]> git.baikalelectronics.ru Git - kernel.git/commit
Revert "android: binder: stop saving a pointer to the VMA"
authorCarlos Llamas <cmllamas@google.com>
Tue, 2 May 2023 20:12:18 +0000 (20:12 +0000)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Tue, 30 May 2023 13:03:19 +0000 (14:03 +0100)
commit3aba0744b0aaf8d6c5219bc77c84fb8a43661a6a
tree8725e9e91c5218d5bdf60bdf39df69a368e686f1
parent404ae5b07c6397b9168fdd8d3e83995535990924
Revert "android: binder: stop saving a pointer to the VMA"

commit c0fd2101781ef761b636769b2f445351f71c3626 upstream.

This reverts commit 6b44656517074b125afb72ca19f63ebc55235761.

This patch fixed an issue reported by syzkaller in [1]. However, this
turned out to be only a band-aid in binder. The root cause, as bisected
by syzkaller, was fixed by commit 8c37e6862a52 ("mm/mmap: undo ->mmap()
when mas_preallocate() fails"). We no longer need the patch for binder.

Reverting such patch allows us to have a lockless access to alloc->vma
in specific cases where the mmap_lock is not required. This approach
avoids the contention that caused a performance regression.

[1] https://lore.kernel.org/all/0000000000004a0dbe05e1d749e0@google.com

[cmllamas: resolved conflicts with rework of alloc->mm and removal of
 binder_alloc_set_vma() also fixed comment section]

Fixes: 6b4465651707 ("android: binder: stop saving a pointer to the VMA")
Cc: Liam Howlett <liam.howlett@oracle.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Link: https://lore.kernel.org/r/20230502201220.1756319-2-cmllamas@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/android/binder_alloc.c
drivers/android/binder_alloc.h
drivers/android/binder_alloc_selftest.c