Re: [PATCH 1/2] mm/mprotect: use mmu_gather

[Peter Zijlstra <peterz@infradead.org>]

On Mon, Oct 04, 2021 at 12:24:14PM -0700, Nadav Amit wrote:
> 
> 
> > On Oct 3, 2021, at 5:10 AM, Peter Zijlstra <peterz@infradead.org> wrote:
> > 
> > On Sat, Sep 25, 2021 at 01:54:22PM -0700, Nadav Amit wrote:
> > 
> >> @@ -338,25 +344,25 @@ static unsigned long change_protection_range(struct vm_area_struct *vma,
> >> 	struct mm_struct *mm = vma->vm_mm;
> >> 	pgd_t *pgd;
> >> 	unsigned long next;
> >> -	unsigned long start = addr;
> >> 	unsigned long pages = 0;
> >> +	struct mmu_gather tlb;
> >> 
> >> 	BUG_ON(addr >= end);
> >> 	pgd = pgd_offset(mm, addr);
> >> 	flush_cache_range(vma, addr, end);
> >> 	inc_tlb_flush_pending(mm);
> > 
> > That seems unbalanced...
> 
> Bad rebase. Thanks for catching it!
> 
> > 
> >> +	tlb_gather_mmu(&tlb, mm);
> >> +	tlb_start_vma(&tlb, vma);
> >> 	do {
> >> 		next = pgd_addr_end(addr, end);
> >> 		if (pgd_none_or_clear_bad(pgd))
> >> 			continue;
> >> -		pages += change_p4d_range(vma, pgd, addr, next, newprot,
> >> +		pages += change_p4d_range(&tlb, vma, pgd, addr, next, newprot,
> >> 					  cp_flags);
> >> 	} while (pgd++, addr = next, addr != end);
> >> 
> >> -	/* Only flush the TLB if we actually modified any entries: */
> >> -	if (pages)
> >> -		flush_tlb_range(vma, start, end);
> >> -	dec_tlb_flush_pending(mm);
> > 
> > ... seeing you do remove the extra decrement.
> 
> Is it really needed? We do not put this comment elsewhere for
> tlb_finish_mmu(). But no problem, I’ll keep it.

-ENOPARSE, did you read decrement as comment? In any case, I don't
particularly care about the comment, and tlb_*_mmu() imply the inc/dec
thingies.

All I tried to do is point out that removing the dec but leaving the inc
is somewhat inconsistent :-)