linux-toradex.git/mm/mprotect.c, branch v4.9.107

mm, mprotect: flush TLB if potentially racing with a parallel reclaim leaving stale TLB entries

2017-08-11T15:49:29+00:00

commit 3ea277194daaeaa84ce75180ec7c7a2075027a68 upstream.

Nadav Amit identified a theoritical race between page reclaim and
mprotect due to TLB flushes being batched outside of the PTL being held.

He described the race as follows:

        CPU0                            CPU1
        ----                            ----
                                        user accesses memory using RW PTE
                                        [PTE now cached in TLB]
        try_to_unmap_one()
        ==> ptep_get_and_clear()
        ==> set_tlb_ubc_flush_pending()
                                        mprotect(addr, PROT_READ)
                                        ==> change_pte_range()
                                        ==> [ PTE non-present - no flush ]

                                        user writes using cached RW PTE
        ...

        try_to_unmap_flush()

The same type of race exists for reads when protecting for PROT_NONE and
also exists for operations that can leave an old TLB entry behind such
as munmap, mremap and madvise.

For some operations like mprotect, it's not necessarily a data integrity
issue but it is a correctness issue as there is a window where an
mprotect that limits access still allows access.  For munmap, it's
potentially a data integrity issue although the race is massive as an
munmap, mmap and return to userspace must all complete between the
window when reclaim drops the PTL and flushes the TLB.  However, it's
theoritically possible so handle this issue by flushing the mm if
reclaim is potentially currently batching TLB flushes.

Other instances where a flush is required for a present pte should be ok
as either the page lock is held preventing parallel reclaim or a page
reference count is elevated preventing a parallel free leading to
corruption.  In the case of page_mkclean there isn't an obvious path
that userspace could take advantage of without using the operations that
are guarded by this patch.  Other users such as gup as a race with
reclaim looks just at PTEs.  huge page variants should be ok as they
don't race with reclaim.  mincore only looks at PTEs.  userfault also
should be ok as if a parallel reclaim takes place, it will either fault
the page back in or read some of the data before the flush occurs
triggering a fault.

Note that a variant of this patch was acked by Andy Lutomirski but this
was for the x86 parts on top of his PCID work which didn't make the 4.13
merge window as expected.  His ack is dropped from this version and
there will be a follow-on patch on top of PCID that will include his
ack.

[akpm@linux-foundation.org: tweak comments]
[akpm@linux-foundation.org: fix spello]
Link: http://lkml.kernel.org/r/20170717155523.emckq2esjro6hf3z@suse.de
Reported-by: Nadav Amit 
Signed-off-by: Mel Gorman 
Cc: Andy Lutomirski 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

mm/numa: Remove duplicated include from mprotect.c

2016-10-19T15:28:48+00:00

Signed-off-by: Wei Yongjun 
Cc: Dave Hansen 
Cc: linux-mm@kvack.org
Cc: Andrew Morton 
Cc: Mel Gorman 
Cc: "Kirill A. Shutemov" 
Link: http://lkml.kernel.org/r/1476719259-6214-1-git-send-email-weiyj.lk@gmail.com
Signed-off-by: Thomas Gleixner

Merge branch 'mm-pkeys-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

2016-10-10T18:01:51+00:00

Pull protection keys syscall interface from Thomas Gleixner:
 "This is the final step of Protection Keys support which adds the
  syscalls so user space can actually allocate keys and protect memory
  areas with them. Details and usage examples can be found in the
  documentation.

  The mm side of this has been acked by Mel"

* 'mm-pkeys-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/pkeys: Update documentation
  x86/mm/pkeys: Do not skip PKRU register if debug registers are not used
  x86/pkeys: Fix pkeys build breakage for some non-x86 arches
  x86/pkeys: Add self-tests
  x86/pkeys: Allow configuration of init_pkru
  x86/pkeys: Default to a restrictive init PKRU
  pkeys: Add details of system call use to Documentation/
  generic syscalls: Wire up memory protection keys syscalls
  x86: Wire up protection keys system calls
  x86/pkeys: Allocation/free syscalls
  x86/pkeys: Make mprotect_key() mask off additional vm_flags
  mm: Implement new pkey_mprotect() system call
  x86/pkeys: Add fault handling for PF_PK page fault bit

mm: vma_merge: fix vm_page_prot SMP race condition against rmap_walk

2016-10-08T01:46:29+00:00

The rmap_walk can access vm_page_prot (and potentially vm_flags in the
pte/pmd manipulations).  So it's not safe to wait the caller to update
the vm_page_prot/vm_flags after vma_merge returned potentially removing
the "next" vma and extending the "current" vma over the
next->vm_start,vm_end range, but still with the "current" vma
vm_page_prot, after releasing the rmap locks.

