[PATCH] Add /proc/PID/{smaps, numa_maps} support for DAX

[PATCH] Add /proc/PID/{smaps, numa_maps} support for DAX

[Fan Du]

So user could check those interface for more detailed
information about how much DAX mappings are currently
created.

Here we use vma_is_dax method to find specific page
struture with DAX {huge, normal}page mappings,
vm_normal_page routine works as before without any
impact on the existing logical where _vm_normal_page
are called.

Signed-off-by: Fan Du <fan.du@intel.com>
---
 fs/proc/task_mmu.c | 25 +++++++++++++++++++------
 1 file changed, 19 insertions(+), 6 deletions(-)

diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index 5589b4b..ba2e58c 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -528,7 +528,11 @@ static void smaps_pte_entry(pte_t *pte, unsigned long addr,
 	struct page *page = NULL;
 
 	if (pte_present(*pte)) {
-		page = vm_normal_page(vma, addr, *pte);
+		if (!vma_is_dax(vma))
+			page = vm_normal_page(vma, addr, *pte);
+		else
+			page = pte_page(*pte);
+
 	} else if (is_swap_pte(*pte)) {
 		swp_entry_t swpent = pte_to_swp_entry(*pte);
 
@@ -579,7 +583,11 @@ static void smaps_pmd_entry(pmd_t *pmd, unsigned long addr,
 	struct page *page;
 
 	/* FOLL_DUMP will return -EFAULT on huge zero page */
-	page = follow_trans_huge_pmd(vma, addr, pmd, FOLL_DUMP);
+	if (!vma_is_dax(vma))
+		page = follow_trans_huge_pmd(vma, addr, pmd, FOLL_DUMP);
+	else
+		page = pmd_page(*pmd);
+
 	if (IS_ERR_OR_NULL(page))
 		return;
 	if (PageAnon(page))
@@ -1668,13 +1676,15 @@ static int gather_pte_stats(pmd_t *pmd, unsigned long addr,
 	spinlock_t *ptl;
 	pte_t *orig_pte;
 	pte_t *pte;
+	struct page *page;
 
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
 	ptl = pmd_trans_huge_lock(pmd, vma);
 	if (ptl) {
-		struct page *page;
-
-		page = can_gather_numa_stats_pmd(*pmd, vma, addr);
+		if (!vma_is_dax(vma))
+			page = can_gather_numa_stats_pmd(*pmd, vma, addr);
+		else
+			page = pmd_page(*pmd);
 		if (page)
 			gather_stats(page, md, pmd_dirty(*pmd),
 				     HPAGE_PMD_SIZE/PAGE_SIZE);
@@ -1687,7 +1697,10 @@ static int gather_pte_stats(pmd_t *pmd, unsigned long addr,
 #endif
 	orig_pte = pte = pte_offset_map_lock(walk->mm, pmd, addr, &ptl);
 	do {
-		struct page *page = can_gather_numa_stats(*pte, vma, addr);
+		if (!vma_is_dax(vma))
+			page = can_gather_numa_stats(*pte, vma, addr);
+		else
+			page = pte_page(*pte);
 		if (!page)
 			continue;
 		gather_stats(page, md, pte_dirty(*pte), 1);
-- 
1.8.3.1

Re: [PATCH] Add /proc/PID/{smaps, numa_maps} support for DAX

[Dan Williams]

On Tue, Oct 24, 2017 at 12:52 AM, Fan Du <fan.du@intel.com> wrote:
>
> So user could check those interface for more detailed
> information about how much DAX mappings are currently
> created.
>
> Here we use vma_is_dax method to find specific page
> struture with DAX {huge, normal}page mappings,
> vm_normal_page routine works as before without any
> impact on the existing logical where _vm_normal_page
> are called.
>
> Signed-off-by: Fan Du <fan.du@intel.com>
> ---
>  fs/proc/task_mmu.c | 25 +++++++++++++++++++------
>  1 file changed, 19 insertions(+), 6 deletions(-)
>
> diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
> index 5589b4b..ba2e58c 100644
> --- a/fs/proc/task_mmu.c
> +++ b/fs/proc/task_mmu.c
> @@ -528,7 +528,11 @@ static void smaps_pte_entry(pte_t *pte, unsigned long addr,
>         struct page *page = NULL;
>
>         if (pte_present(*pte)) {
> -               page = vm_normal_page(vma, addr, *pte);
> +               if (!vma_is_dax(vma))
> +                       page = vm_normal_page(vma, addr, *pte);
> +               else
> +                       page = pte_page(*pte);

