Well, if you have it ported over, then that's good...something I don't have to mess with. All I need to do is add my RAM Mut table and any other maps to RAM.

Also, fyi, John posted the latest livemap exe that includes ms in the logging. I have tested that and it's working fine.

 

I think that I get it now! You are comparing the load versus your MAP output in kpa. Right?

In some areas, it becomes linear but on others the correlation is kinda skewed as expected.

Do you mind sharing the final tables for reference?

Thanks in advanced!

 

Sure here are my tables from another post

 

I'll have to give it a spin; did he include source with the new version? (I seem to recall there was some problem with the charting stuff and redistribution.) If there's no source, I'll have to hack that into the (old) version I'm using, since his builds don't play nicely with small (800x600) displays. It was a bummer realizing that the swanky new HP/TQ estimator didn't work with my old 1-second-resolution logs.

Testing of DMA logging at lunch was a wash; no normal functionality broke, but the ECU reboots whenever the client initiates a DMA connection init. I fat-fingered at least tephra's DEAD address, so there's probably at least a couple of other typos staring at me right now.

 

Yes, John included the source code. I think it's in the 96530006 livemap thread, towards the end.

 

Okay, it's still randomly rebooting on me during comm initialization (MUT logging still works fine, and everything else looks peachy; I'm still impressed at how low-impact this addition is). Here's what I've changed in 96530706 to add John's DMA code, for anyone else who might want to peek at this as well:

Interrupt handlers:

• 0x138: DMAC3 DEI3 vector changed from 0x9c84 to 0xed10.
• 0x324: SCI0 RXI0 vector changed from 0xd5d4 to 0x3ec000 (location of John's new code).

DMA-interruption code changes:

• 0xd46e: Changed "tst #1, r0" (c8 01 8b) to "cmp/eq #0x37, r0" (88 37 89).
• 0xd6dc: Changed 0xffffeccc reference to 0xffff8480 (DMAOPFLAG).
• 0xdc26: Changed "mov.b r0, @r10" (2a 00) to "nop" (00 09).
• 0xe9e6: Changed "mov.l r10, @r11"(2b a2) to "nop" (00 09).

Entry-point adjustment:

• 0xefac: Changed 0x3e000 (tephra's entry point) to 0x3eeb0 (John's entry point).

New code:

• Copy John's chunk of code at 3ec00 through 3ef5f. I've actually gone to the trouble of re-assembling this now, and the assembled version matches what I ended up with through hex editing. Go figure. If you don't re-assemble his source, you need to make the following changes:
• 0x3ef18: Changed 0xffffa800 to 0xffff841e (tephra's DEAD address).
• 0x3ef30: Changed 0x3f000 to 0x3e000 (tephra's entry point).
• 0x3ef34: Changed 0x38b00 to 0x37b00 (new start of alt-map location).

Tephra adjustments:

• 0x3e1d4: Changed 0xffff8984 reference to 0xffff898a (this is a WTF for me, but it was changed in John's version of 96530006).
• 0x3e640: Changed 0xffff841c reference to 0xffffa800 (another WTF, same thing).

Alt-map relocation to RAM:

• 0x3e7d4: Changed 0x37b40 to 0xffffa040 (alt injector scaling). This isn't strictly necessary, since I don't think the client actually supports scaling adjustments right now, does it? (Anyone want to write a feedback-driven injector latency and scaling tuning tool using this? )
• 0x3e7dc: Changed 0x37b42 to 0xffffa042 (alt high-octane fuel).
• 0x3e7e8: Changed 0x37c82 to 0xffffa182 (alt high-octane timing).
• 0x3e7f4: Changed 0x37e42 to 0xffffa342 (alt BWGDC).
• 0x3e7fc: Changed 0x37ec2 to 0xffffa3c2 (alt BDEL).

Other stuff:

• MUT0 (or something similar) should probably be changed to 1-byte load.
• In the client, address.xml needs to be changed so that the high-octane timing map is at a182, rather than a2cd.
• In the client, blockaddress.xml needs to be changed so that the "ROM address" is at 0x37b0, not 0x038b0.

I've attached the assembly source I'm working with, if anyone wants to see it. (It's not much different from what John originally posted, just reformatted a bit so the binutils sh assembler likes it.) I'm not providing a pre-patched ROM for obvious reasons right now. jcsbanks, l2r99gst, mrfred, tephra, or anyone else who feels like taking a peek at this, can you see anything obvious that I've missed? I'm particularly curious about the changes to those two locations in tephra's code; I haven't dug into what's happening there enough to know why he changed that.

 

Where is this being done in the v5.1 dma-livemap ROM? I am looking at that in IDA Pro, but don't seem to see it. I thought John just copied this whole block to RAM and didn't think the alt maps in this location needed to be explicitly pointed to RAM.

I know if you want to copy another map to RAM, like injector latency, then you have to change the pointer to the RAM address based on where it is copied to that block, though. I just asked John that question actually.

Of course, I'm not that good with this stuff, so if this was a dumb question, feel free to let me know.

 

In 96530006, those changes were at:

• 0x3f650 (injector scaling)
• 0x3f654 (high-octane fuel)
• 0x3f65c (low-octane fuel)
• 0x3f664 (high-octane timing)
• 0x3f66c (low-octane timing)
• 0x3f674 (BWGDC)
• 0x3f67c (BDEL)

He copies that block of ROM to RAM, but then you still have to fix up any individual references. The addresses change a bit from v5.10 to v7, because we lost the low-octane maps.

If you're wondering how I found that, I just did a diff against a hexdump of a stock v5.10 vs. John's version, and walked through the list of changes to see which was which. That's where I found the two WTFs in my list above.

 

Thanks. I never did the hex compare yet. That's what I was going to do to compare John's changes, but he said he was going to send me an XML with just the livemap/DMA changes, so I never got around to that yet.

If I have some time, I will try to walk through what you have done to see if I can see anything obvious. I doubt I will, but sometimes I surpirse myself, but for the most part I'm just dangerous enough to mess around with the ROM in IDA.

I have to fix my damn omni power sensor first, since I got one of the 'bad' ones where the sensor plug wires need to be switched.

Edit: you mean alt injector scaling, not latency.

 

Well, I never did get this working; E0 seems to be the cause of the random reboots. I'm out of time to look at any of this until next Tuesday, as I'm out of the country, so if someone else wants to pick up where I left off, you won't be duplicating effort.

And it occurs to me that I shouldn't be clogging up mrfred's thread about speed density with this, should I?

 

Have you tried Mr Freds patch for SD yet, if so how is it working?
Ray

 

I've been very busy at work. I'm sending him a patched ROM this evening to try.

 

Great!!!! thanks for all your hard work.

 

No rush, mrfred. I won't be able to test until the weekend anyway. If I get a ROM by then, I will be happy to test it out.

I will be out of town mid-week until the next week, so if the ROM isn't ready by then, feel free to send to another eager tester.


Eric

 

I will gladly test a tephra v5.10 version of this rom if you'd like to email it to me or pm me a link. My email is tom04841@gmail.com