|
From: Lev I. <lis...@ci...> - 2005-11-11 15:41:00
|
Hi Julian, I found that place (in dispatch-ppc32.S) which sets the FPU and AltiVec to default modes, and commented out those instructions. Now I get a bit further in the process - but any program I run SEGV's. I did a gdb, and got the following disassembly at the SEGV point : 0x71401bd0: bctrl 0x71401bd4: stw r16,0(r17) <--- SEGV here (beginning of function?) 0x71401bd8: lwz r15,1852(r31) 0x71401bdc: lwz r16,900(r31) 0x71401be0: stw r15,952(r31) 0x71401be4: stw r16,0(r31) 0x71401be8: lis r18,4487 0x71401bec: ori r18,r18,52804 0x71401bf0: stw r18,896(r31) 0x71401bf4: addi r18,r17,144 0x71401bf8: lwz r21,112(r31) 0x71401bfc: mr r3,r18 0x71401c00: lwz r22,1064(r31) 0x71401c04: mr r4,r22 0x71401c08: lis r12,28976 0x71401c0c: ori r12,r12,52048 0x71401c10: mtctr r12 0x71401c14: bctrl Looks like some generated code, because the address is the same from any program I run. Any suggestions, or does this look to complicated to be easily fixable? --Lev Julian Seward wrote: >It turns out you are the fourth person we know of to try V on >a ppc405 or similar. I believe the root problem is the 405 does >not support floating point, and so the system SIGILLs when it >attempts to set the FPU control word before running instrumented >code. > >I finally have access to such a system and so will look into >whether there is an easy fix, in the next day or so. > >J > > > |