See usage of pte_devmap, pmd_devmap, and pud_devmap in mm/gup.c. You
need to check for that bit to know that there is a 'struct page' entry
associated with a DAX mapping.  I believe we just need to fix up and
resend his approach from last year:

    https://marc.info/?l=linux-kernel&m=148045064103357&w=2

RE: [PATCH] Add /proc/PID/{smaps, numa_maps} support for DAX

[Du, Fan]

>-----Original Message-----
>From: Dan Williams [mailto:dan.j.williams@intel.com]
>Sent: Tuesday, October 24, 2017 4:47 PM
>To: Du, Fan <fan.du@intel.com>
>Cc: Andrew Morton <akpm@linux-foundation.org>; Christoph Hellwig
><hch@lst.de>; linux-kernel@vger.kernel.org; Dave Hansen
><dave.hansen@linux.intel.com>
>Subject: Re: [PATCH] Add /proc/PID/{smaps, numa_maps} support for DAX
>
>On Tue, Oct 24, 2017 at 12:52 AM, Fan Du <fan.du@intel.com> wrote:
>>
>> So user could check those interface for more detailed
>> information about how much DAX mappings are currently
>> created.
>>
>> Here we use vma_is_dax method to find specific page
>> struture with DAX {huge, normal}page mappings,
>> vm_normal_page routine works as before without any
>> impact on the existing logical where _vm_normal_page
>> are called.
>>
>> Signed-off-by: Fan Du <fan.du@intel.com>
>> ---
>>  fs/proc/task_mmu.c | 25 +++++++++++++++++++------
>>  1 file changed, 19 insertions(+), 6 deletions(-)
>>
>> diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
>> index 5589b4b..ba2e58c 100644
>> --- a/fs/proc/task_mmu.c
>> +++ b/fs/proc/task_mmu.c
>> @@ -528,7 +528,11 @@ static void smaps_pte_entry(pte_t *pte, unsigned
>long addr,
>>         struct page *page = NULL;
>>
>>         if (pte_present(*pte)) {
>> -               page = vm_normal_page(vma, addr, *pte);
>> +               if (!vma_is_dax(vma))
>> +                       page = vm_normal_page(vma, addr, *pte);
>> +               else
>> +                       page = pte_page(*pte);
>
>See usage of pte_devmap, pmd_devmap, and pud_devmap in mm/gup.c. You
>need to check for that bit to know that there is a 'struct page' entry
>associated with a DAX mapping.  

Thanks Dan for your review, got it!


>I believe we just need to fix up and
>resend his approach from last year:
>
>    https://marc.info/?l=linux-kernel&m=148045064103357&w=2

Sure, I will rebase Dave's patch and send v2 soon.

Re: [PATCH] Add /proc/PID/{smaps, numa_maps} support for DAX

[Michal Hocko]

On Tue 24-10-17 15:52:01, Fan Du wrote:
> So user could check those interface for more detailed
> information about how much DAX mappings are currently
> created.
> 
> Here we use vma_is_dax method to find specific page
> struture with DAX {huge, normal}page mappings,
> vm_normal_page routine works as before without any
> impact on the existing logical where _vm_normal_page
> are called.

And like for any other user visible changes it should be CC to the
linux-api mailing list and a usecase which requires this information
provided as well.
-- 
Michal Hocko
SUSE Labs

RE: [PATCH] Add /proc/PID/{smaps, numa_maps} support for DAX

[Du, Fan]

>-----Original Message-----
>From: Michal Hocko [mailto:mhocko@kernel.org]
>Sent: Tuesday, October 24, 2017 8:02 PM
>To: Du, Fan <fan.du@intel.com>
>Cc: akpm@linux-foundation.org; hch@lst.de; Williams, Dan J
><dan.j.williams@intel.com>; linux-kernel@vger.kernel.org
>Subject: Re: [PATCH] Add /proc/PID/{smaps, numa_maps} support for DAX
>
>On Tue 24-10-17 15:52:01, Fan Du wrote:
>> So user could check those interface for more detailed
>> information about how much DAX mappings are currently
>> created.
>>
>> Here we use vma_is_dax method to find specific page
>> struture with DAX {huge, normal}page mappings,
>> vm_normal_page routine works as before without any
>> impact on the existing logical where _vm_normal_page
>> are called.
>
>And like for any other user visible changes it should be CC to the
>linux-api mailing list and a usecase which requires this information
>provided as well.

Done!
Please kindly review v3.


>--
>Michal Hocko
>SUSE Labs