| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
This makes read O(1) and poll that was already O(1) becomes lockless.
Signed-off-by: Andrea Arcangeli <[email protected]>
Acked-by: Pavel Emelyanov <[email protected]>
Cc: Sanidhya Kashyap <[email protected]>
Cc: [email protected]
Cc: "Kirill A. Shutemov" <[email protected]>
Cc: Andres Lagar-Cavilla <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Peter Feiner <[email protected]>
Cc: "Dr. David Alan Gilbert" <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: "Huangpeng (Peter)" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
This is an optimization but it's a userland visible one and it affects
the API.
The downside of this optimization is that if you call poll() and you
get POLLIN, read(ufd) may still return -EAGAIN. The blocked userfault
may be waken by a different thread, before read(ufd) comes
around. This in short means that poll() isn't really usable if the
userfaultfd is opened in blocking mode.
userfaults won't wait in "pending" state to be read anymore and any
UFFDIO_WAKE or similar operations that has the objective of waking
userfaults after their resolution, will wake all blocked userfaults
for the resolved range, including those that haven't been read() by
userland yet.
The behavior of poll() becomes not standard, but this obviates the
need of "spurious" UFFDIO_WAKE and it lets the userland threads to
restart immediately without requiring an UFFDIO_WAKE. This is even
more significant in case of repeated faults on the same address from
multiple threads.
This optimization is justified by the measurement that the number of
spurious UFFDIO_WAKE accounts for 5% and 10% of the total
userfaults for heavy workloads, so it's worth optimizing those away.
Signed-off-by: Andrea Arcangeli <[email protected]>
Acked-by: Pavel Emelyanov <[email protected]>
Cc: Sanidhya Kashyap <[email protected]>
Cc: [email protected]
Cc: "Kirill A. Shutemov" <[email protected]>
Cc: Andres Lagar-Cavilla <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Peter Feiner <[email protected]>
Cc: "Dr. David Alan Gilbert" <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: "Huangpeng (Peter)" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
I had requests to return the full address (not the page aligned one) to
userland.
It's not entirely clear how the page offset could be relevant because
userfaults aren't like SIGBUS that can sigjump to a different place and it
actually skip resolving the fault depending on a page offset. There's
currently no real way to skip the fault especially because after a
UFFDIO_COPY|ZEROPAGE, the fault is optimized to be retried within the
kernel without having to return to userland first (not even self modifying
code replacing the .text that touched the faulting address would prevent
the fault to be repeated). Userland cannot skip repeating the fault even
more so if the fault was triggered by a KVM secondary page fault or any
get_user_pages or any copy-user inside some syscall which will return to
kernel code. The second time FAULT_FLAG_RETRY_NOWAIT won't be set leading
to a SIGBUS being raised because the userfault can't wait if it cannot
release the mmap_map first (and FAULT_FLAG_RETRY_NOWAIT is required for
that).
Still returning userland a proper structure during the read() on the uffd,
can allow to use the current UFFD_API for the future non-cooperative
extensions too and it looks cleaner as well. Once we get additional
fields there's no point to return the fault address page aligned anymore
to reuse the bits below PAGE_SHIFT.
The only downside is that the read() syscall will read 32bytes instead of
8bytes but that's not going to be measurable overhead.
The total number of new events that can be extended or of new future bits
for already shipped events, is limited to 64 by the features field of the
uffdio_api structure. If more will be needed a bump of UFFD_API will be
required.
[[email protected]: use __packed]
Signed-off-by: Andrea Arcangeli <[email protected]>
Acked-by: Pavel Emelyanov <[email protected]>
Cc: Sanidhya Kashyap <[email protected]>
Cc: [email protected]
Cc: "Kirill A. Shutemov" <[email protected]>
Cc: Andres Lagar-Cavilla <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Peter Feiner <[email protected]>
Cc: "Dr. David Alan Gilbert" <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: "Huangpeng (Peter)" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
This is (seems to be) the minimal thing that is required to unblock
standard uffd usage from the non-cooperative one. Now more bits can be
added to the features field indicating e.g. UFFD_FEATURE_FORK and others
needed for the latter use-case.
Signed-off-by: Pavel Emelyanov <[email protected]>
Signed-off-by: Andrea Arcangeli <[email protected]>
Cc: Sanidhya Kashyap <[email protected]>
Cc: [email protected]
Cc: "Kirill A. Shutemov" <[email protected]>
Cc: Andres Lagar-Cavilla <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Peter Feiner <[email protected]>
Cc: "Dr. David Alan Gilbert" <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: "Huangpeng (Peter)" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
Once an userfaultfd has been created and certain region of the process
virtual address space have been registered into it, the thread responsible
for doing the memory externalization can manage the page faults in
userland by talking to the kernel using the userfaultfd protocol.
poll() can be used to know when there are new pending userfaults to be
read (POLLIN).
Signed-off-by: Andrea Arcangeli <[email protected]>
Acked-by: Pavel Emelyanov <[email protected]>
Cc: Sanidhya Kashyap <[email protected]>
Cc: [email protected]
Cc: "Kirill A. Shutemov" <[email protected]>
Cc: Andres Lagar-Cavilla <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Peter Feiner <[email protected]>
Cc: "Dr. David Alan Gilbert" <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: "Huangpeng (Peter)" <[email protected]>
Cc: Dan Carpenter <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
If userfaultfd is armed on a certain vma we can't "fill" the holes with
zeroes or we'll break the userland on demand paging. The holes if the
userfault is armed, are really missing information (not zeroes) that the
userland has to load from network or elsewhere.
The same issue happens for wrprotected ptes that we can't just convert
into a single writable pmd_trans_huge.
We could however in theory still merge across zeropages if only
VM_UFFD_MISSING is set (so if VM_UFFD_WP is not set)... that could be
slightly improved but it'd be much more complex code for a tiny corner
case.
Signed-off-by: Andrea Arcangeli <[email protected]>
Acked-by: Pavel Emelyanov <[email protected]>
Cc: Sanidhya Kashyap <[email protected]>
Cc: [email protected]
Cc: "Kirill A. Shutemov" <[email protected]>
Cc: Andres Lagar-Cavilla <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Peter Feiner <[email protected]>
Cc: "Dr. David Alan Gilbert" <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: "Huangpeng (Peter)" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
vma->vm_userfaultfd_ctx is yet another vma parameter that vma_merge
must be aware about so that we can merge vmas back like they were
originally before arming the userfaultfd on some memory range.
Signed-off-by: Andrea Arcangeli <[email protected]>
Acked-by: Pavel Emelyanov <[email protected]>
Cc: Sanidhya Kashyap <[email protected]>
Cc: [email protected]
Cc: "Kirill A. Shutemov" <[email protected]>
Cc: Andres Lagar-Cavilla <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Peter Feiner <[email protected]>
Cc: "Dr. David Alan Gilbert" <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: "Huangpeng (Peter)" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
This is where the page faults must be modified to call
handle_userfault() if userfaultfd_missing() is true (so if the
vma->vm_flags had VM_UFFD_MISSING set).
handle_userfault() then takes care of blocking the page fault and
delivering it to userland.
The fault flags must also be passed as parameter so the "read|write"
kind of fault can be passed to userland.
