Re: x86/pti: smp_processor_id() called while preemptible in resume-from-sleep

[Thomas Gleixner]

On Sat, 30 Dec 2017, Linus Torvalds wrote:
> On Sat, Dec 30, 2017 at 10:20 AM, Thomas Gleixner <t...@linutronix.de> wrote:
> So I'd suggest moving the preempt_disable() up to the top of that
> function, regardless of whether we could then remove that seemingly
> stale TLB flush in that crazy
> smpboot_setup/restore_warm_reset_vector() dance...

Regardless of the other bogosity, I did some digging and had to go back
into the 2.1 aera to find the reason for this local_flush_tlb(). That old
code manipulated swapper_pg_dir in do_boot_cpu.

+       local_flush_tlb();
+       SMP_PRINTK(("1.\n"));
+       *((volatile unsigned short *) phys_to_virt(0x469)) = ((unsigned 
long)stack)>>4;
+       SMP_PRINTK(("2.\n"));
+       *((volatile unsigned short *) phys_to_virt(0x467)) = 0;
+       SMP_PRINTK(("3.\n"));
+
+       maincfg=swapper_pg_dir[0];
+       ((unsigned long *)swapper_pg_dir)[0]=0x102007;

        .....

+       swapper_pg_dir[0]=maincfg;
+       local_flush_tlb();

That pagetable fiddling dissappeared long ago, but the flush stayed around
forever.

Thanks,

        tglx