linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Matthew Wilcox <willy@infradead.org>
To: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Andrew Morton <akpm@linux-foundation.org>,
	David Hildenbrand <david@redhat.com>,
	Oscar Salvador <osalvador@suse.de>,
	Miaohe Lin <linmiaohe@huawei.com>,
	Naoya Horiguchi <nao.horiguchi@gmail.com>,
	linux-mm@kvack.org, dan.carpenter@linaro.org,
	Jonathan Cameron <Jonathan.Cameron@huawei.com>
Subject: Re: [PATCH v3 1/5] mm: memory_hotplug: remove head variable in do_migrate_range()
Date: Sat, 28 Sep 2024 05:55:56 +0100	[thread overview]
Message-ID: <ZveMXByLE9LvO2z3@casper.infradead.org> (raw)
In-Reply-To: <20240827114728.3212578-2-wangkefeng.wang@huawei.com>

On Tue, Aug 27, 2024 at 07:47:24PM +0800, Kefeng Wang wrote:
> Directly use a folio for HugeTLB and THP when calculate the next pfn, then
> remove unused head variable.

I just noticed this got merged.  You're going to hit BUG_ON with it.

> -		if (PageHuge(page)) {
> -			pfn = page_to_pfn(head) + compound_nr(head) - 1;
> -			isolate_hugetlb(folio, &source);
> -			continue;
> -		} else if (PageTransHuge(page))
> -			pfn = page_to_pfn(head) + thp_nr_pages(page) - 1;
> +		/*
> +		 * No reference or lock is held on the folio, so it might
> +		 * be modified concurrently (e.g. split).  As such,
> +		 * folio_nr_pages() may read garbage.  This is fine as the outer
> +		 * loop will revisit the split folio later.
> +		 */
> +		if (folio_test_large(folio)) {

But it's not fine.  Look at the implementation of folio_test_large():

static inline bool folio_test_large(const struct folio *folio)
{
        return folio_test_head(folio);
}

That's going to be provided by:

#define FOLIO_TEST_FLAG(name, page)                                     \
static __always_inline bool folio_test_##name(const struct folio *folio) \
{ return test_bit(PG_##name, const_folio_flags(folio, page)); }

and here's the BUG:

static const unsigned long *const_folio_flags(const struct folio *folio,
                unsigned n)
{
        const struct page *page = &folio->page;

        VM_BUG_ON_PGFLAGS(PageTail(page), page);
        VM_BUG_ON_PGFLAGS(n > 0 && !test_bit(PG_head, &page->flags), page);
        return &page[n].flags;
}

(this page can be transformed from a head page to a tail page because,
as the comment notes, we don't hold a reference.

Please back this out.



  reply	other threads:[~2024-09-28  4:56 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-08-27 11:47 [PATCH v3 0/5] mm: memory_hotplug: improve do_migrate_range() Kefeng Wang
2024-08-27 11:47 ` [PATCH v3 1/5] mm: memory_hotplug: remove head variable in do_migrate_range() Kefeng Wang
2024-09-28  4:55   ` Matthew Wilcox [this message]
2024-09-28  8:34     ` David Hildenbrand
2024-09-28  8:39       ` David Hildenbrand
2024-09-29  1:16         ` Miaohe Lin
2024-09-29  2:04           ` Kefeng Wang
2024-09-29  2:19             ` Miaohe Lin
2024-09-30  9:25               ` David Hildenbrand
2024-10-09  7:27                 ` Kefeng Wang
2024-08-27 11:47 ` [PATCH v3 2/5] mm: memory-failure: add unmap_poisoned_folio() Kefeng Wang
2024-08-31  8:16   ` Miaohe Lin
2024-08-27 11:47 ` [PATCH v3 3/5] mm: memory_hotplug: check hwpoisoned page firstly in do_migrate_range() Kefeng Wang
2024-08-31  8:36   ` Miaohe Lin
2024-08-27 11:47 ` [PATCH v3 4/5] mm: migrate: add isolate_folio_to_list() Kefeng Wang
2024-08-27 11:47 ` [PATCH v3 5/5] mm: memory_hotplug: unify Huge/LRU/non-LRU movable folio isolation Kefeng Wang
2024-08-29 15:05   ` [PATCH v3 5-fix/5] mm: memory_hotplug: unify Huge/LRU/non-LRU movable folio isolation fix Kefeng Wang
2024-08-29 15:19     ` David Hildenbrand
2024-08-30  1:23       ` Kefeng Wang
2024-08-31  9:01   ` [PATCH v3 5/5] mm: memory_hotplug: unify Huge/LRU/non-LRU movable folio isolation Miaohe Lin

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=ZveMXByLE9LvO2z3@casper.infradead.org \
    --to=willy@infradead.org \
    --cc=Jonathan.Cameron@huawei.com \
    --cc=akpm@linux-foundation.org \
    --cc=dan.carpenter@linaro.org \
    --cc=david@redhat.com \
    --cc=linmiaohe@huawei.com \
    --cc=linux-mm@kvack.org \
    --cc=nao.horiguchi@gmail.com \
    --cc=osalvador@suse.de \
    --cc=wangkefeng.wang@huawei.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