<PaulFertser>
PlasmaHH: I can provide some background on what's happening.
<PlasmaHH>
ah well... I have been sleeping most of the time, always when I Have time to relax I get some nice virus and fever :P ...
<PaulFertser>
PlasmaHH: so it runs an "async" algorithm for flashing on target. That means there's a circular buffer with two pointers. One pointer is adjusted by the algorithm as it reads and flashes the data from it, and another is updated by OpenOCD as it uploads the data to RAM in parallel.
<PlasmaHH>
PaulFertser: that would be nice... at about line 155 or so it seems that it is stuck and times out and I think somewhere in there was a hint of a hardfault
<PaulFertser>
PlasmaHH: and I see this target is more trickyng than most.
<PaulFertser>
PlasmaHH: btw do you get consistent results if you use "program" in OpenOCD and not "load" in GDB?
<PlasmaHH>
PaulFertser: I can test that, I had the feeling that program always works better, but that might have been because I always used it right after a failing load (where the second load would normally succesd too)
<PaulFertser>
PlasmaHH: it's supposed that target algorithm is run directly after "reset init" (which is "reset halt" and then the reset init Tcl proc). But in this case gdb-flash-erase-start event is overridden and I do not see it including reset init.
<PaulFertser>
But this script isn't doing that at all.
<PaulFertser>
I would expect target/stm32x5x_common.cfg "gdb-flash-erase-start" handler to begin with "reset init".
<PlasmaHH>
so the loader will be copied into 0x20000000 and then started... where can I find the code for it so I can look in the controller if its there? to figure out what state the controller is currently in
<PaulFertser>
The code is originally from c9d40366ad55ea3a83f2ff438aab1e62da653169
<PaulFertser>
PlasmaHH: I suggest you first try with doing "mon reset halt" prior to "load" to see if it makes any difference.
<PaulFertser>
Because it feels like the current config is plain incorrect.
<PaulFertser>
"mon reset init" even.
<PlasmaHH>
I have now tried 5 times a sequence of operations with "load" that always fails, then three times with "monitor program" which always succeeded
<PlasmaHH>
alright will try
<PlasmaHH>
indeed, doing a "mon reset init" will let the first load succeed
<PaulFertser>
PlasmaHH: feel free to propose a patch to hahve target/stm32x5x_common.cfg "gdb-flash-erase-start" handler to begin with "reset init"
<PlasmaHH>
right as the very first instruction?
<PaulFertser>
PlasmaHH: yeah
<PlasmaHH>
sweet, thhat works fine
<PlasmaHH>
whats the preferred method, through the bugtracker?
<PaulFertser>
PlasmaHH: no, through the Gerrit instance, see HACKING
<PlasmaHH>
alright *adds yet another account to the giant pile*
<PaulFertser>
Hm, don't want to use github oauth for that?
<PlasmaHH>
yeah, didnt have a look at first, oauth is still not being used widely enough
<PlasmaHH>
so much projects doing things in weird ways...
<PlasmaHH>
some even stuck in the 90s where you send bug reports to mailing lists and those are forgotten forever
<PaulFertser>
Many people complain about OpenOCD not providing auth services for Gerrit.
<Haohmaru>
M$ github
<Haohmaru>
i recently submitted a wishlist thing to GCC, i think it's a mailing list
<PlasmaHH>
gcc has a butracker (bugzilla based)
<PlasmaHH>
I think git is still using mailing lists only
<PlasmaHH>
which explains why some bugs stick around for decades tehre
<PaulFertser>
I do not think mailing lists are problematic, it's just a tool and a very usable one, and these days you have Patchwork and b4 and whatsnot.
<PaulFertser>
The bugs stick around not because mailing lists are anyhow bad.
<PaulFertser>
Haohmaru: our Gerrit supports some other popular OAuth too and also any OpenID instance, you can run your own (and I do).
<PlasmaHH>
PaulFertser: how do bugs sent to mailinglists months or years ago get any visibility? There is no "unresolved bugs" overview like in bug trackers
<PlasmaHH>
esepciallyin bigger projects things easily get drowned in the noise
<PaulFertser>
PlasmaHH: bugs are supposed to be picked up by bugtracking systems like Debian's. And patches by Patchwork.
<PaulFertser>
PlasmaHH: I do not remember any big project which lacks some issue tracking system.
<PaulFertser>
I know ncurses but there the maintainer is very attentive and keeps track of everything from the ML.
<PlasmaHH>
a lot of stuff in linus orbit is unusual ;)
<PaulFertser>
PlasmaHH: guess this project is going forward in a special way where if someone doesn't insist on bug reporting and fixing it's just not happening and that's fine for them.
<PlasmaHH>
possibly, nevertheless very annoying
<PlasmaHH>
especially if in the company at least once per week someone is struck by that bug but people are often not proficient enough in git to get out of that situation themelves
<PaulFertser>
Well then probably one should be sending a patch rather than a bug report.
<PaulFertser>
Your patch got it to Gerrit (and the mailing list), thank you!
<PaulFertser>
You might want to add Tarek to reviewers. Not sure if he's still active though.
<PlasmaHH>
PaulFertser: thats one of the things I dislike about how oss developed in the last one or two decades. back then people were thankful for bugreports and feature ideas and we wall knew that people only had so much time on both sides. these days its so often "send patches instead of bugreport or feature request" and people get all annoyed by you,
<PlasmaHH>
forgetting that also the reporters have such much time, no experience with the project or programming language (or just outright dislike that language) etc... the world has become so entitled and needy on all sides, and it got worse by the increasing number of projects. There are times when I do stuff that I create one bugreport per day for various projects
<PlasmaHH>
As a new contributor I rather not try to attach work to someone specific, don't want him to get mad, even though its just a small thing ;)
crabbedhaloablut has quit []
crabbedhaloablut has joined #openocd
PlasmaHH has quit [Read error: Connection reset by peer]
PlasmaHH has joined #openocd
PlasmaHH has quit [Ping timeout: 252 seconds]
PlasmaHH has joined #openocd
<PaulFertser>
PlasmaHH: I do not think there are many sane developers who do not understand how much time and effort a good bug report takes and do not appreciate that.
<PaulFertser>
PlasmaHH: but with many projects they indeed often do not have the capacity to work on something that's not affecting their own immediate goals. While still having some/enough time to review contributions.