On Wed, 11 May 2011 15:48:59 +0200, Luca Tettamanti wrote: > On Wed, May 11, 2011 at 02:43:35PM +0200, Luca Tettamanti wrote: > > Jeff's CPU does indeed lack lahf (see cpuinfo), so the driver should not > > unconditionally use the intruction. > > The driver should check X86_FEATURE_LAHF_LM or use use a different > > method to read EFLAFS (pushf?). > > The driver uses lahf to copy the lowest byte of EFLAGS to eax, then > shifts it to right and test the lowest bit (CF). > Using pushf instead we get the whole EFLAGS and test the lowest bit. > Does it make sense? Someone should double check the patch, I don't trust > my assembly skills ;) > > diff --git a/drivers/char/i8k.c b/drivers/char/i8k.c > index d72433f..3554d10 100644 > --- a/drivers/char/i8k.c > +++ b/drivers/char/i8k.c > @@ -139,8 +139,8 @@ static int i8k_smm(struct smm_regs *regs) > "movl %%edi,20(%%rax)\n\t" > "popq %%rdx\n\t" > "movl %%edx,0(%%rax)\n\t" > - "lahf\n\t" > - "shrl $8,%%eax\n\t" > + "pusfh\n\t" I think you meant pushf, not pusfh, right? > + "popl %%eax\n\t" > "andl $1,%%eax\n" > :"=a"(rc) > : "a"(regs) I am no asm expert, but I don't get why you removed the shrl instruction. As I understand it, your pushf+popl replaces only lahf, the bit shifting is still needed (I assume the code wants to test bit 8 in the flag register.) -- Jean Delvare _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
