openwrt/staging/blogic.git
5 years agomm/hugetlb: get rid of NODEMASK_ALLOC
Oscar Salvador [Tue, 14 May 2019 00:19:23 +0000 (17:19 -0700)]
mm/hugetlb: get rid of NODEMASK_ALLOC

NODEMASK_ALLOC is used to allocate a nodemask bitmap, and it does it by
first determining whether it should be allocated on the stack or
dynamically, depending on NODES_SHIFT.  Right now, it goes the dynamic
path whenever the nodemask_t is above 32 bytes.

Although we could bump it to a reasonable value, the largest a nodemask_t
can get is 128 bytes, so since __nr_hugepages_store_common is called from
a rather short stack we can just get rid of the NODEMASK_ALLOC call here.

This reduces some code churn and complexity.

Link: http://lkml.kernel.org/r/20190402133415.21983-1-osalvador@suse.de
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Alex Ghiti <alex@ghiti.fr>
Cc: David Rientjes <rientjes@google.com>
Cc: Jing Xiangfeng <jingxiangfeng@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agohugetlbfs: fix potential over/underflow setting node specific nr_hugepages
Mike Kravetz [Tue, 14 May 2019 00:19:20 +0000 (17:19 -0700)]
hugetlbfs: fix potential over/underflow setting node specific nr_hugepages

The number of node specific huge pages can be set via a file such as:
/sys/devices/system/node/node1/hugepages/hugepages-2048kB/nr_hugepages
When a node specific value is specified, the global number of huge pages
must also be adjusted.  This adjustment is calculated as the specified
node specific value + (global value - current node value).  If the node
specific value provided by the user is large enough, this calculation
could overflow an unsigned long leading to a smaller than expected number
of huge pages.

To fix, check the calculation for overflow.  If overflow is detected, use
ULONG_MAX as the requested value.  This is inline with the user request to
allocate as many huge pages as possible.

It was also noticed that the above calculation was done outside the
hugetlb_lock.  Therefore, the values could be inconsistent and result in
underflow.  To fix, the calculation is moved within the routine
set_max_huge_pages() where the lock is held.

In addition, the code in __nr_hugepages_store_common() which tries to
handle the case of not being able to allocate a node mask would likely
result in incorrect behavior.  Luckily, it is very unlikely we will ever
take this path.  If we do, simply return ENOMEM.

Link: http://lkml.kernel.org/r/20190328220533.19884-1-mike.kravetz@oracle.com
Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com>
Reported-by: Jing Xiangfeng <jingxiangfeng@huawei.com>
Reviewed-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Cc: David Rientjes <rientjes@google.com>
Cc: Alex Ghiti <alex@ghiti.fr>
Cc: Jing Xiangfeng <jingxiangfeng@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomem-hotplug: fix node spanned pages when we have a node with only ZONE_MOVABLE
Linxu Fang [Tue, 14 May 2019 00:19:17 +0000 (17:19 -0700)]
mem-hotplug: fix node spanned pages when we have a node with only ZONE_MOVABLE

342332e6a925 ("mm/page_alloc.c: introduce kernelcore=mirror option") and
later patches rewrote the calculation of node spanned pages.

e506b99696a2 ("mem-hotplug: fix node spanned pages when we have a movable
node"), but the current code still has problems,

When we have a node with only zone_movable and the node id is not zero,
the size of node spanned pages is double added.

That's because we have an empty normal zone, and zone_start_pfn or
zone_end_pfn is not between arch_zone_lowest_possible_pfn and
arch_zone_highest_possible_pfn, so we need to use clamp to constrain the
range just like the commit <96e907d13602> (bootmem: Reimplement
__absent_pages_in_range() using for_each_mem_pfn_range()).

e.g.
Zone ranges:
  DMA      [mem 0x0000000000001000-0x0000000000ffffff]
  DMA32    [mem 0x0000000001000000-0x00000000ffffffff]
  Normal   [mem 0x0000000100000000-0x000000023fffffff]
Movable zone start for each node
  Node 0: 0x0000000100000000
  Node 1: 0x0000000140000000
Early memory node ranges
  node   0: [mem 0x0000000000001000-0x000000000009efff]
  node   0: [mem 0x0000000000100000-0x00000000bffdffff]
  node   0: [mem 0x0000000100000000-0x000000013fffffff]
  node   1: [mem 0x0000000140000000-0x000000023fffffff]

node 0 DMA spanned:0xfff   present:0xf9e   absent:0x61
node 0 DMA32 spanned:0xff000 present:0xbefe0 absent:0x40020
node 0 Normal spanned:0 present:0 absent:0
node 0 Movable spanned:0x40000 present:0x40000 absent:0
On node 0 totalpages(node_present_pages): 1048446
node_spanned_pages:1310719
node 1 DMA spanned:0     present:0 absent:0
node 1 DMA32 spanned:0     present:0 absent:0
node 1 Normal spanned:0x100000    present:0x100000 absent:0
node 1 Movable spanned:0x100000    present:0x100000 absent:0
On node 1 totalpages(node_present_pages): 2097152
node_spanned_pages:2097152
Memory: 6967796K/12582392K available (16388K kernel code, 3686K rwdata,
4468K rodata, 2160K init, 10444K bss, 5614596K reserved, 0K
cma-reserved)

It shows that the current memory of node 1 is double added.
After this patch, the problem is fixed.

node 0 DMA spanned:0xfff   present:0xf9e   absent:0x61
node 0 DMA32 spanned:0xff000 present:0xbefe0 absent:0x40020
node 0 Normal spanned:0 present:0 absent:0
node 0 Movable spanned:0x40000 present:0x40000 absent:0
On node 0 totalpages(node_present_pages): 1048446
node_spanned_pages:1310719
node 1 DMA spanned:0     present:0 absent:0
node 1 DMA32 spanned:0     present:0 absent:0
node 1 Normal spanned:0     present:0 absent:0
node 1 Movable spanned:0x100000    present:0x100000 absent:0
On node 1 totalpages(node_present_pages): 1048576
node_spanned_pages:1048576
memory: 6967796K/8388088K available (16388K kernel code, 3686K rwdata,
4468K rodata, 2160K init, 10444K bss, 1420292K reserved, 0K
cma-reserved)

Link: http://lkml.kernel.org/r/1554178276-10372-1-git-send-email-fanglinxu@huawei.com
Signed-off-by: Linxu Fang <fanglinxu@huawei.com>
Cc: Taku Izumi <izumi.taku@jp.fujitsu.com>
Cc: Xishi Qiu <qiuxishi@huawei.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Pavel Tatashin <pavel.tatashin@microsoft.com>
Cc: Oscar Salvador <osalvador@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/vmscan: drop may_writepage and classzone_idx from direct reclaim begin template
Yafang Shao [Tue, 14 May 2019 00:19:14 +0000 (17:19 -0700)]
mm/vmscan: drop may_writepage and classzone_idx from direct reclaim begin template

There are three tracepoints using this template, which are
mm_vmscan_direct_reclaim_begin,
mm_vmscan_memcg_reclaim_begin,
mm_vmscan_memcg_softlimit_reclaim_begin.

Regarding mm_vmscan_direct_reclaim_begin,
sc.may_writepage is !laptop_mode, that's a static setting, and
reclaim_idx is derived from gfp_mask which is already show in this
tracepoint.

Regarding mm_vmscan_memcg_reclaim_begin,
may_writepage is !laptop_mode too, and reclaim_idx is (MAX_NR_ZONES-1),
which are both static value.

mm_vmscan_memcg_softlimit_reclaim_begin is the same with
mm_vmscan_memcg_reclaim_begin.

So we can drop them all.

Link: http://lkml.kernel.org/r/1553736322-32235-1-git-send-email-laoar.shao@gmail.com
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: page_mkclean vs MADV_DONTNEED race
Aneesh Kumar K.V [Tue, 14 May 2019 00:19:11 +0000 (17:19 -0700)]
mm: page_mkclean vs MADV_DONTNEED race

MADV_DONTNEED is handled with mmap_sem taken in read mode.  We call
page_mkclean without holding mmap_sem.

MADV_DONTNEED implies that pages in the region are unmapped and subsequent
access to the pages in that range is handled as a new page fault.  This
implies that if we don't have parallel access to the region when
MADV_DONTNEED is run we expect those range to be unallocated.

w.r.t page_mkclean() we need to make sure that we don't break the
MADV_DONTNEED semantics.  MADV_DONTNEED check for pmd_none without holding
pmd_lock.  This implies we skip the pmd if we temporarily mark pmd none.
Avoid doing that while marking the page clean.

Keep the sequence same for dax too even though we don't support
MADV_DONTNEED for dax mapping

The bug was noticed by code review and I didn't observe any failures w.r.t
test run.  This is similar to

commit 58ceeb6bec86d9140f9d91d71a710e963523d063
Author: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Date:   Thu Apr 13 14:56:26 2017 -0700

    thp: fix MADV_DONTNEED vs. MADV_FREE race

commit ced108037c2aa542b3ed8b7afd1576064ad1362a
Author: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Date:   Thu Apr 13 14:56:20 2017 -0700

    thp: fix MADV_DONTNEED vs. numa balancing race

Link: http://lkml.kernel.org/r/20190321040610.14226-1-aneesh.kumar@linux.ibm.com
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc:"Kirill A . Shutemov" <kirill@shutemov.name>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: introduce put_user_page*(), placeholder versions
John Hubbard [Tue, 14 May 2019 00:19:08 +0000 (17:19 -0700)]
mm: introduce put_user_page*(), placeholder versions

A discussion of the overall problem is below.

As mentioned in patch 0001, the steps are to fix the problem are:

1) Provide put_user_page*() routines, intended to be used
   for releasing pages that were pinned via get_user_pages*().

2) Convert all of the call sites for get_user_pages*(), to
   invoke put_user_page*(), instead of put_page(). This involves dozens of
   call sites, and will take some time.

3) After (2) is complete, use get_user_pages*() and put_user_page*() to
   implement tracking of these pages. This tracking will be separate from
   the existing struct page refcounting.

4) Use the tracking and identification of these pages, to implement
   special handling (especially in writeback paths) when the pages are
   backed by a filesystem.

Overview
========

Some kernel components (file systems, device drivers) need to access
memory that is specified via process virtual address.  For a long time,
the API to achieve that was get_user_pages ("GUP") and its variations.
However, GUP has critical limitations that have been overlooked; in
particular, GUP does not interact correctly with filesystems in all
situations.  That means that file-backed memory + GUP is a recipe for
potential problems, some of which have already occurred in the field.

GUP was first introduced for Direct IO (O_DIRECT), allowing filesystem
code to get the struct page behind a virtual address and to let storage
hardware perform a direct copy to or from that page.  This is a
short-lived access pattern, and as such, the window for a concurrent
writeback of GUP'd page was small enough that there were not (we think)
any reported problems.  Also, userspace was expected to understand and
accept that Direct IO was not synchronized with memory-mapped access to
that data, nor with any process address space changes such as munmap(),
mremap(), etc.

Over the years, more GUP uses have appeared (virtualization, device
drivers, RDMA) that can keep the pages they get via GUP for a long period
of time (seconds, minutes, hours, days, ...).  This long-term pinning
makes an underlying design problem more obvious.

In fact, there are a number of key problems inherent to GUP:

Interactions with file systems
==============================

File systems expect to be able to write back data, both to reclaim pages,
and for data integrity.  Allowing other hardware (NICs, GPUs, etc) to gain
write access to the file memory pages means that such hardware can dirty
the pages, without the filesystem being aware.  This can, in some cases
(depending on filesystem, filesystem options, block device, block device
options, and other variables), lead to data corruption, and also to kernel
bugs of the form:

    kernel BUG at /build/linux-fQ94TU/linux-4.4.0/fs/ext4/inode.c:1899!
    backtrace:
        ext4_writepage
        __writepage
        write_cache_pages
        ext4_writepages
        do_writepages
        __writeback_single_inode
        writeback_sb_inodes
        __writeback_inodes_wb
        wb_writeback
        wb_workfn
        process_one_work
        worker_thread
        kthread
        ret_from_fork

...which is due to the file system asserting that there are still buffer
heads attached:

        ({                                                      \
                BUG_ON(!PagePrivate(page));                     \
                ((struct buffer_head *)page_private(page));     \
        })

Dave Chinner's description of this is very clear:

    "The fundamental issue is that ->page_mkwrite must be called on every
    write access to a clean file backed page, not just the first one.
    How long the GUP reference lasts is irrelevant, if the page is clean
    and you need to dirty it, you must call ->page_mkwrite before it is
    marked writeable and dirtied. Every. Time."

This is just one symptom of the larger design problem: real filesystems
that actually write to a backing device, do not actually support
get_user_pages() being called on their pages, and letting hardware write
directly to those pages--even though that pattern has been going on since
about 2005 or so.

Long term GUP
=============

Long term GUP is an issue when FOLL_WRITE is specified to GUP (so, a
writeable mapping is created), and the pages are file-backed.  That can
lead to filesystem corruption.  What happens is that when a file-backed
page is being written back, it is first mapped read-only in all of the CPU
page tables; the file system then assumes that nobody can write to the
page, and that the page content is therefore stable.  Unfortunately, the
GUP callers generally do not monitor changes to the CPU pages tables; they
instead assume that the following pattern is safe (it's not):

    get_user_pages()

    Hardware can keep a reference to those pages for a very long time,
    and write to it at any time.  Because "hardware" here means "devices
    that are not a CPU", this activity occurs without any interaction with
    the kernel's file system code.

    for each page
        set_page_dirty
        put_page()

In fact, the GUP documentation even recommends that pattern.

Anyway, the file system assumes that the page is stable (nothing is
writing to the page), and that is a problem: stable page content is
necessary for many filesystem actions during writeback, such as checksum,
encryption, RAID striping, etc.  Furthermore, filesystem features like COW
(copy on write) or snapshot also rely on being able to use a new page for
as memory for that memory range inside the file.

Corruption during write back is clearly possible here.  To solve that, one
idea is to identify pages that have active GUP, so that we can use a
bounce page to write stable data to the filesystem.  The filesystem would
work on the bounce page, while any of the active GUP might write to the
original page.  This would avoid the stable page violation problem, but
note that it is only part of the overall solution, because other problems
remain.

Other filesystem features that need to replace the page with a new one can
be inhibited for pages that are GUP-pinned.  This will, however, alter and
limit some of those filesystem features.  The only fix for that would be
to require GUP users to monitor and respond to CPU page table updates.
Subsystems such as ODP and HMM do this, for example.  This aspect of the
problem is still under discussion.

Direct IO
=========

Direct IO can cause corruption, if userspace does Direct-IO that writes to
a range of virtual addresses that are mmap'd to a file.  The pages written
to are file-backed pages that can be under write back, while the Direct IO
is taking place.  Here, Direct IO races with a write back: it calls GUP
before page_mkclean() has replaced the CPU pte with a read-only entry.
The race window is pretty small, which is probably why years have gone by
before we noticed this problem: Direct IO is generally very quick, and
tends to finish up before the filesystem gets around to do anything with
the page contents.  However, it's still a real problem.  The solution is
to never let GUP return pages that are under write back, but instead,
force GUP to take a write fault on those pages.  That way, GUP will
properly synchronize with the active write back.  This does not change the
required GUP behavior, it just avoids that race.

Details
=======

Introduces put_user_page(), which simply calls put_page().  This provides
a way to update all get_user_pages*() callers, so that they call
put_user_page(), instead of put_page().

Also introduces put_user_pages(), and a few dirty/locked variations, as a
replacement for release_pages(), and also as a replacement for open-coded
loops that release multiple pages.  These may be used for subsequent
performance improvements, via batching of pages to be released.

This is the first step of fixing a problem (also described in [1] and [2])
with interactions between get_user_pages ("gup") and filesystems.

Problem description: let's start with a bug report.  Below, is what
happens sometimes, under memory pressure, when a driver pins some pages
via gup, and then marks those pages dirty, and releases them.  Note that
the gup documentation actually recommends that pattern.  The problem is
that the filesystem may do a writeback while the pages were gup-pinned,
and then the filesystem believes that the pages are clean.  So, when the
driver later marks the pages as dirty, that conflicts with the
filesystem's page tracking and results in a BUG(), like this one that I
experienced:

    kernel BUG at /build/linux-fQ94TU/linux-4.4.0/fs/ext4/inode.c:1899!
    backtrace:
        ext4_writepage
        __writepage
        write_cache_pages
        ext4_writepages
        do_writepages
        __writeback_single_inode
        writeback_sb_inodes
        __writeback_inodes_wb
        wb_writeback
        wb_workfn
        process_one_work
        worker_thread
        kthread
        ret_from_fork

...which is due to the file system asserting that there are still buffer
heads attached:

        ({                                                      \
                BUG_ON(!PagePrivate(page));                     \
                ((struct buffer_head *)page_private(page));     \
        })

Dave Chinner's description of this is very clear:

    "The fundamental issue is that ->page_mkwrite must be called on
    every write access to a clean file backed page, not just the first
    one.  How long the GUP reference lasts is irrelevant, if the page is
    clean and you need to dirty it, you must call ->page_mkwrite before it
    is marked writeable and dirtied.  Every.  Time."

This is just one symptom of the larger design problem: real filesystems
that actually write to a backing device, do not actually support
get_user_pages() being called on their pages, and letting hardware write
directly to those pages--even though that pattern has been going on since
about 2005 or so.

The steps are to fix it are:

1) (This patch): provide put_user_page*() routines, intended to be used
   for releasing pages that were pinned via get_user_pages*().

2) Convert all of the call sites for get_user_pages*(), to
   invoke put_user_page*(), instead of put_page(). This involves dozens of
   call sites, and will take some time.

3) After (2) is complete, use get_user_pages*() and put_user_page*() to
   implement tracking of these pages. This tracking will be separate from
   the existing struct page refcounting.

4) Use the tracking and identification of these pages, to implement
   special handling (especially in writeback paths) when the pages are
   backed by a filesystem.

