Fred's firmware development diary

[Fred]

I learned a bit more about this same issue tonight. I still need to do some tests though to confirm my theories and I'll probably post to the hc11 mailing list too and see what the resident gurus think about it. The linker builds an elf file that knows that there is a flash version and ram version. It knows how big each is, but it seems to use the wrong size to decide if there is enough flash space or not...

I note that two of my vars are 0x0000 and 0xFFFF and both are written out to the flash in the newly included section. This points to it being a fault in the linker or configuration thereof.

Bed!

Fred.

[Fred]

Solved! I closed my first mantis bug this morning. I found the issue in the linker script on the train and we now have the extra flash that I said we had. This is good for many reasons :-) I also moved one function to one of the fixed config blocks as an experiment and that (of course) works fine too. So, it turns out we have about 5kB spare flash right now :-)

No stress!

Fred.

[Fred]

So, I've got about 4.5k of free flash right now for development before we need to start paging and a fixed or new loader is required. This is good as it is about 35% more than we are currently using. I should be OK with flash space for at least a month or two more.

Plenty of time for Sean to get busy with the loader project.

Fred.

[Fred]

Huston, we have ignition!

I just used the serial monitor manually with cutecom to blow away a page of paged flash! This means the bug IS with hcs12mem, so the only work needed is on that, or on a GUI replacement!

It's a pretty simple app required, anyone up for the task?

I reckon one of you could knock it over pretty quickly. Maybe later we can do something flash with robust data transfer etc, but for now the quick fix is a patch for hcs12mem or a standalone app to do the job for it.

This will be required in the next few weeks to months, so if someone got stuck in now that would be awesome!

I'll log a task in mantis and we'll see if we can get someone to pick it up :-) Maybe Sean ;-)

Fred.

[Fred]

Ahhh HUH, so, the issue is that hcs12mem uses the erase all function which *should* work, but doesn't.

http://freeems.aaronb.info/tracker/view.php?id=21

It should be an easyish app to write. The ONLY difference from hcs12mem behaviour would be the erase all via many erase page thing.

Fred.

[Fred]

I've just been through my list of serial issues and I've completed about 1/4 of them along with doing a bunch of other unrelated but important stuff today. I think it's possible that I could do a release tomorrow night with full serial comms and documentation for it. Then it's down to business with the hardware interface. I've got both serial docs and hardware docs as number 1 priorities in mantis right now as they both block others work. I'm doing serial first because I'll be held up by a lack of tuning tool soon so the earlier I get others working on that the better. The hardware is lower priority because it will be no use until both firmware and tuning app are fairly complete. It's still pretty high up though, but not as high as the serial stuff.

Post those two issues, the task will be expanding the serial stuff to actually do things such as tune the tables and burn the thermistor lookups etc. The flash burning bug is blocking that work to some extent though. I can also spend more time on math, config, build stuff, actual features, wheel decoding, scheduling etc.

I'm pretty excited about the future right now as the ground work has largely been done and we are getting quite close to something that barely works (as opposed to something that doesn't work).

I'm off to bed and will be working hard on this tomorrow till I get some results or have to sleep again.

Fred.

[Fred]

Today was fairly depressing for a variety of reasons, however I did get some good work done on the serial stuff. I've been working on it all day and redoing significant parts of it to make it right. I should have plenty to do on the train tomorrow morning. Looking forward to meeting Dave in the evening and then some more coding on the way home, and maybe some testing once I get back. If testing goes well I'll be writing up the documents on the train on tuesday and releasing tuesday night. This is highly dependent on the ground work completed tonight and tomorrow though, so we'll see how I go for time and progress...

I'll be annoyed if the release is pushed out back past the 5th as that will mean more than one month :-(

Still, lots of progress wrapped up in it I guess. I'd like to knock this back to smaller chunks of work per release if possible, but sometimes it's just not I guess.

Fred.

[Fred]

1am, dishes, bed.

Getting closer with the serial re-write (yes, I know I only just wrote it). I know what I need to do, just a matter of getting it done and ensuring it is correct.

Tomorrow on the train should see significant work done on it.

Fred.

[Fred]

Got the ISR stuff pretty well sorted tonight. Just need to go through the core stuff and figure out a better error / debug mechanism to use. Should get some testing done tomorrow night hopefully.

Was good to catch up with Dave (bmw) tonight for a beer/wine. I had a small wine. He had 3 pints of guiness :-) I'm getting soft in my old age ;-)

I'll have to catch him again for a bbq with some FreeEMS engines in the medium term!

Fred.

[Fred]

Well, today the penny dropped! I put 2 and 2 together and got 4.

hcs12mem sends the erase all command and the serial monitor erases all .... of one flash block. not all of them....

SO, we now have 128k of flash, not 48k and HEAPS of room to grow. In fact ms2 doesn't use a full 128k anyway, so we should be fine for absolutely ages.

I guess the priority of the loader app just dropped a few notches :-)

Paging of the table data works and works well. I don't like the warnings it gives, but I'll figure out a way to quiet them down or make less of them.

Serial kinda sorta works, which is quite good considering I haven't finished making my changes and its half one way and half the other. I need to work on it some more, maybe tonight.

Fred.