[Xen-devel] [For Xen-4.10 RFC PATCH 0/3] Reduce unnecessary icache maintenance operations

[Punit Agrawal <punit.agrawal@xxxxxxx>]

Hi,

Although icache maintenance operations are required when changing page
table mappings, in certain situations they are being called far more
frequently than necessary.

As icache maintenance has a performance penalty, one that increases
with the number of cores in the system, this RFC attempts to improve
the situation by hoisting knowledge of required icache maintenance
operations to higher layers.

Patch 1 adds a parameter to flush_page_to_ram() to prevent performing
icache maintenance per page. Current calls to flush_page_to_ram() loop
over pages and performing a full icache flush for each page is
excessive.

Patch 2 hoists icache maintenance from flush_page_to_ram() to
p2m_cache_flush().

Patch 3 introduces a new MEMF_ flag to indicate to alloc_heap_pages()
that icache maintenance will be performed by the caller. The icache
maintenance operations are performed in populate_physmap(). As I
couldn't icache maintenance operations for x86, an empty helper is
introduced.

Is it reasonable to move cache maintenance operation to higher layers?
The alternative would be perform cache maintenance operations on
ranges larger than page size but the benefit (reduced icache
operations) might not be equivalent.

This is my first contribution to Xen, so please go gentle if I've
missed something obvious.

Thanks,
Punit

Punit Agrawal (3):
  Allow control of icache invalidations when calling flush_page_to_ram()
  arm: p2m: Prevent redundant icache flushes
  Prevent redundant icache flushes in populate_physmap()

 xen/arch/arm/mm.c              | 5 +++--
 xen/arch/arm/p2m.c             | 4 +++-
 xen/common/memory.c            | 6 ++++++
 xen/common/page_alloc.c        | 2 +-
 xen/include/asm-arm/page.h     | 2 +-
 xen/include/asm-x86/flushtlb.h | 2 +-
 xen/include/asm-x86/page.h     | 4 ++++
 xen/include/xen/mm.h           | 2 ++
 8 files changed, 21 insertions(+), 6 deletions(-)

-- 
2.11.0


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel