[SOLVED]KEXTs & Altivec stack corruption

Subject: [SOLVED]KEXTs & Altivec stack corruption
From: "Mike Vannorsdel" <email@hidden>
Date: Sat, 12 Oct 2002 05:12:07 -0600

I have somewhat solved my problem with stack corruption. It seems that some AltiVec-less sources compiled with the -faltivec flag can have bad side effects. What happens with some of these sources is that the compiler creates four functions (saveVEC, restVEC, saveVEC_vr11, restVEC_vr10). Some of my functions in the sources will have calls to these (usually saveVEC_vr11 and restVEC_vr10) inserted into them. These new functions are what seem to be corrupting the stack.

So my solution was to compile all non-AltiVec sources without the -faltivec flag and the AltiVec sources with it. This prevented the extra functions from being created and inserted into my functions. And lots of testing has shown the AltiVec code running very well without problems in my KEXT with this method.

I'd be interested to know why these functions are created and inserted into other functions (so far functions with no vector types or instructions). It's definitely a GCC problem.
_______________________________________________
darwin-kernel mailing list | email@hidden
Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel
Do not post admin requests to the list. They will be ignored.

Prev by Date: More on KEXTs & Altivec stack corruption
Next by Date: sharing custom filesystem
Previous by thread: More on KEXTs & Altivec stack corruption
Next by thread: sharing custom filesystem
Index(es):
- Date
- Thread

Home