[1] https://lwn.net/Articles/774411/ : "DMA and get_user_pages()"
[2] https://lwn.net/Articles/753027/ : "The Trouble with get_user_pages()"

Link: http://lkml.kernel.org/r/20190327023632.13307-2-jhubbard@nvidia.com
Signed-off-by: John Hubbard <jhubbard@nvidia.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Mike Rapoport <rppt@linux.ibm.com> [docs]
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Jérôme Glisse <jglisse@redhat.com>
Reviewed-by: Christoph Lameter <cl@linux.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Ralph Campbell <rcampbell@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agohugetlb: allow to free gigantic pages regardless of the configuration
Alexandre Ghiti [Tue, 14 May 2019 00:19:04 +0000 (17:19 -0700)]
hugetlb: allow to free gigantic pages regardless of the configuration

On systems without CONTIG_ALLOC activated but that support gigantic pages,
boottime reserved gigantic pages can not be freed at all.  This patch
simply enables the possibility to hand back those pages to memory
allocator.

Link: http://lkml.kernel.org/r/20190327063626.18421-5-alex@ghiti.fr
Signed-off-by: Alexandre Ghiti <alex@ghiti.fr>
Acked-by: David S. Miller <davem@davemloft.net> [sparc]
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Andy Lutomirsky <luto@kernel.org>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: "H . Peter Anvin" <hpa@zytor.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: simplify MEMORY_ISOLATION && COMPACTION || CMA into CONTIG_ALLOC
Alexandre Ghiti [Tue, 14 May 2019 00:19:00 +0000 (17:19 -0700)]
mm: simplify MEMORY_ISOLATION && COMPACTION || CMA into CONTIG_ALLOC

This condition allows to define alloc_contig_range, so simplify it into a
more accurate naming.

Link: http://lkml.kernel.org/r/20190327063626.18421-4-alex@ghiti.fr
Signed-off-by: Alexandre Ghiti <alex@ghiti.fr>
Suggested-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andy Lutomirsky <luto@kernel.org>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: "H . Peter Anvin" <hpa@zytor.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agosparc: advertise gigantic page support
Alexandre Ghiti [Tue, 14 May 2019 00:18:56 +0000 (17:18 -0700)]
sparc: advertise gigantic page support

sparc actually supports gigantic pages and selecting
ARCH_HAS_GIGANTIC_PAGE allows it to allocate and free gigantic pages at
runtime.

sparc allows configuration such as huge pages of 16GB, pages of 8KB and
MAX_ORDER = 13 (default): HPAGE_SHIFT (34) - PAGE_SHIFT (13) = 21 >=
MAX_ORDER (13)

Link: http://lkml.kernel.org/r/20190327063626.18421-3-alex@ghiti.fr
Signed-off-by: Alexandre Ghiti <alex@ghiti.fr>
Acked-by: David S. Miller <davem@davemloft.net>
Cc: Andy Lutomirsky <luto@kernel.org>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: "H . Peter Anvin" <hpa@zytor.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agosh: advertise gigantic page support
Alexandre Ghiti [Tue, 14 May 2019 00:18:53 +0000 (17:18 -0700)]
sh: advertise gigantic page support

Patch series "Fix free/allocation of runtime gigantic pages", v8.

This series fixes sh and sparc that did not advertise their gigantic page
support and then were not able to allocate and free those pages at
runtime.  It renames MEMORY_ISOLATION && COMPACTION || CMA condition into
the more accurate CONTIG_ALLOC, since it allows the definition of
alloc_contig_range function.

Finally, it then fixes the wrong definition of ARCH_HAS_GIGANTIC_PAGE
config that, without MEMORY_ISOLATION && COMPACTION || CMA defined, did
not allow architectures to free boottime allocated gigantic pages although
unrelated.

This patch (of 4):

sh actually supports gigantic pages and selecting ARCH_HAS_GIGANTIC_PAGE
allows it to allocate and free gigantic pages at runtime.

At least sdk7786_defconfig exposes such a configuration with huge pages of
64MB, pages of 4KB and MAX_ORDER = 11: HPAGE_SHIFT (26) - PAGE_SHIFT (12)
= 14 >= MAX_ORDER (11)

Link: http://lkml.kernel.org/r/20190327063626.18421-2-alex@ghiti.fr
Signed-off-by: Alexandre Ghiti <alex@ghiti.fr>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Rich Felker <dalias@libc.org>
Cc: David S. Miller <davem@davemloft.net>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: "H . Peter Anvin" <hpa@zytor.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andy Lutomirsky <luto@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoriscv: switch over to generic free_initmem()
Mike Rapoport [Tue, 14 May 2019 00:18:49 +0000 (17:18 -0700)]
riscv: switch over to generic free_initmem()

The riscv version of free_initmem() differs from the generic one only in
that it sets the freed memory to zero.

Make ricsv use the generic version and poison the freed memory.

Link: http://lkml.kernel.org/r/1550515285-17446-5-git-send-email-rppt@linux.ibm.com
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Reviewed-by: Palmer Dabbelt <palmer@sifive.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Richard Kuo <rkuo@codeaurora.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoinit: free_initmem: poison freed init memory
Mike Rapoport [Tue, 14 May 2019 00:18:46 +0000 (17:18 -0700)]
init: free_initmem: poison freed init memory

Various architectures including x86 poison the freed init memory.  Do the
same in the generic free_initmem implementation and switch sparc32
architecture that is identical to the generic code over to it now.

Link: http://lkml.kernel.org/r/1550515285-17446-4-git-send-email-rppt@linux.ibm.com
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Palmer Dabbelt <palmer@sifive.com>
Cc: Richard Kuo <rkuo@codeaurora.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agohexagon: switch over to generic free_initmem()
Mike Rapoport [Tue, 14 May 2019 00:18:43 +0000 (17:18 -0700)]
hexagon: switch over to generic free_initmem()

hexagon implementation of free_initmem() is currently empty and marked
with comment

 * Todo:  free pages between __init_begin and __init_end; possibly
 * some devtree related stuff as well.

Switch it to the generic implementation.

Link: http://lkml.kernel.org/r/1550515285-17446-3-git-send-email-rppt@linux.ibm.com
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Palmer Dabbelt <palmer@sifive.com>
Cc: Richard Kuo <rkuo@codeaurora.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoinit: provide a generic free_initmem implementation
Mike Rapoport [Tue, 14 May 2019 00:18:40 +0000 (17:18 -0700)]
init: provide a generic free_initmem implementation

Patch series "provide a generic free_initmem implementation", v2.

Many architectures implement free_initmem() in exactly the same or very
similar way: they wrap the call to free_initmem_default() with sometimes
different 'poison' parameter.

These patches switch those architectures to use a generic implementation
that does free_initmem_default(POISON_FREE_INITMEM).

This was inspired by Christoph's patches for free_initrd_mem [1] and I
shamelessly copied changelog entries from his patches :)

[1] https://lore.kernel.org/lkml/20190213174621.29297-1-hch@lst.de/

This patch (of 2):

For most architectures free_initmem just a wrapper for the same
free_initmem_default(-1) call.  Provide that as a generic implementation
marked __weak.

Link: http://lkml.kernel.org/r/1550515285-17446-2-git-send-email-rppt@linux.ibm.com
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Palmer Dabbelt <palmer@sifive.com>
Cc: Richard Kuo <rkuo@codeaurora.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoinitramfs: poison freed initrd memory
Christoph Hellwig [Tue, 14 May 2019 00:18:37 +0000 (17:18 -0700)]
initramfs: poison freed initrd memory

Various architectures including x86 poison the freed initrd memory.  Do
the same in the generic free_initrd_mem implementation and switch a few
more architectures that are identical to the generic code over to it now.

Link: http://lkml.kernel.org/r/20190213174621.29297-9-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com> [arm64]
Cc: Geert Uytterhoeven <geert@linux-m68k.org> [m68k]
Cc: Steven Price <steven.price@arm.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Guan Xuetao <gxt@pku.edu.cn>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoinitramfs: provide a generic free_initrd_mem implementation
Christoph Hellwig [Tue, 14 May 2019 00:18:34 +0000 (17:18 -0700)]
initramfs: provide a generic free_initrd_mem implementation

For most architectures free_initrd_mem just expands to the same
free_reserved_area call.  Provide that as a generic implementation marked
__weak.

Link: http://lkml.kernel.org/r/20190213174621.29297-8-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Geert Uytterhoeven <geert@linux-m68k.org> [m68k]
Acked-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com> [arm64]
Cc: Steven Price <steven.price@arm.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Guan Xuetao <gxt@pku.edu.cn>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoinitramfs: move the legacy keepinitrd parameter to core code
Christoph Hellwig [Tue, 14 May 2019 00:18:30 +0000 (17:18 -0700)]
initramfs: move the legacy keepinitrd parameter to core code

No need to handle the freeing disable in arch code when we already have a
core hook (and a different name for the option) for it.

Link: http://lkml.kernel.org/r/20190213174621.29297-7-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Catalin Marinas <catalin.marinas@arm.com> [arm64]
Acked-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org> [m68k]
Cc: Steven Price <steven.price@arm.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Guan Xuetao <gxt@pku.edu.cn>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoinitramfs: cleanup populate_rootfs
Christoph Hellwig [Tue, 14 May 2019 00:18:27 +0000 (17:18 -0700)]
initramfs: cleanup populate_rootfs

The code for kernels that support ramdisks or not is mostly the same.
Unify it by using an IS_ENABLED for the info message, and moving the error
message into a stub for populate_initrd_image.

[cai@lca.pw: fix a compilation error]
Link: http://lkml.kernel.org/r/20190328014806.36375-1-cai@lca.pw
Link: http://lkml.kernel.org/r/20190213174621.29297-6-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Qian Cai <cai@lca.pw>
Acked-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com> [arm64]
Cc: Geert Uytterhoeven <geert@linux-m68k.org> [m68k]
Cc: Steven Price <steven.price@arm.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Guan Xuetao <gxt@pku.edu.cn>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoinitramfs: factor out a helper to populate the initrd image
Christoph Hellwig [Tue, 14 May 2019 00:18:24 +0000 (17:18 -0700)]
initramfs: factor out a helper to populate the initrd image

This will allow for cleaner code sharing in the caller.

Link: http://lkml.kernel.org/r/20190213174621.29297-5-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com> [arm64]
Cc: Geert Uytterhoeven <geert@linux-m68k.org> [m68k]
Cc: Steven Price <steven.price@arm.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Guan Xuetao <gxt@pku.edu.cn>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoinitramfs: cleanup initrd freeing
Christoph Hellwig [Tue, 14 May 2019 00:18:21 +0000 (17:18 -0700)]
initramfs: cleanup initrd freeing

Factor the kexec logic into a separate helper, and then inline the rest of
free_initrd into the only caller.

Link: http://lkml.kernel.org/r/20190213174621.29297-4-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com> [arm64]
Cc: Geert Uytterhoeven <geert@linux-m68k.org> [m68k]
Cc: Steven Price <steven.price@arm.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Guan Xuetao <gxt@pku.edu.cn>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoinitramfs: free initrd memory if opening /initrd.image fails
Christoph Hellwig [Tue, 14 May 2019 00:18:17 +0000 (17:18 -0700)]
initramfs: free initrd memory if opening /initrd.image fails

Patch series "initramfs tidyups".

I've spent some time chasing down behavior in initramfs and found
plenty of opportunity to improve the code.  A first stab on that is
contained in this series.

This patch (of 7):

We free the initrd memory for all successful or error cases except for the
case where opening /initrd.image fails, which looks like an oversight.

Steven said:

: This also changes the behaviour when CONFIG_INITRAMFS_FORCE is enabled
: - specifically it means that the initrd is freed (previously it was
: ignored and never freed).  But that seems like reasonable behaviour and
: the previous behaviour looks like another oversight.

Link: http://lkml.kernel.org/r/20190213174621.29297-3-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Steven Price <steven.price@arm.com>
Acked-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com> [arm64]
Cc: Geert Uytterhoeven <geert@linux-m68k.org> [m68k]
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Guan Xuetao <gxt@pku.edu.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/cma.c: fix crash on CMA allocation if bitmap allocation fails
Yue Hu [Tue, 14 May 2019 00:18:14 +0000 (17:18 -0700)]
mm/cma.c: fix crash on CMA allocation if bitmap allocation fails

f022d8cb7ec7 ("mm: cma: Don't crash on allocation if CMA area can't be
activated") fixes the crash issue when activation fails via setting
cma->count as 0, same logic exists if bitmap allocation fails.

Link: http://lkml.kernel.org/r/20190325081309.6004-1-zbestahu@gmail.com
Signed-off-by: Yue Hu <huyue2@yulong.com>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Laura Abbott <labbott@redhat.com>
Cc: Mike Rapoport <rppt@linux.vnet.ibm.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: memcontrol: quarantine the mem_cgroup_[node_]nr_lru_pages() API
Johannes Weiner [Tue, 14 May 2019 00:18:11 +0000 (17:18 -0700)]
mm: memcontrol: quarantine the mem_cgroup_[node_]nr_lru_pages() API

Only memcg_numa_stat_show() uses those wrappers and the lru bitmasks,
group them together.

Link: http://lkml.kernel.org/r/20190228163020.24100-7-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Roman Gushchin <guro@fb.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: memcontrol: push down mem_cgroup_nr_lru_pages()
Johannes Weiner [Tue, 14 May 2019 00:18:08 +0000 (17:18 -0700)]
mm: memcontrol: push down mem_cgroup_nr_lru_pages()

mem_cgroup_nr_lru_pages() is just a convenience wrapper around
memcg_page_state() that takes bitmasks of lru indexes and aggregates the
counts for those.

Replace callsites where the bitmask is simple enough with direct
memcg_page_state() call(s).

Link: http://lkml.kernel.org/r/20190228163020.24100-6-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Roman Gushchin <guro@fb.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: memcontrol: push down mem_cgroup_node_nr_lru_pages()
Johannes Weiner [Tue, 14 May 2019 00:18:05 +0000 (17:18 -0700)]
mm: memcontrol: push down mem_cgroup_node_nr_lru_pages()

mem_cgroup_node_nr_lru_pages() is just a convenience wrapper around
lruvec_page_state() that takes bitmasks of lru indexes and aggregates the
counts for those.

Replace callsites where the bitmask is simple enough with direct
lruvec_page_state() calls.

This removes the last extern user of mem_cgroup_node_nr_lru_pages(), so
make that function private again, too.

Link: http://lkml.kernel.org/r/20190228163020.24100-5-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Roman Gushchin <guro@fb.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: memcontrol: replace node summing with memcg_page_state()
Johannes Weiner [Tue, 14 May 2019 00:18:03 +0000 (17:18 -0700)]
mm: memcontrol: replace node summing with memcg_page_state()

Instead of adding up the node counters, use memcg_page_state() to get the
memcg state directly.  This is a bit cheaper and more stream-lined.

Link: http://lkml.kernel.org/r/20190228163020.24100-4-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Roman Gushchin <guro@fb.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: memcontrol: replace zone summing with lruvec_page_state()
Johannes Weiner [Tue, 14 May 2019 00:18:00 +0000 (17:18 -0700)]
mm: memcontrol: replace zone summing with lruvec_page_state()

Instead of adding up the zone counters, use lruvec_page_state() to get the
node state directly.  This is a bit cheaper and more stream-lined.

Link: http://lkml.kernel.org/r/20190228163020.24100-3-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Roman Gushchin <guro@fb.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: memcontrol: track LRU counts in the vmstats array
Johannes Weiner [Tue, 14 May 2019 00:17:57 +0000 (17:17 -0700)]
mm: memcontrol: track LRU counts in the vmstats array

Patch series "mm: memcontrol: clean up the LRU counts tracking".

The memcg LRU stats usage is currently a bit messy.  Memcg has private
per-zone counters because reclaim needs zone granularity sometimes, but we
also have plenty of users that need to awkwardly sum them up to node or
memcg granularity.  Meanwhile the canonical per-memcg vmstats do not track
the LRU counts (NR_INACTIVE_ANON etc.) as you'd expect.

This series enables LRU count tracking in the per-memcg vmstats array such
that lruvec_page_state() and memcg_page_state() work on the enum
node_stat_item items for the LRU counters.  Then it converts all the
callers that don't specifically need per-zone numbers over to that.

This patch (of 6):

The memcg code currently maintains private per-zone breakdowns of the LRU
counters.  This is necessary for reclaim decisions which are still
zone-based, but there are a variety of users of these counters that only
want the aggregate per-lruvec or per-memcg LRU counts, and they need to
painfully sum up the zone counters on each request for that.

These would be better served using the memcg vmstats arrays, which track
VM statistics at the desired scope already.  They just don't have the LRU
counts right now.

So to kick off the conversion, begin tracking LRU counts in those.

Link: http://lkml.kernel.org/r/20190228163020.24100-2-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Roman Gushchin <guro@fb.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/vmscan: add tracepoints for node reclaim
Yafang Shao [Tue, 14 May 2019 00:17:53 +0000 (17:17 -0700)]
mm/vmscan: add tracepoints for node reclaim

The page alloc fast path it may perform node reclaim, which may cause a
latency spike.  We should add tracepoint for this event, and also measure
the latency it causes.

So bellow two tracepoints are introduced,
mm_vmscan_node_reclaim_begin
mm_vmscan_node_reclaim_end

Link: http://lkml.kernel.org/r/1551421452-5385-1-git-send-email-laoar.shao@gmail.com
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Souptick Joarder <jrdr.linux@gmail.com>
Cc: <shaoyafang@didiglobal.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/page_isolation.c: remove redundant pfn_valid_within() in __first_valid_page()
Anshuman Khandual [Tue, 14 May 2019 00:17:50 +0000 (17:17 -0700)]
mm/page_isolation.c: remove redundant pfn_valid_within() in __first_valid_page()

pfn_valid_within() calls pfn_valid() when CONFIG_HOLES_IN_ZONE making it
redundant for both definitions (w/wo CONFIG_MEMORY_HOTPLUG) of the helper
pfn_to_online_page() which either calls pfn_valid() or pfn_valid_within().
pfn_valid_within() being 1 when !CONFIG_HOLES_IN_ZONE is irrelevant
either way.  This does not change functionality.

