linux-toradex.git/kernel/memremap.c, branch v4.4.93

mm: fix devm_memremap_pages crash, use mem_hotplug_{begin, done}

2017-01-19T19:17:18+00:00

commit f931ab479dd24cf7a2c6e2df19778406892591fb upstream.

Both arch_add_memory() and arch_remove_memory() expect a single threaded
context.

For example, arch/x86/mm/init_64.c::kernel_physical_mapping_init() does
not hold any locks over this check and branch:

    if (pgd_val(*pgd)) {
    	pud = (pud_t *)pgd_page_vaddr(*pgd);
    	paddr_last = phys_pud_init(pud, __pa(vaddr),
    				   __pa(vaddr_end),
    				   page_size_mask);
    	continue;
    }

    pud = alloc_low_page();
    paddr_last = phys_pud_init(pud, __pa(vaddr), __pa(vaddr_end),
    			   page_size_mask);

The result is that two threads calling devm_memremap_pages()
simultaneously can end up colliding on pgd initialization.  This leads
to crash signatures like the following where the loser of the race
initializes the wrong pgd entry:

    BUG: unable to handle kernel paging request at ffff888ebfff0000
    IP: memcpy_erms+0x6/0x10
    PGD 2f8e8fc067 PUD 0 /* <---- Invalid PUD */
    Oops: 0000 [#1] SMP DEBUG_PAGEALLOC
    CPU: 54 PID: 3818 Comm: systemd-udevd Not tainted 4.6.7+ #13
    task: ffff882fac290040 ti: ffff882f887a4000 task.ti: ffff882f887a4000
    RIP: memcpy_erms+0x6/0x10
    [..]
    Call Trace:
      ? pmem_do_bvec+0x205/0x370 [nd_pmem]
      ? blk_queue_enter+0x3a/0x280
      pmem_rw_page+0x38/0x80 [nd_pmem]
      bdev_read_page+0x84/0xb0

Hold the standard memory hotplug mutex over calls to
arch_{add,remove}_memory().

Fixes: 41e94a851304 ("add devm_memremap_pages")
Link: http://lkml.kernel.org/r/148357647831.9498.12606007370121652979.stgit@dwillia2-desk3.amr.corp.intel.com
Signed-off-by: Dan Williams 
Cc: Christoph Hellwig 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

devm_memremap: Fix error value when memremap failed

2016-03-03T23:07:08+00:00

commit 93f834df9c2d4e362dfdc4b05daa0a4e18814836 upstream.

devm_memremap() returns an ERR_PTR() value in case of error.
However, it returns NULL when memremap() failed.  This causes
the caller, such as the pmem driver, to proceed and oops later.

Change devm_memremap() to return ERR_PTR(-ENXIO) when memremap()
failed.

Signed-off-by: Toshi Kani 
Cc: Andrew Morton 
Reviewed-by: Ross Zwisler 
Signed-off-by: Dan Williams 
Signed-off-by: Greg Kroah-Hartman

devm_memremap_release(): fix memremap'd addr handling

2016-02-25T20:01:23+00:00

commit 9273a8bbf58a15051e53a777389a502420ddc60e upstream.

The pmem driver calls devm_memremap() to map a persistent memory range.
When the pmem driver is unloaded, this memremap'd range is not released
so the kernel will leak a vma.

Fix devm_memremap_release() to handle a given memremap'd address
properly.

Signed-off-by: Toshi Kani 
Acked-by: Dan Williams 
Cc: Christoph Hellwig 
Cc: Ross Zwisler 
Cc: Matthew Wilcox 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

Merge tag 'libnvdimm-for-4.4' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm

2015-11-10T20:07:22+00:00

Pull libnvdimm updates from Dan Williams:
 "Outside of the new ACPI-NFIT hot-add support this pull request is more
  notable for what it does not contain, than what it does.  There were a
  handful of development topics this cycle, dax get_user_pages, dax
  fsync, and raw block dax, that need more more iteration and will wait
  for 4.5.

  The patches to make devm and the pmem driver NUMA aware have been in
  -next for several weeks.  The hot-add support has not, but is
  contained to the NFIT driver and is passing unit tests.  The coredump
  support is straightforward and was looked over by Jeff.  All of it has
  received a 0day build success notification across 107 configs.

  Summary:

   - Add support for the ACPI 6.0 NFIT hot add mechanism to process
     updates of the NFIT at runtime.

   - Teach the coredump implementation how to filter out DAX mappings.

   - Introduce NUMA hints for allocations made by the pmem driver, and
     as a side effect all devm allocations now hint their NUMA node by
     default"

* tag 'libnvdimm-for-4.4' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm:
  coredump: add DAX filtering for FDPIC ELF coredumps
  coredump: add DAX filtering for ELF coredumps
  acpi: nfit: Add support for hot-add
  nfit: in acpi_nfit_init, break on a 0-length table
  pmem, memremap: convert to numa aware allocations
  devm_memremap_pages: use numa_mem_id
  devm: make allocations numa aware by default
  devm_memremap: convert to return ERR_PTR
  devm_memunmap: use devres_release()
  pmem: kill memremap_pmem()
  x86, mm: quiet arch_add_memory()

memremap: fix highmem support

2015-10-26T20:55:56+00:00

Currently memremap checks if the range is "System RAM" and returns the
kernel linear address.  This is broken for highmem platforms where a
range may be "System RAM", but is not part of the kernel linear mapping.
Fallback to ioremap_cache() in these cases, to let the arch code attempt
to handle it.

Note that ARM ioremap will WARN when attempting to remap ram, and in
that case the caller needs to be fixed.  For this reason, existing
ioremap_cache() usages for ARM are already trained to avoid attempts to
remap ram.

The impact of this bug is low for now since the pmem driver is the only
user of memremap(), but this is important to fix before more conversions
to memremap arrive in 4.4.

Cc: Rafael J. Wysocki 
Reported-by: Ard Biesheuvel 
Acked-by: Ard Biesheuvel 
Signed-off-by: Dan Williams

pmem, memremap: convert to numa aware allocations

2015-10-09T21:00:33+00:00

Given that pmem ranges come with numa-locality hints, arrange for the
resulting driver objects to be obtained from node-local memory.

Reviewed-by: Tejun Heo 
Reviewed-by: Christoph Hellwig 
Signed-off-by: Dan Williams

devm_memremap_pages: use numa_mem_id

2015-10-09T21:00:33+00:00

Hint to closest numa node for the placement of newly allocated pages.
As that is where the device's other allocations will originate by
default when it does not specify a NUMA node.

Cc: Ross Zwisler 
Reviewed-by: Tejun Heo 
Reviewed-by: Christoph Hellwig 
Signed-off-by: Dan Williams

devm_memremap: convert to return ERR_PTR

2015-10-09T21:00:33+00:00

Make devm_memremap consistent with the error return scheme of
devm_memremap_pages to remove special casing in the pmem driver.

Cc: Christoph Hellwig 
Cc: Ross Zwisler 
Reviewed-by: Christoph Hellwig 
Signed-off-by: Dan Williams

devm_memunmap: use devres_release()

2015-10-09T21:00:33+00:00

Remove open coded call to memunmap.

Cc: Christoph Hellwig 
Cc: Ross Zwisler 
Reviewed-by: Christoph Hellwig 
Signed-off-by: Dan Williams

add devm_memremap_pages

2015-08-27T23:40:58+00:00

This behaves like devm_memremap except that it ensures we have page
structures available that can back the region.

Signed-off-by: Christoph Hellwig 
[djbw: catch attempts to remap RAM, drop flags]
Signed-off-by: Dan Williams