Skip to content

B
Bluefield Linux

mellanox / Bluefield Linux
11 个月 前同步成功

通知 3
Star 0
Fork 0
B
Bluefield Linux

体验新版 GitCode，发现更多精彩内容 >>

切换分支/标签
  1. 07 3月, 2010 1 次提交
    • R
      mm: change anon_vma linking to fix multi-process server scalability issue · 5beb4930
      由 Rik van Riel 提交于 
      The old anon_vma code can lead to scalability issues with heavily forking
workloads.  Specifically, each anon_vma will be shared between the parent
process and all its child processes.

In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes.  However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.

This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock.  This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands.  Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.

This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA.  At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated.  The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.

This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
 This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.

The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations.  This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures.  This in
turn means error handling needs to be added to the calling functions.

A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock.  To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag.  This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.

Some test results:

Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.

With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time.  The anon_vma lock contention appears to be resolved.

[akpm@linux-foundation.org: cleanups]
Signed-off-by: NRik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>
      5beb4930
  3. 09 2月, 2010 1 次提交
    • T
      [IA64] Remove COMPAT_IA32 support · 32974ad4
      由 Tony Luck 提交于 
      This has been broken since May 2008 when Al Viro killed altroot support.
Since nobody has complained, it would appear that there are no users of
this code (A plausible theory since the main OSVs that support ia64 prefer
to use the IA32-EL software emulation).
Signed-off-by: NTony Luck <tony.luck@intel.com>
      32974ad4
  5. 07 1月, 2010 1 次提交
  7. 02 10月, 2009 1 次提交
    • T
      ia64: don't alias VMALLOC_END to vmalloc_end · 126b3fcd
      由 Tejun Heo 提交于 
      If CONFIG_VIRTUAL_MEM_MAP is enabled, ia64 defines macro VMALLOC_END
as unsigned long variable vmalloc_end which is adjusted to prepare
room for vmemmap.  This becomes probnlematic if a local variables
vmalloc_end is defined in some function (not very unlikely) and
VMALLOC_END is used in the function - the function thinks its
referencing the global VMALLOC_END value but would be referencing its
own local vmalloc_end variable.

There's no reason VMALLOC_END should be a macro.  Just define it as an
unsigned long variable if CONFIG_VIRTUAL_MEM_MAP is set to avoid nasty
surprises.
Signed-off-by: NTejun Heo <tj@kernel.org>
Acked-by: NTony Luck <tony.luck@intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: linux-ia64 <linux-ia64@vger.kernel.org>
Cc: Christoph Lameter <cl@linux-foundation.org>
      126b3fcd
  9. 23 9月, 2009 4 次提交
  11. 22 9月, 2009 1 次提交
  13. 18 6月, 2009 1 次提交
    • M
      [IA64] Convert ia64 to use int-ll64.h · e088a4ad
      由 Matthew Wilcox 提交于 
      It is generally agreed that it would be beneficial for u64 to be an
unsigned long long on all architectures.  ia64 (in common with several
other 64-bit architectures) currently uses unsigned long.  Migrating
piecemeal is too painful; this giant patch fixes all compilation warnings
and errors that come as a result of switching to use int-ll64.h.

Note that userspace will still see __u64 defined as unsigned long.  This
is important as it affects C++ name mangling.

[Updated by Tony Luck to change efi.h:efi_freemem_callback_t to use
 u64 for start/end rather than unsigned long]
Signed-off-by: NMatthew Wilcox <willy@linux.intel.com>
Signed-off-by: NTony Luck <tony.luck@intel.com>
      e088a4ad
  15. 02 4月, 2009 1 次提交
  17. 27 3月, 2009 2 次提交
  19. 07 1月, 2009 1 次提交
    • G
      mm: show node to memory section relationship with symlinks in sysfs · c04fc586
      由 Gary Hade 提交于 
      Show node to memory section relationship with symlinks in sysfs

Add /sys/devices/system/node/nodeX/memoryY symlinks for all
the memory sections located on nodeX.  For example:
/sys/devices/system/node/node1/memory135 -> ../../memory/memory135
indicates that memory section 135 resides on node1.

Also revises documentation to cover this change as well as updating
Documentation/ABI/testing/sysfs-devices-memory to include descriptions
of memory hotremove files 'phys_device', 'phys_index', and 'state'
that were previously not described there.