Signed-off-by: Andrea Arcangeli <[email protected]>
Acked-by: Pavel Emelyanov <[email protected]>
Cc: Sanidhya Kashyap <[email protected]>
Cc: [email protected]
Cc: "Kirill A. Shutemov" <[email protected]>
Cc: Andres Lagar-Cavilla <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Peter Feiner <[email protected]>
Cc: "Dr. David Alan Gilbert" <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: "Huangpeng (Peter)" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
These two flags gets set in vma->vm_flags to tell the VM common code
if the userfaultfd is armed and in which mode (only tracking missing
faults, only tracking wrprotect faults or both). If neither flags is
set it means the userfaultfd is not armed on the vma.
Signed-off-by: Andrea Arcangeli <[email protected]>
Acked-by: Pavel Emelyanov <[email protected]>
Cc: Sanidhya Kashyap <[email protected]>
Cc: [email protected]
Cc: "Kirill A. Shutemov" <[email protected]>
Cc: Andres Lagar-Cavilla <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Peter Feiner <[email protected]>
Cc: "Dr. David Alan Gilbert" <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: "Huangpeng (Peter)" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
This adds the vm_userfaultfd_ctx to the vm_area_struct.
Signed-off-by: Andrea Arcangeli <[email protected]>
Acked-by: Pavel Emelyanov <[email protected]>
Cc: Sanidhya Kashyap <[email protected]>
Cc: [email protected]
Cc: "Kirill A. Shutemov" <[email protected]>
Cc: Andres Lagar-Cavilla <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Peter Feiner <[email protected]>
Cc: "Dr. David Alan Gilbert" <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: "Huangpeng (Peter)" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
Kernel header defining the methods needed by the VM common code to
interact with the userfaultfd.
Signed-off-by: Andrea Arcangeli <[email protected]>
Acked-by: Pavel Emelyanov <[email protected]>
Cc: Sanidhya Kashyap <[email protected]>
Cc: [email protected]
Cc: "Kirill A. Shutemov" <[email protected]>
Cc: Andres Lagar-Cavilla <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Peter Feiner <[email protected]>
Cc: "Dr. David Alan Gilbert" <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: "Huangpeng (Peter)" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
Defines the uAPI of the userfaultfd, notably the ioctl numbers and protocol.
Signed-off-by: Andrea Arcangeli <[email protected]>
Acked-by: Pavel Emelyanov <[email protected]>
Cc: Sanidhya Kashyap <[email protected]>
Cc: [email protected]
Cc: "Kirill A. Shutemov" <[email protected]>
Cc: Andres Lagar-Cavilla <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Peter Feiner <[email protected]>
Cc: "Dr. David Alan Gilbert" <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: "Huangpeng (Peter)" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
userfaultfd needs to wake all waitqueues (pass 0 as nr parameter), instead
of the current hardcoded 1 (that would wake just the first waitqueue in
the head list).
Signed-off-by: Andrea Arcangeli <[email protected]>
Acked-by: Pavel Emelyanov <[email protected]>
Cc: Sanidhya Kashyap <[email protected]>
Cc: [email protected]
Cc: "Kirill A. Shutemov" <[email protected]>
Cc: Andres Lagar-Cavilla <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Peter Feiner <[email protected]>
Cc: "Dr. David Alan Gilbert" <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: "Huangpeng (Peter)" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
This is the latest userfaultfd patchset. The postcopy live migration
feature on the qemu side is mostly ready to be merged and it entirely
depends on the userfaultfd syscall to be merged as well. So it'd be great
if this patchset could be reviewed for merging in -mm.
Userfaults allow to implement on demand paging from userland and more
generally they allow userland to more efficiently take control of the
behavior of page faults than what was available before (PROT_NONE +
SIGSEGV trap).
The use cases are:
1) KVM postcopy live migration (one form of cloud memory
externalization).
KVM postcopy live migration is the primary driver of this work:
http://blog.zhaw.ch/icclab/setting-up-post-copy-live-migration-in-openstack/
http://lists.gnu.org/archive/html/qemu-devel/2015-02/msg04873.html
2) postcopy live migration of binaries inside linux containers:
http://thread.gmane.org/gmane.linux.kernel.mm/132662
3) KVM postcopy live snapshotting (allowing to limit/throttle the
memory usage, unlike fork would, plus the avoidance of fork
overhead in the first place).
While the wrprotect tracking is not implemented yet, the syscall API is
already contemplating the wrprotect fault tracking and it's generic enough
to allow its later implementation in a backwards compatible fashion.
4) KVM userfaults on shared memory. The UFFDIO_COPY lowlevel method
should be extended to work also on tmpfs and then the
uffdio_register.ioctls will notify userland that UFFDIO_COPY is
available even when the registered virtual memory range is tmpfs
backed.
5) alternate mechanism to notify web browsers or apps on embedded
devices that volatile pages have been reclaimed. This basically
avoids the need to run a syscall before the app can access with the
CPU the virtual regions marked volatile. This depends on point 4)
to be fulfilled first, as volatile pages happily apply to tmpfs.
Even though there wasn't a real use case requesting it yet, it also
allows to implement distributed shared memory in a way that readonly
shared mappings can exist simultaneously in different hosts and they
can be become exclusive at the first wrprotect fault.
This patch (of 22):
Add documentation.
Signed-off-by: Andrea Arcangeli <[email protected]>
Acked-by: Pavel Emelyanov <[email protected]>
Cc: Sanidhya Kashyap <[email protected]>
Cc: [email protected]
Cc: "Kirill A. Shutemov" <[email protected]>
Cc: Andres Lagar-Cavilla <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Paolo Bonzini <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Andy Lutomirski <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Peter Feiner <[email protected]>
Cc: "Dr. David Alan Gilbert" <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: "Huangpeng (Peter)" <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
While debugging a networking issue, I hit a condition that triggered an
object to be freed into the wrong kmem cache, and thus triggered the
warning in cache_from_obj().
The arguments in the error message are in wrong order: the location
of the object's kmem cache is in cachep, not s.
Signed-off-by: Daniel Borkmann <[email protected]>
Cc: Christoph Lameter <[email protected]>
Cc: Pekka Enberg <[email protected]>
Cc: David Rientjes <[email protected]>
Cc: Joonsoo Kim <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
Description is almost copied from commit fb05e7a89f50 ("net: don't wait
for order-3 page allocation").
I saw excessive direct memory reclaim/compaction triggered by slub. This
causes performance issues and add latency. Slub uses high-order
allocation to reduce internal fragmentation and management overhead. But,
direct memory reclaim/compaction has high overhead and the benefit of
high-order allocation can't compensate the overhead of both work.
This patch makes auxiliary high-order allocation atomic. If there is no
memory pressure and memory isn't fragmented, the alloction will still
success, so we don't sacrifice high-order allocation's benefit here. If
the atomic allocation fails, direct memory reclaim/compaction will not be
triggered, allocation fallback to low-order immediately, hence the direct
memory reclaim/compaction overhead is avoided. In the allocation failure
case, kswapd is waken up and trying to make high-order freepages, so
allocation could success next time.
Following is the test to measure effect of this patch.
System: QEMU, CPU 8, 512 MB
Mem: 25% memory is allocated at random position to make fragmentation.
Memory-hogger occupies 150 MB memory.
Workload: hackbench -g 20 -l 1000
Average result by 10 runs (Base va Patched)
elapsed_time(s): 4.3468 vs 2.9838
compact_stall: 461.7 vs 73.6
pgmigrate_success: 28315.9 vs 7256.1
Signed-off-by: Joonsoo Kim <[email protected]>
Cc: Christoph Lameter <[email protected]>
Cc: Pekka Enberg <[email protected]>
Acked-by: David Rientjes <[email protected]>
Cc: Shaohua Li <[email protected]>
Cc: Vlastimil Babka <[email protected]>
Cc: Michal Hocko <[email protected]>
Cc: Eric Dumazet <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
sysfs_slab_add() shouldn't call kobject_put at error path: this puts last
reference of kmem-cache kobject and frees it. Kmem cache will be freed
second time at error path in kmem_cache_create().
For example this happens when slub debug was enabled in runtime and
somebody creates new kmem cache:
# echo 1 | tee /sys/kernel/slab/*/sanity_checks
# modprobe configfs
"configfs_dir_cache" cannot be merged because existing slab have debug and
cannot create new slab because unique name ":t-0000096" already taken.
Signed-off-by: Konstantin Khlebnikov <[email protected]>
Acked-by: Christoph Lameter <[email protected]>
Cc: Pekka Enberg <[email protected]>
Acked-by: David Rientjes <[email protected]>
Cc: Joonsoo Kim <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
Initializing a new slab can introduce rather large latencies because most
of the initialization runs always with interrupts disabled.
There is no point in doing so. The newly allocated slab is not visible
yet, so there is no reason to protect it against concurrent alloc/free.
Move the expensive parts of the initialization into allocate_slab(), so
for all allocations with GFP_WAIT set, interrupts are enabled.
Signed-off-by: Thomas Gleixner <[email protected]>
Acked-by: Christoph Lameter <[email protected]>
Cc: Pekka Enberg <[email protected]>
Acked-by: David Rientjes <[email protected]>
Cc: Joonsoo Kim <[email protected]>
Cc: Sebastian Andrzej Siewior <[email protected]>
Cc: Steven Rostedt <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
Per request of Joonsoo Kim adding kmem debug support.
I've tested that when debugging is disabled, then there is almost no
performance impact as this code basically gets removed by the compiler.
Need some guidance in enabling and testing this.
bulk- PREVIOUS - THIS-PATCH
1 - 43 cycles(tsc) 10.811 ns - 44 cycles(tsc) 11.236 ns improved -2.3%
2 - 27 cycles(tsc) 6.867 ns - 28 cycles(tsc) 7.019 ns improved -3.7%
3 - 21 cycles(tsc) 5.496 ns - 22 cycles(tsc) 5.526 ns improved -4.8%
4 - 24 cycles(tsc) 6.038 ns - 19 cycles(tsc) 4.786 ns improved 20.8%
8 - 17 cycles(tsc) 4.280 ns - 18 cycles(tsc) 4.572 ns improved -5.9%
16 - 17 cycles(tsc) 4.483 ns - 18 cycles(tsc) 4.658 ns improved -5.9%
30 - 18 cycles(tsc) 4.531 ns - 18 cycles(tsc) 4.568 ns improved 0.0%
32 - 58 cycles(tsc) 14.586 ns - 65 cycles(tsc) 16.454 ns improved -12.1%
34 - 53 cycles(tsc) 13.391 ns - 63 cycles(tsc) 15.932 ns improved -18.9%
48 - 65 cycles(tsc) 16.268 ns - 50 cycles(tsc) 12.506 ns improved 23.1%
64 - 53 cycles(tsc) 13.440 ns - 63 cycles(tsc) 15.929 ns improved -18.9%
128 - 79 cycles(tsc) 19.899 ns - 86 cycles(tsc) 21.583 ns improved -8.9%
158 - 90 cycles(tsc) 22.732 ns - 90 cycles(tsc) 22.552 ns improved 0.0%
250 - 95 cycles(tsc) 23.916 ns - 98 cycles(tsc) 24.589 ns improved -3.2%
Signed-off-by: Jesper Dangaard Brouer <[email protected]>
Cc: Christoph Lameter <[email protected]>
Cc: Pekka Enberg <[email protected]>
Cc: David Rientjes <[email protected]>
Cc: Joonsoo Kim <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
This implements SLUB specific kmem_cache_free_bulk(). SLUB allocator now
both have bulk alloc and free implemented.
Choose to reenable local IRQs while calling slowpath __slab_free(). In
worst case, where all objects hit slowpath call, the performance should
still be faster than fallback function __kmem_cache_free_bulk(), because
local_irq_{disable+enable} is very fast (7-cycles), while the fallback
invokes this_cpu_cmpxchg() which is slightly slower (9-cycles).
Nitpicking, this should be faster for N>=4, due to the entry cost of
local_irq_{disable+enable}.
Do notice that the save+restore variant is very expensive, this is key to
why this optimization works.
CPU: i7-4790K CPU @ 4.00GHz
* local_irq_{disable,enable}: 7 cycles(tsc) - 1.821 ns
* local_irq_{save,restore} : 37 cycles(tsc) - 9.443 ns
Measurements on CPU CPU i7-4790K @ 4.00GHz
Baseline normal fastpath (alloc+free cost): 43 cycles(tsc) 10.834 ns
Bulk- fallback - this-patch
1 - 58 cycles(tsc) 14.542 ns - 43 cycles(tsc) 10.811 ns improved 25.9%
2 - 50 cycles(tsc) 12.659 ns - 27 cycles(tsc) 6.867 ns improved 46.0%
3 - 48 cycles(tsc) 12.168 ns - 21 cycles(tsc) 5.496 ns improved 56.2%
4 - 47 cycles(tsc) 11.987 ns - 24 cycles(tsc) 6.038 ns improved 48.9%
8 - 46 cycles(tsc) 11.518 ns - 17 cycles(tsc) 4.280 ns improved 63.0%
16 - 45 cycles(tsc) 11.366 ns - 17 cycles(tsc) 4.483 ns improved 62.2%
30 - 45 cycles(tsc) 11.433 ns - 18 cycles(tsc) 4.531 ns improved 60.0%
32 - 75 cycles(tsc) 18.983 ns - 58 cycles(tsc) 14.586 ns improved 22.7%
34 - 71 cycles(tsc) 17.940 ns - 53 cycles(tsc) 13.391 ns improved 25.4%
48 - 80 cycles(tsc) 20.077 ns - 65 cycles(tsc) 16.268 ns improved 18.8%
64 - 71 cycles(tsc) 17.799 ns - 53 cycles(tsc) 13.440 ns improved 25.4%
128 - 91 cycles(tsc) 22.980 ns - 79 cycles(tsc) 19.899 ns improved 13.2%
158 - 100 cycles(tsc) 25.241 ns - 90 cycles(tsc) 22.732 ns improved 10.0%
250 - 102 cycles(tsc) 25.583 ns - 95 cycles(tsc) 23.916 ns improved 6.9%
Signed-off-by: Jesper Dangaard Brouer <[email protected]>
Cc: Christoph Lameter <[email protected]>
Cc: Pekka Enberg <[email protected]>
Cc: David Rientjes <[email protected]>
Cc: Joonsoo Kim <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
Call slowpath __slab_alloc() from within the bulk loop, as the side-effect
of this call likely repopulates c->freelist.
Choose to reenable local IRQs while calling slowpath.
Saving some optimizations for later. E.g. it is possible to extract
parts of __slab_alloc() and avoid the unnecessary and expensive (37
cycles) local_irq_{save,restore}. For now, be happy calling
__slab_alloc() this lower icache impact of this func and I don't have to
worry about correctness.
Measurements on CPU CPU i7-4790K @ 4.00GHz
Baseline normal fastpath (alloc+free cost): 42 cycles(tsc) 10.601 ns
Bulk- fallback - this-patch
1 - 58 cycles(tsc) 14.516 ns - 49 cycles(tsc) 12.459 ns improved 15.5%
2 - 51 cycles(tsc) 12.930 ns - 38 cycles(tsc) 9.605 ns improved 25.5%
3 - 49 cycles(tsc) 12.274 ns - 34 cycles(tsc) 8.525 ns improved 30.6%
4 - 48 cycles(tsc) 12.058 ns - 32 cycles(tsc) 8.036 ns improved 33.3%
8 - 46 cycles(tsc) 11.609 ns - 31 cycles(tsc) 7.756 ns improved 32.6%
16 - 45 cycles(tsc) 11.451 ns - 32 cycles(tsc) 8.148 ns improved 28.9%
30 - 79 cycles(tsc) 19.865 ns - 68 cycles(tsc) 17.164 ns improved 13.9%
32 - 76 cycles(tsc) 19.212 ns - 66 cycles(tsc) 16.584 ns improved 13.2%
34 - 74 cycles(tsc) 18.600 ns - 63 cycles(tsc) 15.954 ns improved 14.9%
48 - 88 cycles(tsc) 22.092 ns - 77 cycles(tsc) 19.373 ns improved 12.5%
64 - 80 cycles(tsc) 20.043 ns - 68 cycles(tsc) 17.188 ns improved 15.0%
128 - 99 cycles(tsc) 24.818 ns - 89 cycles(tsc) 22.404 ns improved 10.1%
158 - 99 cycles(tsc) 24.977 ns - 92 cycles(tsc) 23.089 ns improved 7.1%
250 - 106 cycles(tsc) 26.552 ns - 99 cycles(tsc) 24.785 ns improved 6.6%
Signed-off-by: Jesper Dangaard Brouer <[email protected]>
Cc: Christoph Lameter <[email protected]>
Cc: Pekka Enberg <[email protected]>
Cc: David Rientjes <[email protected]>
Cc: Joonsoo Kim <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
First piece: acceleration of retrieval of per cpu objects
If we are allocating lots of objects then it is advantageous to disable
interrupts and avoid the this_cpu_cmpxchg() operation to get these objects
faster.
Note that we cannot do the fast operation if debugging is enabled, because
we would have to add extra code to do all the debugging checks. And it
would not be fast anyway.
Note also that the requirement of having interrupts disabled avoids having
to do processor flag operations.
Allocate as many objects as possible in the fast way and then fall back to
the generic implementation for the rest of the objects.
Measurements on CPU CPU i7-4790K @ 4.00GHz
Baseline normal fastpath (alloc+free cost): 42 cycles(tsc) 10.554 ns
Bulk- fallback - this-patch
1 - 57 cycles(tsc) 14.432 ns - 48 cycles(tsc) 12.155 ns improved 15.8%
2 - 50 cycles(tsc) 12.746 ns - 37 cycles(tsc) 9.390 ns improved 26.0%
3 - 48 cycles(tsc) 12.180 ns - 33 cycles(tsc) 8.417 ns improved 31.2%
4 - 48 cycles(tsc) 12.015 ns - 32 cycles(tsc) 8.045 ns improved 33.3%
8 - 46 cycles(tsc) 11.526 ns - 30 cycles(tsc) 7.699 ns improved 34.8%
16 - 45 cycles(tsc) 11.418 ns - 32 cycles(tsc) 8.205 ns improved 28.9%
30 - 80 cycles(tsc) 20.246 ns - 73 cycles(tsc) 18.328 ns improved 8.8%
32 - 79 cycles(tsc) 19.946 ns - 72 cycles(tsc) 18.208 ns improved 8.9%
34 - 78 cycles(tsc) 19.659 ns - 71 cycles(tsc) 17.987 ns improved 9.0%
48 - 86 cycles(tsc) 21.516 ns - 82 cycles(tsc) 20.566 ns improved 4.7%
64 - 93 cycles(tsc) 23.423 ns - 89 cycles(tsc) 22.480 ns improved 4.3%
128 - 100 cycles(tsc) 25.170 ns - 99 cycles(tsc) 24.871 ns improved 1.0%
158 - 102 cycles(tsc) 25.549 ns - 101 cycles(tsc) 25.375 ns improved 1.0%
250 - 101 cycles(tsc) 25.344 ns - 100 cycles(tsc) 25.182 ns improved 1.0%
Signed-off-by: Christoph Lameter <[email protected]>
Signed-off-by: Jesper Dangaard Brouer <[email protected]>
Cc: Pekka Enberg <[email protected]>
Cc: David Rientjes <[email protected]>
Cc: Joonsoo Kim <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
Add the basic infrastructure for alloc/free operations on pointer arrays.
It includes a generic function in the common slab code that is used in
this infrastructure patch to create the unoptimized functionality for slab
bulk operations.
Allocators can then provide optimized allocation functions for situations
in which large numbers of objects are needed. These optimization may
avoid taking locks repeatedly and bypass metadata creation if all objects
in slab pages can be used to provide the objects required.
Allocators can extend the skeletons provided and add their own code to the
bulk alloc and free functions. They can keep the generic allocation and
freeing and just fall back to those if optimizations would not work (like
for example when debugging is on).
Signed-off-by: Christoph Lameter <[email protected]>
Signed-off-by: Jesper Dangaard Brouer <[email protected]>
Cc: Pekka Enberg <[email protected]>
Cc: David Rientjes <[email protected]>
Cc: Joonsoo Kim <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
With this patchset the SLUB allocator now has both bulk alloc and free
implemented.
This patchset mostly optimizes the "fastpath" where objects are available
on the per CPU fastpath page. This mostly amortize the less-heavy
none-locked cmpxchg_double used on fastpath.
The "fallback" bulking (e.g __kmem_cache_free_bulk) provides a good basis
for comparison. Measurements[1] of the fallback functions
__kmem_cache_{free,alloc}_bulk have been copied from slab_common.c and
forced "noinline" to force a function call like slab_common.c.
Measurements on CPU CPU i7-4790K @ 4.00GHz
Baseline normal fastpath (alloc+free cost): 42 cycles(tsc) 10.601 ns
Measurements last-patch with disabled debugging:
Bulk- fallback - this-patch
1 - 57 cycles(tsc) 14.448 ns - 44 cycles(tsc) 11.236 ns improved 22.8%
2 - 51 cycles(tsc) 12.768 ns - 28 cycles(tsc) 7.019 ns improved 45.1%
3 - 48 cycles(tsc) 12.232 ns - 22 cycles(tsc) 5.526 ns improved 54.2%
4 - 48 cycles(tsc) 12.025 ns - 19 cycles(tsc) 4.786 ns improved 60.4%
8 - 46 cycles(tsc) 11.558 ns - 18 cycles(tsc) 4.572 ns improved 60.9%
16 - 45 cycles(tsc) 11.458 ns - 18 cycles(tsc) 4.658 ns improved 60.0%
30 - 45 cycles(tsc) 11.499 ns - 18 cycles(tsc) 4.568 ns improved 60.0%
32 - 79 cycles(tsc) 19.917 ns - 65 cycles(tsc) 16.454 ns improved 17.7%
34 - 78 cycles(tsc) 19.655 ns - 63 cycles(tsc) 15.932 ns improved 19.2%
48 - 68 cycles(tsc) 17.049 ns - 50 cycles(tsc) 12.506 ns improved 26.5%
64 - 80 cycles(tsc) 20.009 ns - 63 cycles(tsc) 15.929 ns improved 21.3%
128 - 94 cycles(tsc) 23.749 ns - 86 cycles(tsc) 21.583 ns improved 8.5%
158 - 97 cycles(tsc) 24.299 ns - 90 cycles(tsc) 22.552 ns improved 7.2%
250 - 102 cycles(tsc) 25.681 ns - 98 cycles(tsc) 24.589 ns improved 3.9%
Benchmarking shows impressive improvements in the "fastpath" with a small
number of objects in the working set. Once the working set increases,
resulting in activating the "slowpath" (that contains the heavier locked
cmpxchg_double) the improvement decreases.
