linux-kernel - [PATCH v7 0/9] binder: faster page installations

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <20241210143114.661252-1-cmllamas@google.com>
Date: Tue, 10 Dec 2024 14:30:56 +0000
From: Carlos Llamas <cmllamas@...gle.com>
To: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Cc: linux-kernel@...r.kernel.org, kernel-team@...roid.com, 
	Carlos Llamas <cmllamas@...gle.com>, Alice Ryhl <aliceryhl@...gle.com>, 
	Barry Song <v-songbaohua@...o.com>, David Hildenbrand <david@...hat.com>, 
	"Liam R. Howlett" <Liam.Howlett@...cle.com>, Matthew Wilcox <willy@...radead.org>, 
	Minchan Kim <minchan@...nel.org>, Nhat Pham <nphamcs@...il.com>, 
	Johannes Weiner <hannes@...xchg.org>, Suren Baghdasaryan <surenb@...gle.com>, Todd Kjos <tkjos@...gle.com>, 
	Viktor Martensson <vmartensson@...gle.com>, Hillf Danton <hdanton@...a.com>, 
	Lorenzo Stoakes <lorenzo.stoakes@...cle.com>
Subject: [PATCH v7 0/9] binder: faster page installations

The main focus of these patches is to improve the performance of binder
page installations, primarily by reducing contention on the mmap_lock.
The idea is to allow concurrent page insertion by leveraging per-vma
locking and get_user_pages_remote().

Unfortunately, this required reverting the alloc->lock spinlock back
into a mutex in order to serialize with the shrinker. At least until
finding a better solution e.g. support page zapping with a spinlock.
The trade off is still quite worth it though.

Other patches are also included that remove unsafe and redundant things
such as the alloc->vma pointer or the struct binder_lru_page concept.

Note: I'll work on setting up a page fault handler for binder next.
I believe an idea from Alice Ryhl to deferred the page insertions will
make this finally feasible. I only need to figure out a few performance
bits but if/when done most of the manual page insertion code in binder
could be dropped. :)

Changelog:

v7:
* cast page->private to struct binder_shrinker_mdata before kfree()

v6:
https://lore.kernel.org/all/20241203215452.2820071-1-cmllamas@google.com/
* rework the usage of page->lru and page->index into only page->private
* expanded comments aobut vma validation using alloc->mapped per Suren

v5:
https://lore.kernel.org/all/20241126184021.45292-1-cmllamas@google.com/
* rebase and resolve conflicts with commit da0c02516c50 ("mm/list_lru:
  simplify the list_lru walk callback function")
* add comment to binder_alloc_is_mapped() about valid vma check
* add note to commit log about future page->lru removal
* collect Reviewed-by tags from Suren

v4:
https://lore.kernel.org/all/20241119183250.3497547-1-cmllamas@google.com/
 * add explicit FOLL_NOFAULT to get_user_pages_remote()
 * per-vma locking to shrinker path (per Liam)

v3:
https://lore.kernel.org/all/20241108191057.3288442-1-cmllamas@google.com/
 * collect "Reviewed-by" tags from Suren
 * use full commit subject in revert (not only sha1)
 * add "goto unlock" label for vma_lookup() failures
 * address -ENOMEM error override in separate patch
 * squash "remove alloc->vma" patch into alloc->mapped patch
 * pass 'struct binder_alloc *' to  binder_page_lookup() too
 * factor out individual mmget_not_zero()/mmput_async() calls
 * cleanup binder_page_insert() to avoid using goto
 * document only one mapping allowed per binder instance
 * check binder_alloc_is_mapped() in binder_page_lookup()
 * remove no longer need local page_to_free pointer

v2:
https://lore.kernel.org/all/20241107040239.2847143-1-cmllamas@google.com/
 * fix locking order when upgrading from vma lock to mmap lock
 * switch folio_walk_start() for get_user_pages_remote()
 * release vma/mmap locks and mmput() right after vm_insert_page()
 * add binder_page_alloc() helper for binder_install_single_page()

v1:
https://lore.kernel.org/all/20241105200258.2380168-1-cmllamas@google.com/

Cc: Alice Ryhl <aliceryhl@...gle.com>
Cc: Barry Song <v-songbaohua@...o.com>
Cc: David Hildenbrand <david@...hat.com>
Cc: Liam R. Howlett <Liam.Howlett@...cle.com>
Cc: Matthew Wilcox <willy@...radead.org>
Cc: Minchan Kim <minchan@...nel.org>
Cc: Nhat Pham <nphamcs@...il.com>
Cc: Johannes Weiner <hannes@...xchg.org>
Cc: Suren Baghdasaryan <surenb@...gle.com>
Cc: Todd Kjos <tkjos@...gle.com>
Cc: Viktor Martensson <vmartensson@...gle.com>
Cc: Hillf Danton <hdanton@...a.com>
Cc: Lorenzo Stoakes <lorenzo.stoakes@...cle.com>

Carlos Llamas (9):
  Revert "binder: switch alloc->mutex to spinlock_t"
  binder: concurrent page installation
  binder: select correct nid for pages in LRU
  binder: store shrinker metadata under page->private
  binder: replace alloc->vma with alloc->mapped
  binder: rename alloc->buffer to vm_start
  binder: use per-vma lock in page installation
  binder: propagate vm_insert_page() errors
  binder: use per-vma lock in page reclaiming

 drivers/android/binder.c                |   2 +-
 drivers/android/binder_alloc.c          | 366 +++++++++++++++---------
 drivers/android/binder_alloc.h          |  45 +--
 drivers/android/binder_alloc_selftest.c |  18 +-
 drivers/android/binder_trace.h          |   2 +-
 5 files changed, 269 insertions(+), 164 deletions(-)

-- 
2.47.0.338.g60cca15819-goog