In addition to it always being a good policy to provide users with
the maximum possible amount of physical location information for
resources that can be hot-added and/or hot-removed, the following
are some (but likely not all) of the user benefits provided by
this change.
Immediate:
  - Provides information needed to determine the specific node
    on which a defective DIMM is located.  This will reduce system
    downtime when the node or defective DIMM is swapped out.
  - Prevents unintended onlining of a memory section that was
    previously offlined due to a defective DIMM.  This could happen
    during node hot-add when the user or node hot-add assist script
    onlines _all_ offlined sections due to user or script inability
    to identify the specific memory sections located on the hot-added
    node.  The consequences of reintroducing the defective memory
    could be ugly.
  - Provides information needed to vary the amount and distribution
    of memory on specific nodes for testing or debugging purposes.
Future:
  - Will provide information needed to identify the memory
    sections that need to be offlined prior to physical removal
    of a specific node.

Symlink creation during boot was tested on 2-node x86_64, 2-node
ppc64, and 2-node ia64 systems.  Symlink creation during physical
memory hot-add tested on a 2-node x86_64 system.
Signed-off-by: NGary Hade <garyhade@us.ibm.com>
Signed-off-by: NBadari Pulavarty <pbadari@us.ibm.com>
Acked-by: NIngo Molnar <mingo@elte.hu>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>
      c04fc586
  21. 20 10月, 2008 1 次提交
    • B
      mm: cleanup to make remove_memory() arch-neutral · 71088785
      由 Badari Pulavarty 提交于 
      There is nothing architecture specific about remove_memory().
remove_memory() function is common for all architectures which support
hotplug memory remove.  Instead of duplicating it in every architecture,
collapse them into arch neutral function.

[akpm@linux-foundation.org: fix the export]
Signed-off-by: NBadari Pulavarty <pbadari@us.ibm.com>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Gary Hade <garyhade@us.ibm.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>
      71088785
  23. 07 9月, 2008 1 次提交
  25. 16 5月, 2008 1 次提交
    • H
      [IA64] fix personality(PER_LINUX32) performance issue · 839052d2
      由 Huang, Xiaolan 提交于 
      The patch aims to fix a performance issue for the syscall
personality(PER_LINUX32).

On IA-64 box, the syscall personality (PER_LINUX32) has poor performance
because it failed to find the Linux/x86 execution domain. Then it tried
to load the kernel module however it failed always and it used the default
execution domain PER_LINUX instead. Requesting kernel modules is very
expensive. It caused the performance issue. (see the function
lookup_exec_domain in kernel/exec_domain.c).

To resolve the issue, execution domain Linux/x86 is always registered in
initialization time for IA-64 architecture.
Signed-off-by: NXiaolan Huang <xiaolan.huang@intel.com>
Signed-off-by: NTony Luck <tony.luck@intel.com>
      839052d2
  27. 28 4月, 2008 1 次提交
  29. 12 4月, 2008 1 次提交
    • Z
      [IA64] Fix NUMA configuration issue · 98075d24
      由 Zoltan Menyhart 提交于 
      There is a NUMA memory configuration issue in 2.6.24:

A 2-node machine of ours has got the following memory layout:

Node 0:	0 - 2 Gbytes
Node 0:	4 - 8 Gbytes
Node 1:	8 - 16 Gbytes
Node 0:	16 - 18 Gbytes

"efi_memmap_init()" merges the three last ranges into one.

"register_active_ranges()" is called as follows:

efi_memmap_walk(register_active_ranges, NULL);

i.e. once for the 4 - 18 Gbytes range. It picks up the node
number from the start address, and registers all the memory for
the node #0.

"register_active_ranges()" should be called as follows to
make sure there is no merged address range at its entry:

efi_memmap_walk(filter_memory, register_active_ranges);

"filter_memory()" is similar to "filter_rsvd_memory()",
but the reserved memory ranges are not filtered out.
Signed-off-by: NZoltan Menyhart <Zoltan.Menyhart@bull.net>
Signed-off-by: NTony Luck <tony.luck@intel.com>
      98075d24
  31. 10 4月, 2008 1 次提交
  33. 07 3月, 2008 1 次提交
  35. 30 10月, 2007 1 次提交
    • A
      [IA64] ia64/mm/init.c: fix section mismatches · 18b8befd
      由 Adrian Bunk 提交于 
      This patch fixes the following section mismatches:

<--  snip  -->

...
WARNING: vmlinux.o(.text+0x5b5c2): Section mismatch: reference to .init.text:memmap_init_zone (between 'memmap_init' and 'virtual_memmap_init')
WARNING: vmlinux.o(.text+0x5b842): Section mismatch: reference to .init.text:memmap_init_zone (between 'virtual_memmap_init' and 'ia64_mmu_init')
...