I'm currently working on also optimizing the "slowpath" (as network stack
use-case hits this), but this patchset should provide a good foundation
for further improvements. Rest of my patch queue in this area needs some
more work, but preliminary results are good. I'm attending Netfilter
Workshop[2] next week, and I'll hopefully return working on further
improvements in this area.
This patch (of 6):
s/succedd/succeed/
Signed-off-by: Jesper Dangaard Brouer <[email protected]>
Cc: Christoph Lameter <[email protected]>
Cc: Pekka Enberg <[email protected]>
Cc: David Rientjes <[email protected]>
Cc: Joonsoo Kim <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
Rename watchdog_suspend() to lockup_detector_suspend() and
watchdog_resume() to lockup_detector_resume() to avoid confusion with the
watchdog subsystem and to be consistent with the existing name
lockup_detector_init().
Also provide comment blocks to explain the watchdog_running and
watchdog_suspended variables and their relationship.
Signed-off-by: Ulrich Obergfell <[email protected]>
Reviewed-by: Aaron Tomlin <[email protected]>
Cc: Guenter Roeck <[email protected]>
Cc: Don Zickus <[email protected]>
Cc: Ulrich Obergfell <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Michal Hocko <[email protected]>
Cc: Stephane Eranian <[email protected]>
Cc: Chris Metcalf <[email protected]>
Cc: Frederic Weisbecker <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Ingo Molnar <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
Remove watchdog_nmi_disable_all() and watchdog_nmi_enable_all() since
these functions are no longer needed. If a subsystem has a need to
deactivate the watchdog temporarily, it should utilize the
watchdog_suspend() and watchdog_resume() functions.
[[email protected]: fix build with CONFIG_LOCKUP_DETECTOR=m]
Signed-off-by: Ulrich Obergfell <[email protected]>
Reviewed-by: Aaron Tomlin <[email protected]>
Cc: Guenter Roeck <[email protected]>
Cc: Don Zickus <[email protected]>
Cc: Ulrich Obergfell <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Michal Hocko <[email protected]>
Cc: Stephane Eranian <[email protected]>
Cc: Chris Metcalf <[email protected]>
Cc: Frederic Weisbecker <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Ingo Molnar <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
Remove update_watchdog() and restart_watchdog_hrtimer() since these
functions are no longer needed. Changes of parameters such as the sample
period are honored at the time when the watchdog threads are being
unparked.
Signed-off-by: Ulrich Obergfell <[email protected]>
Reviewed-by: Aaron Tomlin <[email protected]>
Cc: Guenter Roeck <[email protected]>
Cc: Don Zickus <[email protected]>
Cc: Ulrich Obergfell <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Michal Hocko <[email protected]>
Cc: Stephane Eranian <[email protected]>
Cc: Chris Metcalf <[email protected]>
Cc: Frederic Weisbecker <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Ingo Molnar <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
This interface can be utilized to deactivate the hard and soft lockup
detector temporarily. Callers are expected to minimize the duration of
deactivation. Multiple deactivations are allowed to occur in parallel but
should be rare in practice.
[[email protected]: remove unneeded static initialization]
Signed-off-by: Ulrich Obergfell <[email protected]>
Reviewed-by: Aaron Tomlin <[email protected]>
Cc: Guenter Roeck <[email protected]>
Cc: Don Zickus <[email protected]>
Cc: Ulrich Obergfell <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Michal Hocko <[email protected]>
Cc: Stephane Eranian <[email protected]>
Cc: Chris Metcalf <[email protected]>
Cc: Frederic Weisbecker <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Ingo Molnar <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
Originally watchdog_nmi_enable(cpu) and watchdog_nmi_disable(cpu) were
only called in watchdog thread context. However, the following commits
utilize these functions outside of watchdog thread context too.
commit 9809b18fcf6b8d8ec4d3643677345907e6b50eca
Author: Michal Hocko <[email protected]>
Date: Tue Sep 24 15:27:30 2013 -0700
watchdog: update watchdog_thresh properly
commit b3738d29323344da3017a91010530cf3a58590fc
Author: Stephane Eranian <[email protected]>
Date: Mon Nov 17 20:07:03 2014 +0100
watchdog: Add watchdog enable/disable all functions
Hence, it is now possible that these functions execute concurrently with
the same 'cpu' argument. This concurrency is problematic because per-cpu
'watchdog_ev' can be accessed/modified without adequate synchronization.
The patch series aims to address the above problem. However, instead of
introducing locks to protect per-cpu 'watchdog_ev' a different approach is
taken: Invoke these functions by parking and unparking the watchdog
threads (to ensure they are always called in watchdog thread context).
static struct smp_hotplug_thread watchdog_threads = {
...
.park = watchdog_disable, // calls watchdog_nmi_disable()
.unpark = watchdog_enable, // calls watchdog_nmi_enable()
};
Both previously mentioned commits call these functions in a similar way
and thus in principle contain some duplicate code. The patch series also
avoids this duplication by providing a commonly usable mechanism.
- Patch 1/4 introduces the watchdog_{park|unpark}_threads functions that
park/unpark all watchdog threads specified in 'watchdog_cpumask'. They
are intended to be called inside of kernel/watchdog.c only.
- Patch 2/4 introduces the watchdog_{suspend|resume} functions which can
be utilized by external callers to deactivate the hard and soft lockup
detector temporarily.
- Patch 3/4 utilizes watchdog_{park|unpark}_threads to replace some code
that was introduced by commit 9809b18fcf6b8d8ec4d3643677345907e6b50eca.
- Patch 4/4 utilizes watchdog_{suspend|resume} to replace some code that
was introduced by commit b3738d29323344da3017a91010530cf3a58590fc.
A few corner cases should be mentioned here for completeness.
- kthread_park() of watchdog/N could hang if cpu N is already locked up.
However, if watchdog is enabled the lockup will be detected anyway.
- kthread_unpark() of watchdog/N could hang if cpu N got locked up after
kthread_park(). The occurrence of this scenario should be _very_ rare
in practice, in particular because it is not expected that temporary
deactivation will happen frequently, and if it happens at all it is
expected that the duration of deactivation will be short.
This patch (of 4): introduce watchdog_park_threads() and watchdog_unpark_threads()
These functions are intended to be used only from inside kernel/watchdog.c
to park/unpark all watchdog threads that are specified in
watchdog_cpumask.
