#69 amdstatus() broken?
While trying to program connected to a BF537 processor, I noticed it takes _ages_ and seems to spend a lot of time in the loop in amdstatus() up to line 188 (that polls for the status bit after writing to flash). At least that's where the debugger always stops when interrupting the program by keypress.
It takes between 300 and 800 iterations of the loop before the status is accepted. For each iteration, an usleep(100) is made. That means even if usleep takes only the (guaranteed) minimum time of 100 us to execute, it's 30 to 80 milliseconds per written word, or no more than 33 words per second.
Does it really take that long until the write process is completed or do we have a problem with the detection of this toggling bit?
Might also be related to bug 1915338?
Might also be related to JTAG activity queuing?
Logged In: YES
user_id=478715
Originator: YES
Maybe a board problem, happens after a while only. But this is a verified(?) BF537 STAMP from ADSP...
It seems that in some cases (not always at the same address) the toggling bits are almost NEVER detected
Logged In: YES
user_id=478715
Originator: YES
Doesn't work with 0.8 either
Logged In: YES
user_id=478715
Originator: YES
It works with another (otherwise similar) cable and lower TCK speed. So it is a hardware related and not an UrJTAG bug.