<--  snip  -->
Signed-off-by: NAdrian Bunk <bunk@kernel.org>
Signed-off-by: NTony Luck <tony.luck@intel.com>
      18b8befd
  37. 20 10月, 2007 1 次提交
  39. 17 10月, 2007 3 次提交
  41. 12 5月, 2007 1 次提交
  43. 08 5月, 2007 1 次提交
    • C
      Make page->private usable in compound pages · d85f3385
      由 Christoph Lameter 提交于 
      If we add a new flag so that we can distinguish between the first page and the
tail pages then we can avoid to use page->private in the first page.
page->private == page for the first page, so there is no real information in
there.

Freeing up page->private makes the use of compound pages more transparent.
They become more usable like real pages.  Right now we have to be careful f.e.
 if we are going beyond PAGE_SIZE allocations in the slab on i386 because we
can then no longer use the private field.  This is one of the issues that
cause us not to support debugging for page size slabs in SLAB.

Having page->private available for SLUB would allow more meta information in
the page struct.  I can probably avoid the 16 bit ints that I have in there
right now.

Also if page->private is available then a compound page may be equipped with
buffer heads.  This may free up the way for filesystems to support larger
blocks than page size.

We add PageTail as an alias of PageReclaim.  Compound pages cannot currently
be reclaimed.  Because of the alias one needs to check PageCompound first.

The RFC for the this approach was discussed at
http://marc.info/?t=117574302800001&r=1&w=2

[nacc@us.ibm.com: fix hugetlbfs]
Signed-off-by: NChristoph Lameter <clameter@sgi.com>
Signed-off-by: NNishanth Aravamudan <nacc@us.ibm.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>
      d85f3385
  45. 30 3月, 2007 1 次提交
    • K
      [IA64] bugfix stack layout upside-down · 83d2cd3d
      由 KAMEZAWA Hiroyuki 提交于 
      ia64 expects following vm layout:

== low memory
[register-stack grows up]
[memory-stack grows down]
== high memory

But the code assigns the base of the register stack at the
maximum stack size offset from the fixed address where the
stack *might* start.  Stack randomization will result in the
memory stack starting at a lower address than this, and if the
user has set a low stack limit with "ulimit -s", then you can
end up with the register stack above the memory stack (or if
you were very unlucky right on top of it!).

Fix: Calculate the base address for the register stack starting
from the actual address of the memory stack.
Signed-off-by: NKAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: NTony Luck <tony.luck@intel.com>
      83d2cd3d
  47. 21 3月, 2007 1 次提交
    • Z
      [IA64] min_low_pfn and max_low_pfn calculation fix · a3f5c338
      由 Zou Nan hai 提交于 
      We have seen bad_pte_print when testing crashdump on an SN machine in
recent 2.6.20 kernel.  There are tons of bad pte print (pfn < max_low_pfn)
reports when the crash kernel boots up, all those reported bad pages
are inside initmem range; That is because if the crash kernel code and
data happens to be at the beginning of the 1st node. build_node_maps in
discontig.c will bypass reserved regions with filter_rsvd_memory. Since
min_low_pfn is calculated in build_node_map, so in this case, min_low_pfn
will be greater than kernel code and data.

Because pages inside initmem are freed and reused later, we saw
pfn_valid check fail on those pages.

I think this theoretically happen on a normal kernel. When I check
min_low_pfn and max_low_pfn calculation in contig.c and discontig.c.
I found more issues than this.

1. min_low_pfn and max_low_pfn calculation is inconsistent between
contig.c and discontig.c,
min_low_pfn is calculated as the first page number of boot memmap in
contig.c (Why? Though this may work at the most of the time, I don't
think it is the right logic). It is calculated as the lowest physical
memory page number bypass reserved regions in discontig.c.
max_low_pfn is calculated include reserved regions in contig.c. It is
calculated exclude reserved regions in discontig.c.

2. If kernel code and data region is happen to be at the begin or the
end of physical memory, when min_low_pfn and max_low_pfn calculation is
bypassed kernel code and data, pages in initmem will report bad.

3. initrd is also in reserved regions, if it is at the begin or at the
end of physical memory, kernel will refuse to reuse the memory. Because
the virt_addr_valid check in free_initrd_mem.

