Basic compatibility for STLink-v3 programmer

[Nightwalker-87]

Added support for JTAG command API v2:
--> compatibility with the STLink-v3 firmware which dropped support for the previous API v1
--> as of firmware version J11 the ST-LINK-V1 supports API v2 commands as well

As described by @martonmiklos in #863: "Apart from utilising the V2 command API there are some other minor differences in the v3 handling like extended version query, which generally seems to do it's job. At the moment most of the commands are working, however the flash loaded never hits a breakpoint after loading as I mentioned in #820."

(Closes #1023)

[Nightwalker-87]

This PR is going to be merged next as soon as it is finally approved.

[chenguokai]

Mostly good, at least backward compatibility has been kept.
There is one warning at the end:
libusb: warning [libusb_exit] application left some devices open

Not sure if this warning is caused by this PR

[Nightwalker-87]

I don't know really - if so usb.c would be of interest, but I haven't found anything there yet.
@chenguokai does it happen on develop as well in your test case?

[chenguokai]

does it happen on develop as well in your test case?

No, at least with commit ID 4c1f5b9575a90178a9f8fd29b64ccbbb00953ec9

[Nightwalker-87]

I'll have a look. What did you do exactly? How can I try to reproduce?
The flash and erase procedures (for example) conclude successfully here without any additional notifications.

[chenguokai]

How can I try to reproduce?

Edit: Warning shows even without --debug, my test command is

./st-flash write ~/Documents/stm32/dev/stm32_clang/build/stm32.bin 0x8000000 

st-flash 1.6.0-302-g4c1f5b9
2020-05-16T20:30:17 INFO common.c: F1xx Medium-density: 20 KiB SRAM, 64 KiB flash in at least 1 KiB pages.
file /Users/cgk/Documents/stm32/dev/stm32_clang/build/stm32.bin md5 checksum: 8f79575b232c1ffe40a2f9afff75a7a6, stlink checksum: 0x000a8934
2020-05-16T20:30:17 INFO common.c: Attempting to write 9272 (0x2438) bytes to stm32 address: 134217728 (0x8000000)
2020-05-16T20:30:17 INFO common.c: Flash page at addr: 0x08000000 erased
2020-05-16T20:30:17 INFO common.c: Flash page at addr: 0x08000400 erased
2020-05-16T20:30:17 INFO common.c: Flash page at addr: 0x08000800 erased
2020-05-16T20:30:17 INFO common.c: Flash page at addr: 0x08000c00 erased
2020-05-16T20:30:17 INFO common.c: Flash page at addr: 0x08001000 erased
2020-05-16T20:30:17 INFO common.c: Flash page at addr: 0x08001400 erased
2020-05-16T20:30:17 INFO common.c: Flash page at addr: 0x08001800 erased
2020-05-16T20:30:17 INFO common.c: Flash page at addr: 0x08001c00 erased
2020-05-16T20:30:17 INFO common.c: Flash page at addr: 0x08002000 erased
2020-05-16T20:30:17 INFO common.c: Flash page at addr: 0x08002400 erased
2020-05-16T20:30:17 INFO common.c: Finished erasing 10 pages of 1024 (0x400) bytes
2020-05-16T20:30:17 INFO common.c: Starting Flash write for VL/F0/F3/F1_XL core id
2020-05-16T20:30:17 INFO flash_loader.c: Successfully loaded flash loader in sram
 10/10 pages written
2020-05-16T20:30:17 INFO common.c: Starting verification of write complete
2020-05-16T20:30:17 INFO common.c: Flash written and verified! jolly good!
libusb: warning [libusb_exit] application left some devices open

FYI, my libusb version is 1.0.23 on macOS.

[Nightwalker-87]

Uh, I can't reproduce that. This is exactly what I tried here. Could it be that this is related to clang somehow? I use gcc on Debian. That is the only difference I see currently (or it is OS-specific).

[chenguokai]

Trying gcc on macOS only to get a compile error🤔

stlink/src/flash_loader.c:194:9: error: 'size' may be used uninitialized in this function [-Werror=maybe-uninitialized]
  194 |  size_t size;
      |         ^~~~
cc1: all warnings being treated as errors

Edit: with some quick fix, I compiled stlink with gcc, the warning persists.

I will try on my raspberry pi with raspbian.

Edit: on my raspbian, the error still exists.

st-flash 1.6.0-307-gd827ce1e-dirty
2020-05-16T21:36:02 INFO common.c: F1xx Medium-density: 20 KiB SRAM, 64 KiB flash in at least 1 KiB pages.
file /home/pi/stm32.bin md5 checksum: 8f79575b232c1ffe40a2f9afff75a7a6, stlink checksum: 0x000a8934
2020-05-16T21:36:02 INFO common.c: Attempting to write 9272 (0x2438) bytes to stm32 address: 134217728 (0x8000000)
2020-05-16T21:36:03 INFO common.c: Flash page at addr: 0x08000000 erased
2020-05-16T21:36:03 INFO common.c: Flash page at addr: 0x08000400 erased
2020-05-16T21:36:03 INFO common.c: Flash page at addr: 0x08000800 erased
2020-05-16T21:36:03 INFO common.c: Flash page at addr: 0x08000c00 erased
2020-05-16T21:36:03 INFO common.c: Flash page at addr: 0x08001000 erased
2020-05-16T21:36:03 INFO common.c: Flash page at addr: 0x08001400 erased
2020-05-16T21:36:03 INFO common.c: Flash page at addr: 0x08001800 erased
2020-05-16T21:36:03 INFO common.c: Flash page at addr: 0x08001c00 erased
2020-05-16T21:36:03 INFO common.c: Flash page at addr: 0x08002000 erased
2020-05-16T21:36:03 INFO common.c: Flash page at addr: 0x08002400 erased
2020-05-16T21:36:03 INFO common.c: Finished erasing 10 pages of 1024 (0x400) bytes
2020-05-16T21:36:03 INFO common.c: Starting Flash write for VL/F0/F3/F1_XL core id
2020-05-16T21:36:03 INFO flash_loader.c: Successfully loaded flash loader in sram
 10/10 pages written
2020-05-16T21:36:03 INFO common.c: Starting verification of write complete
2020-05-16T21:36:04 INFO common.c: Flash written and verified! jolly good!
libusb: warning [libusb_exit] application left some devices open

[Nightwalker-87]

Ha, I got it now! There appeared to be some remnants of an old install somewhere which are now cleaned up. It happens with st-info --probe also. I found this on the web:

https://stackoverflow.com/questions/44919347/libusb-cant-claim-device

[chenguokai]

I am a bit confusing. So under what circumstances will the warning appear? And what does the old install mean, a registration of stlink hardware from another driver?

[Nightwalker-87]

I am a bit confusing. So under what circumstances will the warning appear? And what does the old install mean, a registration of stlink hardware from another driver?

I had an old commit installed locally with sudo make install what I lost track of. So when I did rebuild a later version, the terminal commands were still linked to that installed version of the toolset, so I did not notice at first.

The warning seems to appear whenever there is a cmd sent via usb. I haven't found the commit when it first appeared. It is somewhere in the stlink-v3_pre branch.

[Nightwalker-87]

The problem occurred in commit 0a91936.