#64871 - Ultima2876 - Tue Dec 27, 2005 11:33 pm
When I call kragInit it seems to generate 2 no$gba errors... anyone know why, what the errors might be, and most importantly how I can sort it out?
I would really love to have a perfect no$ rating ;P
#64878 - kusma - Wed Dec 28, 2005 12:30 am
indeed krawall does, atleast in the free sdk. i believe this is caused by some unaligned memory accesses or something like that, and you cannot resolve those without either having the source code or doing some serious reverse engeneering.
#64879 - Ultima2876 - Wed Dec 28, 2005 12:37 am
Is there an older version that doesn't cause the errors?
#64886 - gauauu - Wed Dec 28, 2005 2:18 am
Just a note though, that errors or no, in my experience, it works fine on hardware.
#64890 - wintermute - Wed Dec 28, 2005 3:07 am
The registered Krawall is worse, unfortunately.
Apex Audio doesn't suffer the same problems and there's a sample for this library in the libgba examples.
#64912 - Peter - Wed Dec 28, 2005 1:59 pm
| wintermute wrote: |
| The registered Krawall is worse, unfortunately. |
Why is it worse? I used Krawall quite a while and was pretty comfortable with it, but if there is something wrong with Krawall I would like to know what and if there are better solutions.
Thanks in advance
#64915 - keldon - Wed Dec 28, 2005 2:24 pm
| Quote: |
| Just a note though, that errors or no, in my experience, it works fine on hardware. |
Isn't this then more of a nocash problem and not a krawal problem? Not to mention the fact that no gba code should be crashing nocash whether or not the code is correct. Maybe you need to make Martin aware of this.
#64917 - Peter - Wed Dec 28, 2005 2:36 pm
| keldon wrote: |
| Maybe you need to make Martin aware of this. |
Martin is aware of that though. I wrote him about a year ago. I got this reply (in german):
| Quote: |
Die freeware version z?hlt einfach die errors, anstelle der debug warning messages, besonders hohe error-counts kann man mit dem goldroad assembler erzielen, oder wahrscheinlich auch wenn man mirrors benutzt, zB stackpointer=3FFFFXX statt 307FFXX.
Die einzelnen warnings/errors sind,
Bad I/O (reserved port addresses, reserved port bits)
Changing Reserved CPSR Bits == Show Warning Message
Writes to readonly Memory (sehr beliebt)
8bit VRAM access, 16/32bit SRAM access
Memory Range Checks (unused/reserved memory, inklusive mirrors) Mis-aligned Memory Reads and Writes Suspicious/Meaningless Opcodes (zB reserved bits, zB Goldroad) Bad Jump Addresses (jumps wo man nicht hin soll)
Ein paar von den Sachen kann man eventuell absichtlich als cleveren programming trick benutzen, in dem fall w?r's dann nicht wirklich ein error.
|
So, not everything that the no$gba freeware version counts as an error, is one.
#64930 - wintermute - Wed Dec 28, 2005 5:36 pm
| keldon wrote: |
| Quote: | | Just a note though, that errors or no, in my experience, it works fine on hardware. |
Isn't this then more of a nocash problem and not a krawal problem?
|
No, No$ is reporting problems with the code, writes to read only memory, misaligned reads/writes and things of that nature.
| Quote: |
Not to mention the fact that no gba code should be crashing nocash whether or not the code is correct. Maybe you need to make Martin aware of this. |
No$ does *not* crash. It simply reports that the code is doing something it shouldn't
#64933 - keldon - Wed Dec 28, 2005 5:56 pm
Yes I misread that completey. When he said errors I completely forgot about the nocash error system.
#65524 - Miked0801 - Tue Jan 03, 2006 8:32 pm
If I remember right, this has to do with the master set volume function and a GCC optimization error. It doesn't actually hurt anything as the value is set again later with a seperate piece of code. I also believe that newer versions of Krawall fix this.
Or this could be a new exception