Link: http://lkml.kernel.org/r/1553141595-26907-1-git-send-email-anshuman.khandual@arm.com
Signed-off-by: Anshuman Khandual <anshuman.khandual@arm.com>
Reviewed-by: Zi Yan <ziy@nvidia.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm, compaction: some tracepoints should be defined only when CONFIG_COMPACTION is set
Yafang Shao [Tue, 14 May 2019 00:17:47 +0000 (17:17 -0700)]
mm, compaction: some tracepoints should be defined only when CONFIG_COMPACTION is set

Only mm_compaction_isolate_{free, migrate}pages may be used when
CONFIG_COMPACTION is not set.  All others are used only when
CONFIG_COMPACTION is set.

After this change, if CONFIG_COMPACTION is not set, the tracepoints that
only work when CONFIG_COMPACTION is set will not be exposed to userspace.
Without this change, they will always be exposed in debugfs whether
CONFIG_COMPACTION is set or not.  This is an improvement.

Link: http://lkml.kernel.org/r/1552440403-11780-1-git-send-email-laoar.shao@gmail.com
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: compaction: show gfp flag names in try_to_compact_pages tracepoint
Yafang Shao [Tue, 14 May 2019 00:17:44 +0000 (17:17 -0700)]
mm: compaction: show gfp flag names in try_to_compact_pages tracepoint

Showing the gfp flag names instead of the gfp_mask makes trace more
convenient.

Link: http://lkml.kernel.org/r/1552527998-13162-1-git-send-email-laoar.shao@gmail.com
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/cma.c: fix the bitmap status to show failed allocation reason
Yue Hu [Tue, 14 May 2019 00:17:41 +0000 (17:17 -0700)]
mm/cma.c: fix the bitmap status to show failed allocation reason

Currently one bit in cma bitmap represents number of pages rather than
one page, cma->count means cma size in pages. So to find available pages
via find_next_zero_bit()/find_next_bit() we should use cma size not in
pages but in bits although current free pages number is correct due to
zero value of order_per_bit. Once order_per_bit is changed the bitmap
status will be incorrect.

The size input in cma_debug_show_areas() is not correct.  It will
affect the available pages at some position to debug the failure issue.

This is an example with order_per_bit = 1

Before this change:
[    4.120060] cma: number of available pages: 1@93+4@108+7@121+7@137+7@153+7@169+7@185+7@201+3@213+3@221+3@229+3@237+3@245+3@253+3@261+3@269+3@277+3@285+3@293+3@301+3@309+3@317+3@325+19@333+15@369+512@512=> 638 free of 1024 total pages

After this change:
[    4.143234] cma: number of available pages: 2@93+8@108+14@121+14@137+14@153+14@169+14@185+14@201+6@213+6@221+6@229+6@237+6@245+6@253+6@261+6@269+6@277+6@285+6@293+6@301+6@309+6@317+6@325+38@333+30@369=> 252 free of 1024 total pages

Obviously the bitmap status before is incorrect.

Link: http://lkml.kernel.org/r/20190320060829.9144-1-zbestahu@gmail.com
Signed-off-by: Yue Hu <huyue2@yulong.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Mike Rapoport <rppt@linux.vnet.ibm.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Laura Abbott <labbott@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/compaction.c: fix an undefined behaviour
Qian Cai [Tue, 14 May 2019 00:17:38 +0000 (17:17 -0700)]
mm/compaction.c: fix an undefined behaviour

In a low-memory situation, cc->fast_search_fail can keep increasing as it
is unable to find an available page to isolate in
fast_isolate_freepages().  As the result, it could trigger an error below,
so just compare with the maximum bits can be shifted first.

UBSAN: Undefined behaviour in mm/compaction.c:1160:30
shift exponent 64 is too large for 64-bit type 'unsigned long'
CPU: 131 PID: 1308 Comm: kcompactd1 Kdump: loaded Tainted: G
W    L    5.0.0+ #17
Call trace:
 dump_backtrace+0x0/0x450
 show_stack+0x20/0x2c
 dump_stack+0xc8/0x14c
 __ubsan_handle_shift_out_of_bounds+0x7e8/0x8c4
 compaction_alloc+0x2344/0x2484
 unmap_and_move+0xdc/0x1dbc
 migrate_pages+0x274/0x1310
 compact_zone+0x26ec/0x43bc
 kcompactd+0x15b8/0x1a24
 kthread+0x374/0x390
 ret_from_fork+0x10/0x18

[akpm@linux-foundation.org: code cleanup]
Link: http://lkml.kernel.org/r/20190320203338.53367-1-cai@lca.pw
Fixes: 70b44595eafe ("mm, compaction: use free lists to quickly locate a migration source")
Signed-off-by: Qian Cai <cai@lca.pw>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/memory_hotplug.c: fix the wrong usage of N_HIGH_MEMORY
Baoquan He [Tue, 14 May 2019 00:17:35 +0000 (17:17 -0700)]
mm/memory_hotplug.c: fix the wrong usage of N_HIGH_MEMORY

In node_states_check_changes_online(), N_HIGH_MEMORY is used to substitute
ZONE_HIGHMEM directly.  This is not right.  N_HIGH_MEMORY is to mark the
memory state of node.  Here zone index is checked, which should be
compared with 'ZONE_HIGHMEM' accordingly.

Replace it with ZONE_HIGHMEM.

This is a code cleanup - no known runtime effects.

Link: http://lkml.kernel.org/r/20190320080732.14933-1-bhe@redhat.com
Fixes: 8efe33f40f3e ("mm/memory_hotplug.c: simplify node_states_check_changes_online")
Signed-off-by: Baoquan He <bhe@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Cc: Wei Yang <richard.weiyang@gmail.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm,memory_hotplug: drop redundant hugepage_migration_supported check
Oscar Salvador [Tue, 14 May 2019 00:17:32 +0000 (17:17 -0700)]
mm,memory_hotplug: drop redundant hugepage_migration_supported check

has_unmovable_pages() already checks whether the hugetlb page supports
migration, so all non-migratable hugetlb pages should have been caught
there.  Let us drop the check from scan_movable_pages() as is redundant.

Link: http://lkml.kernel.org/r/20190320152658.10855-3-osalvador@suse.de
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm,memory_hotplug: unlock 1GB-hugetlb on x86_64
Oscar Salvador [Tue, 14 May 2019 00:17:29 +0000 (17:17 -0700)]
mm,memory_hotplug: unlock 1GB-hugetlb on x86_64

On x86_64, 1GB-hugetlb pages could never be offlined due to the fact
that hugepage_migration_supported() returned false for PUD_SHIFT.
So whenever we wanted to offline a memblock containing a gigantic
hugetlb page, we never got beyond has_unmovable_pages() check.
This changed with [1], where now we also return true for PUD_SHIFT.

After that patch, the check in has_unmovable_pages() and scan_movable_pages()
returned true, but we still had a final barrier in do_migrate_range():

if (compound_order(head) > PFN_SECTION_SHIFT) {
ret = -EBUSY;
break;
}

This is not really nice, and we do not really need it.
It is perfectly possible to migrate a gigantic page as long as another node has
a spare gigantic page for us.
In alloc_huge_page_nodemask(), we calculate the __real__ number of free pages,
and if any, we try to dequeue one from another node.

This all works fine when we do have another node with a spare gigantic page,
but if that is not the case, alloc_huge_page_nodemask() ends up calling
alloc_migrate_huge_page() which bails out if the wanted page is gigantic.
That is mainly because finding a 1GB (or even 16GB on powerpc) contiguous
memory is quite unlikely when the system has been running for a while.

In that situation, we will keep looping forever because scan_movable_pages()
will give us the same page and we will fail again because there is no node
where we can dequeue a gigantic page from.
This is not nice, and it has been raised that we might want to treat -ENOMEM
as a fatal error in do_migrate_range(), but this has to be checked further.

Anyway, I would tend say that this is the administrator's job, to make sure
that the system can keep up with the memory to be offlined, so that would mean
that if we want to use gigantic pages, make sure that the other nodes have at
least enough gigantic pages to keep up in case we need to offline memory.

Just for the sake of completeness, this is one of the tests done:

 # echo 1 > /sys/devices/system/node/node1/hugepages/hugepages-1048576kB/nr_hugepages
 # echo 1 > /sys/devices/system/node/node2/hugepages/hugepages-1048576kB/nr_hugepages

 # cat /sys/devices/system/node/node1/hugepages/hugepages-1048576kB/nr_hugepages
   1
 # cat /sys/devices/system/node/node1/hugepages/hugepages-1048576kB/free_hugepages
   1

 # cat /sys/devices/system/node/node2/hugepages/hugepages-1048576kB/nr_hugepages
   1
 # cat /sys/devices/system/node/node2/hugepages/hugepages-1048576kB/free_hugepages
   1

 (hugetlb1gb is a program that maps 1GB region using MAP_HUGE_1GB)

 # numactl -m 1 ./hugetlb1gb
 # cat /sys/devices/system/node/node1/hugepages/hugepages-1048576kB/free_hugepages
   0
 # cat /sys/devices/system/node/node2/hugepages/hugepages-1048576kB/free_hugepages
   1

 # offline node1 memory
 # cat /sys/devices/system/node/node2/hugepages/hugepages-1048576kB/free_hugepages
   0

[1] https://lore.kernel.org/patchwork/patch/998796/

Link: http://lkml.kernel.org/r/20190320152658.10855-2-osalvador@suse.de
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoIB/mthca: use the new FOLL_LONGTERM flag to get_user_pages_fast()
Ira Weiny [Tue, 14 May 2019 00:17:25 +0000 (17:17 -0700)]
IB/mthca: use the new FOLL_LONGTERM flag to get_user_pages_fast()

Use the new FOLL_LONGTERM to get_user_pages_fast() to protect against FS
DAX pages being mapped.

Link: http://lkml.kernel.org/r/20190328084422.29911-8-ira.weiny@intel.com
Link: http://lkml.kernel.org/r/20190317183438.2057-8-ira.weiny@intel.com
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Hogan <jhogan@kernel.org>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Mike Marshall <hubcap@omnibond.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoIB/qib: use the new FOLL_LONGTERM flag to get_user_pages_fast()
Ira Weiny [Tue, 14 May 2019 00:17:21 +0000 (17:17 -0700)]
IB/qib: use the new FOLL_LONGTERM flag to get_user_pages_fast()

Use the new FOLL_LONGTERM to get_user_pages_fast() to protect against FS
DAX pages being mapped.

Link: http://lkml.kernel.org/r/20190328084422.29911-7-ira.weiny@intel.com
Link: http://lkml.kernel.org/r/20190317183438.2057-7-ira.weiny@intel.com
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Hogan <jhogan@kernel.org>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Mike Marshall <hubcap@omnibond.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoIB/hfi1: use the new FOLL_LONGTERM flag to get_user_pages_fast()
Ira Weiny [Tue, 14 May 2019 00:17:18 +0000 (17:17 -0700)]
IB/hfi1: use the new FOLL_LONGTERM flag to get_user_pages_fast()

Use the new FOLL_LONGTERM to get_user_pages_fast() to protect against FS
DAX pages being mapped.

[ira.weiny@intel.com: v3]
Link: http://lkml.kernel.org/r/20190328084422.29911-6-ira.weiny@intel.com
Link: http://lkml.kernel.org/r/20190328084422.29911-6-ira.weiny@intel.com
Link: http://lkml.kernel.org/r/20190317183438.2057-6-ira.weiny@intel.com
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Hogan <jhogan@kernel.org>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Mike Marshall <hubcap@omnibond.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/gup: add FOLL_LONGTERM capability to GUP fast
Ira Weiny [Tue, 14 May 2019 00:17:14 +0000 (17:17 -0700)]
mm/gup: add FOLL_LONGTERM capability to GUP fast

DAX pages were previously unprotected from longterm pins when users called
get_user_pages_fast().

Use the new FOLL_LONGTERM flag to check for DEVMAP pages and fall back to
regular GUP processing if a DEVMAP page is encountered.

[ira.weiny@intel.com: v3]
Link: http://lkml.kernel.org/r/20190328084422.29911-5-ira.weiny@intel.com
Link: http://lkml.kernel.org/r/20190328084422.29911-5-ira.weiny@intel.com
Link: http://lkml.kernel.org/r/20190317183438.2057-5-ira.weiny@intel.com
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Hogan <jhogan@kernel.org>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Mike Marshall <hubcap@omnibond.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/gup: change GUP fast to use flags rather than a write 'bool'
Ira Weiny [Tue, 14 May 2019 00:17:11 +0000 (17:17 -0700)]
mm/gup: change GUP fast to use flags rather than a write 'bool'

To facilitate additional options to get_user_pages_fast() change the
singular write parameter to be gup_flags.

This patch does not change any functionality.  New functionality will
follow in subsequent patches.

Some of the get_user_pages_fast() call sites were unchanged because they
already passed FOLL_WRITE or 0 for the write parameter.

NOTE: It was suggested to change the ordering of the get_user_pages_fast()
arguments to ensure that callers were converted.  This breaks the current
GUP call site convention of having the returned pages be the final
parameter.  So the suggestion was rejected.

Link: http://lkml.kernel.org/r/20190328084422.29911-4-ira.weiny@intel.com
Link: http://lkml.kernel.org/r/20190317183438.2057-4-ira.weiny@intel.com
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Mike Marshall <hubcap@omnibond.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Hogan <jhogan@kernel.org>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/gup: change write parameter to flags in fast walk
Ira Weiny [Tue, 14 May 2019 00:17:07 +0000 (17:17 -0700)]
mm/gup: change write parameter to flags in fast walk

In order to support more options in the GUP fast walk, change the write
parameter to flags throughout the call stack.

This patch does not change functionality and passes FOLL_WRITE where write
was previously used.

Link: http://lkml.kernel.org/r/20190328084422.29911-3-ira.weiny@intel.com
Link: http://lkml.kernel.org/r/20190317183438.2057-3-ira.weiny@intel.com
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Hogan <jhogan@kernel.org>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Mike Marshall <hubcap@omnibond.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/gup: replace get_user_pages_longterm() with FOLL_LONGTERM
Ira Weiny [Tue, 14 May 2019 00:17:03 +0000 (17:17 -0700)]
mm/gup: replace get_user_pages_longterm() with FOLL_LONGTERM

Pach series "Add FOLL_LONGTERM to GUP fast and use it".

HFI1, qib, and mthca, use get_user_pages_fast() due to its performance
advantages.  These pages can be held for a significant time.  But
get_user_pages_fast() does not protect against mapping FS DAX pages.

Introduce FOLL_LONGTERM and use this flag in get_user_pages_fast() which
retains the performance while also adding the FS DAX checks.  XDP has also
shown interest in using this functionality.[1]

In addition we change get_user_pages() to use the new FOLL_LONGTERM flag
and remove the specialized get_user_pages_longterm call.

[1] https://lkml.org/lkml/2019/3/19/939

"longterm" is a relative thing and at this point is probably a misnomer.
This is really flagging a pin which is going to be given to hardware and
can't move.  I've thought of a couple of alternative names but I think we
have to settle on if we are going to use FL_LAYOUT or something else to
solve the "longterm" problem.  Then I think we can change the flag to a
better name.

Secondly, it depends on how often you are registering memory.  I have
spoken with some RDMA users who consider MR in the performance path...
For the overall application performance.  I don't have the numbers as the
tests for HFI1 were done a long time ago.  But there was a significant
advantage.  Some of which is probably due to the fact that you don't have
to hold mmap_sem.

Finally, architecturally I think it would be good for everyone to use
*_fast.  There are patches submitted to the RDMA list which would allow
the use of *_fast (they reworking the use of mmap_sem) and as soon as they
are accepted I'll submit a patch to convert the RDMA core as well.  Also
to this point others are looking to use *_fast.

As an aside, Jasons pointed out in my previous submission that *_fast and
*_unlocked look very much the same.  I agree and I think further cleanup
will be coming.  But I'm focused on getting the final solution for DAX at
the moment.

This patch (of 7):

This patch starts a series which aims to support FOLL_LONGTERM in
get_user_pages_fast().  Some callers who would like to do a longterm (user
controlled pin) of pages with the fast variant of GUP for performance
purposes.

Rather than have a separate get_user_pages_longterm() call, introduce
FOLL_LONGTERM and change the longterm callers to use it.

This patch does not change any functionality.  In the short term
"longterm" or user controlled pins are unsafe for Filesystems and FS DAX
in particular has been blocked.  However, callers of get_user_pages_fast()
were not "protected".

FOLL_LONGTERM can _only_ be supported with get_user_pages[_fast]() as it
requires vmas to determine if DAX is in use.

NOTE: In merging with the CMA changes we opt to change the
get_user_pages() call in check_and_migrate_cma_pages() to a call of
__get_user_pages_locked() on the newly migrated pages.  This makes the
code read better in that we are calling __get_user_pages_locked() on the
pages before and after a potential migration.

As a side affect some of the interfaces are cleaned up but this is not the
primary purpose of the series.

In review[1] it was asked:

<quote>
> This I don't get - if you do lock down long term mappings performance
> of the actual get_user_pages call shouldn't matter to start with.
>
> What do I miss?

A couple of points.

First "longterm" is a relative thing and at this point is probably a
misnomer.  This is really flagging a pin which is going to be given to
hardware and can't move.  I've thought of a couple of alternative names
but I think we have to settle on if we are going to use FL_LAYOUT or
something else to solve the "longterm" problem.  Then I think we can
change the flag to a better name.

Second, It depends on how often you are registering memory.  I have spoken
with some RDMA users who consider MR in the performance path...  For the
overall application performance.  I don't have the numbers as the tests
for HFI1 were done a long time ago.  But there was a significant
advantage.  Some of which is probably due to the fact that you don't have
to hold mmap_sem.

Finally, architecturally I think it would be good for everyone to use
*_fast.  There are patches submitted to the RDMA list which would allow
the use of *_fast (they reworking the use of mmap_sem) and as soon as they
are accepted I'll submit a patch to convert the RDMA core as well.  Also
to this point others are looking to use *_fast.

As an asside, Jasons pointed out in my previous submission that *_fast and
*_unlocked look very much the same.  I agree and I think further cleanup
will be coming.  But I'm focused on getting the final solution for DAX at
the moment.

</quote>

[1] https://lore.kernel.org/lkml/20190220180255.GA12020@iweiny-DESK2.sc.intel.com/T/#md6abad2569f3bf6c1f03686c8097ab6563e94965

[ira.weiny@intel.com: v3]
Link: http://lkml.kernel.org/r/20190328084422.29911-2-ira.weiny@intel.com
Link: http://lkml.kernel.org/r/20190328084422.29911-2-ira.weiny@intel.com
Link: http://lkml.kernel.org/r/20190317183438.2057-2-ira.weiny@intel.com
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Rich Felker <dalias@libc.org>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: James Hogan <jhogan@kernel.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Mike Marshall <hubcap@omnibond.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: generalize putback scan functions
Kirill Tkhai [Tue, 14 May 2019 00:17:00 +0000 (17:17 -0700)]
mm: generalize putback scan functions

This combines two similar functions move_active_pages_to_lru() and
putback_inactive_pages() into single move_pages_to_lru().  This remove
duplicate code and makes object file size smaller.

Before:
   text    data     bss     dec     hex filename
  57082    4732     128   61942    f1f6 mm/vmscan.o
After:
   text    data     bss     dec     hex filename
  55112    4600     128   59840    e9c0 mm/vmscan.o

Note, that now we are checking for !page_evictable() coming from
shrink_active_list(), which shouldn't change any behavior since that path
works with evictable pages only.

Link: http://lkml.kernel.org/r/155290129627.31489.8321971028677203248.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: remove pages_to_free argument of move_active_pages_to_lru()
Kirill Tkhai [Tue, 14 May 2019 00:16:57 +0000 (17:16 -0700)]
mm: remove pages_to_free argument of move_active_pages_to_lru()

We may use input argument list as output argument too.  This makes the
function more similar to putback_inactive_pages().

Link: http://lkml.kernel.org/r/155290129079.31489.16180612694090502942.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: move nr_deactivate accounting to shrink_active_list()
Kirill Tkhai [Tue, 14 May 2019 00:16:54 +0000 (17:16 -0700)]
mm: move nr_deactivate accounting to shrink_active_list()

We know which LRU is not active.

[chris@chrisdown.name: fix build on !CONFIG_MEMCG]
Link: http://lkml.kernel.org/r/20190322150513.GA22021@chrisdown.name
Link: http://lkml.kernel.org/r/155290128498.31489.18250485448913338607.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Chris Down <chris@chrisdown.name>
Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: move recent_rotated pages calculation to shrink_inactive_list()
Kirill Tkhai [Tue, 14 May 2019 00:16:51 +0000 (17:16 -0700)]
mm: move recent_rotated pages calculation to shrink_inactive_list()

Patch series "mm: Generalize putback functions"]

putback_inactive_pages() and move_active_pages_to_lru() are almost
similar, so this patchset merges them ina single function.

This patch (of 4):

The patch moves the calculation from putback_inactive_pages() to
shrink_inactive_list().  This makes putback_inactive_pages() looking more
similar to move_active_pages_to_lru().

To do that, we account activated pages in reclaim_stat::nr_activate.
Since a page may change its LRU type from anon to file cache inside
shrink_page_list() (see ClearPageSwapBacked()), we have to account pages
for the both types.  So, nr_activate becomes an array.

Previously we used nr_activate to account PGACTIVATE events, but now we
account them into pgactivate variable (since they are about number of
pages in general, not about sum of hpage_nr_pages).

Link: http://lkml.kernel.org/r/155290127956.31489.3393586616054413298.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm, page_alloc: disallow __GFP_COMP in alloc_pages_exact()
Vlastimil Babka [Tue, 14 May 2019 00:16:47 +0000 (17:16 -0700)]
mm, page_alloc: disallow __GFP_COMP in alloc_pages_exact()

alloc_pages_exact*() allocates a page of sufficient order and then splits
it to return only the number of pages requested.  That makes it
incompatible with __GFP_COMP, because compound pages cannot be split.

As shown by [1] things may silently work until the requested size
(possibly depending on user) stops being power of two.  Then for
CONFIG_DEBUG_VM, BUG_ON() triggers in split_page().  Without
CONFIG_DEBUG_VM, consequences are unclear.

There are several options here, none of them great:

1) Don't do the splitting when __GFP_COMP is passed, and return the
   whole compound page.  However if caller then returns it via
   free_pages_exact(), that will be unexpected and the freeing actions
   there will be wrong.

2) Warn and remove __GFP_COMP from the flags.  But the caller may have
   really wanted it, so things may break later somewhere.

3) Warn and return NULL.  However NULL may be unexpected, especially
   for small sizes.

This patch picks option 2, because as Michal Hocko put it: "callers wanted
it" is much less probable than "caller is simply confused and more gfp
flags is surely better than fewer".

[1] https://lore.kernel.org/lkml/20181126002805.GI18977@shao2-debian/T/#u

Link: http://lkml.kernel.org/r/0c6393eb-b28d-4607-c386-862a71f09de6@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Takashi Iwai <tiwai@suse.de>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: page cache: store only head pages in i_pages
Matthew Wilcox [Tue, 14 May 2019 00:16:44 +0000 (17:16 -0700)]
mm: page cache: store only head pages in i_pages

Transparent Huge Pages are currently stored in i_pages as pointers to
consecutive subpages.  This patch changes that to storing consecutive
pointers to the head page in preparation for storing huge pages more
efficiently in i_pages.

Large parts of this are "inspired" by Kirill's patch
https://lore.kernel.org/lkml/20170126115819.58875-2-kirill.shutemov@linux.intel.com/

[willy@infradead.org: fix swapcache pages]
Link: http://lkml.kernel.org/r/20190324155441.GF10344@bombadil.infradead.org
[kirill@shutemov.name: hugetlb stores pages in page cache differently]
Link: http://lkml.kernel.org/r/20190404134553.vuvhgmghlkiw2hgl@kshutemo-mobl1
Link: http://lkml.kernel.org/r/20190307153051.18815-1-willy@infradead.org
Signed-off-by: Matthew Wilcox <willy@infradead.org>
Acked-by: Jan Kara <jack@suse.cz>
Reviewed-by: Kirill Shutemov <kirill@shutemov.name>
Reviewed-and-tested-by: Song Liu <songliubraving@fb.com>
Tested-by: William Kucharski <william.kucharski@oracle.com>
Reviewed-by: William Kucharski <william.kucharski@oracle.com>
Tested-by: Qian Cai <cai@lca.pw>
Cc: Hugh Dickins <hughd@google.com>
Cc: Song Liu <liu.song.a23@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agouserfaultfd/sysctl: add vm.unprivileged_userfaultfd
Peter Xu [Tue, 14 May 2019 00:16:41 +0000 (17:16 -0700)]
userfaultfd/sysctl: add vm.unprivileged_userfaultfd

Userfaultfd can be misued to make it easier to exploit existing
use-after-free (and similar) bugs that might otherwise only make a
short window or race condition available.  By using userfaultfd to
stall a kernel thread, a malicious program can keep some state that it
wrote, stable for an extended period, which it can then access using an
existing exploit.  While it doesn't cause the exploit itself, and while
it's not the only thing that can stall a kernel thread when accessing a
memory location, it's one of the few that never needs privilege.

We can add a flag, allowing userfaultfd to be restricted, so that in
general it won't be useable by arbitrary user programs, but in
environments that require userfaultfd it can be turned back on.

Add a global sysctl knob "vm.unprivileged_userfaultfd" to control
whether userfaultfd is allowed by unprivileged users.  When this is
set to zero, only privileged users (root user, or users with the
CAP_SYS_PTRACE capability) will be able to use the userfaultfd
syscalls.

Andrea said:

: The only difference between the bpf sysctl and the userfaultfd sysctl
: this way is that the bpf sysctl adds the CAP_SYS_ADMIN capability
: requirement, while userfaultfd adds the CAP_SYS_PTRACE requirement,
: because the userfaultfd monitor is more likely to need CAP_SYS_PTRACE
: already if it's doing other kind of tracking on processes runtime, in
: addition of userfaultfd.  In other words both syscalls works only for
: root, when the two sysctl are opt-in set to 1.

[dgilbert@redhat.com: changelog additions]
[akpm@linux-foundation.org: documentation tweak, per Mike]
Link: http://lkml.kernel.org/r/20190319030722.12441-2-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Suggested-by: Andrea Arcangeli <aarcange@redhat.com>
Suggested-by: Mike Rapoport <rppt@linux.ibm.com>
Reviewed-by: Mike Rapoport <rppt@linux.ibm.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Luis Chamberlain <mcgrof@kernel.org>
Cc: Maxime Coquelin <maxime.coquelin@redhat.com>
Cc: Maya Gokhale <gokhale2@llnl.gov>
Cc: Jerome Glisse <jglisse@redhat.com>
Cc: Pavel Emelyanov <xemul@virtuozzo.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Martin Cracauer <cracauer@cons.org>
Cc: Denis Plotnikov <dplotnikov@virtuozzo.com>
Cc: Marty McFadden <mcfadden8@llnl.gov>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: "Kirill A . Shutemov" <kirill@shutemov.name>
Cc: "Dr . David Alan Gilbert" <dgilbert@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/cma_debug.c: fix the break condition in cma_maxchunk_get()
Yue Hu [Tue, 14 May 2019 00:16:37 +0000 (17:16 -0700)]
mm/cma_debug.c: fix the break condition in cma_maxchunk_get()

If not find zero bit in find_next_zero_bit(), it will return the size
parameter passed in, so the start bit should be compared with bitmap_maxno
rather than cma->count.  Although getting maxchunk is working fine due to
zero value of order_per_bit currently, the operation will be stuck if
order_per_bit is set as non-zero.

Link: http://lkml.kernel.org/r/20190319092734.276-1-zbestahu@gmail.com
Signed-off-by: Yue Hu <huyue2@yulong.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Joe Perches <joe@perches.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Dmitry Safonov <d.safonov@partner.samsung.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoinclude/trace/events/vmscan.h: drop zone id from kswapd tracepoints
Yafang Shao [Tue, 14 May 2019 00:16:34 +0000 (17:16 -0700)]
include/trace/events/vmscan.h: drop zone id from kswapd tracepoints

It is not clear how the zone id is useful in kswapd tracepoints and the id
itself is not really easy to process because it depends on the
configuration (available zones).  Let's drop the id for now.  If somebody
really needs that information then the zone name should be used instead.

[mhocko@suse.com: new changelog]
Link: http://lkml.kernel.org/r/1552451813-10833-1-git-send-email-laoar.shao@gmail.com
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/slab.c: fix an infinite loop in leaks_show()
Qian Cai [Tue, 14 May 2019 00:16:31 +0000 (17:16 -0700)]
mm/slab.c: fix an infinite loop in leaks_show()

"cat /proc/slab_allocators" could hang forever on SMP machines with
kmemleak or object debugging enabled due to other CPUs running do_drain()
will keep making kmemleak_object or debug_objects_cache dirty and unable
to escape the first loop in leaks_show(),

do {
set_store_user_clean(cachep);
drain_cpu_caches(cachep);
...

} while (!is_store_user_clean(cachep));

For example,

do_drain
  slabs_destroy
    slab_destroy
      kmem_cache_free
        __cache_free
          ___cache_free
            kmemleak_free_recursive
              delete_object_full
                __delete_object
                  put_object
                    free_object_rcu
                      kmem_cache_free
                        cache_free_debugcheck --> dirty kmemleak_object

One approach is to check cachep->name and skip both kmemleak_object and
debug_objects_cache in leaks_show().  The other is to set store_user_clean
after drain_cpu_caches() which leaves a small window between
drain_cpu_caches() and set_store_user_clean() where per-CPU caches could
be dirty again lead to slightly wrong information has been stored but
could also speed up things significantly which sounds like a good
compromise.  For example,

 # cat /proc/slab_allocators
 0m42.778s # 1st approach
 0m0.737s  # 2nd approach

[akpm@linux-foundation.org: tweak comment]
Link: http://lkml.kernel.org/r/20190411032635.10325-1-cai@lca.pw
Fixes: d31676dfde25 ("mm/slab: alternative implementation for DEBUG_SLAB_LEAK")
Signed-off-by: Qian Cai <cai@lca.pw>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/slub.c: update the comment about slab frozen
Liu Xiang [Tue, 14 May 2019 00:16:28 +0000 (17:16 -0700)]
mm/slub.c: update the comment about slab frozen

Now frozen slab can only be on the per cpu partial list.

Link: http://lkml.kernel.org/r/1554022325-11305-1-git-send-email-liu.xiang6@zte.com.cn
Signed-off-by: Liu Xiang <liu.xiang6@zte.com.cn>
Cc: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/slab.c: remove unneed check in cpuup_canceled
Li RongQing [Tue, 14 May 2019 00:16:25 +0000 (17:16 -0700)]
mm/slab.c: remove unneed check in cpuup_canceled

nc is a member of percpu allocation memory, and cannot be NULL.

Link: http://lkml.kernel.org/r/1553159353-5056-1-git-send-email-lirongqing@baidu.com
Signed-off-by: Li RongQing <lirongqing@baidu.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoslub: remove useless kmem_cache_debug() before remove_full()
Liu Xiang [Tue, 14 May 2019 00:16:22 +0000 (17:16 -0700)]
slub: remove useless kmem_cache_debug() before remove_full()

When CONFIG_SLUB_DEBUG is not enabled, remove_full() is empty.
While CONFIG_SLUB_DEBUG is enabled, remove_full() can check
s->flags by itself. So kmem_cache_debug() is useless and
can be removed.

Link: http://lkml.kernel.org/r/1552577313-2830-1-git-send-email-liu.xiang6@zte.com.cn
Signed-off-by: Liu Xiang <liu.xiang6@zte.com.cn>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: remove stale comment from page struct
Tobin C. Harding [Tue, 14 May 2019 00:16:19 +0000 (17:16 -0700)]
mm: remove stale comment from page struct

We now use the slab_list list_head instead of the lru list_head.  This
comment has become stale.

Remove stale comment from page struct slab_list list_head.

Link: http://lkml.kernel.org/r/20190402230545.2929-8-tobin@kernel.org
Signed-off-by: Tobin C. Harding <tobin@kernel.org>
Acked-by: Christoph Lameter <cl@linux.com>
Reviewed-by: Roman Gushchin <guro@fb.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Pekka Enberg <penberg@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoslab: use slab_list instead of lru
Tobin C. Harding [Tue, 14 May 2019 00:16:15 +0000 (17:16 -0700)]
slab: use slab_list instead of lru

Currently we use the page->lru list for maintaining lists of slabs.  We
have a list in the page structure (slab_list) that can be used for this
purpose.  Doing so makes the code cleaner since we are not overloading the
lru list.

Use the slab_list instead of the lru list for maintaining lists of slabs.

Link: http://lkml.kernel.org/r/20190402230545.2929-7-tobin@kernel.org
Signed-off-by: Tobin C. Harding <tobin@kernel.org>
Acked-by: Christoph Lameter <cl@linux.com>
Reviewed-by: Roman Gushchin <guro@fb.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Pekka Enberg <penberg@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoslub: use slab_list instead of lru
Tobin C. Harding [Tue, 14 May 2019 00:16:12 +0000 (17:16 -0700)]
slub: use slab_list instead of lru

Currently we use the page->lru list for maintaining lists of slabs.  We
have a list in the page structure (slab_list) that can be used for this
purpose.  Doing so makes the code cleaner since we are not overloading the
lru list.

Use the slab_list instead of the lru list for maintaining lists of slabs.

Link: http://lkml.kernel.org/r/20190402230545.2929-6-tobin@kernel.org
Signed-off-by: Tobin C. Harding <tobin@kernel.org>
Acked-by: Christoph Lameter <cl@linux.com>
Reviewed-by: Roman Gushchin <guro@fb.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Pekka Enberg <penberg@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoslub: add comments to endif pre-processor macros
Tobin C. Harding [Tue, 14 May 2019 00:16:09 +0000 (17:16 -0700)]
slub: add comments to endif pre-processor macros

SLUB allocator makes heavy use of ifdef/endif pre-processor macros.  The
pairing of these statements is at times hard to follow e.g.  if the pair
are further than a screen apart or if there are nested pairs.  We can
reduce cognitive load by adding a comment to the endif statement of form

       #ifdef CONFIG_FOO
       ...
       #endif /* CONFIG_FOO */

Add comments to endif pre-processor macros if ifdef/endif pair is not
immediately apparent.

Link: http://lkml.kernel.org/r/20190402230545.2929-5-tobin@kernel.org
Signed-off-by: Tobin C. Harding <tobin@kernel.org>
Acked-by: Christoph Lameter <cl@linux.com>
Reviewed-by: Roman Gushchin <guro@fb.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Pekka Enberg <penberg@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoslob: use slab_list instead of lru
Tobin C. Harding [Tue, 14 May 2019 00:16:06 +0000 (17:16 -0700)]
slob: use slab_list instead of lru

Currently we use the page->lru list for maintaining lists of slabs.  We
have a list_head in the page structure (slab_list) that can be used for
this purpose.  Doing so makes the code cleaner since we are not
overloading the lru list.

The slab_list is part of a union within the page struct (included here
stripped down):

union {
struct { /* Page cache and anonymous pages */
struct list_head lru;
...
};
struct {
dma_addr_t dma_addr;
};
struct { /* slab, slob and slub */
union {
struct list_head slab_list;
struct { /* Partial pages */
struct page *next;
int pages; /* Nr of pages left */
int pobjects; /* Approximate count */
};
};
...