So it is better to fix and clean up those issues.
Calculate min_low_pfn and max_low_pfn in a consistent way.
Signed-off-by: NZou Nan hai <nanhai.zou@intel.com>
Acked-by: NJay Lan <jlan@sgi.com>
Signed-off-by: NTony Luck <tony.luck@intel.com>
      a3f5c338
  49. 12 2月, 2007 2 次提交
  51. 07 2月, 2007 1 次提交
    • C
      [IA64] relax per-cpu TLB requirement to DTC · 00b65985
      由 Chen, Kenneth W 提交于 
      Instead of pinning per-cpu TLB into a DTR, use DTC.  This will free up
one TLB entry for application, or even kernel if access pattern to
per-cpu data area has high temporal locality.

Since per-cpu is mapped at the top of region 7 address, we just need to
add special case in alt_dtlb_miss.  The physical address of per-cpu data
is already conveniently stored in IA64_KR(PER_CPU_DATA).  Latency for
alt_dtlb_miss is not affected as we can hide all the latency.  It was
measured that alt_dtlb_miss handler has 23 cycles latency before and
after the patch.

The performance effect is massive for applications that put lots of tlb
pressure on CPU.  Workload environment like database online transaction
processing or application uses tera-byte of memory would benefit the most.
Measurement with industry standard database benchmark shown an upward
of 1.6% gain.  While smaller workloads like cpu, java also showing small
improvement.
Signed-off-by: NKen Chen <kenneth.w.chen@intel.com>
Signed-off-by: NTony Luck <tony.luck@intel.com>
      00b65985
  53. 06 2月, 2007 2 次提交
    • J
      [IA64] swiotlb bug fixes · cde14bbf
      由 Jan Beulich 提交于 
      This patch fixes
- marking I-cache clean of pages DMAed to now only done for IA64
- broken multiple inclusion in include/asm-x86_64/swiotlb.h
- missing call to mark_clean in swiotlb_sync_sg()
- a (perhaps only theoretical) issue in swiotlb_dma_supported() when
io_tlb_end is exactly at the end of memory
Signed-off-by: NJan Beulich <jbeulich@novell.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NTony Luck <tony.luck@intel.com>
      cde14bbf
    • B
      [IA64] register memory ranges in a consistent manner · 139b8304
      由 Bob Picco 提交于 
      While pursuing and unrelated issue with 64Mb granules I noticed a problem
related to inconsistent use of add_active_range.  There doesn't appear any
reason to me why FLATMEM versus DISCONTIG_MEM should register memory to
add_active_range with different code.  So I've changed the code into a
common implementation.

The other subtle issue fixed by this patch was calling add_active_range in
count_node_pages before granule aligning is performed.  We were lucky with
16MB granules but not so with 64MB granules.  count_node_pages has reserved
regions filtered out and as a consequence linked kernel text and data
aren't covered by calls to count_node_pages.  So linked kernel regions
wasn't reported to add_active_regions.  This resulted in free_initmem
causing numerous bad_page reports.  This won't occur with this patch
because now all known memory regions are reported by
register_active_ranges.
Acked-by: NMel Gorman <mel@csn.ul.ie>
Signed-off-by: NBob Picco <bob.picco@hp.com>
Acked-by: NSimon Horman <horms@verge.net.au>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NTony Luck <tony.luck@intel.com>
      139b8304
  55. 12 1月, 2007 1 次提交
  57. 13 12月, 2006 1 次提交
  59. 08 12月, 2006 1 次提交
  61. 27 9月, 2006 1 次提交
  63. 04 8月, 2006 1 次提交
    • B
      [IA64] fix show_mem for VIRTUAL_MEM_MAP+FLATMEM · e44e41d0
      由 Bob Picco 提交于 
      contig.c (FLATMEM) requires the same optimization as in discontig.c for show_mem
when VIRTUAL_MEM_MAP is in use. Otherwise FLATMEM has softlockup timeouts.
This was boot tested for memory configuration: SPARSEMEM,
DISCONTIG+VIRTUAL_MEM_MAP, FLATMEM, FLATMEM+VIRTUAL_MEM_MAP and
FLATMEM+VIRTUAL_MEM_MAP with largest memory gap less than LARGE_GAP by
using boot parameter "mem=".

This was boot tested and "echo m >/proc/sysrq-trigger" output evaluated for
: FLATMEM, FLATMEM+VIRTUAL_MEM_MAP, DISCONTIGMEM+VIRTUAL_MEM_MAP and
SPARSEMEM.
Signed-off-by: NBob Picco <bob.picco@hp.com>
Signed-off-by: NTony Luck <tony.luck@intel.com>
      e44e41d0