The vm_page_prot/vm_flags must be transferred from the "next" vma to the
current vma while vma_merge still holds the rmap locks.

The side effect of this race condition is pte corruption during migrate
as remove_migration_ptes when run on a address of the "next" vma that
got removed, used the vm_page_prot of the current vma.

  migrate   	      	        mprotect
  ------------			-------------
  migrating in "next" vma
				vma_merge() # removes "next" vma and
			        	    # extends "current" vma
					    # current vma is not with
					    # vm_page_prot updated
  remove_migration_ptes
  read vm_page_prot of current "vma"
  establish pte with wrong permissions
				vm_set_page_prot(vma) # too late!
				change_protection in the old vma range
				only, next range is not updated

This caused segmentation faults and potentially memory corruption in
heavy mprotect loads with some light page migration caused by compaction
in the background.

Hugh Dickins pointed out the comment about the Odd case 8 in vma_merge
which confirms the case 8 is only buggy one where the race can trigger,
in all other vma_merge cases the above cannot happen.

This fix removes the oddness factor from case 8 and it converts it from:

      AAAA
  PPPPNNNNXXXX -> PPPPNNNNNNNN

to:

      AAAA
  PPPPNNNNXXXX -> PPPPXXXXXXXX

XXXX has the right vma properties for the whole merged vma returned by
vma_adjust, so it solves the problem fully.  It has the added benefits
that the callers could stop updating vma properties when vma_merge
succeeds however the callers are not updated by this patch (there are
bits like VM_SOFTDIRTY that still need special care for the whole range,
as the vma merging ignores them, but as long as they're not processed by
rmap walks and instead they're accessed with the mmap_sem at least for
reading, they are fine not to be updated within vma_adjust before
releasing the rmap_locks).

Link: http://lkml.kernel.org/r/1474309513-20313-1-git-send-email-aarcange@redhat.com
Signed-off-by: Andrea Arcangeli 
Reported-by: Aditya Mandaleeka 
Cc: Rik van Riel 
Cc: Hugh Dickins 
Cc: Mel Gorman 
Cc: Jan Vorlicek 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

mm: vm_page_prot: update with WRITE_ONCE/READ_ONCE

2016-10-08T01:46:29+00:00

vma->vm_page_prot is read lockless from the rmap_walk, it may be updated
concurrently and this prevents the risk of reading intermediate values.

Link: http://lkml.kernel.org/r/1474660305-19222-1-git-send-email-aarcange@redhat.com
Signed-off-by: Andrea Arcangeli 
Cc: Rik van Riel 
Cc: Hugh Dickins 
Cc: Mel Gorman 
Cc: Jan Vorlicek 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

x86/pkeys: Allocation/free syscalls

2016-09-09T11:02:27+00:00

This patch adds two new system calls:

	int pkey_alloc(unsigned long flags, unsigned long init_access_rights)
	int pkey_free(int pkey);

These implement an "allocator" for the protection keys
themselves, which can be thought of as analogous to the allocator
that the kernel has for file descriptors.  The kernel tracks
which numbers are in use, and only allows operations on keys that
are valid.  A key which was not obtained by pkey_alloc() may not,
for instance, be passed to pkey_mprotect().

These system calls are also very important given the kernel's use
of pkeys to implement execute-only support.  These help ensure
that userspace can never assume that it has control of a key
unless it first asks the kernel.  The kernel does not promise to
preserve PKRU (right register) contents except for allocated
pkeys.

The 'init_access_rights' argument to pkey_alloc() specifies the
rights that will be established for the returned pkey.  For
instance:

	pkey = pkey_alloc(flags, PKEY_DENY_WRITE);

will allocate 'pkey', but also sets the bits in PKRU[1] such that
writing to 'pkey' is already denied.

The kernel does not prevent pkey_free() from successfully freeing
in-use pkeys (those still assigned to a memory range by
pkey_mprotect()).  It would be expensive to implement the checks
for this, so we instead say, "Just don't do it" since sane
software will never do it anyway.

Any piece of userspace calling pkey_alloc() needs to be prepared
for it to fail.  Why?  pkey_alloc() returns the same error code
(ENOSPC) when there are no pkeys and when pkeys are unsupported.
They can be unsupported for a whole host of reasons, so apps must
be prepared for this.  Also, libraries or LD_PRELOADs might steal
keys before an application gets access to them.

This allocation mechanism could be implemented in userspace.
Even if we did it in userspace, we would still need additional
user/kernel interfaces to tell userspace which keys are being
used by the kernel internally (such as for execute-only
mappings).  Having the kernel provide this facility completely
removes the need for these additional interfaces, or having an
implementation of this in userspace at all.

