From: David Hildenbrand <david@redhat.com>
To: Baoquan He <bhe@redhat.com>
Cc: linux-kernel@vger.kernel.org, linux-mm@kvack.org,
akpm@linux-foundation.org, dan.j.williams@intel.com,
richardw.yang@linux.intel.com
Subject: Re: [PATCH 1/7] mm/sparse.c: Introduce new function fill_subsection_map()
Date: Tue, 11 Feb 2020 15:44:14 +0100 [thread overview]
Message-ID: <659d16c9-79f7-a1d6-ca05-d6164c9c11a6@redhat.com> (raw)
In-Reply-To: <20200211124648.GF8965@MiWiFi-R3L-srv>
On 11.02.20 13:46, Baoquan He wrote:
> On 02/10/20 at 10:49am, David Hildenbrand wrote:
>> On 09.02.20 11:48, Baoquan He wrote:
>>> Wrap the codes filling subsection map in section_activate() into
>>> fill_subsection_map(), this makes section_activate() cleaner and
>>> easier to follow.
>>>
>>> Signed-off-by: Baoquan He <bhe@redhat.com>
>>> ---
>>> mm/sparse.c | 45 ++++++++++++++++++++++++++++++++++-----------
>>> 1 file changed, 34 insertions(+), 11 deletions(-)
>>>
>>> diff --git a/mm/sparse.c b/mm/sparse.c
>>> index c184b69460b7..9ad741ccbeb6 100644
>>> --- a/mm/sparse.c
>>> +++ b/mm/sparse.c
>>> @@ -788,24 +788,28 @@ static void section_deactivate(unsigned long pfn, unsigned long nr_pages,
>>> depopulate_section_memmap(pfn, nr_pages, altmap);
>>> }
>>>
>>> -static struct page * __meminit section_activate(int nid, unsigned long pfn,
>>> - unsigned long nr_pages, struct vmem_altmap *altmap)
>>> +/**
>>> + * fill_subsection_map - fill subsection map of a memory region
>>> + * @pfn - start pfn of the memory range
>>> + * @nr_pages - number of pfns to add in the region
>>> + *
>>> + * This clears the related subsection map inside one section, and only
>>> + * intended for hotplug.
>>> + *
>>> + * Return:
>>> + * * 0 - On success.
>>> + * * -EINVAL - Invalid memory region.
>>> + * * -EEXIST - Subsection map has been set.
>>> + */
>>> +static int fill_subsection_map(unsigned long pfn, unsigned long nr_pages)
>>> {
>>> - DECLARE_BITMAP(map, SUBSECTIONS_PER_SECTION) = { 0 };
>>> struct mem_section *ms = __pfn_to_section(pfn);
>>> - struct mem_section_usage *usage = NULL;
>>> + DECLARE_BITMAP(map, SUBSECTIONS_PER_SECTION) = { 0 };
>>> unsigned long *subsection_map;
>>> - struct page *memmap;
>>> int rc = 0;
>>>
>>> subsection_mask_set(map, pfn, nr_pages);
>>>
>>> - if (!ms->usage) {
>>> - usage = kzalloc(mem_section_usage_size(), GFP_KERNEL);
>>> - if (!usage)
>>> - return ERR_PTR(-ENOMEM);
>>> - ms->usage = usage;
>>> - }
>>> subsection_map = &ms->usage->subsection_map[0];
>>>
>>> if (bitmap_empty(map, SUBSECTIONS_PER_SECTION))
>>> @@ -816,6 +820,25 @@ static struct page * __meminit section_activate(int nid, unsigned long pfn,
>>> bitmap_or(subsection_map, map, subsection_map,
>>> SUBSECTIONS_PER_SECTION);
>>>
>>> + return rc;
>>> +}
>>> +
>>> +static struct page * __meminit section_activate(int nid, unsigned long pfn,
>>> + unsigned long nr_pages, struct vmem_altmap *altmap)
>>> +{
>>> + struct mem_section *ms = __pfn_to_section(pfn);
>>> + struct mem_section_usage *usage = NULL;
>>> + struct page *memmap;
>>> + int rc = 0;
>>> +
>>> + if (!ms->usage) {
>>> + usage = kzalloc(mem_section_usage_size(), GFP_KERNEL);
>>> + if (!usage)
>>> + return ERR_PTR(-ENOMEM);
>>> + ms->usage = usage;
>>> + }
>>> +
>>> + rc = fill_subsection_map(pfn, nr_pages);
>>> if (rc) {
>>> if (usage)
>>> ms->usage = NULL;
>>>
>>
>> What about having two variants of
>> section_activate()/section_deactivate() instead? Then we don't have any
>> subsection related stuff in !subsection code.
