[PATCH] mm: pmd dirty emulation in page fault handler

[Michal Hocko]

On Thu 22-12-16 23:52:03, Minchan Kim wrote:
[...]
> >From b3ec95c0df91ad113525968a4a6b53030fd0b48d Mon Sep 17 00:00:00 2001
> From: Minchan Kim <minchan at kernel.org>
> Date: Thu, 22 Dec 2016 23:43:49 +0900
> Subject: [PATCH v2] mm: pmd dirty emulation in page fault handler
> 
> Andreas reported [1] made a test in jemalloc hang in THP mode in arm64.
> http://lkml.kernel.org/r/mvmmvfy37g1.fsf@hawking.suse.de
> 
> The problem is page fault handler supports only accessed flag emulation
> for THP page of SW-dirty/accessed architecture.
> 
> This patch enables dirty-bit emulation for those architectures.
> Without it, MADV_FREE makes application hang by repeated fault forever.

The changelog is rather terse and considering the issue is rather subtle
and it aims the stable tree I think it could see more information. How
do we end up looping in the page fault and why the dirty pmd stops it.
Could you update the changelog to be more verbose, please? I am still
digesting this patch but I believe it is correct fwiw...

Thanks!

> [1] b8d3c4c3009d, mm/huge_memory.c: don't split THP page when MADV_FREE syscall is called
> 
> Cc: Jason Evans <je at fb.com>
> Cc: Kirill A. Shutemov <kirill.shutemov at linux.intel.com>
> Cc: Will Deacon <will.deacon at arm.com>
> Cc: Catalin Marinas <catalin.marinas at arm.com>
> Cc: linux-arch at vger.kernel.org
> Cc: linux-arm-kernel at lists.infradead.org
> Cc: <stable at vger.kernel.org> [4.5+]
> Fixes: b8d3c4c3009d ("mm/huge_memory.c: don't split THP page when MADV_FREE syscall is called")
> Reported-by: Andreas Schwab <schwab at suse.de>
> Signed-off-by: Minchan Kim <minchan at kernel.org>
> ---
> * from v1
>   * Remove __handle_mm_fault part - Kirill
> 
>  mm/huge_memory.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index 10eedbf..29ec8a4 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -883,15 +883,17 @@ void huge_pmd_set_accessed(struct vm_fault *vmf, pmd_t orig_pmd)
>  {
>  	pmd_t entry;
>  	unsigned long haddr;
> +	bool write = vmf->flags & FAULT_FLAG_WRITE;
>  
>  	vmf->ptl = pmd_lock(vmf->vma->vm_mm, vmf->pmd);
>  	if (unlikely(!pmd_same(*vmf->pmd, orig_pmd)))
>  		goto unlock;
>  
>  	entry = pmd_mkyoung(orig_pmd);
> +	if (write)
> +		entry = pmd_mkdirty(entry);
>  	haddr = vmf->address & HPAGE_PMD_MASK;
> -	if (pmdp_set_access_flags(vmf->vma, haddr, vmf->pmd, entry,
> -				vmf->flags & FAULT_FLAG_WRITE))
> +	if (pmdp_set_access_flags(vmf->vma, haddr, vmf->pmd, entry, write))
>  		update_mmu_cache_pmd(vmf->vma, vmf->address, vmf->pmd);
>  
>  unlock:
> -- 
> 2.7.4
> 
> --
> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> the body to majordomo at kvack.org.  For more info on Linux MM,
> see: http://www.linux-mm.org/ .
> Don't email: <a href=mailto:"dont at kvack.org"> email at kvack.org </a>

-- 
Michal Hocko
SUSE Labs