[PATCH 0/3] zsmalloc: remove x86 dependency

[Seth Jennings]

This patchset continues/adapts Minchan Kim's work
to remove the x86 dependency from zsmalloc.

However, instead of whitelisting archs with support for
local_tlb_flush_kernel_range() in the zsmalloc Kconfig,
this patchset allows zsmalloc to work with all archs
through the addition of a generic/portable page
mapping methods (i.e. memcpy) when the required tlb
flushing functionality is not supported by the arch.

The arch advertises support for local_tlb_flush_kernel_range()
by defining __HAVE_LOCAL_FLUSH_TLB_KERNEL_RANGE

The third patch in the set adds local_tlb_flush_kernel_range()
support to x86.  In my single-threaded tests using zcache,
using the pte/tlb mapping method was 40% faster than the generic
method. So while the third patch is optional, it is highly
recommended.

Alex Shi is working on a large x86 patchset that includes
functionality similar to the third patch, however, it seems
that this patchset is getting very little attention and
includes much more than is needed for zsmalloc's purposes.

https://lkml.org/lkml/2012/6/12/116

Future work:
 - Add __HAVE_LOCAL_FLUSH_TLB_KERNEL_RANGE definition to
   archs that already have local_tlb_flush_kernel_range()
 - Add mapping mode flags (RO, WO, RW) to zs_map_object()
   to avoid unnecessary copies in the generic case

Based on Greg's staging-next.

Seth Jennings (3):
  zram/zcache: swtich Kconfig dependency from X86 to ZSMALLOC
  zsmalloc: add generic path and remove x86 dependency
  x86: add local_tlb_flush_kernel_range()

 arch/x86/include/asm/tlbflush.h          |   21 +++++
 drivers/staging/zcache/Kconfig           |    5 +-
 drivers/staging/zram/Kconfig             |    5 +-
 drivers/staging/zsmalloc/Kconfig         |    4 -
 drivers/staging/zsmalloc/zsmalloc-main.c |  136 ++++++++++++++++++++++++------
 drivers/staging/zsmalloc/zsmalloc_int.h  |    5 +-
 6 files changed, 138 insertions(+), 38 deletions(-)

-- 
1.7.9.5

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/