Skip to content

x86, PAT: PAT patchset

Venkatesh Pallipadi lkml 1 version New

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 Latest10 patches 0 repliesNew

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/10x86, PAT: keep identity maps consistent with mmaps even when pat_disabledarch/x86/mm/pat.c+3 -100 replies
2/10x86, PAT: ioremap to follow same PAT restrictions as other PAT usersarch/x86/mm/ioremap.c+3 -140 replies
3/10x86, PAT: New i/f for driver to request memtype for IO regionsarch/x86/include/asm/pat.h arch/x86/mm/pat.c+54 -00 replies
4/10x86, PAT: Add PAT reserve free to io_mapping* APIsarch/x86/include/asm/iomap.h arch/x86/mm/iomap_32.c include/linux/io-mapping.h+43 -100 replies
5/10x86, PAT: Add rbtree to do quick lookup in memtype trackingarch/x86/mm/pat.c+86 -200 replies
6/10x86, PAT: Generalize the use of page flag PG_uncachedarch/ia64/Kconfig arch/x86/Kconfig include/linux/page-flags.h+10 -20 replies
7/10x86, PAT: Use page flags to track memtypes of RAM pagesarch/x86/include/asm/cacheflush.h arch/x86/mm/pat.c+102 -430 replies
8/10x86, PAT: Add lookup_memtype to get the current memtype of a paddrarch/x86/mm/pat.c+45 -00 replies
9/10x86, PAT: lookup the protection from memtype list on vm_insert_pfn()arch/x86/mm/pat.c+9 -150 replies
10/10x86, PAT: Sanity check remap_pfn_range for RAM regionarch/x86/mm/pat.c+21 -30 replies

Reviews trailers per patch

No Reviewed-by, Acked-by or Tested-by trailers for v1 yet.

Read the v1 thread

Cover letter [PATCH 00/10] x86, PAT: PAT patchset

Another PAT patchset. Apart from minor cleanups, there are 3 relatively big changes here.
* Adding memtype tracking to io_mapping_* APIs. Currently these APIs hardcoded the type to WC.
* An rbtree added to keep track of PAT entries. This will avoid worst case latency with PAT entry lookup.
* Tracking of RAM pages using 2 bits in page struct. This will give us the ability to lookup the type of any physical address.

This patchset covers almost all the items in our PAT todo list, except for:
* Make PAT APIs to take try MTRR setup internally when pat is disabled.
* Need a set_pages_array_wc interface.

These patches are targeted for .32 and hoping to get some extended test until the merge upstream. I have tested this with some of my test programs and also with regressions tests in Eric's intel-gpu-tools here
http://cgit.freedesktop.org/xorg/app/intel-gpu-tools/

Venkatesh Pallipadi (10):
x86, PAT: keep identity maps consistent with mmaps even when
pat_disabled
x86, PAT: ioremap to follow same PAT restrictions as other PAT users x86, PAT: New i/f for driver to request memtype for IO regions x86, PAT: Add PAT reserve free to io_mapping* APIs
x86, PAT: Add rbtree to do quick lookup in memtype tracking x86, PAT: Generalize the use of page flag PG_uncached x86, PAT: Use page flags to track memtypes of RAM pages x86, PAT: Add lookup_memtype to get the current memtype of a paddr x86, PAT: lookup the protection from memtype list on vm_insert_pfn() x86, PAT: Sanity check remap_pfn_range for RAM region

 arch/ia64/Kconfig                 |    4 +
 arch/x86/Kconfig                  |    4 +
 arch/x86/include/asm/cacheflush.h |   54 ++++++-
 arch/x86/include/asm/iomap.h      |    9 +-
 arch/x86/include/asm/pat.h        |    5 +
 arch/x86/mm/iomap_32.c            |   27 +++-
 arch/x86/mm/ioremap.c             |   17 +--
 arch/x86/mm/pat.c                 |  352 +++++++++++++++++++++++++++----------
 include/linux/io-mapping.h        |   17 ++-
 include/linux/page-flags.h        |    4 +-
 10 files changed, 376 insertions(+), 117 deletions(-)