Signed-off-by: Ulrich Obergfell <[email protected]>
Reviewed-by: Aaron Tomlin <[email protected]>
Cc: Guenter Roeck <[email protected]>
Cc: Don Zickus <[email protected]>
Cc: Ulrich Obergfell <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Michal Hocko <[email protected]>
Cc: Stephane Eranian <[email protected]>
Cc: Chris Metcalf <[email protected]>
Cc: Frederic Weisbecker <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Ingo Molnar <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
The kernel's NMI watchdog has nothing to do with the watchdog subsystem.
Its header declarations should be in linux/nmi.h, not linux/watchdog.h.
The code provided two sets of dummy functions if HARDLOCKUP_DETECTOR is
not configured, one in the include file and one in kernel/watchdog.c.
Remove the dummy functions from kernel/watchdog.c and use those from the
include file.
Signed-off-by: Guenter Roeck <[email protected]>
Cc: Stephane Eranian <[email protected]>
Cc: Peter Zijlstra (Intel) <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Don Zickus <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
housekeeping_mask gathers all the CPUs that aren't part of the nohz_full
set. This is exactly what we want the watchdog to be affine to without
the need to use complicated cpumask operations.
Signed-off-by: Frederic Weisbecker <[email protected]>
Reviewed-by: Chris Metcalf <[email protected]>
Cc: Thomas Gleixner <[email protected]>
Cc: Chris Metcalf <[email protected]>
Cc: Don Zickus <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Ulrich Obergfell <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
It makes the registration cheaper and simpler for the smpboot per-cpu
kthread users that don't need to always update the cpumask after threads
creation.
[[email protected]: fix for allow passing the cpumask on per-cpu thread registration]
Signed-off-by: Frederic Weisbecker <[email protected]>
Reviewed-by: Chris Metcalf <[email protected]>
Reviewed-by: Thomas Gleixner <[email protected]>
Cc: Chris Metcalf <[email protected]>
Cc: Don Zickus <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Ulrich Obergfell <[email protected]>
Signed-off-by: Stephen Rothwell <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
The per-cpu kthread cleanup() callback is the mirror of the setup()
callback. When the per-cpu kthread is started, it first calls setup()
to initialize the resources which are then released by cleanup() when
the kthread exits.
Now since the introduction of a per-cpu kthread cpumask, the kthreads
excluded by the cpumask on boot may happen to be parked immediately
after their creation without taking the setup() stage, waiting to be
asked to unpark to do so. Then when smpboot_unregister_percpu_thread()
is later called, the kthread is stopped without having ever called
setup().
But this triggers a bug as the kthread unconditionally calls cleanup()
on exit but this doesn't mirror any setup(). Thus the kernel crashes
because we try to free resources that haven't been initialized, as in
the watchdog case:
WATCHDOG disable 0
WATCHDOG disable 1
WATCHDOG disable 2
BUG: unable to handle kernel NULL pointer dereference at (null)
IP: hrtimer_active+0x26/0x60
[...]
Call Trace:
hrtimer_try_to_cancel+0x1c/0x280
hrtimer_cancel+0x1d/0x30
watchdog_disable+0x56/0x70
watchdog_cleanup+0xe/0x10
smpboot_thread_fn+0x23c/0x2c0
kthread+0xf8/0x110
ret_from_fork+0x3f/0x70
This bug is currently masked with explicit kthread unparking before
kthread_stop() on smpboot_destroy_threads(). This forces a call to
setup() and then unpark().
We could fix this by unconditionally calling setup() on kthread entry.
But setup() isn't always cheap. In the case of watchdog it launches
hrtimer, perf events, etc... So we may as well like to skip it if there
are chances the kthread will never be used, as in a reduced cpumask value.
So let's simply do a state machine check before calling cleanup() that
makes sure setup() has been called before mirroring it.
And remove the nasty hack workaround.
Signed-off-by: Frederic Weisbecker <[email protected]>
Reviewed-by: Chris Metcalf <[email protected]>
Reviewed-by: Thomas Gleixner <[email protected]>
Cc: Chris Metcalf <[email protected]>
Cc: Don Zickus <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Ulrich Obergfell <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
The cpumask is allocated before threads get created. If the latter step
fails, we need to free the cpumask.
Signed-off-by: Frederic Weisbecker <[email protected]>
Reviewed-by: Chris Metcalf <[email protected]>
Reviewed-by: Thomas Gleixner <[email protected]>
Cc: Chris Metcalf <[email protected]>
Cc: Don Zickus <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Ulrich Obergfell <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
Many file systems that implement the show_options hook fail to correctly
escape their output which could lead to unescaped characters (e.g. new
lines) leaking into /proc/mounts and /proc/[pid]/mountinfo files. This
could lead to confusion, spoofed entries (resulting in things like
systemd issuing false d-bus "mount" notifications), and who knows what
else. This looks like it would only be the root user stepping on
themselves, but it's possible weird things could happen in containers or
in other situations with delegated mount privileges.
Here's an example using overlay with setuid fusermount trusting the
contents of /proc/mounts (via the /etc/mtab symlink). Imagine the use
of "sudo" is something more sneaky:
$ BASE="ovl"
$ MNT="$BASE/mnt"
$ LOW="$BASE/lower"
$ UP="$BASE/upper"
$ WORK="$BASE/work/ 0 0
none /proc fuse.pwn user_id=1000"
$ mkdir -p "$LOW" "$UP" "$WORK"
$ sudo mount -t overlay -o "lowerdir=$LOW,upperdir=$UP,workdir=$WORK" none /mnt
$ cat /proc/mounts
none /root/ovl/mnt overlay rw,relatime,lowerdir=ovl/lower,upperdir=ovl/upper,workdir=ovl/work/ 0 0
none /proc fuse.pwn user_id=1000 0 0
$ fusermount -u /proc
$ cat /proc/mounts
cat: /proc/mounts: No such file or directory
This fixes the problem by adding new seq_show_option and
seq_show_option_n helpers, and updating the vulnerable show_option
handlers to use them as needed. Some, like SELinux, need to be open
coded due to unusual existing escape mechanisms.
[[email protected]: add lost chunk, per Kees]
[[email protected]: seq_show_option should be using const parameters]
Signed-off-by: Kees Cook <[email protected]>
Acked-by: Serge Hallyn <[email protected]>
Acked-by: Jan Kara <[email protected]>
Acked-by: Paul Moore <[email protected]>
Cc: J. R. Okajima <[email protected]>
Signed-off-by: Kees Cook <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
NULL check before kfree is redundant and so clean them up.
Signed-off-by: Joseph Qi <[email protected]>
Reviewed-by: Mark Fasheh <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
These uses sometimes do and sometimes don't have '\n' terminations. Make
the uses consistently use '\n' terminations and remove the newline from
the functions.
Miscellanea:
o Coalesce formats
o Realign arguments
Signed-off-by: Joe Perches <[email protected]>
Reviewed-by: Mark Fasheh <[email protected]>
Cc: Joel Becker <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
While appending an extent to a file, it will call these functions:
ocfs2_insert_extent
-> call ocfs2_grow_tree() if there's no free rec
-> ocfs2_add_branch add a new branch to extent tree,
now rec[0] in the leaf of rightmost path is empty
-> ocfs2_do_insert_extent
-> ocfs2_rotate_tree_right
-> ocfs2_extend_rotate_transaction
-> jbd2_journal_restart if jbd2_journal_extend fail
-> ocfs2_insert_path
-> ocfs2_extend_trans
-> jbd2_journal_restart if jbd2_journal_extend fail
-> ocfs2_insert_at_leaf
-> ocfs2_et_update_clusters
Function jbd2_journal_restart() may be called and it may happened that
buffers dirtied in ocfs2_add_branch() are committed
while buffers dirtied in ocfs2_insert_at_leaf() and
ocfs2_et_update_clusters() are not.