Here we see that slab_list and lru are the same bits.  We can verify that
this change is safe to do by examining the object file produced from
slob.c before and after this patch is applied.

Steps taken to verify:

 1. checkout current tip of Linus' tree

    commit a667cb7a94d4 ("Merge branch 'akpm' (patches from Andrew)")

 2. configure and build (select SLOB allocator)

    CONFIG_SLOB=y
    CONFIG_SLAB_MERGE_DEFAULT=y

 3. dissasemble object file `objdump -dr mm/slub.o > before.s
 4. apply patch
 5. build
 6. dissasemble object file `objdump -dr mm/slub.o > after.s
 7. diff before.s after.s

Use slab_list list_head instead of the lru list_head for maintaining
lists of slabs.

Link: http://lkml.kernel.org/r/20190402230545.2929-4-tobin@kernel.org
Signed-off-by: Tobin C. Harding <tobin@kernel.org>
Reviewed-by: Roman Gushchin <guro@fb.com>
Acked-by: Christoph Lameter <cl@linux.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Pekka Enberg <penberg@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoslob: respect list_head abstraction layer
Tobin C. Harding [Tue, 14 May 2019 00:16:03 +0000 (17:16 -0700)]
slob: respect list_head abstraction layer

Currently we reach inside the list_head.  This is a violation of the layer
of abstraction provided by the list_head.  It makes the code fragile.
More importantly it makes the code wicked hard to understand.

The code reaches into the list_head structure to counteract the fact that
the list _may_ have been changed during slob_page_alloc().  Instead of
this we can add a return parameter to slob_page_alloc() to signal that the
list was modified (list_del() called with page->lru to remove page from
the freelist).

This code is concerned with an optimisation that counters the tendency for
first fit allocation algorithm to fragment memory into many small chunks
at the front of the memory pool.  Since the page is only removed from the
list when an allocation uses _all_ the remaining memory in the page then
in this special case fragmentation does not occur and we therefore do not
need the optimisation.

Add a return parameter to slob_page_alloc() to signal that the allocation
used up the whole page and that the page was removed from the free list.
After calling slob_page_alloc() check the return value just added and only
attempt optimisation if the page is still on the list.

Use list_head API instead of reaching into the list_head structure to
check if sp is at the front of the list.

Link: http://lkml.kernel.org/r/20190402230545.2929-3-tobin@kernel.org
Signed-off-by: Tobin C. Harding <tobin@kernel.org>
Acked-by: Christoph Lameter <cl@linux.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Roman Gushchin <guro@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agolist: add function list_rotate_to_front()
Tobin C. Harding [Tue, 14 May 2019 00:15:59 +0000 (17:15 -0700)]
list: add function list_rotate_to_front()

Patch series "mm: Use slab_list list_head instead of lru", v5.

Currently the slab allocators (ab)use the struct page 'lru' list_head.  We
have a list head for slab allocators to use, 'slab_list'.

During v2 it was noted by Christoph that the SLOB allocator was reaching
into a list_head, this version adds 2 patches to the front of the set to
fix that.

Clean up all three allocators by using the 'slab_list' list_head instead
of overloading the 'lru' list_head.

This patch (of 7):

Currently if we wish to rotate a list until a specific item is at the
front of the list we can call list_move_tail(head, list).  Note that the
arguments are the reverse way to the usual use of list_move_tail(list,
head).  This is a hack, it depends on the developer knowing how the
list_head operates internally which violates the layer of abstraction
offered by the list_head.  Also, it is not intuitive so the next developer
to come along must study list.h in order to fully understand what is meant
by the call, while this is 'good for' the developer it makes reading the
code harder.  We should have an function appropriately named that does
this if there are users for it intree.

By grep'ing the tree for list_move_tail() and list_tail() and attempting
to guess the argument order from the names it seems there is only one
place currently in the tree that does this - the slob allocatator.

Add function list_rotate_to_front() to rotate a list until the specified
item is at the front of the list.

Link: http://lkml.kernel.org/r/20190402230545.2929-2-tobin@kernel.org
Signed-off-by: Tobin C. Harding <tobin@kernel.org>
Reviewed-by: Christoph Lameter <cl@linux.com>
Reviewed-by: Roman Gushchin <guro@fb.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoocfs2: fix ocfs2 read inode data panic in ocfs2_iget
Shuning Zhang [Tue, 14 May 2019 00:15:56 +0000 (17:15 -0700)]
ocfs2: fix ocfs2 read inode data panic in ocfs2_iget

In some cases, ocfs2_iget() reads the data of inode, which has been
deleted for some reason.  That will make the system panic.  So We should
judge whether this inode has been deleted, and tell the caller that the
inode is a bad inode.

For example, the ocfs2 is used as the backed of nfs, and the client is
nfsv3.  This issue can be reproduced by the following steps.

on the nfs server side,
..../patha/pathb

Step 1: The process A was scheduled before calling the function fh_verify.

Step 2: The process B is removing the 'pathb', and just completed the call
to function dput.  Then the dentry of 'pathb' has been deleted from the
dcache, and all ancestors have been deleted also.  The relationship of
dentry and inode was deleted through the function hlist_del_init.  The
following is the call stack.
dentry_iput->hlist_del_init(&dentry->d_u.d_alias)

At this time, the inode is still in the dcache.

Step 3: The process A call the function ocfs2_get_dentry, which get the
inode from dcache.  Then the refcount of inode is 1.  The following is the
call stack.
nfsd3_proc_getacl->fh_verify->exportfs_decode_fh->fh_to_dentry(ocfs2_get_dentry)

Step 4: Dirty pages are flushed by bdi threads.  So the inode of 'patha'
is evicted, and this directory was deleted.  But the inode of 'pathb'
can't be evicted, because the refcount of the inode was 1.

Step 5: The process A keep running, and call the function
reconnect_path(in exportfs_decode_fh), which call function
ocfs2_get_parent of ocfs2.  Get the block number of parent
directory(patha) by the name of ...  Then read the data from disk by the
block number.  But this inode has been deleted, so the system panic.

Process A                                             Process B
1. in nfsd3_proc_getacl                   |
2.                                        |        dput
3. fh_to_dentry(ocfs2_get_dentry)         |
4. bdi flush dirty cache                  |
5. ocfs2_iget                             |

[283465.542049] OCFS2: ERROR (device sdp): ocfs2_validate_inode_block:
Invalid dinode #580640: OCFS2_VALID_FL not set

[283465.545490] Kernel panic - not syncing: OCFS2: (device sdp): panic forced
after error

[283465.546889] CPU: 5 PID: 12416 Comm: nfsd Tainted: G        W
4.1.12-124.18.6.el6uek.bug28762940v3.x86_64 #2
[283465.548382] Hardware name: VMware, Inc. VMware Virtual Platform/440BX
Desktop Reference Platform, BIOS 6.00 09/21/2015
[283465.549657]  0000000000000000 ffff8800a56fb7b8 ffffffff816e839c
ffffffffa0514758
[283465.550392]  000000000008dc20 ffff8800a56fb838 ffffffff816e62d3
0000000000000008
[283465.551056]  ffff880000000010 ffff8800a56fb848 ffff8800a56fb7e8
ffff88005df9f000
[283465.551710] Call Trace:
[283465.552516]  [<ffffffff816e839c>] dump_stack+0x63/0x81
[283465.553291]  [<ffffffff816e62d3>] panic+0xcb/0x21b
[283465.554037]  [<ffffffffa04e66b0>] ocfs2_handle_error+0xf0/0xf0 [ocfs2]
[283465.554882]  [<ffffffffa04e7737>] __ocfs2_error+0x67/0x70 [ocfs2]
[283465.555768]  [<ffffffffa049c0f9>] ocfs2_validate_inode_block+0x229/0x230
[ocfs2]
[283465.556683]  [<ffffffffa047bcbc>] ocfs2_read_blocks+0x46c/0x7b0 [ocfs2]
[283465.557408]  [<ffffffffa049bed0>] ? ocfs2_inode_cache_io_unlock+0x20/0x20
[ocfs2]
[283465.557973]  [<ffffffffa049f0eb>] ocfs2_read_inode_block_full+0x3b/0x60
[ocfs2]
[283465.558525]  [<ffffffffa049f5ba>] ocfs2_iget+0x4aa/0x880 [ocfs2]
[283465.559082]  [<ffffffffa049146e>] ocfs2_get_parent+0x9e/0x220 [ocfs2]
[283465.559622]  [<ffffffff81297c05>] reconnect_path+0xb5/0x300
[283465.560156]  [<ffffffff81297f46>] exportfs_decode_fh+0xf6/0x2b0
[283465.560708]  [<ffffffffa062faf0>] ? nfsd_proc_getattr+0xa0/0xa0 [nfsd]
[283465.561262]  [<ffffffff810a8196>] ? prepare_creds+0x26/0x110
[283465.561932]  [<ffffffffa0630860>] fh_verify+0x350/0x660 [nfsd]
[283465.562862]  [<ffffffffa0637804>] ? nfsd_cache_lookup+0x44/0x630 [nfsd]
[283465.563697]  [<ffffffffa063a8b9>] nfsd3_proc_getattr+0x69/0xf0 [nfsd]
[283465.564510]  [<ffffffffa062cf60>] nfsd_dispatch+0xe0/0x290 [nfsd]
[283465.565358]  [<ffffffffa05eb892>] ? svc_tcp_adjust_wspace+0x12/0x30
[sunrpc]
[283465.566272]  [<ffffffffa05ea652>] svc_process_common+0x412/0x6a0 [sunrpc]
[283465.567155]  [<ffffffffa05eaa03>] svc_process+0x123/0x210 [sunrpc]
[283465.568020]  [<ffffffffa062c90f>] nfsd+0xff/0x170 [nfsd]
[283465.568962]  [<ffffffffa062c810>] ? nfsd_destroy+0x80/0x80 [nfsd]
[283465.570112]  [<ffffffff810a622b>] kthread+0xcb/0xf0
[283465.571099]  [<ffffffff810a6160>] ? kthread_create_on_node+0x180/0x180
[283465.572114]  [<ffffffff816f11b8>] ret_from_fork+0x58/0x90
[283465.573156]  [<ffffffff810a6160>] ? kthread_create_on_node+0x180/0x180

Link: http://lkml.kernel.org/r/1554185919-3010-1-git-send-email-sunny.s.zhang@oracle.com
Signed-off-by: Shuning Zhang <sunny.s.zhang@oracle.com>
Reviewed-by: Joseph Qi <jiangqi903@gmail.com>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Changwei Ge <gechangwei@live.cn>
Cc: piaojun <piaojun@huawei.com>
Cc: "Gang He" <ghe@suse.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoocfs2: use common file type conversion
Phillip Potter [Tue, 14 May 2019 00:15:53 +0000 (17:15 -0700)]
ocfs2: use common file type conversion

Deduplicate the ocfs2 file type conversion implementation and remove
OCFS2_FT_* definitions - file systems that use the same file types as
defined by POSIX do not need to define their own versions and can use the
common helper functions decared in fs_types.h and implemented in
fs_types.c

Common implementation can be found via bbe7449e2599 ("fs: common
implementation of file type").

Link: http://lkml.kernel.org/r/20190326213919.GA20878@pathfinder
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Phillip Potter <phil@philpotter.co.uk>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <jiangqi903@gmail.com>
Cc: Changwei Ge <gechangwei@live.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoMAINTAINERS: add Joseph as ocfs2 co-maintainer
Joseph Qi [Tue, 14 May 2019 00:15:49 +0000 (17:15 -0700)]
MAINTAINERS: add Joseph as ocfs2 co-maintainer

I have been contributing and reviewing to the ocfs2 filesystem for recent
years and I'm willing to continue doing so.  Volunteer as a co-maintainer
for ocfs2 filesystem.

Link: http://lkml.kernel.org/r/f56d75b3-2be5-25c2-51f2-c3f5423d4f14@gmail.com
Signed-off-by: Joseph Qi <joseph.qi@linux.alibaba.com>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Mark Fasheh <mfasheh@suse.com>
Cc: piaojun <piaojun@huawei.com>
Cc: "Gang He" <ghe@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <jiangqi903@gmail.com>
Cc: Changwei Ge <gechangwei@live.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoarch/sh/boards/mach-dreamcast/irq.c: Remove duplicate header
Sabyasachi Gupta [Tue, 14 May 2019 00:15:46 +0000 (17:15 -0700)]
arch/sh/boards/mach-dreamcast/irq.c: Remove duplicate header

Remove linux/irq.h which is included more than once.

Link: http://lkml.kernel.org/r/5c8682ef.1c69fb81.5a1ea.2e7f@mx.google.com
Signed-off-by: Sabyasachi Gupta <sabyasachi.linux@gmail.com>
Acked-by: Souptick Joarder <jrdr.linux@gmail.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Rich Felker <dalias@libc.org>
Cc: Mukesh Ojha <mojha@codeaurora.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agokernel/sys.c: prctl: fix false positive in validate_prctl_map()
Cyrill Gorcunov [Tue, 14 May 2019 00:15:40 +0000 (17:15 -0700)]
kernel/sys.c: prctl: fix false positive in validate_prctl_map()

While validating new map we require the @start_data to be strictly less
than @end_data, which is fine for regular applications (this is why this
nit didn't trigger for that long).  These members are set from executable
loaders such as elf handers, still it is pretty valid to have a loadable
data section with zero size in file, in such case the start_data is equal
to end_data once kernel loader finishes.

As a result when we're trying to restore such programs the procedure fails
and the kernel returns -EINVAL.  From the image dump of a program:

 | "mm_start_code": "0x400000",
 | "mm_end_code": "0x8f5fb4",
 | "mm_start_data": "0xf1bfb0",
 | "mm_end_data": "0xf1bfb0",

Thus we need to change validate_prctl_map from strictly less to less or
equal operator use.

Link: http://lkml.kernel.org/r/20190408143554.GY1421@uranus.lan
Fixes: f606b77f1a9e3 ("prctl: PR_SET_MM -- introduce PR_SET_MM_MAP operation")
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Andrey Vagin <avagin@gmail.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/hugetlb.c: don't put_page in lock of hugetlb_lock
Kai Shen [Tue, 14 May 2019 00:15:37 +0000 (17:15 -0700)]
mm/hugetlb.c: don't put_page in lock of hugetlb_lock

spinlock recursion happened when do LTP test:
#!/bin/bash
./runltp -p -f hugetlb &
./runltp -p -f hugetlb &
./runltp -p -f hugetlb &
./runltp -p -f hugetlb &
./runltp -p -f hugetlb &

The dtor returned by get_compound_page_dtor in __put_compound_page may be
the function of free_huge_page which will lock the hugetlb_lock, so don't
put_page in lock of hugetlb_lock.

 BUG: spinlock recursion on CPU#0, hugemmap05/1079
  lock: hugetlb_lock+0x0/0x18, .magic: dead4ead, .owner: hugemmap05/1079, .owner_cpu: 0
 Call trace:
  dump_backtrace+0x0/0x198
  show_stack+0x24/0x30
  dump_stack+0xa4/0xcc
  spin_dump+0x84/0xa8
  do_raw_spin_lock+0xd0/0x108
  _raw_spin_lock+0x20/0x30
  free_huge_page+0x9c/0x260
  __put_compound_page+0x44/0x50
  __put_page+0x2c/0x60
  alloc_surplus_huge_page.constprop.19+0xf0/0x140
  hugetlb_acct_memory+0x104/0x378
  hugetlb_reserve_pages+0xe0/0x250
  hugetlbfs_file_mmap+0xc0/0x140
  mmap_region+0x3e8/0x5b0
  do_mmap+0x280/0x460
  vm_mmap_pgoff+0xf4/0x128
  ksys_mmap_pgoff+0xb4/0x258
  __arm64_sys_mmap+0x34/0x48
  el0_svc_common+0x78/0x130
  el0_svc_handler+0x38/0x78
  el0_svc+0x8/0xc

Link: http://lkml.kernel.org/r/b8ade452-2d6b-0372-32c2-703644032b47@huawei.com
Fixes: 9980d744a0 ("mm, hugetlb: get rid of surplus page accounting tricks")
Signed-off-by: Kai Shen <shenkai8@huawei.com>
Signed-off-by: Feilong Lin <linfeilong@huawei.com>
Reported-by: Wang Wang <wangwang2@huawei.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/huge_memory: fix vmf_insert_pfn_{pmd, pud}() crash, handle unaligned addresses
Dan Williams [Tue, 14 May 2019 00:15:33 +0000 (17:15 -0700)]
mm/huge_memory: fix vmf_insert_pfn_{pmd, pud}() crash, handle unaligned addresses

Starting with c6f3c5ee40c1 ("mm/huge_memory.c: fix modifying of page
protection by insert_pfn_pmd()") vmf_insert_pfn_pmd() internally calls
pmdp_set_access_flags().  That helper enforces a pmd aligned @address
argument via VM_BUG_ON() assertion.

Update the implementation to take a 'struct vm_fault' argument directly
and apply the address alignment fixup internally to fix crash signatures
like:

    kernel BUG at arch/x86/mm/pgtable.c:515!
    invalid opcode: 0000 [#1] SMP NOPTI
    CPU: 51 PID: 43713 Comm: java Tainted: G           OE     4.19.35 #1
    [..]
    RIP: 0010:pmdp_set_access_flags+0x48/0x50
    [..]
    Call Trace:
     vmf_insert_pfn_pmd+0x198/0x350
     dax_iomap_fault+0xe82/0x1190
     ext4_dax_huge_fault+0x103/0x1f0
     ? __switch_to_asm+0x40/0x70
     __handle_mm_fault+0x3f6/0x1370
     ? __switch_to_asm+0x34/0x70
     ? __switch_to_asm+0x40/0x70
     handle_mm_fault+0xda/0x200
     __do_page_fault+0x249/0x4f0
     do_page_fault+0x32/0x110
     ? page_fault+0x8/0x30
     page_fault+0x1e/0x30

Link: http://lkml.kernel.org/r/155741946350.372037.11148198430068238140.stgit@dwillia2-desk3.amr.corp.intel.com
Fixes: c6f3c5ee40c1 ("mm/huge_memory.c: fix modifying of page protection by insert_pfn_pmd()")
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Reported-by: Piotr Balcer <piotr.balcer@intel.com>
Tested-by: Yan Ma <yan.ma@intel.com>
Tested-by: Pankaj Gupta <pagupta@redhat.com>
Reviewed-by: Matthew Wilcox <willy@infradead.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Chandan Rajendra <chandan@linux.ibm.com>
Cc: Souptick Joarder <jrdr.linux@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoMerge tag 'iommu-updates-v5.2' of ssh://gitolite.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 13 May 2019 13:23:18 +0000 (09:23 -0400)]
Merge tag 'iommu-updates-v5.2' of ssh://gitolite./linux/kernel/git/joro/iommu

Pull IOMMU updates from Joerg Roedel:

 - ATS support for ARM-SMMU-v3.

