Re: apparent regressions from TLB range flushing page set

[Alex Shi <alex.shi@intel.com>]

On 08/20/2012 10:12 PM, Jan Beulich wrote:

> Alex,
> 
> without even having run that code yet, I think I see two bugs here,
> both of which I'm pretty sure I pointed out at least once during the
> review cycle:


I was thought you have 'Agreed' for xen part code. :)

> 
> For one, while TLB_FLUSH_ALL gets passed as 'end' argument to
> flush_tlb_others(), the Xen code was made to check its 'start'
> parameter.


Do you mean need the following change? --untested.

>From the logical of flush_tlb_others, the old code should cause unflushed TLB issue
(when end == -1, xen just executed INVLPG_MULTI, not whole TLB_FLUSH_MULTI). but we
 didn't find this problem in xen test. So, guess the xen code has other bug cover this. 

=========
diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
index b65a761..5141d80 100644
--- a/arch/x86/xen/mmu.c
+++ b/arch/x86/xen/mmu.c
@@ -1283,7 +1283,7 @@ static void xen_flush_tlb_others(const struct cpumask *cpus,
        cpumask_clear_cpu(smp_processor_id(), to_cpumask(args->mask));
 
        args->op.cmd = MMUEXT_TLB_FLUSH_MULTI;
-       if (start != TLB_FLUSH_ALL && (end - start) <= PAGE_SIZE) {
+       if (end != TLB_FLUSH_ALL && (end - start) <= PAGE_SIZE) {
                args->op.cmd = MMUEXT_INVLPG_MULTI;
                args->op.arg1.linear_addr = start;
        }