So an empty rec[0] is left in rightmost path which will cause
read-only filesystem when call ocfs2_commit_truncate()
with the error message: "Inode %lu has an empty extent record".
This is not a serious problem, so remove the rightmost path when call
ocfs2_commit_truncate().
Signed-off-by: joyce.xue <[email protected]>
Reviewed-by: Mark Fasheh <[email protected]>
Cc: Joel Becker <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
ocfs2_write_end_nolock()
1: After we call ocfs2_journal_access_di() in ocfs2_write_begin(),
jbd2_journal_restart() may also be called, in this function transaction
A's t_updates-- and obtains a new transaction B. If
jbd2_journal_commit_transaction() is happened to commit transaction A,
when t_updates==0, it will continue to complete commit and unfile
buffer.
So when jbd2_journal_dirty_metadata(), the handle is pointed a new
transaction B, and the buffer head's journal head is already freed,
jh->b_transaction == NULL, jh->b_next_transaction == NULL, it returns
EINVAL, So it triggers the BUG_ON(status).
thread 1 jbd2
ocfs2_write_begin jbd2_journal_commit_transaction
ocfs2_write_begin_nolock
ocfs2_start_trans
jbd2__journal_start(t_updates+1,
transaction A)
ocfs2_journal_access_di
ocfs2_write_cluster_by_desc
ocfs2_mark_extent_written
ocfs2_change_extent_flag
ocfs2_split_extent
ocfs2_extend_rotate_transaction
jbd2_journal_restart
(t_updates-1,transaction B) t_updates==0
__jbd2_journal_refile_buffer
(jh->b_transaction = NULL)
ocfs2_write_end
ocfs2_write_end_nolock
ocfs2_journal_dirty
jbd2_journal_dirty_metadata(bug)
ocfs2_commit_trans
2. In ext4, I found that: jbd2_journal_get_write_access() called by
ext4_write_end.
ext4_write_begin
ext4_journal_start
__ext4_journal_start_sb
ext4_journal_check_start
jbd2__journal_start
ext4_write_end
ext4_mark_inode_dirty
ext4_reserve_inode_write
ext4_journal_get_write_access
jbd2_journal_get_write_access
ext4_mark_iloc_dirty
ext4_do_update_inode
ext4_handle_dirty_metadata
jbd2_journal_dirty_metadata
3. So I think we should put ocfs2_journal_access_di before
ocfs2_journal_dirty in the ocfs2_write_end. and it works well after my
modification.
Signed-off-by: vicky <[email protected]>
Reviewed-by: Mark Fasheh <[email protected]>
Cc: Joel Becker <[email protected]>
Cc: Zhangguanghui <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
o2hb_elapsed_msecs computes the time taken for a disk heartbeat.
'struct timeval' variables are used to store start and end times. On
32-bit systems, the 'tv_sec' component of 'struct timeval' will overflow
in year 2038 and beyond.
This patch solves the overflow with the following:
1. Replace o2hb_elapsed_msecs using 'ktime_t' values to measure start
and end time, and built-in function 'ktime_ms_delta' to compute the
elapsed time. ktime_get_real() is used since the code prints out the
wallclock time.
2. Changes format string to print time as a single 64-bit nanoseconds
value ("%lld") instead of seconds and microseconds. This simplifies
the code since converting ktime_t to that format would need expensive
computation. However, the debug log string is less readable than the
previous format.
Signed-off-by: Tina Ruchandani <[email protected]>
Suggested by: Arnd Bergmann <[email protected]>
Reviewed-by: Mark Fasheh <[email protected]>
Cc: Joel Becker <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
There is a race case between crashed dio and rm, which will lead to
OCFS2_VALID_FL not set read-only.
N1 N2
------------------------------------------------------------------------
dd with direct flag
rm file
crashed with an dio entry left
in orphan dir
clear OCFS2_VALID_FL in
ocfs2_remove_inode
recover N1 and read the corrupted inode,
and set filesystem read-only
So we skip the inode deletion this time and wait for dio entry recovered
first.
Signed-off-by: Joseph Qi <[email protected]>
Reviewed-by: Mark Fasheh <[email protected]>
Cc: Joel Becker <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
The following case will lead to a lockres is freed but is still in use.
cat /sys/kernel/debug/o2dlm/locking_state dlm_thread
lockres_seq_start
-> lock dlm->track_lock
-> get resA
resA->refs decrease to 0,
call dlm_lockres_release,
and wait for "cat" unlock.
Although resA->refs is already set to 0,
increase resA->refs, and then unlock
lock dlm->track_lock
-> list_del_init()
-> unlock
-> free resA
In such a race case, invalid address access may occurs. So we should
delete list res->tracking before resA->refs decrease to 0.
Signed-off-by: Yiwen Jiang <[email protected]>
Reviewed-by: Joseph Qi <[email protected]>
Cc: Joel Becker <[email protected]>
Signed-off-by: Mark Fasheh <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
This bug in mainline code is pointed out by Mark Fasheh. When
ocfs2_iop_set_acl() and ocfs2_iop_get_acl() are entered from VFS layer,
inode lock is not held. This seems to be regression from older kernels.
The patch is to fix that.
Orabug: 20189959
Signed-off-by: Tariq Saeed <[email protected]>
Reviewed-by: Mark Fasheh <[email protected]>
Cc: Joel Becker <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
PID: 614 TASK: ffff882a739da580 CPU: 3 COMMAND: "ocfs2dc"
#0 [ffff882ecc3759b0] machine_kexec at ffffffff8103b35d
#1 [ffff882ecc375a20] crash_kexec at ffffffff810b95b5
#2 [ffff882ecc375af0] oops_end at ffffffff815091d8
#3 [ffff882ecc375b20] die at ffffffff8101868b
#4 [ffff882ecc375b50] do_trap at ffffffff81508bb0
#5 [ffff882ecc375ba0] do_invalid_op at ffffffff810165e5
#6 [ffff882ecc375c40] invalid_op at ffffffff815116fb
[exception RIP: ocfs2_ci_checkpointed+208]
RIP: ffffffffa0a7e940 RSP: ffff882ecc375cf0 RFLAGS: 00010002
RAX: 0000000000000001 RBX: 000000000000654b RCX: ffff8812dc83f1f8
RDX: 00000000000017d9 RSI: ffff8812dc83f1f8 RDI: ffffffffa0b2c318
RBP: ffff882ecc375d20 R8: ffff882ef6ecfa60 R9: ffff88301f272200
R10: 0000000000000000 R11: 0000000000000000 R12: ffffffffffffffff
R13: ffff8812dc83f4f0 R14: 0000000000000000 R15: ffff8812dc83f1f8
ORIG_RAX: ffffffffffffffff CS: 0010 SS: 0018
#7 [ffff882ecc375d28] ocfs2_check_meta_downconvert at ffffffffa0a7edbd [ocfs2]
#8 [ffff882ecc375d38] ocfs2_unblock_lock at ffffffffa0a84af8 [ocfs2]
#9 [ffff882ecc375dc8] ocfs2_process_blocked_lock at ffffffffa0a85285 [ocfs2]
#10 [ffff882ecc375e18] ocfs2_downconvert_thread_do_work at ffffffffa0a85445 [ocfs2]
#11 [ffff882ecc375e68] ocfs2_downconvert_thread at ffffffffa0a854de [ocfs2]
#12 [ffff882ecc375ee8] kthread at ffffffff81090da7
#13 [ffff882ecc375f48] kernel_thread_helper at ffffffff81511884
assert is tripped because the tran is not checkpointed and the lock level is PR.
