Memory hot-delete a memory range present at boot causes an
error message in __release_region(), such as:

 Trying to free nonexistent resource <0000000070000000-0000000077ffffff>

Hot-delete operation still continues since __release_region() is 
a void function, but the target memory range is not freed from
iomem_resource as the result.  This also leads a failure in a 
subsequent hot-add operation to the same memory range since the
address range is still in-use in iomem_resource.

This problem happens because the granularity of memory resource ranges
may be different between boot and hot-delete.  During bootup,
iomem_resource is set up from the boot descriptor table, such as EFI
Memory Table and e820.  Each resource entry usually covers the whole
contiguous memory range.  Hot-delete request, on the other hand, may
target to a particular range of memory resource, and its size can be
much smaller than the whole contiguous memory.  Since the existing
release interfaces like __release_region() require a requested region
to be exactly matched to a resource entry, they do not allow a partial
resource to be released.

This patchset introduces release_mem_region_adjustable() for memory
hot-delete operations, which allows releasing a partial memory range
and adjusts remaining resource accordingly.  This patchset makes no
changes to the existing interfaces since their restriction is still
valid for I/O resources.

---
Toshi Kani (3):
 resource: Add __adjust_resource() for internal use
 resource: Add release_mem_region_adjustable()
 mm: Change __remove_pages() to call release_mem_region_adjustable()

---
 include/linux/ioport.h |   2 +
 kernel/resource.c      | 122 +++++++++++++++++++++++++++++++++++++++++++------
 mm/memory_hotplug.c    |  11 ++++-
 3 files changed, 120 insertions(+), 15 deletions(-)
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Reply via email to