EFIDynoTuning Forum

Note that 5.0.5 at the moment is a debug build, it creates a file called "<name>_dbg.txt. I was going to fix it, but my virtualbox just stopped working after a minor linux upgrade (this is never supposed to happen...) so it may take me a while to sort out the mess. It won't remove cleanly as it gets an error, and it won't install/reinstall as it gets an error..... so stuck in the middle somewhere.... web doesn't have a solution (yet).

crash in funcs -

jamie,wwhite - I have a 6DBX but not a 6DBD. I tried jamie's dir with 6bdx on my linux system. Doesn't crash, BUT it doesn't list properly, with no error reported about the command - so something weird is happening... looks like table at cd94 just continues to run over the funcs. (I can see that 6dbx bin has funcs at ce22 and ce3e.) So something to find in there. Maybe a win versus linux difference for the crash (happens sometimes) Will look ASAP.

jsa - lookup funcs were being recognised, but subr names got replaced with standard names when they should not have been - fixed in 5.0.5 I hope, didn't fully test as pushed out 5.0.5 to fix my stupid mistake (above)

wwhite wrote: ↑2024 Jan 09, 20:48 I get exact same seg fault as before using 5.0.4

Code: Select all

Reading symbols from SAD_5.0.4.exe...
(gdb) run
Starting program: C:\Users\wwhite\Programming\SAD\SAD_5.0.4.exe
[New Thread 14704.0x3b80]
KID2

Thread 1 received signal SIGSEGV, Segmentation fault.
0x0040c25a in add_spf_register (p=0xce1c90, dest=58, sce=0, popaddr=622135)
    at C:/CLProjects/SAD/WSAD/Core.cpp:6205
warning: Source file is more recent than executable.
6205
(gdb) bt
#0  0x0040c25a in add_spf_register (p=0xce1c90, dest=58, sce=0,
    popaddr=622135) at C:/CLProjects/SAD/WSAD/Core.cpp:6205
<snip>
(gdb)

Fabulous and many thanks - that's recognisable to me as to where the crash is, and it's... A SCREW UP by me.

I put a temp check in (as in the release docs), and then... I managed to remove it again before rebuilding SAD on Windows.... <swear words>

Sorry about that . OK, I'll fix that and do a 5.0.5 quickly right now.

"well hey, we all make mistakes" said the Dalek climbing off the dustbin......... (from a British comedian...)

Guys,

5.0.4 released. I've already spotted a problem with comments causing bit/field names to get messed up. It fixes one crash, but sounds like there's others lurking in there too.

General Note - I always start testing with AA (it's probably the simplest bin in existence) and then to A9L (test subroutine arguments) and then XDT2 (a 2 bank bin) as a progression of more features which have to work, and 'get basics right first' approach. So the bigger 4 banks get to wait a bit....

I will have a look at RZASA and see if I can spot why lots of code is missed - and see if v4 does it too.

Again - I did have a 'gap' scanner to try to pick up lost code, I have not put it back yet....maybe in next release after crashes fixed. I will also put in a "press any key to continue" for windows (temporary) to see if you get any useful info for crashes (I reckon I know how to do that...).

BOOSTEDEVERYTHING wrote: ↑2024 Jan 07, 09:30

Code: Select all

FUN 12254 122CB "uuwFn036_MAF_Transfer" :W V 13107 :W V +1 P +5      # Mass Air Flow meter transfer function. CDAN 24-6

Can't see anything wrong in there - or with jamie's commands.
tried them out but it doesn't crash for me ........
Guys, do you know if SAD starts processing and THEN fails ?? if so it's doing something wrong in the data processing, not the command processing.

......OK more digging to do.

NB. I think I've found crash reported by jsa, KID2 etc.

You mean like a "name2" is duplicate of "name1" ?? yep, could look at that..........
I was wondering if it should apply for other commands too, like " Word 1234 overlaps with byte 1235" , where it makes sense.
(note that code -> code or word ->word etc. will just get merged together)

General - if you get a crash with a command (dir file), and not if you remove it, it's possible the crash is when parsing the command, or what you're asking for breaks SAD. So I might ask for your dir file if I can't find why. With Windows, a program can just disappear... I may have to add a "type any char to continue" so you can read what happened..

I want SAD to never break - even if you feed it garbage it should reject it with an error, not crash.

missed parts - SAD can't always get all the code and data out by scanning, as there are what I call 'sneaky' tricks. (it's perfectly legal code to write, just very hard to process automatically). I did do a scan of the 'gaps' in some v4, but I switched it off because it produced a lot of 'false code' which was actually data. (no it's not always easy to tell the difference...). So for now, yes, may miss bits. Intend to put that back later, also some better table fixes.
if you are sure a block is code, you can do
SCAN <address>
to kick of processing for that block, and anything else that block CALLS as subroutines.... if it's not code it (should) will mark it as invalid and continue.

Still working though bugs............

Thanks guys,

All bugs welcome. Don't apologize for finding them !

I have found a bug which can cause a crash, and several bit flag listings are wrong. Other stuff ongoing too.

Will get rid of the stars in the file in github (didn't realise it screws up git operations....)

Note that the command

SYM 0015 "AICE_A_D_CTRL" [W ][B4 ]

IS WRONG AND ALWAYS HAS BEEN - It's just that v4 didn't check properly and v5 does, so it sneaked through and I didn't spot it.

It should be

SYM 0015 "AICE_A_D_CTRL" [W B4 ] # ONE extra definition -> bit flag 4 is a WRITE symbol
or
SYM 0015 "AICE_A_D_CTRL" : W B4 # fixed in original format , drop the second colon

Considering I rewrote a LOT of SAD code, it's doin' OK, but not great.....I thought I had tested stuff better, but obviously not....

Thanks again - trying to get more data so that SAD can recognise the timer structure in more bins.
The calculation, ' IO_Timer to millisecs' is the first foundation to build on.
So far, looks like early bins do 1/1000 (true millisecs) and later bins do 1/1024 to make maths easier/quicker to get to other fractions of secs.

(released 5.0.3, see other thread)

Fixed a list of stuff reported, and released it as 5.0.3. More to do, but hope that fixes make SAD more useful right now.

Thanks for your responses to help make this a better product.

Guys,
does anyone have a list of clock speeds for the various units/bins ? There are 9 speeds in the Ford manuals, but I reckon there are more.

I am trying to decide what to do with SAD for the timer setup. It's a really important part, as EVERYTHING has timers all over the place. So any information welcome, even if it's just "My XXXX has a x Mhz crystal in it" If you have any pictures of boards with the crystal on them, will be useful.

jsa has suggested that later bins like RZASA may have a 27Mhz crystal, which would be normal progression from the 20Mhz maximum in the manuals.

Thanks.