>
> Thanks for looking into this, David.
>
> Having two variants of section_activate()/section_deactivate() is also
> good. Just not like memmap handling which is very different between classic
> sparse and vmemmap, makes having two variants very attractive, the code
> and logic in section_activate()/section_deactivate() is not too much,
> and both of them basically can share the most of code, these make the
> variants way not so necessary. I personally prefer the current way, what
> do you think?
I was looking at
if (nr_pages < PAGES_PER_SECTION && early_section(ms))
return pfn_to_page(pfn);
and thought that it is also specific to sub-section handling. I wonder
if we can simply move that into the VMEMMAP variant of
populate_section_memmap()?
But apart from that I agree that the end result with the current
approach is also nice.
Can you reshuffle the patches, moving the fixes to the very front so we
can backport more easily?
--
Thanks,
David / dhildenb
next prev parent reply other threads:[~2020-02-11 14:44 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-02-09 10:48 [PATCH 0/7] mm/hotplug: Only use subsection in VMEMMAP case and fix hot add/remove failure in SPARSEMEM|!VMEMMAP case Baoquan He
2020-02-09 10:48 ` [PATCH 1/7] mm/sparse.c: Introduce new function fill_subsection_map() Baoquan He
2020-02-09 23:05 ` Wei Yang
2020-02-09 23:11 ` Wei Yang
2020-02-10 3:25 ` Baoquan He
2020-02-10 9:49 ` David Hildenbrand
2020-02-11 12:46 ` Baoquan He
2020-02-11 14:44 ` David Hildenbrand [this message]
2020-02-12 11:21 ` Baoquan He
2020-02-09 10:48 ` [PATCH 2/7] mm/sparse.c: Introduce a new function clear_subsection_map() Baoquan He
2020-02-09 23:07 ` Wei Yang
2020-02-10 3:36 ` Baoquan He
2020-02-10 6:02 ` Wei Yang
2020-02-09 10:48 ` [PATCH 3/7] mm/sparse.c: only use subsection map in VMEMMAP case Baoquan He
2020-02-09 23:23 ` Wei Yang
2020-02-11 14:43 ` David Hildenbrand
2020-02-12 11:26 ` Baoquan He
2020-02-11 20:14 ` Dan Williams
2020-02-12 9:39 ` David Hildenbrand
2020-02-12 11:20 ` Baoquan He
2020-02-12 15:48 ` Dan Williams
2020-02-09 10:48 ` [PATCH 4/7] mm/sparse.c: Use __get_free_pages() instead in populate_section_memmap() Baoquan He
2020-02-09 23:39 ` Wei Yang
2020-02-09 10:48 ` [PATCH 5/7] mm/sparse.c: update code comment about section activate/deactivate Baoquan He
2020-02-09 10:48 ` [PATCH 6/7] mm/sparsemem: pfn_to_page is not valid yet on SPARSEMEM Baoquan He
2020-02-10 3:45 ` Baoquan He
2020-02-09 10:48 ` [PATCH 7/7] mm/hotplug: fix hot remove failure in SPARSEMEM|!VMEMMAP case Baoquan He
2020-02-09 23:52 ` Wei Yang
2020-02-10 3:41 ` Baoquan He
2020-02-10 6:08 ` Wei Yang
2020-02-10 7:54 ` Baoquan He
2020-02-10 23:05 ` Wei Yang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=659d16c9-79f7-a1d6-ca05-d6164c9c11a6@redhat.com \
--to=david@redhat.com \
--cc=akpm@linux-foundation.org \
--cc=bhe@redhat.com \
--cc=dan.j.williams@intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=richardw.yang@linux.intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).