Note that we have to make changes to all of the architectures
that do not use mman-common.h because we use the new
PKEY_DENY_ACCESS/WRITE macros in arch-independent code.

1. PKRU is the Protection Key Rights User register.  It is a
   usermode-accessible register that controls whether writes
   and/or access to each individual pkey is allowed or denied.

Signed-off-by: Dave Hansen 
Acked-by: Mel Gorman 
Cc: linux-arch@vger.kernel.org
Cc: Dave Hansen 
Cc: arnd@arndb.de
Cc: linux-api@vger.kernel.org
Cc: linux-mm@kvack.org
Cc: luto@kernel.org
Cc: akpm@linux-foundation.org
Cc: torvalds@linux-foundation.org
Link: http://lkml.kernel.org/r/20160729163015.444FE75F@viggo.jf.intel.com
Signed-off-by: Thomas Gleixner

x86/pkeys: Make mprotect_key() mask off additional vm_flags

2016-09-09T11:02:26+00:00

Today, mprotect() takes 4 bits of data: PROT_READ/WRITE/EXEC/NONE.
Three of those bits: READ/WRITE/EXEC get translated directly in to
vma->vm_flags by calc_vm_prot_bits().  If a bit is unset in
mprotect()'s 'prot' argument then it must be cleared in vma->vm_flags
during the mprotect() call.

We do this clearing today by first calculating the VMA flags we
want set, then clearing the ones we do not want to inherit from
the original VMA:

	vm_flags = calc_vm_prot_bits(prot, key);
	...
	newflags = vm_flags;
	newflags |= (vma->vm_flags & ~(VM_READ | VM_WRITE | VM_EXEC));

However, we *also* want to mask off the original VMA's vm_flags in
which we store the protection key.

To do that, this patch adds a new macro:

	ARCH_VM_PKEY_FLAGS

which allows the architecture to specify additional bits that it would
like cleared.  We use that to ensure that the VM_PKEY_BIT* bits get
cleared.

Signed-off-by: Dave Hansen 
Acked-by: Mel Gorman 
Reviewed-by: Thomas Gleixner 
Cc: linux-arch@vger.kernel.org
Cc: Dave Hansen 
Cc: arnd@arndb.de
Cc: linux-api@vger.kernel.org
Cc: linux-mm@kvack.org
Cc: luto@kernel.org
Cc: akpm@linux-foundation.org
Cc: torvalds@linux-foundation.org
Link: http://lkml.kernel.org/r/20160729163013.E48D6981@viggo.jf.intel.com
Signed-off-by: Thomas Gleixner

mm: Implement new pkey_mprotect() system call

2016-09-09T11:02:26+00:00

pkey_mprotect() is just like mprotect, except it also takes a
protection key as an argument.  On systems that do not support
protection keys, it still works, but requires that key=0.
Otherwise it does exactly what mprotect does.

I expect it to get used like this, if you want to guarantee that
any mapping you create can *never* be accessed without the right
protection keys set up.

	int real_prot = PROT_READ|PROT_WRITE;
	pkey = pkey_alloc(0, PKEY_DENY_ACCESS);
	ptr = mmap(NULL, PAGE_SIZE, PROT_NONE, MAP_ANONYMOUS|MAP_PRIVATE, -1, 0);
	ret = pkey_mprotect(ptr, PAGE_SIZE, real_prot, pkey);

This way, there is *no* window where the mapping is accessible
since it was always either PROT_NONE or had a protection key set
that denied all access.

We settled on 'unsigned long' for the type of the key here.  We
only need 4 bits on x86 today, but I figured that other
architectures might need some more space.

Semantically, we have a bit of a problem if we combine this
syscall with our previously-introduced execute-only support:
What do we do when we mix execute-only pkey use with
pkey_mprotect() use?  For instance:

	pkey_mprotect(ptr, PAGE_SIZE, PROT_WRITE, 6); // set pkey=6
	mprotect(ptr, PAGE_SIZE, PROT_EXEC);  // set pkey=X_ONLY_PKEY?
	mprotect(ptr, PAGE_SIZE, PROT_WRITE); // is pkey=6 again?

To solve that, we make the plain-mprotect()-initiated execute-only
support only apply to VMAs that have the default protection key (0)
set on them.