Some time ago, chmod command had been executed. As result, the following call
chain left the inode cluster lock in PR state, latter on causing the assert.
system_call_fastpath
-> my_chmod
-> sys_chmod
-> sys_fchmodat
-> notify_change
-> ocfs2_setattr
-> posix_acl_chmod
-> ocfs2_iop_set_acl
-> ocfs2_set_acl
-> ocfs2_acl_set_mode
Here is how.
1119 int ocfs2_setattr(struct dentry *dentry, struct iattr *attr)
1120 {
1247 ocfs2_inode_unlock(inode, 1); <<< WRONG thing to do.
..
1258 if (!status && attr->ia_valid & ATTR_MODE) {
1259 status = posix_acl_chmod(inode, inode->i_mode);
519 posix_acl_chmod(struct inode *inode, umode_t mode)
520 {
..
539 ret = inode->i_op->set_acl(inode, acl, ACL_TYPE_ACCESS);
287 int ocfs2_iop_set_acl(struct inode *inode, struct posix_acl *acl, ...
288 {
289 return ocfs2_set_acl(NULL, inode, NULL, type, acl, NULL, NULL);
224 int ocfs2_set_acl(handle_t *handle,
225 struct inode *inode, ...
231 {
..
252 ret = ocfs2_acl_set_mode(inode, di_bh,
253 handle, mode);
168 static int ocfs2_acl_set_mode(struct inode *inode, struct buffer_head ...
170 {
183 if (handle == NULL) {
>>> BUG: inode lock not held in ex at this point <<<
184 handle = ocfs2_start_trans(OCFS2_SB(inode->i_sb),
185 OCFS2_INODE_UPDATE_CREDITS);
ocfs2_setattr.#1247 we unlock and at #1259 call posix_acl_chmod. When we reach
ocfs2_acl_set_mode.#181 and do trans, the inode cluster lock is not held in EX
mode (it should be). How this could have happended?
We are the lock master, were holding lock EX and have released it in
ocfs2_setattr.#1247. Note that there are no holders of this lock at
this point. Another node needs the lock in PR, and we downconvert from
EX to PR. So the inode lock is PR when do the trans in
ocfs2_acl_set_mode.#184. The trans stays in core (not flushed to disc).
Now another node want the lock in EX, downconvert thread gets kicked
(the one that tripped assert abovt), finds an unflushed trans but the
lock is not EX (it is PR). If the lock was at EX, it would have flushed
the trans ocfs2_ci_checkpointed -> ocfs2_start_checkpoint before
downconverting (to NULL) for the request.
ocfs2_setattr must not drop inode lock ex in this code path. If it
does, takes it again before the trans, say in ocfs2_set_acl, another
cluster node can get in between, execute another setattr, overwriting
the one in progress on this node, resulting in a mode acl size combo
that is a mix of the two.
Orabug: 20189959
Signed-off-by: Tariq Saeed <[email protected]>
Reviewed-by: Mark Fasheh <[email protected]>
Cc: Joel Becker <[email protected]>
Cc: Joseph Qi <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
Currently error handling in dlm_request_join is a little obscure, so
optimize it to promote readability.
If packet.code is invalid, reset it to JOIN_DISALLOW to keep it
meaningful. It only influences the log printing.
Signed-off-by: Norton.Zhu <[email protected]>
Cc: Srinivas Eeda <[email protected]>
Reviewed-by: Mark Fasheh <[email protected]>
Cc: Joel Becker <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
When running dirop_fileop_racer we found a case that inode
can not removed.
Two nodes, say Node A and Node B, mount the same ocfs2 volume. Create
two dirs /race/1/ and /race/2/ in the filesystem.
Node A Node B
rm -r /race/2/
mv /race/1/ /race/2/
call ocfs2_unlink(), get
the EX mode of /race/2/
wait for B unlock /race/2/
decrease i_nlink of /race/2/ to 0,
and add inode of /race/2/ into
orphan dir, unlock /race/2/
got EX mode of /race/2/. because
/race/1/ is dir, so inc i_nlink
of /race/2/ and update into disk,
unlock /race/2/
because i_nlink of /race/2/
is not zero, this inode will
always remain in orphan dir
This patch fixes this case by test whether i_nlink of new dir is zero.
Signed-off-by: Yiwen Jiang <[email protected]>
Reviewed-by: Mark Fasheh <[email protected]>
Cc: Joel Becker <[email protected]>
Cc: Joseph Qi <[email protected]>
Cc: Xue jiufei <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
In ocfs2, ip_alloc_sem is used to protect allocation changes on the
node. In direct IO, we add ip_alloc_sem to protect date consistent
between direct-io and ocfs2_truncate_file race (buffer io use
ip_alloc_sem already). Although inode->i_mutex lock is used to avoid
concurrency of above situation, i think ip_alloc_sem is still needed
because protect allocation changes is significant.
Other filesystem like ext4 also uses rw_semaphore to protect data
consistent between get_block-vs-truncate race by other means, So
ip_alloc_sem in ocfs2 direct io is needed.
Signed-off-by: Weiwei Wang <[email protected]>
Signed-off-by: Mark Fasheh <[email protected]>
Cc: Joel Becker <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
In case a validation fails, clear the rest of the buffers and return the
error to the calling function.
This also facilitates bubbling up the error originating from ocfs2_error
to calling functions.
Signed-off-by: Goldwyn Rodrigues <[email protected]>
Reviewed-by: Mark Fasheh <[email protected]>
Cc: Joel Becker <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
Caveat: This may return -EROFS for a read case, which seems wrong. This
is happening even without this patch series though. Should we convert
EROFS to EIO?
Signed-off-by: Goldwyn Rodrigues <[email protected]>
Cc: Mark Fasheh <[email protected]>
Cc: Joel Becker <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
OCFS2 is often used in high-availaibility systems. However, ocfs2
converts the filesystem to read-only at the drop of the hat. This may
not be necessary, since turning the filesystem read-only would affect
other running processes as well, decreasing availability.
This attempt is to add errors=continue, which would return the EIO to
the calling process and terminate furhter processing so that the
filesystem is not corrupted further. However, the filesystem is not
converted to read-only.
As a future plan, I intend to create a small utility or extend
fsck.ocfs2 to fix small errors such as in the inode. The input to the
utility such as the inode can come from the kernel logs so we don't have
to schedule a downtime for fixing small-enough errors.
The patch changes the ocfs2_error to return an error. The error
returned depends on the mount option set. If none is set, the default
is to turn the filesystem read-only.
Perhaps errors=continue is not the best option name. Historically it is
used for making an attempt to progress in the current process itself.
Should we call it errors=eio? or errors=killproc? Suggestions/Comments
welcome.
Sources are available at:
https://github.com/goldwynr/linux/tree/error-cont
Signed-off-by: Goldwyn Rodrigues <[email protected]>
Signed-off-by: Mark Fasheh <[email protected]>
Cc: Joel Becker <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|