Per-VMA lock support for swap and userfaults
From: | Suren Baghdasaryan <surenb-AT-google.com> | |
To: | akpm-AT-linux-foundation.org | |
Subject: | [PATCH v6 0/6] Per-VMA lock support for swap and userfaults | |
Date: | Thu, 29 Jun 2023 19:04:29 -0700 | |
Message-ID: | <20230630020436.1066016-1-surenb@google.com> | |
Cc: | willy-AT-infradead.org, hannes-AT-cmpxchg.org, mhocko-AT-suse.com, josef-AT-toxicpanda.com, jack-AT-suse.cz, ldufour-AT-linux.ibm.com, laurent.dufour-AT-fr.ibm.com, michel-AT-lespinasse.org, liam.howlett-AT-oracle.com, jglisse-AT-google.com, vbabka-AT-suse.cz, minchan-AT-google.com, dave-AT-stgolabs.net, punit.agrawal-AT-bytedance.com, lstoakes-AT-gmail.com, hdanton-AT-sina.com, apopple-AT-nvidia.com, peterx-AT-redhat.com, ying.huang-AT-intel.com, david-AT-redhat.com, yuzhao-AT-google.com, dhowells-AT-redhat.com, hughd-AT-google.com, viro-AT-zeniv.linux.org.uk, brauner-AT-kernel.org, pasha.tatashin-AT-soleen.com, surenb-AT-google.com, linux-mm-AT-kvack.org, linux-fsdevel-AT-vger.kernel.org, linux-kernel-AT-vger.kernel.org, kernel-team-AT-android.com | |
Archive-link: | Article |
When per-VMA locks were introduced in [1] several types of page faults would still fall back to mmap_lock to keep the patchset simple. Among them are swap and userfault pages. The main reason for skipping those cases was the fact that mmap_lock could be dropped while handling these faults and that required additional logic to be implemented. Implement the mechanism to allow per-VMA locks to be dropped for these cases. First, change handle_mm_fault to drop per-VMA locks when returning VM_FAULT_RETRY or VM_FAULT_COMPLETED to be consistent with the way mmap_lock is handled. Then change folio_lock_or_retry to accept vm_fault and return vm_fault_t which simplifies later patches. Finally allow swap and uffd page faults to be handled under per-VMA locks by dropping per-VMA and retrying, the same way it's done under mmap_lock. Naturally, once VMA lock is dropped that VMA should be assumed unstable and can't be used. Changes since v5 posted at [2] - 6/6 moved changes in sanitize_fault_flags into 3/6, per Peter Xu - rebased over Linus' ToT Note: patch 3/6 will cause a trivial merge conflict in arch/arm64/mm/fault.c when applied over mm-unstable branch due to a patch from ARM64 tree [3] which is missing in mm-unstable. [1] https://lore.kernel.org/all/20230227173632.3292573-1-sure... [2] https://lore.kernel.org/all/20230628172529.744839-1-suren... [3] https://lore.kernel.org/all/20230524131305.2808-1-jszhang... Suren Baghdasaryan (6): swap: remove remnants of polling from read_swap_cache_async mm: add missing VM_FAULT_RESULT_TRACE name for VM_FAULT_COMPLETED mm: drop per-VMA lock when returning VM_FAULT_RETRY or VM_FAULT_COMPLETED mm: change folio_lock_or_retry to use vm_fault directly mm: handle swap page faults under per-VMA lock mm: handle userfaults under VMA lock arch/arm64/mm/fault.c | 3 ++- arch/powerpc/mm/fault.c | 3 ++- arch/s390/mm/fault.c | 3 ++- arch/x86/mm/fault.c | 3 ++- fs/userfaultfd.c | 34 ++++++++++++---------------- include/linux/mm.h | 37 ++++++++++++++++++++++++++++++ include/linux/mm_types.h | 3 ++- include/linux/pagemap.h | 9 ++++---- mm/filemap.c | 37 +++++++++++++++--------------- mm/madvise.c | 4 ++-- mm/memory.c | 49 ++++++++++++++++++++++------------------ mm/swap.h | 1 - mm/swap_state.c | 12 ++++------ 13 files changed, 118 insertions(+), 80 deletions(-) -- 2.41.0.255.g8b1d071c50-goog