 - AUX domain support in the IOMMU-API and the Intel VT-d driver. This
   adds support for multiple DMA address spaces per (PCI-)device. The
   use-case is to multiplex devices between host and KVM guests in a
   more flexible way than supported by SR-IOV.

 - the rest are smaller cleanups and fixes, two of which needed to be
   reverted after testing in linux-next.

* tag 'iommu-updates-v5.2' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/joro/iommu: (45 commits)
  Revert "iommu/amd: Flush not present cache in iommu_map_page"
  Revert "iommu/amd: Remove the leftover of bypass support"
  iommu/vt-d: Fix leak in intel_pasid_alloc_table on error path
  iommu/vt-d: Make kernel parameter igfx_off work with vIOMMU
  iommu/vt-d: Set intel_iommu_gfx_mapped correctly
  iommu/amd: Flush not present cache in iommu_map_page
  iommu/vt-d: Cleanup: no spaces at the start of a line
  iommu/vt-d: Don't request page request irq under dmar_global_lock
  iommu/vt-d: Use struct_size() helper
  iommu/mediatek: Fix leaked of_node references
  iommu/amd: Remove amd_iommu_pd_list
  iommu/arm-smmu: Log CBFRSYNRA register on context fault
  iommu/arm-smmu-v3: Don't disable SMMU in kdump kernel
  iommu/arm-smmu-v3: Disable tagged pointers
  iommu/arm-smmu-v3: Add support for PCI ATS
  iommu/arm-smmu-v3: Link domains and devices
  iommu/arm-smmu-v3: Add a master->domain pointer
  iommu/arm-smmu-v3: Store SteamIDs in master
  iommu/arm-smmu-v3: Rename arm_smmu_master_data to arm_smmu_master
  ACPI/IORT: Check ATS capability in root complex nodes
  ...

5 years agoMerge tag 'linux-watchdog-5.2-rc1' of git://www.linux-watchdog.org/linux-watchdog
Linus Torvalds [Mon, 13 May 2019 13:20:42 +0000 (09:20 -0400)]
Merge tag 'linux-watchdog-5.2-rc1' of git://linux-watchdog.org/linux-watchdog

Pull watchdog updates from Wim Van Sebroeck:

 - a new watchdog driver for the ROHM BD70528 watchdog block

 - a new watchdog driver for the i.MX system controller watchdog

 - conversions to use device managed functions and other improvements

 - refactor watchdog_init_timeout

 - make watchdog core configurable as module

 - pretimeout governors improvements

 - a lot of other fixes

* tag 'linux-watchdog-5.2-rc1' of git://www.linux-watchdog.org/linux-watchdog: (114 commits)
  watchdog: Enforce that at least one pretimeout governor is enabled
  watchdog: stm32: add dynamic prescaler support
  watchdog: Improve Kconfig entry ordering and dependencies
  watchdog: npcm: Enable modular builds
  watchdog: Make watchdog core configurable as module
  watchdog: Move pretimeout governor configuration up
  watchdog: Use depends instead of select for pretimeout governors
  watchdog: rtd119x: drop unused module.h include
  watchdog: intel_scu: make it explicitly non-modular
  watchdog: coh901327: make it explicitly non-modular
  watchdog: ziirave_wdt: drop warning after calling watchdog_init_timeout
  watchdog: xen_wdt: drop warning after calling watchdog_init_timeout
  watchdog: stm32_iwdg: drop warning after calling watchdog_init_timeout
  watchdog: st_lpc_wdt: drop warning after calling watchdog_init_timeout
  watchdog: sp5100_tco: drop warning after calling watchdog_init_timeout
  watchdog: renesas_wdt: drop warning after calling watchdog_init_timeout
  watchdog: nic7018_wdt: drop warning after calling watchdog_init_timeout
  watchdog: ni903x_wdt: drop warning after calling watchdog_init_timeout
  watchdog: imx_sc_wdt: drop warning after calling watchdog_init_timeout
  watchdog: i6300esb: drop warning after calling watchdog_init_timeout
  ...

5 years agoMerge tag 'upstream-5.2-rc1' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 12 May 2019 22:16:31 +0000 (18:16 -0400)]
Merge tag 'upstream-5.2-rc1' of ssh://gitolite./linux/kernel/git/rw/ubifs

Pull UBI/UBIFS updates from Richard Weinberger:

 - fscrypt framework usage updates

 - One huge fix for xattr unlink

 - Cleanup of fscrypt ifdefs

 - Fix for our new UBIFS auth feature

* tag 'upstream-5.2-rc1' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/rw/ubifs:
  ubi: wl: Fix uninitialized variable
  ubifs: Drop unnecessary setting of zbr->znode
  ubifs: Remove ifdefs around CONFIG_UBIFS_ATIME_SUPPORT
  ubifs: Remove #ifdef around CONFIG_FS_ENCRYPTION
  ubifs: Limit number of xattrs per inode
  ubifs: orphan: Handle xattrs like files
  ubifs: journal: Handle xattrs like files
  ubifs: find.c: replace swap function with built-in one
  ubifs: Do not skip hash checking in data nodes
  ubifs: work around high stack usage with clang
  ubifs: remove unused function __ubifs_shash_final
  ubifs: remove unnecessary #ifdef around fscrypt_ioctl_get_policy()
  ubifs: remove unnecessary calls to set up directory key

5 years agoMerge tag 'mtd/for-5.2' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/mtd...
Linus Torvalds [Sun, 12 May 2019 21:57:52 +0000 (17:57 -0400)]
Merge tag 'mtd/for-5.2' of ssh://gitolite./linux/kernel/git/mtd/linux

Pull MTD updates from Richard Weinberger:
 "MTD core changes:
   - New AFS partition parser
   - Update MAINTAINERS entry
   - Use of fall-throughs markers

  NAND core changes:
   - Support having the bad block markers in either the first, second or
     last page of a block. The combination of all three location is now
     possible.
   - Constification of NAND_OP_PARSER(_PATTERN) elements.
   - Generic NAND DT bindings changed to yaml format (can be used to
     check the proposed bindings. First platform to be fully supported:
     sunxi.
   - Stopped using several legacy hooks.
   - Preparation to use the generic NAND layer with the addition of
     several helpers and the removal of the struct nand_chip from
     generic functions.
   - Kconfig cleanup to prepare the introduction of external ECC engines
     support.
   - Fallthrough comments.
   - Introduction of the SPI-mem dirmap API for SPI-NAND devices.

  Raw NAND controller drivers changes:
   - nandsim:
      - Switch to ->exec-op().
   - meson:
      - Misc cleanups and fixes.
      - New OOB layout.
   - Sunxi:
      - A23/A33 NAND DMA support.
   - Ingenic:
      - Full reorganization and cleanup.
      - Clear separation between NAND controller and ECC engine.
      - Support JZ4740 an JZ4725B.
   - Denali:
      - Clear controller/chip separation.
      - ->exec_op() migration.
      - Various cleanups.
   - fsl_elbc:
      - Enable software ECC support.
   - Atmel:
      - Sam9x60 support.
   - GPMI:
      - Introduce the GPMI_IS_MXS() macro.
   - Various trivial/spelling/coding style fixes.

  SPI NOR core changes:
   - Print all JEDEC ID bytes on error
   - Fix comment of spi_nor_find_best_erase_type()
   - Add region locking flags for s25fl512s

  SPI NOR controller drivers changes:
   - intel-spi:
      - Avoid crossing 4K address boundary on read/write
      - Add support for Intel Comet Lake SPI serial flash"

* tag 'mtd/for-5.2' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/mtd/linux: (120 commits)
  mtd: part: fix incorrect format specifier for an unsigned long long
  mtd: lpddr_cmds: Mark expected switch fall-through
  mtd: phram: Mark expected switch fall-throughs
  mtd: cfi_cmdset_0002: Mark expected switch fall-throughs
  mtd: cfi_util: mark expected switch fall-throughs
  MAINTAINERS: MTD Git repository is hosted on kernel.org
  MAINTAINERS: Update jffs2 entry
  mtd: afs: add v2 partition parsing
  mtd: afs: factor the IIS read into partition parser
  mtd: afs: factor footer parsing into the v1 part parsing
  mtd: factor out v1 partition parsing
  mtd: afs: simplify partition detection
  mtd: afs: simplify partition parsing
  mtd: partitions: Add OF support to AFS partitions
  mtd: partitions: Add AFS partitions DT bindings
  mtd: afs: Move AFS partition parser to parsers subdir
  mtd: maps: Make uclinux_ram_map static
  mtd: maps: Allow MTD_PHYSMAP with MTD_RAM
  MAINTAINERS: Add myself as MTD maintainer
  MAINTAINERS: Remove my name from the MTD and NAND entries
  ...

5 years agoMerge tag 'for-linus-5.2-rc1' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 12 May 2019 21:52:13 +0000 (17:52 -0400)]
Merge tag 'for-linus-5.2-rc1' of ssh://gitolite./linux/kernel/git/rw/uml

Pull UML updates from Richard Weinberger:

 - Kconfig cleanups

 - Fix cpu_all_mask() usage

 - Various bug fixes

* tag 'for-linus-5.2-rc1' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/rw/uml:
  um: irq: don't set the chip for all irqs
  um: define set_pte_at() as a static inline function, not a macro
  um: remove uses of variable length arrays
  um: remove unused variable
  uml: fix a boot splat wrt use of cpu_all_mask
  um: Do not unlock mutex that is not hold.
  hostfs: fix mismatch between link_file definition and declaration
  arch: um: drivers: Kconfig: pedantic formatting
  arch: um: Kconfig: pedantic indention cleanups
  um: Revert to using stack for pt_regs in signal handling

5 years agoMerge tag 'tag-chrome-platform-for-v5.2' of ssh://gitolite.kernel.org/pub/scm/linux...
Linus Torvalds [Sun, 12 May 2019 11:00:21 +0000 (07:00 -0400)]
Merge tag 'tag-chrome-platform-for-v5.2' of ssh://gitolite./linux/kernel/git/chrome-platform/linux

Pull chrome platform updates from Benson Leung:
 "CrOS EC:
   - Add EC host command support using rpmsg
   - Add new CrOS USB PD logging driver
   - Transfer spi messages at high priority
   - Add support to trace CrOS EC commands
   - Minor fixes and cleanups in protocol and debugfs

  Wilco EC:
   - Standardize Wilco EC mailbox interface
   - Add h1_gpio status to debugfs"

* tag 'tag-chrome-platform-for-v5.2' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/chrome-platform/linux:
  platform/chrome: cros_ec_proto: Add trace event to trace EC commands
  platform/chrome: cros_ec_debugfs: Use cros_ec_cmd_xfer_status helper
  platform/chrome: cros_ec: Add EC host command support using rpmsg
  platform/chrome: wilco_ec: Add h1_gpio status to debugfs
  platform/chrome: wilco_ec: Standardize mailbox interface
  platform/chrome: cros_ec_proto: check for NULL transfer function
  platform/chrome: Add CrOS USB PD logging driver
  platform/chrome: cros_ec_spi: Transfer messages at high priority
  platform/chrome: cros_ec_debugfs: no need to check return value of debugfs_create functions
  platform/chrome: cros_ec_debugfs: Remove dev_warn when console log is not supported

5 years agoMerge tag 'gpio-v5.2-1' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux...
Linus Torvalds [Sat, 11 May 2019 14:54:43 +0000 (10:54 -0400)]
Merge tag 'gpio-v5.2-1' of git://git./linux/kernel/git/linusw/linux-gpio

Pull gpio updates from Linus Walleij:
 "This is the bulk of the GPIO changes for the v5.2 kernel cycle. A bit
  later than usual because I was ironing out my own mistakes. I'm
  holding some stuff back for the next kernel as a result, and this
  should be a healthy and well tested batch.

  Core changes:

   - The gpiolib MMIO driver has been enhanced to handle two direction
     registers, i.e. one register to set lines as input and one register
     to set lines as output. It turns out some silicon engineer thinks
     the ability to configure a line as input and output at the same
     time makes sense, this can be debated but includes a lot of analog
     electronics reasoning, and the registers are there and need to be
     handled consistently. Unsurprisingly, we enforce the lines to be
     either inputs or outputs in such schemes.

   - Send in the proper argument value to .set_config() dispatched to
     the pin control subsystem. Nobody used it before, now someone does,
     so fix it to work as expected.

   - The ACPI gpiolib portions can now handle pin bias setting (pull up
     or pull down). This has been in the ACPI spec for years and we
     finally have it properly integrated with Linux GPIOs. It was based
     on an observation from Andy Schevchenko that Thomas Petazzoni's
     changes to the core for biasing the PCA950x GPIO expander actually
     happen to fit hand-in-glove with what the ACPI core needed. Such
     nice synergies happen sometimes.

  New drivers:

   - A new driver for the Mellanox BlueField GPIO controller. This is
     using 64bit MMIO registers and can configure lines as inputs and
     outputs at the same time and after improving the MMIO library we
     handle it just fine. Interesting.

   - A new IXP4xx proper gpiochip driver with hierarchical interrupts
     should be coming in from the ARM SoC tree as well.

  Driver enhancements:

   - The PCA053x driver handles the CAT9554 GPIO expander.

   - The PCA053x driver handles the NXP PCAL6416 GPIO expander.

   - Wake-up support on PCA053x GPIO lines.

   - OMAP now does a nice asynchronous IRQ handling on wake-ups by
     letting everything wake up on edges, and this makes runtime PM work
     as expected too.

  Misc:

   - Several cleanups such as devres fixes.

   - Get rid of some languager comstructs that cause problems when
     compiling with LLVMs clang.

   - Documentation review and update"

* tag 'gpio-v5.2-1' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio: (85 commits)
  gpio: Update documentation
  docs: gpio: convert docs to ReST and rename to *.rst
  gpio: sch: Remove write-only core_base
  gpio: pxa: Make two symbols static
  gpiolib: acpi: Respect pin bias setting
  gpiolib: acpi: Add acpi_gpio_update_gpiod_lookup_flags() helper
  gpiolib: acpi: Set pin value, based on bias, more accurately
  gpiolib: acpi: Change type of dflags
  gpiolib: Introduce GPIO_LOOKUP_FLAGS_DEFAULT
  gpiolib: Make use of enum gpio_lookup_flags consistent
  gpiolib: Indent entry values of enum gpio_lookup_flags
  gpio: pca953x: add support for pca6416
  dt-bindings: gpio: pca953x: document the nxp,pca6416
  gpio: pca953x: add pcal6416 to the of_device_id table
  gpio: gpio-omap: Remove conditional pm_runtime handling for GPIO interrupts
  gpio: gpio-omap: configure edge detection for level IRQs for idle wakeup
  tracing: stop making gpio tracing configurable
  gpio: pca953x: Configure wake-up path when wake-up is enabled
  gpio: of: Optimize quirk checks
  gpio: mmio: Drop bgpio_dir_inverted
  ...

5 years agoMerge tag 'vfio-v5.2-rc1' of git://github.com/awilliam/linux-vfio
Linus Torvalds [Sat, 11 May 2019 14:47:46 +0000 (10:47 -0400)]
Merge tag 'vfio-v5.2-rc1' of git://github.com/awilliam/linux-vfio

Pull VFIO updates from Alex Williamson:

 - Improve dev_printk() usage (Bjorn Helgaas)

 - Fix issue with blocking in !TASK_RUNNING state while waiting for
   userspace to release devices (Farhan Ali)

 - Fix error path cleanup in nvlink setup (Greg Kurz)

 - mdev-core cleanups and fixes in preparation for more use cases (Parav
   Pandit)

 - Cornelia has volunteered as an official vfio reviewer (Cornelia Huck)

* tag 'vfio-v5.2-rc1' of git://github.com/awilliam/linux-vfio:
  vfio: Add Cornelia Huck as reviewer
  vfio/mdev: Avoid inline get and put parent helpers
  vfio/mdev: Fix aborting mdev child device removal if one fails
  vfio/mdev: Follow correct remove sequence
  vfio/mdev: Avoid masking error code to EBUSY
  vfio/mdev: Drop redundant extern for exported symbols
  vfio/mdev: Removed unused kref
  vfio/mdev: Avoid release parent reference during error path
  vfio-pci/nvlink2: Fix potential VMA leak
  vfio: Fix WARNING "do not call blocking ops when !TASK_RUNNING"
  vfio: Use dev_printk() when possible

5 years agoMerge branch 'next-tomoyo2' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris...
Linus Torvalds [Sat, 11 May 2019 14:38:59 +0000 (10:38 -0400)]
Merge branch 'next-tomoyo2' of git://git./linux/kernel/git/jmorris/linux-security

Pull tomoyo updates from James Morris:
 "Fixes to enable fuzz testing, and a fix for calculating whether a
  filesystem is user-modifiable"

* 'next-tomoyo2' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/linux-security:
  tomoyo: Don't emit WARNING: string while fuzzing testing.
  tomoyo: Change pathname calculation for read-only filesystems.
  tomoyo: Check address length before reading address family
  tomoyo: Add a kernel config option for fuzzing testing.