Proposed semantics:
1. protection key 0 is special and represents the default,
   "unassigned" protection key.  It is always allocated.
2. mprotect() never affects a mapping's pkey_mprotect()-assigned
   protection key. A protection key of 0 (even if set explicitly)
   represents an unassigned protection key.
   2a. mprotect(PROT_EXEC) on a mapping with an assigned protection
       key may or may not result in a mapping with execute-only
       properties.  pkey_mprotect() plus pkey_set() on all threads
       should be used to _guarantee_ execute-only semantics if this
       is not a strong enough semantic.
3. mprotect(PROT_EXEC) may result in an "execute-only" mapping. The
   kernel will internally attempt to allocate and dedicate a
   protection key for the purpose of execute-only mappings.  This
   may not be possible in cases where there are no free protection
   keys available.  It can also happen, of course, in situations
   where there is no hardware support for protection keys.

Signed-off-by: Dave Hansen 
Acked-by: Mel Gorman 
Cc: linux-arch@vger.kernel.org
Cc: Dave Hansen 
Cc: arnd@arndb.de
Cc: linux-api@vger.kernel.org
Cc: linux-mm@kvack.org
Cc: luto@kernel.org
Cc: akpm@linux-foundation.org
Cc: torvalds@linux-foundation.org
Link: http://lkml.kernel.org/r/20160729163012.3DDD36C4@viggo.jf.intel.com
Signed-off-by: Thomas Gleixner

mm: thp: check pmd_trans_unstable() after split_huge_pmd()

2016-07-26T23:19:19+00:00

split_huge_pmd() doesn't guarantee that the pmd is normal pmd pointing
to pte entries, which can be checked with pmd_trans_unstable().  Some
callers make this assertion and some do it differently and some not, so
let's do it in a unified manner.

Link: http://lkml.kernel.org/r/1464741400-12143-1-git-send-email-n-horiguchi@ah.jp.nec.com
Signed-off-by: Naoya Horiguchi 
Cc: "Kirill A. Shutemov" 
Cc: Hugh Dickins 
Cc: Mel Gorman 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

mm: make mmap_sem for write waits killable for mm syscalls

2016-05-24T00:04:14+00:00

This is a follow up work for oom_reaper [1].  As the async OOM killing
depends on oom_sem for read we would really appreciate if a holder for
write didn't stood in the way.  This patchset is changing many of
down_write calls to be killable to help those cases when the writer is
blocked and waiting for readers to release the lock and so help
__oom_reap_task to process the oom victim.

Most of the patches are really trivial because the lock is help from a
shallow syscall paths where we can return EINTR trivially and allow the
current task to die (note that EINTR will never get to the userspace as
the task has fatal signal pending).  Others seem to be easy as well as
the callers are already handling fatal errors and bail and return to
userspace which should be sufficient to handle the failure gracefully.
I am not familiar with all those code paths so a deeper review is really
appreciated.

As this work is touching more areas which are not directly connected I
have tried to keep the CC list as small as possible and people who I
believed would be familiar are CCed only to the specific patches (all
should have received the cover though).

This patchset is based on linux-next and it depends on
down_write_killable for rw_semaphores which got merged into tip
locking/rwsem branch and it is merged into this next tree.  I guess it
would be easiest to route these patches via mmotm because of the
dependency on the tip tree but if respective maintainers prefer other
way I have no objections.

I haven't covered all the mmap_write(mm->mmap_sem) instances here

  $ git grep "down_write(.*\)" next/master | wc -l
  98
  $ git grep "down_write(.*\)" | wc -l
  62

I have tried to cover those which should be relatively easy to review in
this series because this alone should be a nice improvement.  Other
places can be changed on top.

[0] http://lkml.kernel.org/r/1456752417-9626-1-git-send-email-mhocko@kernel.org
[1] http://lkml.kernel.org/r/1452094975-551-1-git-send-email-mhocko@kernel.org
[2] http://lkml.kernel.org/r/1456750705-7141-1-git-send-email-mhocko@kernel.org

This patch (of 18):

This is the first step in making mmap_sem write waiters killable.  It
focuses on the trivial ones which are taking the lock early after
entering the syscall and they are not changing state before.

Therefore it is very easy to change them to use down_write_killable and
immediately return with -EINTR.  This will allow the waiter to pass away
without blocking the mmap_sem which might be required to make a forward
progress.  E.g.  the oom reaper will need the lock for reading to
dismantle the OOM victim address space.

The only tricky function in this patch is vm_mmap_pgoff which has many
call sites via vm_mmap.  To reduce the risk keep vm_mmap with the
original non-killable semantic for now.

vm_munmap callers do not bother checking the return value so open code
it into the munmap syscall path for now for simplicity.

Signed-off-by: Michal Hocko 
Acked-by: Vlastimil Babka 
Cc: Mel Gorman 
Cc: "Kirill A. Shutemov" 
Cc: Konstantin Khlebnikov 
Cc: Hugh Dickins 
Cc: Andrea Arcangeli 
Cc: David Rientjes 
Cc: Dave Hansen 
Cc: Johannes Weiner 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds