* Re: [PATCH] mm,vmstat: correct pagetypeinfo statistics when show
2019-05-10 12:36 [PATCH] mm,vmstat: correct pagetypeinfo statistics when show Bruce ZHANG
@ 2019-05-10 18:49 ` Daniel Jordan
2019-05-13 8:53 ` Mel Gorman
2019-05-10 20:13 ` Daniel Jordan
2019-05-10 21:24 ` Daniel Jordan
2 siblings, 1 reply; 6+ messages in thread
From: Daniel Jordan @ 2019-05-10 18:49 UTC (permalink / raw)
To: Bruce ZHANG; +Cc: akpm, guro, mhocko, vbabka, jannh, linux-mm, mgorman
On Fri, May 10, 2019 at 12:36:48PM +0000, Bruce ZHANG wrote:
> The "Free pages count per migrate type at order" are shown with the
> order from 0 ~ (MAX_ORDER-1), while "Page block order" just print
> pageblock_order. If the macro CONFIG_HUGETLB_PAGE is defined, the
> pageblock_order may not be equal to (MAX_ORDER-1).
All of this is true, but why do you think it's wrong?
It makes sense that "Page block order" corresponds to pageblock_order,
regardless of whether pageblock_order == MAX_ORDER-1.
Cc Mel, who added these two lines.
> Signed-off-by: Zhang Bo <bo.zhang@nxp.com>
> ---
> mm/vmstat.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/mm/vmstat.c b/mm/vmstat.c
> index 6389e87..b0089cf 100644
> --- a/mm/vmstat.c
> +++ b/mm/vmstat.c
> @@ -1430,8 +1430,8 @@ static int pagetypeinfo_show(struct seq_file *m, void *arg)
> if (!node_state(pgdat->node_id, N_MEMORY))
> return 0;
>
> - seq_printf(m, "Page block order: %d\n", pageblock_order);
> - seq_printf(m, "Pages per block: %lu\n", pageblock_nr_pages);
> + seq_printf(m, "Page block order: %d\n", MAX_ORDER - 1);
> + seq_printf(m, "Pages per block: %lu\n", MAX_ORDER_NR_PAGES);
> seq_putc(m, '\n');
> pagetypeinfo_showfree(m, pgdat);
> pagetypeinfo_showblockcount(m, pgdat);
> --
> 1.9.1
>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] mm,vmstat: correct pagetypeinfo statistics when show
2019-05-10 18:49 ` Daniel Jordan
@ 2019-05-13 8:53 ` Mel Gorman
2019-05-17 6:44 ` Anshuman Khandual
0 siblings, 1 reply; 6+ messages in thread
From: Mel Gorman @ 2019-05-13 8:53 UTC (permalink / raw)
To: Daniel Jordan; +Cc: Bruce ZHANG, akpm, guro, mhocko, vbabka, jannh, linux-mm
On Fri, May 10, 2019 at 02:49:00PM -0400, Daniel Jordan wrote:
> On Fri, May 10, 2019 at 12:36:48PM +0000, Bruce ZHANG wrote:
> > The "Free pages count per migrate type at order" are shown with the
> > order from 0 ~ (MAX_ORDER-1), while "Page block order" just print
> > pageblock_order. If the macro CONFIG_HUGETLB_PAGE is defined, the
> > pageblock_order may not be equal to (MAX_ORDER-1).
>
> All of this is true, but why do you think it's wrong?
>
Indeed, why is this wrong?
> It makes sense that "Page block order" corresponds to pageblock_order,
> regardless of whether pageblock_order == MAX_ORDER-1.
>
Page block order is related to the PMD huge page size, it's not directly
related to MAX_ORDER other than MAX_ORDER is larger than
pageblock_order.
> Cc Mel, who added these two lines.
>
> > Signed-off-by: Zhang Bo <bo.zhang@nxp.com>
What's there is correct so unless there is a great explanation as to why
it should be different;
Naked-by: Mel Gorman <mgorman@techsingularity.net>
--
Mel Gorman
SUSE Labs
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] mm,vmstat: correct pagetypeinfo statistics when show
2019-05-13 8:53 ` Mel Gorman
@ 2019-05-17 6:44 ` Anshuman Khandual
0 siblings, 0 replies; 6+ messages in thread
From: Anshuman Khandual @ 2019-05-17 6:44 UTC (permalink / raw)
To: Mel Gorman, Daniel Jordan
Cc: Bruce ZHANG, akpm, guro, mhocko, vbabka, jannh, linux-mm
On 05/13/2019 02:23 PM, Mel Gorman wrote:
> On Fri, May 10, 2019 at 02:49:00PM -0400, Daniel Jordan wrote:
>> On Fri, May 10, 2019 at 12:36:48PM +0000, Bruce ZHANG wrote:
>>> The "Free pages count per migrate type at order" are shown with the
>>> order from 0 ~ (MAX_ORDER-1), while "Page block order" just print
>>> pageblock_order. If the macro CONFIG_HUGETLB_PAGE is defined, the
>>> pageblock_order may not be equal to (MAX_ORDER-1).
>>
>> All of this is true, but why do you think it's wrong?