5 years agoMerge tag 'xtensa-20190510' of git://github.com/jcmvbkbc/linux-xtensa
Linus Torvalds [Sat, 11 May 2019 14:27:34 +0000 (10:27 -0400)]
Merge tag 'xtensa-20190510' of git://github.com/jcmvbkbc/linux-xtensa

Pull xtensa updates from Max Filippov:

 - implement atomic operations using exclusive access Xtensa option
   operations

 - add support for Xtensa cores with memory protection unit (MPU)

 - clean up xtensa-specific kernel-only headers

 - fix error path in simdisk_setup

* tag 'xtensa-20190510' of git://github.com/jcmvbkbc/linux-xtensa:
  xtensa: implement initialize_cacheattr for MPU cores
  xtensa: add exclusive atomics support
  xtensa: clean up inline assembly in futex.h
  xtensa: replace variant/core.h with asm/core.h
  xtensa: drop ifdef __KERNEL__ from kernel-only headers
  xtensa: set proper error code for simdisk_setup()
  xtensa: fix incorrect fd close in error case of simdisk_setup()

5 years agotomoyo: Don't emit WARNING: string while fuzzing testing.
Tetsuo Handa [Tue, 7 May 2019 11:34:22 +0000 (20:34 +0900)]
tomoyo: Don't emit WARNING: string while fuzzing testing.

Commit cff0e6c3ec3e6230 ("tomoyo: Add a kernel config option for fuzzing
testing.") enabled the learning mode, but syzkaller is detecting any
"WARNING:" string as a crash. Thus, disable TOMOYO's quota warning if
built for fuzzing testing.

Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: James Morris <jamorris@linux.microsoft.com>
5 years agotomoyo: Change pathname calculation for read-only filesystems.
Tetsuo Handa [Wed, 27 Feb 2019 14:19:24 +0000 (23:19 +0900)]
tomoyo: Change pathname calculation for read-only filesystems.

Commit 5625f2e3266319fd ("TOMOYO: Change pathname for non-rename()able
filesystems.") intended to be applied to filesystems where the content is
not controllable from the userspace (e.g. proc, sysfs, securityfs), based
on an assumption that such filesystems do not support rename() operation.

But it turned out that read-only filesystems also do not support rename()
operation despite the content is controllable from the userspace, and that
commit is annoying TOMOYO users who want to use e.g. squashfs as the root
filesystem due to use of local name which does not start with '/'.

Therefore, based on an assumption that filesystems which require the
device argument upon mount() request is an indication that the content
is controllable from the userspace, do not use local name if a filesystem
does not support rename() operation but requires the device argument upon
mount() request.

Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: James Morris <jamorris@linux.microsoft.com>
5 years agotomoyo: Check address length before reading address family
Tetsuo Handa [Fri, 12 Apr 2019 10:59:36 +0000 (19:59 +0900)]
tomoyo: Check address length before reading address family

KMSAN will complain if valid address length passed to bind()/connect()/
sendmsg() is shorter than sizeof("struct sockaddr"->sa_family) bytes.

Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: James Morris <jamorris@linux.microsoft.com>
5 years agotomoyo: Add a kernel config option for fuzzing testing.
Tetsuo Handa [Fri, 12 Apr 2019 11:04:54 +0000 (20:04 +0900)]
tomoyo: Add a kernel config option for fuzzing testing.

syzbot is reporting kernel panic triggered by memory allocation fault
injection before loading TOMOYO's policy [1]. To make the fuzzing tests
useful, we need to assign a profile other than "disabled" (no-op) mode.
Therefore, let's allow syzbot to load TOMOYO's built-in policy for
"learning" mode using a kernel config option. This option must not be
enabled for kernels built for production system, for this option also
disables domain/program checks when modifying policy configuration via
/sys/kernel/security/tomoyo/ interface.

[1] https://syzkaller.appspot.com/bug?extid=29569ed06425fcf67a95

Reported-by: syzbot <syzbot+e1b8084e532b6ee7afab@syzkaller.appspotmail.com>
Reported-by: syzbot <syzbot+29569ed06425fcf67a95@syzkaller.appspotmail.com>
Reported-by: syzbot <syzbot+2ee3f8974c2e7dc69feb@syzkaller.appspotmail.com>
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: James Morris <jamorris@linux.microsoft.com>
5 years agoMerge tag 'docs-5.2a' of git://git.lwn.net/linux
Linus Torvalds [Fri, 10 May 2019 17:24:53 +0000 (13:24 -0400)]
Merge tag 'docs-5.2a' of git://git.lwn.net/linux

Pull more documentation updates from Jonathan Corbet:
 "Some late arriving documentation changes. In particular, this contains
  the conversion of the x86 docs to RST, which has been in the works for
  some time but needed a couple of final tweaks"

* tag 'docs-5.2a' of git://git.lwn.net/linux: (29 commits)
  Documentation: x86: convert x86_64/machinecheck to reST
  Documentation: x86: convert x86_64/cpu-hotplug-spec to reST
  Documentation: x86: convert x86_64/fake-numa-for-cpusets to reST
  Documentation: x86: convert x86_64/5level-paging.txt to reST
  Documentation: x86: convert x86_64/mm.txt to reST
  Documentation: x86: convert x86_64/uefi.txt to reST
  Documentation: x86: convert x86_64/boot-options.txt to reST
  Documentation: x86: convert i386/IO-APIC.txt to reST
  Documentation: x86: convert usb-legacy-support.txt to reST
  Documentation: x86: convert orc-unwinder.txt to reST
  Documentation: x86: convert resctrl_ui.txt to reST
  Documentation: x86: convert microcode.txt to reST
  Documentation: x86: convert pti.txt to reST
  Documentation: x86: convert amd-memory-encryption.txt to reST
  Documentation: x86: convert intel_mpx.txt to reST
  Documentation: x86: convert protection-keys.txt to reST
  Documentation: x86: convert pat.txt to reST
  Documentation: x86: convert mtrr.txt to reST
  Documentation: x86: convert tlb.txt to reST
  Documentation: x86: convert zero-page.txt to reST
  ...

5 years agoMerge tag 'printk-for-5.2-fixes' of ssh://gitolite.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 10 May 2019 17:14:07 +0000 (13:14 -0400)]
Merge tag 'printk-for-5.2-fixes' of ssh://gitolite./linux/kernel/git/pmladek/printk

Pull printk fixup from Petr Mladek:
 "Replace the problematic probe_kernel_read() with original simple
  pointer checks in vsprintf()"

* tag 'printk-for-5.2-fixes' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/pmladek/printk:
  vsprintf: Do not break early boot with probing addresses

5 years agoMerge tag 'pidfd-fixes-v5.2-rc1' of gitolite.kernel.org:pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 10 May 2019 17:07:55 +0000 (13:07 -0400)]
Merge tag 'pidfd-fixes-v5.2-rc1' of gitolite.pub/scm/linux/kernel/git/brauner/linux

Pull pidfd fixes from Christian Brauner:
 "This fixes two bugs:

   - The first one reported by Linus whereby the pidfd-metadata binary
     was not placed in a .gitignore file.

   - The second one is rather urgent and fixes a locking issue found by
     syzkaller.

     What happened is that during process creation we need to check
     whether the cgroup we are in allows us to fork. To perform this
     check the cgroup needs to guard itself against threadgroup changes
     and takes a lock.

     Prior to CLONE_PIDFD the cleanup target "bad_fork_free_pid" would
     also need to release said lock. That's not true anymore since
     CLONE_PIDFD so this is fixed here.

     Syzkaller has tested the patch and was not able to reproduce the
     issue"

* tag 'pidfd-fixes-v5.2-rc1' of gitolite.kernel.org:pub/scm/linux/kernel/git/brauner/linux:
  fork: do not release lock that wasn't taken
  samples: add .gitignore for pidfd-metadata

5 years agoMerge tag 'platform-drivers-x86-v5.2-1' of git://git.infradead.org/linux-platform...
Linus Torvalds [Fri, 10 May 2019 17:03:47 +0000 (13:03 -0400)]
Merge tag 'platform-drivers-x86-v5.2-1' of git://git.infradead.org/linux-platform-drivers-x86

Pull x86 platform driver updates from Andy Shevchenko:
 "Gathered pile of patches for Platform Drivers x86. No surprises and no
  merge conflicts. Business as usual.

  Summary:

   - New driver of power button for Basin Cove PMIC.

   - ASUS WMI driver has got a Fn lock mode switch support.

   - Resolve a never end story with non working Wi-Fi on newer Lenovo
     Ideapad computers. Now the black list is replaced with white list.

   - New facility to debug S0ix failures on Intel Atom platforms. The
     Intel PMC and accompanying drivers are cleaned up.

   - Mellanox got a new TmFifo driver. Besides tachometer sensor and
     watchdog are enabled on Mellanox platforms.

   - The information of embedded controller is now recognized on new
     Thinkpads. Bluetooth driver on Thinkpads is blacklisted for some
     models.

   - Touchscreen DMI driver extended to support 'jumper ezpad 6 pro b'
     and Myria MY8307 2-in-1.

   - Additionally few small fixes here and there for WMI and ACPI laptop
     drivers.

   - The following is an automated git shortlog grouped by driver:

   - alienware-wmi:
      - printing the wrong error code
      - fix kfree on potentially uninitialized pointer

   - asus-wmi:
      - Add fn-lock mode switch support

   - dell-laptop:
      - fix rfkill functionality

   - dell-rbtn:
      - Add missing #include

   - ideapad-laptop:
      - Remove no_hw_rfkill_list

   - intel_pmc_core:
      - Allow to dump debug registers on S0ix failure
      - Convert to a platform_driver
      - Mark local function static

   - intel_pmc_ipc:
      - Don't map non-used optional resources
      - Apply same width for offset definitions
      - Use BIT() macro
      - adding error handling

   - intel_punit_ipc:
      - Revert "Fix resource ioremap warning"

   - mlx-platform:
      - Add mlx-wdt platform driver activation
      - Add support for tachometer speed register
      - Add TmFifo driver for Mellanox BlueField Soc

   - sony-laptop:
      - Fix unintentional fall-through

   - thinkpad_acpi:
      - cleanup for Thinkpad ACPI led
      - Mark expected switch fall-throughs
      - fix spelling mistake "capabilites" -> "capabilities"
      - Read EC information on newer models
      - Disable Bluetooth for some machines

   - touchscreen_dmi:
      - Add info for 'jumper ezpad 6 pro b' touchscreen
      - Add info for Myria MY8307 2-in-1"

* tag 'platform-drivers-x86-v5.2-1' of git://git.infradead.org/linux-platform-drivers-x86: (26 commits)
  platform/x86: Add support for Basin Cove power button
  platform/x86: asus-wmi: Add fn-lock mode switch support
  platform/x86: ideapad-laptop: Remove no_hw_rfkill_list
  platform/x86: touchscreen_dmi: Add info for 'jumper ezpad 6 pro b' touchscreen
  platform/x86: thinkpad_acpi: cleanup for Thinkpad ACPI led
  platform/x86: thinkpad_acpi: Mark expected switch fall-throughs
  platform/x86: sony-laptop: Fix unintentional fall-through
  platform/x86: alienware-wmi: printing the wrong error code
  platform/x86: intel_pmc_core: Allow to dump debug registers on S0ix failure
  platform/x86: intel_pmc_core: Convert to a platform_driver
  platform/x86: mlx-platform: Add mlx-wdt platform driver activation
  platform/x86: mlx-platform: Add support for tachometer speed register
  platform/mellanox: Add TmFifo driver for Mellanox BlueField Soc
  platform/x86: thinkpad_acpi: fix spelling mistake "capabilites" -> "capabilities"
  platform/x86: intel_punit_ipc: Revert "Fix resource ioremap warning"
  platform/x86: intel_pmc_ipc: Don't map non-used optional resources
  platform/x86: intel_pmc_ipc: Apply same width for offset definitions
  platform/x86: intel_pmc_ipc: Use BIT() macro
  platform/x86: alienware-wmi: fix kfree on potentially uninitialized pointer
  platform/x86: dell-laptop: fix rfkill functionality
  ...

5 years agoMerge tag 'fbdev-v5.2' of git://github.com/bzolnier/linux
Linus Torvalds [Fri, 10 May 2019 16:59:51 +0000 (12:59 -0400)]
Merge tag 'fbdev-v5.2' of git://github.com/bzolnier/linux

Pull fbdev updates from Bartlomiej Zolnierkiewicz:
 "Four small fixes for fb core, updates for udlfb, sm712fb, macfb and
  atafb drivers. Redundant code removals from amba-clcd and atmel_lcdfb
  drivers. Minor fixes/cleanups for other fb drivers

  Detailed summary:

   - fix regression in fbcon logo handling on 'quiet' boots (Andreas
     Schwab)

   - fix divide-by-zero error in fb_var_to_videomode() (Shile Zhang)

   - fix 'WARNING in __alloc_pages_nodemask' bug (Jiufei Xue)

   - list all PCI memory BARs as conflicting apertures (Gerd Hoffmann)

   - update udlfb driver: fix sleeping inside spinlock, add mutex around
     rendering calls and remove redundant code (Mikulas Patocka)

   - update sm712fb driver: fix SM720 support related issues (Yifeng Li)

   - update macfb driver: fix DAFB colour table pointer initialization
     and remove redundant code (Finn Thain)

   - update atafb driver: fix kexec support, use dev_*() calls instead
     of printk() and remove obsolete module support (Geert Uytterhoeven)

   - add support to mxsfb driver for skipping display initialization for
     flicker-free display takeover from bootloader (Melchior Franz)

   - remove Versatile and Nomadik board families support from amba-clcd
     driver as they are handled by DRM driver nowadays (Linus Walleij)

   - remove no longer needed AVR and platform_data support from
     atmel_lcdfb driver (Alexandre Belloni)

   - misc fixes (Colin Ian King, Julia Lawall, Gustavo A. R. Silva,
     Aditya Pakki, Kangjie Lu, YueHaibing)

   - misc cleanups (Enrico Weigelt, Kefeng Wang)"

* tag 'fbdev-v5.2' of git://github.com/bzolnier/linux: (38 commits)
  video: fbdev: Use dev_get_drvdata()
  fbcon: Don't reset logo_shown when logo is currently shown
  video: fbdev: atmel_lcdfb: remove set but not used variable 'pdata'
  video: fbdev: mxsfb: remove set but not used variable 'line_count'
  video: fbdev: pvr2fb: remove set but not used variable 'size'
  fbdev: fix WARNING in __alloc_pages_nodemask bug
  video: amba-clcd: Decomission Versatile and Nomadik
  fbdev: sm712fb: fix memory frequency by avoiding a switch/case fallthrough
  fbdev: fix divide error in fb_var_to_videomode
  fbdev: sm712fb: use 1024x768 by default on non-MIPS, fix garbled display
  fbdev: sm712fb: fix support for 1024x768-16 mode
  fbdev: sm712fb: fix crashes and garbled display during DPMS modesetting
  fbdev: sm712fb: fix crashes during framebuffer writes by correctly mapping VRAM
  fbdev: sm712fb: fix boot screen glitch when sm712fb replaces VGA
  fbdev: sm712fb: fix VRAM detection, don't set SR70/71/74/75
  fbdev: sm712fb: fix brightness control on reboot, don't set SR30
  fbdev: sm712fb: fix white screen of death on reboot, don't set CR3B-CR3F
  video: imsttfb: fix potential NULL pointer dereferences
  video: hgafb: fix potential NULL pointer dereference
  fbdev: list all pci memory bars as conflicting apertures
  ...

5 years agoMerge tag 'pwm/for-5.2-rc1' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 10 May 2019 16:57:15 +0000 (12:57 -0400)]
Merge tag 'pwm/for-5.2-rc1' of ssh://gitolite./linux/kernel/git/thierry.reding/linux-pwm

Pull pwm updates from Thierry Reding:
 "Nothing out of the ordinary this cycle.

  The bulk of this is a collection of fixes for existing drivers and
  some cleanups. There's one new driver for i.MX SoCs and addition of
  support for some new variants to existing drivers"

* tag 'pwm/for-5.2-rc1' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/thierry.reding/linux-pwm:
  pwm: meson: Add clock source configuration for Meson G12A
  dt-bindings: pwm: Update bindings for the Meson G12A Family
  pwm: samsung: Don't uses devm_*() functions in ->request()
  pwm: Clear chip_data in pwm_put()
  pwm: Add i.MX TPM PWM driver support
  dt-bindings: pwm: Add i.MX TPM PWM binding
  pwm: imx27: Use devm_platform_ioremap_resource() to simplify code
  pwm: meson: Use the spin-lock only to protect register modifications
  pwm: meson: Don't disable PWM when setting duty repeatedly
  pwm: meson: Consider 128 a valid pre-divider
  pwm: sysfs: fix typo "its" -> "it's"
  pwm: tiehrpwm: Enable compilation for ARCH_K3
  dt-bindings: pwm: tiehrpwm: Add TI AM654 SoC specific compatible
  pwm: tiehrpwm: Update shadow register for disabling PWMs
  pwm: img: Turn final 'else if' into 'else' in img_pwm_config
  pwm: Fix deadlock warning when removing PWM device

5 years agoMerge tag 'mailbox-v5.2' of git://git.linaro.org/landing-teams/working/fujitsu/integr...
Linus Torvalds [Fri, 10 May 2019 16:55:16 +0000 (12:55 -0400)]
Merge tag 'mailbox-v5.2' of git://git.linaro.org/landing-teams/working/fujitsu/integration

Pull mailbox updates from Jassi Brar:

 - New driver: Armada 37xx mailbox controller

 - Misc: Use devm_ api for imx and platform_get_irq for stm32

* tag 'mailbox-v5.2' of git://git.linaro.org/landing-teams/working/fujitsu/integration:
  mailbox: Add support for Armada 37xx rWTM mailbox
  dt-bindings: mailbox: Document armada-3700-rwtm-mailbox binding
  mailbox: stm32-ipcc: check invalid irq
  mailbox: imx: use devm_platform_ioremap_resource() to simplify code

5 years agovsprintf: Do not break early boot with probing addresses
Petr Mladek [Fri, 10 May 2019 08:42:13 +0000 (10:42 +0200)]
vsprintf: Do not break early boot with probing addresses

The commit 3e5903eb9cff70730 ("vsprintf: Prevent crash when dereferencing
invalid pointers") broke boot on several architectures. The common
pattern is that probe_kernel_read() is not working during early
boot because userspace access framework is not ready.

