Acorn Arcade forums: Programming: Trample all over your CPU vectors...
|
Trample all over your CPU vectors... |
|
adrianl (08:57 3/9/2005) jmb (11:43 3/9/2005) ilcook (12:25 10/9/2005)
|
|
Adrian Lees |
Message #68869, posted by adrianl at 08:57, 3/9/2005 |
Member
Posts: 1637
|
... and get away with it!
Well, if you have an Iyonix anyway.
http://freespace.virgin.net/a.m.lees/downloads/raisepvec.zip
The XScale CPU has the ability to move the CPU vectors to another address, and - thanks to jmb trashing them last night - it occurred to me at 6am this morning that, providing you replace OS_ClaimProcessorVector, you can actually move them without OS support.
So, why do it?
Well, any code that accidentally tramples on them will cause a complete machine hang soon after (by taking out the SWI handler or IRQ handler) which is very difficult to debug. This module should stop the hang, though it won't yet raise a data abort when the trampling occurs.
Better than nothing, I hope; to be improved soon[1]
[1] soon may be either before or after the Second Coming.
[Edited by adrianl at 10:01, 3/9/2005] |
|
[ Log in to reply ] |
|
JMB |
Message #68870, posted by jmb at 11:43, 3/9/2005, in reply to message #68869 |
Member
Posts: 467
|
thanks to jmb trashing them last night It wasn't me! It was the Font Manager that did it! Admittedly, SYS"Font_SwitchOutputToBuffer",0,8 SYS"Font_Paint",0,"C",&b90,0,0,0,0,1 is rather foolish; I blame the documentation
|
|
[ Log in to reply ] |
|
Ian Cook |
Message #69010, posted by ilcook at 12:25, 10/9/2005, in reply to message #68870 |
Resident idiot
Posts: 1077
|
thanks to jmb trashing them last night It wasn't me! It was the Font Manager that did it! Admittedly, SYS"Font_SwitchOutputToBuffer",0,8 SYS"Font_Paint",0,"C",&b90,0,0,0,0,1 is rather foolish; I blame the documentation
SLAP |
|
[ Log in to reply ] |
|
|
Acorn Arcade forums: Programming: Trample all over your CPU vectors... |