Re: [tip:perf/core] perf/x86: Fix USER/KERNEL tagging of samples

[Peter Zijlstra <a.p.zijlstra@chello.nl>]

On Mon, 2012-07-09 at 20:41 +0200, Ingo Molnar wrote:
> > +static unsigned long get_segment_base(unsigned int segment)
> > +{
> > +     struct desc_struct *desc;
> > +     int idx = segment >> 3;
> > +
> > +     if ((segment & SEGMENT_TI_MASK) == SEGMENT_LDT) {
> > +             if (idx > LDT_ENTRIES)
> > +                     return 0;
> > +
> > +             desc = current->active_mm->context.ldt;
> > +     } else {
> > +             if (idx > GDT_ENTRIES)
> > +                     return 0;
> > +
> > +             desc = __this_cpu_ptr(&gdt_page.gdt[0]);
> > +     }
> > +
> > +     return get_desc_base(desc + idx);
> 
> Shouldn't idx be checked against active_mm->context.ldt.size, 
> not LDT_ENTRIES (which is really just an upper limit)?

Ah indeed, fixed that.

> > +static unsigned long code_segment_base(struct pt_regs *regs)
> > +{
> > +#ifdef CONFIG_32BIT
> > +     if (user_mode(regs) && regs->cs != __USER_CS)
> > +             return get_segment_base(regs->cs);
> > +#else
> > +     if (test_thread_flag(TIF_IA32)) {
> > +             if (user_mode(regs) && regs->cs != __USER32_CS)
> > +                     return get_segment_base(regs->cs);
> > +     }
> > +#endif
> > +     return 0;
> > +}
> 
> Will this do the right thing for x32 mode? 

hpa? It looks like x32 has TIF_X32, but from the kernel's POV its really
just another 64bit process, so as long as we don't trigger the TIF_IA32
case we'll just return 0. 

set_personality_ia32() looks like TIF_IA32 and TIF_X32 are mutually
exclusive, so I think we do the right thing.