From: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
To: "Liam R. Howlett" <Liam.Howlett@oracle.com>
Cc: Andrew Morton <akpm@linux-foundation.org>,
maple-tree@lists.infradead.org, linux-mm@kvack.org,
linux-kernel@vger.kernel.org,
David Hildenbrand <david@redhat.com>,
Vlastimil Babka <vbabka@suse.cz>,
Suren Baghdasaryan <surenb@google.com>,
Michal Hocko <mhocko@suse.com>, Jann Horn <jannh@google.com>,
Pedro Falcato <pfalcato@suse.de>,
Charan Teja Kalla <quic_charante@quicinc.com>,
shikemeng@huaweicloud.com, kasong@tencent.com, nphamcs@gmail.com,
bhe@redhat.com, baohua@kernel.org, chrisl@kernel.org,
Matthew Wilcox <willy@infradead.org>
Subject: Re: [PATCH v1 9/9] mm: Use unmap_desc struct for freeing page tables.
Date: Thu, 11 Sep 2025 11:06:29 +0100 [thread overview]
Message-ID: <2157a865-5481-43fd-8139-4b9662239eb4@lucifer.local> (raw)
In-Reply-To: <20250909190945.1030905-10-Liam.Howlett@oracle.com>
On Tue, Sep 09, 2025 at 03:09:45PM -0400, Liam R. Howlett wrote:
> Pass through the unmap_desc to free_pgtables() because it almost has
> everything necessary and is already on the stack.
>
> Updates testing code as necessary.
>
> No functional changes intended.
>
> Signed-off-by: Liam R. Howlett <Liam.Howlett@oracle.com>
To be clear, I _love_ the use of this helper struct. All commentary is more
so details. what you're doing here is great... :)
Other than nits below this looks fine so on basis of addressing them:
Reviewed-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
> ---
> mm/internal.h | 5 +----
> mm/memory.c | 21 ++++++++++-----------
> mm/mmap.c | 6 +++---
> mm/vma.c | 7 ++-----
> tools/testing/vma/vma_internal.h | 11 ++---------
> 5 files changed, 18 insertions(+), 32 deletions(-)
>
> diff --git a/mm/internal.h b/mm/internal.h
> index 1239944f2830a..f22329967e908 100644
> --- a/mm/internal.h
> +++ b/mm/internal.h
> @@ -445,10 +445,7 @@ bool __folio_end_writeback(struct folio *folio);
> void deactivate_file_folio(struct folio *folio);
> void folio_activate(struct folio *folio);
>
> -void free_pgtables(struct mmu_gather *tlb, struct ma_state *mas,
> - struct vm_area_struct *start_vma, unsigned long floor,
> - unsigned long ceiling, unsigned long tree_max,
> - bool mm_wr_locked);
> +void free_pgtables(struct mmu_gather *tlb, struct unmap_desc *desc);
>
> void pmd_install(struct mm_struct *mm, pmd_t *pmd, pgtable_t *pte);
>
> diff --git a/mm/memory.c b/mm/memory.c
> index 8d4d976311037..98c5ffd28a109 100644
> --- a/mm/memory.c
> +++ b/mm/memory.c
> @@ -387,15 +387,14 @@ void free_pgd_range(struct mmu_gather *tlb,
> * The tree_max differs from the ceiling when a dup_mmap() failed and the tree
> * has unrelated data to the mm_struct being torn down.
> */
> -void free_pgtables(struct mmu_gather *tlb, struct ma_state *mas,
> - struct vm_area_struct *vma, unsigned long floor,
> - unsigned long ceiling, unsigned long tree_max,
> - bool mm_wr_locked)
> +void free_pgtables(struct mmu_gather *tlb, struct unmap_desc *desc)
God this is nice.
> {
> struct unlink_vma_file_batch vb;
> + struct ma_state *mas = desc->mas;
> + struct vm_area_struct *vma = desc->first;
>
> /* underflow can happen and is fine */
> - WARN_ON_ONCE(tree_max - 1 > ceiling - 1);
> + WARN_ON_ONCE(desc->tree_max - 1 > desc->last_pgaddr - 1);
>
> tlb_free_vmas(tlb);
>
> @@ -407,13 +406,13 @@ void free_pgtables(struct mmu_gather *tlb, struct ma_state *mas,
> * Note: USER_PGTABLES_CEILING may be passed as ceiling and may
> * be 0. This will underflow and is okay.
