Improve align-to-magic-word

[JuantAldea]

This commit implements a byte-by-byte alignment approach.
It consists on a state machine that consumes the UART output byte-by-byte, and that will only accept input of the form:
0xAB 0xCB, 0xXX, 0x00 .

To me, this has greatly reduced the amount of retries I had to perform before flashing the walkie.

Additionally:

• Corrects several memory issues (segmentation faults & leaks).
• Reduces timeout lengths as they hinder the retry-realign process.

Tested on Linux with

Bootloaders:

• 4.00.01 (UV-5R Plus)
• 2.00.06 (UV-K5)

USB adapters chipsets:

• CH341
• PL2303

[JuantAldea]

This greatly reduces the amount of retries and ceremonies I had to perform for flashing the radio.

[sq5bpf]

There have been already patches that improve syncing with the 0xab 0xcd sequence, so obviously there is some problem, but i can't reproduce it.

Please tell me what you're doing that causes these out of sync problems, so that i can test these patches? So in other words: please help me replicate the errors, so that i can see the improvement from your patches.

I use the usual baofeng-labeled cables with CH341 under debian 12. To flash I do ptt + turn on, connect the cable and run k5prog. And it works every time. How do i reproduce your problems?

[JuantAldea]

I don't do anything special as far as I know. Just turn it on and try to flash. With the patches you mention, patches that I already have, usually I have to either kill the programmer and restart it, or restart the radio, or both.

[JuantAldea]

For instance, although maybe not so canonical:
Start the programmer first, then turn the radio on in flash mode. It won't flash, at least I get a "k5_receive: bad magic number", and stay there forever (until it times out...?).

That scenario works with my changes.

Many cases implying a partial flashing, i.e., cancelling it with control+c, would lead to errors when trying to start another flashing, sometimes would even lead to a chain of errors (maybe something's left in the buffer of the serial converter?).

Stopping a flashing and starting another one is something I often do while developing.

[ntoivola]

For what it's worth, I experience the "bad magic number" issue quite frequently but have not been able to find any logic as to when it happens or what I might have done to trigger it.

However, so far I have not experienced it once since applying this patch, so a big 👍 from here.

I have the 2.00.06 bootloader and my cable shows up as "QinHeng Electronics CH340 serial converter" in lsusb.