>>
>
> Indeed, why is this wrong?
>
>> It makes sense that "Page block order" corresponds to pageblock_order,
>> regardless of whether pageblock_order == MAX_ORDER-1.
>>
>
> Page block order is related to the PMD huge page size, it's not directly
> related to MAX_ORDER other than MAX_ORDER is larger than
> pageblock_order.
Right.
>
>> Cc Mel, who added these two lines.
>>
>>> Signed-off-by: Zhang Bo <bo.zhang@nxp.com>
>
> What's there is correct so unless there is a great explanation as to why
> it should be different;
Agreed.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] mm,vmstat: correct pagetypeinfo statistics when show
2019-05-10 12:36 [PATCH] mm,vmstat: correct pagetypeinfo statistics when show Bruce ZHANG
2019-05-10 18:49 ` Daniel Jordan
@ 2019-05-10 20:13 ` Daniel Jordan
2019-05-10 21:24 ` Daniel Jordan
2 siblings, 0 replies; 6+ messages in thread
From: Daniel Jordan @ 2019-05-10 20:13 UTC (permalink / raw)
To: Bruce ZHANG; +Cc: akpm, guro, mhocko, vbabka, jannh, linux-mm, mgorman
On Fri, May 10, 2019 at 12:36:48PM +0000, Bruce ZHANG wrote:
> The "Free pages count per migrate type at order" are shown with the
> order from 0 ~ (MAX_ORDER-1), while "Page block order" just print
> pageblock_order. If the macro CONFIG_HUGETLB_PAGE is defined, the
> pageblock_order may not be equal to (MAX_ORDER-1).
All of this is true, but why is it wrong?
It makes sense that "Page block order" corresponds to pageblock_order,
regardless of whether pageblock_order == MAX_ORDER-1.
Cc Mel, who added these two lines.
> Signed-off-by: Zhang Bo <bo.zhang@nxp.com>
> ---
> mm/vmstat.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/mm/vmstat.c b/mm/vmstat.c
> index 6389e87..b0089cf 100644
> --- a/mm/vmstat.c
> +++ b/mm/vmstat.c
> @@ -1430,8 +1430,8 @@ static int pagetypeinfo_show(struct seq_file *m, void *arg)
> if (!node_state(pgdat->node_id, N_MEMORY))
> return 0;
>
> - seq_printf(m, "Page block order: %d\n", pageblock_order);
> - seq_printf(m, "Pages per block: %lu\n", pageblock_nr_pages);
> + seq_printf(m, "Page block order: %d\n", MAX_ORDER - 1);
> + seq_printf(m, "Pages per block: %lu\n", MAX_ORDER_NR_PAGES);
> seq_putc(m, '\n');
> pagetypeinfo_showfree(m, pgdat);
> pagetypeinfo_showblockcount(m, pgdat);
> --
> 1.9.1
>
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] mm,vmstat: correct pagetypeinfo statistics when show
2019-05-10 12:36 [PATCH] mm,vmstat: correct pagetypeinfo statistics when show Bruce ZHANG
2019-05-10 18:49 ` Daniel Jordan
2019-05-10 20:13 ` Daniel Jordan
@ 2019-05-10 21:24 ` Daniel Jordan
2 siblings, 0 replies; 6+ messages in thread
From: Daniel Jordan @ 2019-05-10 21:24 UTC (permalink / raw)
To: Bruce ZHANG; +Cc: akpm, guro, mhocko, vbabka, jannh, linux-mm, mgorman
On Fri, May 10, 2019 at 12:36:48PM +0000, Bruce ZHANG wrote:
> The "Free pages count per migrate type at order" are shown with the
> order from 0 ~ (MAX_ORDER-1), while "Page block order" just print
> pageblock_order. If the macro CONFIG_HUGETLB_PAGE is defined, the
> pageblock_order may not be equal to (MAX_ORDER-1).
All of this is true, but why is it wrong?
It makes sense that "Page block order" corresponds to pageblock_order,
regardless of whether pageblock_order == MAX_ORDER-1.
Cc Mel, who added these two lines.
> Signed-off-by: Zhang Bo <bo.zhang@nxp.com>
> ---
> mm/vmstat.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/mm/vmstat.c b/mm/vmstat.c
> index 6389e87..b0089cf 100644
> --- a/mm/vmstat.c
> +++ b/mm/vmstat.c
> @@ -1430,8 +1430,8 @@ static int pagetypeinfo_show(struct seq_file *m, void *arg)
> if (!node_state(pgdat->node_id, N_MEMORY))
> return 0;
>
> - seq_printf(m, "Page block order: %d\n", pageblock_order);
> - seq_printf(m, "Pages per block: %lu\n", pageblock_nr_pages);
> + seq_printf(m, "Page block order: %d\n", MAX_ORDER - 1);
> + seq_printf(m, "Pages per block: %lu\n", MAX_ORDER_NR_PAGES);
> seq_putc(m, '\n');
> pagetypeinfo_showfree(m, pgdat);
> pagetypeinfo_showblockcount(m, pgdat);
> --
> 1.9.1
>
^ permalink raw reply [flat|nested] 6+ messages in thread