> */
> - next = mas_find(mas, tree_max - 1);
> + next = mas_find(mas, desc->tree_max - 1);
>
> /*
> * Hide vma from rmap and truncate_pagecache before freeing
> * pgtables
> */
> - if (mm_wr_locked)
> + if (desc->mm_wr_locked)
> vma_start_write(vma);
> unlink_anon_vmas(vma);
>
> @@ -425,16 +424,16 @@ void free_pgtables(struct mmu_gather *tlb, struct ma_state *mas,
> */
> while (next && next->vm_start <= vma->vm_end + PMD_SIZE) {
> vma = next;
> - next = mas_find(mas, tree_max - 1);
> - if (mm_wr_locked)
> + next = mas_find(mas, desc->tree_max - 1);
> + if (desc->mm_wr_locked)
> vma_start_write(vma);
> unlink_anon_vmas(vma);
> unlink_file_vma_batch_add(&vb, vma);
> }
> unlink_file_vma_batch_final(&vb);
>
> - free_pgd_range(tlb, addr, vma->vm_end,
> - floor, next ? next->vm_start : ceiling);
> + free_pgd_range(tlb, addr, vma->vm_end, desc->first_pgaddr,
> + next ? next->vm_start : desc->last_pgaddr);
Hm, actually, isn't ceiling exclusive? So 'last_pgaddr' seems a bit
misleading?
> vma = next;
> } while (vma);
> }
> diff --git a/mm/mmap.c b/mm/mmap.c
> index 6011f62b0a294..9908481452780 100644
> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -1311,10 +1311,10 @@ void exit_mmap(struct mm_struct *mm)
> */
> mm_flags_set(MMF_OOM_SKIP, mm);
> mmap_write_lock(mm);
> + unmap.mm_wr_locked = true;
> mt_clear_in_rcu(&mm->mm_mt);
> - vma_iter_set(&vmi, vma->vm_end);
> - free_pgtables(&tlb, &vmi.mas, vma, FIRST_USER_ADDRESS,
> - USER_PGTABLES_CEILING, USER_PGTABLES_CEILING, true);
> + vma_iter_set(&vmi, unmap.tree_reset);
> + free_pgtables(&tlb, &unmap);
Lovely!
> tlb_finish_mmu(&tlb);
>
> /*
> diff --git a/mm/vma.c b/mm/vma.c
> index ad64cd9795ef3..ba155a539d160 100644
> --- a/mm/vma.c
> +++ b/mm/vma.c
> @@ -476,16 +476,13 @@ void remove_vma(struct vm_area_struct *vma)
> void unmap_region(struct unmap_desc *desc)
> {
> struct mm_struct *mm = desc->first->vm_mm;
> - struct ma_state *mas = desc->mas;
> struct mmu_gather tlb;
>
> tlb_gather_mmu(&tlb, mm);
> update_hiwater_rss(mm);
> unmap_vmas(&tlb, desc);
> - mas_set(mas, desc->tree_reset);
> - free_pgtables(&tlb, mas, desc->first, desc->first_pgaddr,
> - desc->last_pgaddr, desc->tree_max,
> - desc->mm_wr_locked);
> + mas_set(desc->mas, desc->tree_reset);
> + free_pgtables(&tlb, desc);
> tlb_finish_mmu(&tlb);
> }
>
> diff --git a/tools/testing/vma/vma_internal.h b/tools/testing/vma/vma_internal.h
> index d73ad4747d40a..435c5a24480bc 100644
> --- a/tools/testing/vma/vma_internal.h
> +++ b/tools/testing/vma/vma_internal.h
> @@ -892,17 +892,10 @@ static inline void unmap_vmas(struct mmu_gather *tlb, struct unmap_desc *unmap)
> (void)unmap;
> }
>
> -static inline void free_pgtables(struct mmu_gather *tlb, struct ma_state *mas,
> - struct vm_area_struct *vma, unsigned long floor,
> - unsigned long ceiling, unsigned long tree_max,
> - bool mm_wr_locked)
> +static inline void free_pgtables(struct mmu_gather *tlb, struct unmap_desc *desc)
> {
> (void)tlb;
> - (void)mas;
> - (void)vma;
> - (void)floor;
> - (void)ceiling;
> - (void)mm_wr_locked;
> + (void)desc;
> }
Again, let's drop the (void)'s here. Will let Vlasta fix up the conflict I
