I have custom Cortex silicon here. Our device is not on the enumerated lists and it cannot search for the RTT block. It was a problem to use the RTT Viewer in that it couldn't auto-locate the RTT, but we got it working. We ended up putting the RTT block in an explicitly stated memory location and typing that in when launching RTT Viewer.

Now, we need to run J-Scope from RTT. However, unlike RTT Viewer, I don't see any option to explicitly state where the RTT block is. I even tried manually editing MCU.csv and adding a line for our device bit that didn't work, for whatever reason the .csv doesn't change the list on RTT Viewer/J-Scope.

Can I create a jscope_project file and just edit in the RTT address and reload that? What's the XML for the RTT address?

Do I launch Jlink Commander and just use the connection for the J-Scope? How do I tell the Commander an explicit RTT address?

How can I use J-Scope with RTT?

OK, got it loaded!

What I saw is that initially we had a byte reading out ok at 5Ksamp/sec.

Increasing the sample rate to 10 Ksamp/sec means JScope will NOT start. I load, set up the JTAG, press the red "record" circle, it will find the RTT buffer and crash "JScope.exe has stopped working" in a flash.

MCU code is using SEGGER_RTT_MODE_NO_BLOCK_SKIP with a 5K RTT buffer. My first thought is the buffer overruns, which should only cause lost samples in NO_BLOCK_SKIP mode, but I could see where maybe that breaks JScope.exe.

But:

segger.com/jlink-rtt.html

Say JLink base model (I have JLink OB on my desk through USB, should be USB 2.0) @15MHz is 700KB/sec. USB 2.0 presents a 60MB/sec limit so that's not a factor. Up buffer is 5KB, so that shouldn't be a factor. But 5KB/sec (5000 byte samples) works, 10KB/sec crashes every time for me. Even at the 4 MHz we normally use for JTAG here, no way is bandwidth a limit in this situation, right?

I tried changing to SEGGER_RTT_MODE_BLOCK_IF_FIFO_FULL for the buffer. Same thing, it finds the RTT control block, then immediately crashes JScope.exe.

Any idea why we're crashing at this speed, and/or how to fix it?

2 side questions:
1. Is it possible to add the search range to the project file XML? I see that it did not store anything about the address range yet.
2. Can I specify a project file on the jscope.exe command line?

Trying Jscope v511. Nope, v511 will crash as soon as it discovers the RTT block, if more than 5ksamples/sec (bytes) are being sent. However, there's no "jscope.exe has stopped working" message. The whole jscope window just closes and disappears instantly.

Where is the log file? Nothing was generated in \program files (x86)\SEGGER\JScope_V511\

I send byte data for simplicity.

p->aUp[1].sName = "JScope_u1";

SEGGER_RTT_Write(1, &dataArray[0],1);

I did not know the JLink OB was slower than the JLink Base. I don't have the Base on my desk right now. However, I'm sure it's acting the same way.

Further investigation:

The JScope will crash if the buffer is filled prior to its starting up. That is the normal situation- the MCU is reset, the code runs, it puts samples in the RTT buffer, and it will almost certainly be filled up before JScope starts recording. Then we run JScope.exe. You can't press "record" on JScope until the MCU is running, there will be no RTT buffer to find.

When I added a 5 sec delay before putting samples on the RTT, then JScope.exe didn't crash upon loading (the 5MB/sec transfer problem does not exist, there is no transfer). A couple of seconds later the MCU starts loading the RTT buffer and I saw a 37Msamp/sec transfer rate!

Which is odd... without the startup delay, the 5K RTT buffer is pretty much guaranteed to be filled even with the 5Msamp/sec rate that does NOT crash JScope. JScope finds a 100% full buffer upon launch regardless of sample rate. It's only if the RTT buffer is full AND the incoming new sample rate is >5Msamp/sec that it crashes immediately.

Wait, not quite. This works if the RTT buffer is SMALL, 1KB. When I upped the buffer size to 2KB, then JScope DID start recording, it displayed a short window of samples with 47Ksamp/sec, but then crashed.

Just discovered that even if JScope gets started without crashing, it will still crash within a few minutes, unless the bitrate is kept low.

Weirdly the RTT Sample rate shown is very unstable. 28K, 26k, 30K, 24K, etc. It's off an ADC with a very regular sample rate.