It is a generic problem. We have to avoid any complex external
functions in vsprintf() code, especially in the common path.
They might break printk() easily and are hard to debug.

Replace probe_kernel_read() with some simple checks for obvious
problems.

Details:

1. Report on Power:

Kernel crashes very early during boot with with CONFIG_PPC_KUAP and
CONFIG_JUMP_LABEL_FEATURE_CHECK_DEBUG

The problem is the combination of some new code called via printk(),
check_pointer() which calls probe_kernel_read(). That then calls
allow_user_access() (PPC_KUAP) and that uses mmu_has_feature() too early
(before we've patched features). With the JUMP_LABEL debug enabled that
causes us to call printk() & dump_stack() and we end up recursing and
overflowing the stack.

Because it happens so early you don't get any output, just an apparently
dead system.

The stack trace (which you don't see) is something like:

  ...
  dump_stack+0xdc
  probe_kernel_read+0x1a4
  check_pointer+0x58
  string+0x3c
  vsnprintf+0x1bc
  vscnprintf+0x20
  printk_safe_log_store+0x7c
  printk+0x40
  dump_stack_print_info+0xbc
  dump_stack+0x8
  probe_kernel_read+0x1a4
  probe_kernel_read+0x19c
  check_pointer+0x58
  string+0x3c
  vsnprintf+0x1bc
  vscnprintf+0x20
  vprintk_store+0x6c
  vprintk_emit+0xec
  vprintk_func+0xd4
  printk+0x40
  cpufeatures_process_feature+0xc8
  scan_cpufeatures_subnodes+0x380
  of_scan_flat_dt_subnodes+0xb4
  dt_cpu_ftrs_scan_callback+0x158
  of_scan_flat_dt+0xf0
  dt_cpu_ftrs_scan+0x3c
  early_init_devtree+0x360
  early_setup+0x9c

2. Report on s390:

vsnprintf invocations, are broken on s390. For example, the early boot
output now looks like this where the first (efault) should be
the linux_banner:

[    0.099985] (efault)
[    0.099985] setup: Linux is running as a z/VM guest operating system in 64-bit mode
[    0.100066] setup: The maximum memory size is 8192MB
[    0.100070] cma: Reserved 4 MiB at (efault)
[    0.100100] numa: NUMA mode: (efault)

The reason for this, is that the code assumes that
probe_kernel_address() works very early. This however is not true on
at least s390. Uaccess on KERNEL_DS works only after page tables have
been setup on s390, which happens with setup_arch()->paging_init().

Any probe_kernel_address() invocation before that will return -EFAULT.

Fixes: 3e5903eb9cff70730 ("vsprintf: Prevent crash when dereferencing invalid pointers")
Link: http://lkml.kernel.org/r/20190510084213.22149-1-pmladek@suse.com
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Cc: "Tobin C . Harding" <me@tobin.cc>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: linux-kernel@vger.kernel.org
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: linuxppc-dev@lists.ozlabs.org
Cc: Russell Currey <ruscur@russell.cc>
Cc: Christophe Leroy <christophe.leroy@c-s.fr>
Cc: Stephen Rothwell <sfr@ozlabs.org>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: linux-arch@vger.kernel.org
Cc: linux-s390@vger.kernel.org
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Petr Mladek <pmladek@suse.com>
Reviewed-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Signed-off-by: Petr Mladek <pmladek@suse.com>
5 years agoMerge tag 'powerpc-5.2-1' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/power...
Linus Torvalds [Fri, 10 May 2019 12:29:27 +0000 (05:29 -0700)]
Merge tag 'powerpc-5.2-1' of ssh://gitolite./linux/kernel/git/powerpc/linux

Pull powerpc updates from Michael Ellerman:
 "Slightly delayed due to the issue with printk() calling
  probe_kernel_read() interacting with our new user access prevention
  stuff, but all fixed now.

  The only out-of-area changes are the addition of a cpuhp_state, small
  additions to Documentation and MAINTAINERS updates.

  Highlights:

   - Support for Kernel Userspace Access/Execution Prevention (like
     SMAP/SMEP/PAN/PXN) on some 64-bit and 32-bit CPUs. This prevents
     the kernel from accidentally accessing userspace outside
     copy_to/from_user(), or ever executing userspace.

   - KASAN support on 32-bit.

   - Rework of where we map the kernel, vmalloc, etc. on 64-bit hash to
     use the same address ranges we use with the Radix MMU.

   - A rewrite into C of large parts of our idle handling code for
     64-bit Book3S (ie. power8 & power9).

   - A fast path entry for syscalls on 32-bit CPUs, for a 12-17% speedup
     in the null_syscall benchmark.

   - On 64-bit bare metal we have support for recovering from errors
     with the time base (our clocksource), however if that fails
     currently we hang in __delay() and never crash. We now have support
     for detecting that case and short circuiting __delay() so we at
     least panic() and reboot.

   - Add support for optionally enabling the DAWR on Power9, which had
     to be disabled by default due to a hardware erratum. This has the
     effect of enabling hardware breakpoints for GDB, the downside is a
     badly behaved program could crash the machine by pointing the DAWR
     at cache inhibited memory. This is opt-in obviously.

   - xmon, our crash handler, gets support for a read only mode where
     operations that could change memory or otherwise disturb the system
     are disabled.

  Plus many clean-ups, reworks and minor fixes etc.

  Thanks to: Christophe Leroy, Akshay Adiga, Alastair D'Silva, Alexey
  Kardashevskiy, Andrew Donnellan, Aneesh Kumar K.V, Anju T Sudhakar,
  Anton Blanchard, Ben Hutchings, Bo YU, Breno Leitao, Cédric Le Goater,
  Christopher M. Riedl, Christoph Hellwig, Colin Ian King, David Gibson,
  Ganesh Goudar, Gautham R. Shenoy, George Spelvin, Greg Kroah-Hartman,
  Greg Kurz, Horia Geantă, Jagadeesh Pagadala, Joel Stanley, Joe
  Perches, Julia Lawall, Laurentiu Tudor, Laurent Vivier, Lukas Bulwahn,
  Madhavan Srinivasan, Mahesh Salgaonkar, Mathieu Malaterre, Michael
  Neuling, Mukesh Ojha, Nathan Fontenot, Nathan Lynch, Nicholas Piggin,
  Nick Desaulniers, Oliver O'Halloran, Peng Hao, Qian Cai, Ravi
  Bangoria, Rick Lindsley, Russell Currey, Sachin Sant, Stewart Smith,
  Sukadev Bhattiprolu, Thomas Huth, Tobin C. Harding, Tyrel Datwyler,
  Valentin Schneider, Wei Yongjun, Wen Yang, YueHaibing"

* tag 'powerpc-5.2-1' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/powerpc/linux: (205 commits)
  powerpc/64s: Use early_mmu_has_feature() in set_kuap()
  powerpc/book3s/64: check for NULL pointer in pgd_alloc()
  powerpc/mm: Fix hugetlb page initialization
  ocxl: Fix return value check in afu_ioctl()
  powerpc/mm: fix section mismatch for setup_kup()
  powerpc/mm: fix redundant inclusion of pgtable-frag.o in Makefile
  powerpc/mm: Fix makefile for KASAN
  powerpc/kasan: add missing/lost Makefile
  selftests/powerpc: Add a signal fuzzer selftest
  powerpc/booke64: set RI in default MSR
  ocxl: Provide global MMIO accessors for external drivers
  ocxl: move event_fd handling to frontend
  ocxl: afu_irq only deals with IRQ IDs, not offsets
  ocxl: Allow external drivers to use OpenCAPI contexts
  ocxl: Create a clear delineation between ocxl backend & frontend
  ocxl: Don't pass pci_dev around
  ocxl: Split pci.c
  ocxl: Remove some unused exported symbols
  ocxl: Remove superfluous 'extern' from headers
  ocxl: read_pasid never returns an error, so make it void
  ...

5 years agofork: do not release lock that wasn't taken
Christian Brauner [Fri, 10 May 2019 09:53:46 +0000 (11:53 +0200)]
fork: do not release lock that wasn't taken

Avoid calling cgroup_threadgroup_change_end() without having called
cgroup_threadgroup_change_begin() first.

During process creation we need to check whether the cgroup we are in
allows us to fork. To perform this check the cgroup needs to guard itself
against threadgroup changes and takes a lock.
Prior to CLONE_PIDFD the cleanup target "bad_fork_free_pid" would also need
to call cgroup_threadgroup_change_end() because said lock had already been
taken.
However, this is not the case anymore with the addition of CLONE_PIDFD. We
are now allocating a pidfd before we check whether the cgroup we're in can
fork and thus prior to taking the lock. So when copy_process() fails at the
right step it would release a lock we haven't taken.
This bug is not even very subtle to be honest. It's just not very clear
from the naming of cgroup_threadgroup_change_{begin,end}() that a lock is
taken.

Here's the relevant splat:

entry_SYSENTER_compat+0x70/0x7f arch/x86/entry/entry_64_compat.S:139
RIP: 0023:0xf7fec849
Code: 85 d2 74 02 89 0a 5b 5d c3 8b 04 24 c3 8b 14 24 c3 8b 3c 24 c3 90 90
90 90 90 90 90 90 90 90 90 90 51 52 55 89 e5 0f 34 cd 80 <5d> 5a 59 c3 90
90 90 90 eb 0d 90 90 90 90 90 90 90 90 90 90 90 90
RSP: 002b:00000000ffed5a8c EFLAGS: 00000246 ORIG_RAX: 0000000000000078
RAX: ffffffffffffffda RBX: 0000000000003ffc RCX: 0000000000000000
RDX: 00000000200005c0 RSI: 0000000000000000 RDI: 0000000000000000
RBP: 0000000000000012 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
------------[ cut here ]------------
DEBUG_LOCKS_WARN_ON(depth <= 0)
WARNING: CPU: 1 PID: 7744 at kernel/locking/lockdep.c:4052 __lock_release
kernel/locking/lockdep.c:4052 [inline]
WARNING: CPU: 1 PID: 7744 at kernel/locking/lockdep.c:4052
lock_release+0x667/0xa00 kernel/locking/lockdep.c:4321
Kernel panic - not syncing: panic_on_warn set ...
CPU: 1 PID: 7744 Comm: syz-executor007 Not tainted 5.1.0+ #4
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
Google 01/01/2011
Call Trace:
  __dump_stack lib/dump_stack.c:77 [inline]
  dump_stack+0x172/0x1f0 lib/dump_stack.c:113
  panic+0x2cb/0x65c kernel/panic.c:214
  __warn.cold+0x20/0x45 kernel/panic.c:566
  report_bug+0x263/0x2b0 lib/bug.c:186
  fixup_bug arch/x86/kernel/traps.c:179 [inline]
  fixup_bug arch/x86/kernel/traps.c:174 [inline]
  do_error_trap+0x11b/0x200 arch/x86/kernel/traps.c:272
  do_invalid_op+0x37/0x50 arch/x86/kernel/traps.c:291
  invalid_op+0x14/0x20 arch/x86/entry/entry_64.S:972
RIP: 0010:__lock_release kernel/locking/lockdep.c:4052 [inline]
RIP: 0010:lock_release+0x667/0xa00 kernel/locking/lockdep.c:4321
Code: 0f 85 a0 03 00 00 8b 35 77 66 08 08 85 f6 75 23 48 c7 c6 a0 55 6b 87
48 c7 c7 40 25 6b 87 4c 89 85 70 ff ff ff e8 b7 a9 eb ff <0f> 0b 4c 8b 85
70 ff ff ff 4c 89 ea 4c 89 e6 4c 89 c7 e8 52 63 ff
RSP: 0018:ffff888094117b48 EFLAGS: 00010086
RAX: 0000000000000000 RBX: 1ffff11012822f6f RCX: 0000000000000000
RDX: 0000000000000000 RSI: ffffffff815af236 RDI: ffffed1012822f5b
RBP: ffff888094117c00 R08: ffff888092bfc400 R09: fffffbfff113301d
R10: fffffbfff113301c R11: ffffffff889980e3 R12: ffffffff8a451df8
R13: ffffffff8142e71f R14: ffffffff8a44cc80 R15: ffff888094117bd8
  percpu_up_read.constprop.0+0xcb/0x110 include/linux/percpu-rwsem.h:92
  cgroup_threadgroup_change_end include/linux/cgroup-defs.h:712 [inline]
  copy_process.part.0+0x47ff/0x6710 kernel/fork.c:2222
  copy_process kernel/fork.c:1772 [inline]
  _do_fork+0x25d/0xfd0 kernel/fork.c:2338
  __do_compat_sys_x86_clone arch/x86/ia32/sys_ia32.c:240 [inline]
  __se_compat_sys_x86_clone arch/x86/ia32/sys_ia32.c:236 [inline]
  __ia32_compat_sys_x86_clone+0xbc/0x140 arch/x86/ia32/sys_ia32.c:236
  do_syscall_32_irqs_on arch/x86/entry/common.c:334 [inline]
  do_fast_syscall_32+0x281/0xd54 arch/x86/entry/common.c:405
  entry_SYSENTER_compat+0x70/0x7f arch/x86/entry/entry_64_compat.S:139
RIP: 0023:0xf7fec849
Code: 85 d2 74 02 89 0a 5b 5d c3 8b 04 24 c3 8b 14 24 c3 8b 3c 24 c3 90 90
90 90 90 90 90 90 90 90 90 90 51 52 55 89 e5 0f 34 cd 80 <5d> 5a 59 c3 90
90 90 90 eb 0d 90 90 90 90 90 90 90 90 90 90 90 90
RSP: 002b:00000000ffed5a8c EFLAGS: 00000246 ORIG_RAX: 0000000000000078
RAX: ffffffffffffffda RBX: 0000000000003ffc RCX: 0000000000000000
RDX: 00000000200005c0 RSI: 0000000000000000 RDI: 0000000000000000
RBP: 0000000000000012 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
Kernel Offset: disabled
Rebooting in 86400 seconds..

Reported-and-tested-by: syzbot+3286e58549edc479faae@syzkaller.appspotmail.com
Fixes: b3e583825266 ("clone: add CLONE_PIDFD")
Signed-off-by: Christian Brauner <christian@brauner.io>
5 years agosamples: add .gitignore for pidfd-metadata
Christian Brauner [Wed, 8 May 2019 11:02:30 +0000 (13:02 +0200)]
samples: add .gitignore for pidfd-metadata

Ignore the pidfd-metadata binary so it doesn't show up in unwanted
scenarios.

Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Christian Brauner <christian@brauner.io>
5 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Fri, 10 May 2019 02:35:41 +0000 (19:35 -0700)]
Merge branch 'fixes' of git://git./linux/kernel/git/viro/vfs

Pull vfs mount fix from Al Viro:
 "Fix for umount -l/mount --move race caught by syzbot yesterday..."

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  do_move_mount(): fix an unsafe use of is_anon_ns()

5 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Fri, 10 May 2019 00:00:51 +0000 (17:00 -0700)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:
 "Several bug fixes, many are quick merge-window regression cures:

   - When NLM_F_EXCL is not set, allow same fib rule insertion. From
     Hangbin Liu.

   - Several cures in sja1105 DSA driver (while loop exit condition fix,
     return of negative u8, etc.) from Vladimir Oltean.

   - Handle tx/rx delays in realtek PHY driver properly, from Serge
     Semin.

   - Double free in cls_matchall, from Pieter Jansen van Vuuren.

   - Disable SIOCSHWTSTAMP in macvlan/vlan containers, from Hangbin Liu.

   - Endainness fixes in aqc111, from Oliver Neukum.

   - Handle errors in packet_init properly, from Haibing Yue.

   - Various W=1 warning fixes in kTLS, from Jakub Kicinski"

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (34 commits)
  nfp: add missing kdoc
  net/tls: handle errors from padding_length()
  net/tls: remove set but not used variables
  docs/btf: fix the missing section marks
  nfp: bpf: fix static check error through tightening shift amount adjustment
  selftests: bpf: initialize bpf_object pointers where needed
  packet: Fix error path in packet_init
  net/tcp: use deferred jump label for TCP acked data hook
  net: aquantia: fix undefined devm_hwmon_device_register_with_info reference
  aqc111: fix double endianness swap on BE
  aqc111: fix writing to the phy on BE
  aqc111: fix endianness issue in aqc111_change_mtu
  vlan: disable SIOCSHWTSTAMP in container
  macvlan: disable SIOCSHWTSTAMP in container
  tipc: fix hanging clients using poll with EPOLLOUT flag
  tuntap: synchronize through tfiles array instead of tun->numqueues
  tuntap: fix dividing by zero in ebpf queue selection
  dwmac4_prog_mtl_tx_algorithms() missing write operation
  ptp_qoriq: fix NULL access if ptp dt node missing
  net/sched: avoid double free on matchall reoffload
  ...

5 years agonfp: add missing kdoc
Jakub Kicinski [Thu, 9 May 2019 23:19:34 +0000 (16:19 -0700)]
nfp: add missing kdoc

Add missing kdoc for app member.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: Dirk van der Merwe <dirk.vandermerwe@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'tls-warnings'
David S. Miller [Thu, 9 May 2019 23:37:40 +0000 (16:37 -0700)]
Merge branch 'tls-warnings'

Jakub Kicinski says:

====================
net/tls: fix W=1 build warnings

This small series cleans up two outstanding W=1 build
warnings in tls code.  Both are set but not used variables.
The first case looks fairly straightforward.  In the second
I think it's better to propagate the error code, even if
not doing some does not lead to a crash with current code.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>