guess ;)
>
> static inline void mapping_unmap_writable(struct address_space *)
> --
> 2.47.2
>
prev parent reply other threads:[~2025-09-11 10:11 UTC|newest]
Thread overview: 43+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-09-09 19:09 [PATCH v1 0/9] Remove XA_ZERO from error recovery of dup_mmap() Liam R. Howlett
2025-09-09 19:09 ` [PATCH v1 1/9] mm/mmap: Move exit_mmap() trace point Liam R. Howlett
2025-09-09 20:07 ` Suren Baghdasaryan
2025-09-10 12:51 ` Pedro Falcato
2025-09-11 9:19 ` David Hildenbrand
2025-09-09 19:09 ` [PATCH v1 2/9] mm/mmap: Abstract vma clean up from exit_mmap() Liam R. Howlett
2025-09-09 20:09 ` Suren Baghdasaryan
2025-09-10 12:54 ` Pedro Falcato
2025-09-11 9:21 ` David Hildenbrand
2025-09-09 19:09 ` [PATCH v1 3/9] mm/vma: Add limits to unmap_region() for vmas Liam R. Howlett
2025-09-09 20:09 ` Suren Baghdasaryan
2025-09-10 12:56 ` Pedro Falcato
2025-09-11 8:56 ` Lorenzo Stoakes
2025-09-11 9:22 ` David Hildenbrand
2025-09-09 19:09 ` [PATCH v1 4/9] mm/memory: Add tree limit to free_pgtables() Liam R. Howlett
2025-09-09 21:05 ` Suren Baghdasaryan
2025-09-10 13:08 ` Pedro Falcato
2025-09-11 9:08 ` Lorenzo Stoakes
2025-09-11 9:24 ` David Hildenbrand
2025-09-09 19:09 ` [PATCH v1 5/9] mm/vma: Add page table limit to unmap_region() Liam R. Howlett
2025-09-09 21:29 ` Suren Baghdasaryan
2025-09-10 13:08 ` Pedro Falcato
2025-09-09 19:09 ` [PATCH v1 6/9] mm: Change dup_mmap() recovery Liam R. Howlett
2025-09-09 22:03 ` Suren Baghdasaryan
2025-09-10 13:23 ` Pedro Falcato
2025-09-11 9:13 ` Lorenzo Stoakes
2025-09-11 9:31 ` David Hildenbrand
2025-09-09 19:09 ` [PATCH v1 7/9] mm: Introduce unmap_desc struct to reduce function arguments Liam R. Howlett
2025-09-09 21:44 ` Suren Baghdasaryan
2025-09-11 9:22 ` Lorenzo Stoakes
2025-09-11 16:51 ` Suren Baghdasaryan
2025-09-11 16:56 ` Liam R. Howlett
2025-09-11 17:03 ` Suren Baghdasaryan
2025-09-10 13:10 ` Pedro Falcato
2025-09-11 9:20 ` Lorenzo Stoakes
2025-09-09 19:09 ` [PATCH v1 8/9] mm/vma: Use unmap_desc in vms_clear_ptes() and exit_mmap() Liam R. Howlett
2025-09-09 22:16 ` Suren Baghdasaryan
2025-09-11 16:59 ` Liam R. Howlett
2025-09-11 9:53 ` Lorenzo Stoakes
2025-09-12 5:08 ` kernel test robot
2025-09-09 19:09 ` [PATCH v1 9/9] mm: Use unmap_desc struct for freeing page tables Liam R. Howlett
2025-09-09 22:27 ` Suren Baghdasaryan
2025-09-11 10:06 ` Lorenzo Stoakes [this message]
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=2157a865-5481-43fd-8139-4b9662239eb4@lucifer.local \
--to=lorenzo.stoakes@oracle.com \
--cc=Liam.Howlett@oracle.com \
--cc=akpm@linux-foundation.org \
--cc=baohua@kernel.org \
--cc=bhe@redhat.com \
--cc=chrisl@kernel.org \
--cc=david@redhat.com \
--cc=jannh@google.com \
--cc=kasong@tencent.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=maple-tree@lists.infradead.org \
--cc=mhocko@suse.com \
--cc=nphamcs@gmail.com \
--cc=pfalcato@suse.de \
--cc=quic_charante@quicinc.com \
--cc=shikemeng@huaweicloud.com \
--cc=surenb@google.com \
--cc=vbabka@suse.cz \
--cc=willy@infradead.org \
/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