMicroelectronics)
 INFO probe_rs::probe::common: IR lengths are unambiguous: [4, 5]
 INFO probe_rs::probe::cmsisdap: Configuring JTAG with ir lengths: [4, 5]
 INFO probe_rs::architecture::arm::sequences: Debug port Default is powered down, powering up
 INFO probe_rs::architecture::arm::communication_interface: Debug Port version: DPv0 MinDP: NotImplemented
 WARN probe_rs::architecture::arm::memory::romtable: Component at 0xe0001000: CIDR0 has invalid preamble (expected 0xd, got 0x0)
 WARN probe_rs::architecture::arm::memory::romtable: Component at 0xe0001000: CIDR2 has invalid preamble (expected 0x5, got 0x0)
 WARN probe_rs::architecture::arm::memory::romtable: Component at 0xe0001000: CIDR3 has invalid preamble (expected 0xb1, got 0x0)
 WARN probe_rs::architecture::arm::memory::romtable: Component at 0xe0000000: CIDR0 has invalid preamble (expected 0xd, got 0x1)
 WARN probe_rs::architecture::arm::memory::romtable: Component at 0xe0000000: CIDR1 has invalid preamble (expected 0x0, got 0x1)
 WARN probe_rs::architecture::arm::memory::romtable: Component at 0xe0000000: CIDR2 has invalid preamble (expected 0x5, got 0x1)
 WARN probe_rs::architecture::arm::memory::romtable: Component at 0xe0000000: CIDR3 has invalid preamble (expected 0xb1, got 0x1)
 WARN probe_rs::architecture::arm::memory::romtable: Component at 0xe0040000: CIDR0 has invalid preamble (expected 0xd, got 0x0)
 WARN probe_rs::architecture::arm::memory::romtable: Component at 0xe0040000: CIDR2 has invalid preamble (expected 0x5, got 0x0)
 WARN probe_rs::architecture::arm::memory::romtable: Component at 0xe0040000: CIDR3 has invalid preamble (expected 0xb1, got 0x0)
 INFO probe_rs::architecture::arm::sequences: Powering down debug port Default
ARM Chip with debug port Default:

Debug Port: DPv0, Designer: <unknown>
└── V1(0) MemoryAP
    └── 0 MemoryAP (AmbaAhb3)
        ├── 0xe00ff000 ROM Table (Class 1), Designer: STMicroelectronics
        ├── 0xe0001000 Generic
        ├── 0xe0000000 Generic
        ├── 0xe0040000 Generic
        └── 0xe0041000 Cortex-M4 ETM   (Coresight Component)

Designer: <unknown> indicates something is still not quite right, though - SWD prints Debug Port: DPv1, Designer: ARM Ltd.

[bugadani]

Hmm I think I got this right, just DPIDR comes back as 0 over JTAG for whatever reason. This chip is iffy (SWD seems to be sampled incorrectly) so maybe I should expect anomalies like this.

I was able to run an embedded-test test suite, so I guess it's time to start cleaning this up :)

[bugadani]

Let's see what we can do with this :)

 cargo probe-rs run --chip stm32g431kb --protocol jtag "C:\_Hobby\rpi-smoke-tester-setup\embedded-test\tests\stm32g431kb"
    Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.57s
     Running `target\debug\probe-rs.exe run --chip stm32g431kb --protocol jtag C:\_Hobby\rpi-smoke-tester-setup\embedded-test\tests\stm32g431kb`
      Erasing ✔ 100% [####################]  18.00 KiB @  42.22 KiB/s (took 0s)
  Programming ✔ 100% [####################]  18.00 KiB @  24.01 KiB/s (took 1s)                                                                                                                                                                
     Finished in 1.18s

running 1 test
test test::test ... ok

test result: ok. 1 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.06s

 cargo probe-rs run --chip stm32g431kb --protocol swd "C:\_Hobby\rpi-smoke-tester-setup\embedded-test\tests\stm32g431kb" 
    Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.61s
     Running `target\debug\probe-rs.exe run --chip stm32g431kb --protocol swd C:\_Hobby\rpi-smoke-tester-setup\embedded-test\tests\stm32g431kb`
      Erasing ✔ 100% [####################]  18.00 KiB @  45.91 KiB/s (took 0s)
  Programming ✔ 100% [####################]  18.00 KiB @  28.35 KiB/s (took 1s)                                                                                                                                                                
     Finished in 1.03s

running 1 test
test test::test ... ok

test result: ok. 1 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.15s

[korken89]

Very nice! Is there also a reference implementation in e.g. the Rusty Probe's firmware? I'm thinking about how I'll test the PR.

[korken89]

Do you have a reference implementation I can port to the Rusty Probe? I think that should be enough to validate this as working. :)

[bugadani]

https://github.com/bugadani/fruitfly

[korken89]

I've started the implementation here: probe-rs/rusty-probe-firmware#31

[bugadani]

Sorry Emil, I don't really want to merge PRs without you, but I don't want to force you into writing code for this, either. I also trust the PR enough to think it's useful in the state it is - I've used this to interface with multiple devices using multiple debug probe implementations and I'd like to move on to other work. I don't expect this code to be perfect, but I think I've reasonably matched ARM's implementation. I believe we can improve everything iteratively as we need to.