This adds some basic interface tests. It makes use of the --dry-run
option to determine which variables have been set and to which values.
We also check for the various warnings generated.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
tests/Makefile | 10 +++-
tests/hugect
During testing of the new multi page size support in hugectl I found a bug
in the options handling which had been missed. So I have been putting together
a new test harness for host tools. This is an example of that and includes
support for interface testing of hugectl.
-apw
Andy Whitcroft (1
possible result in warnings as normal.
The sizes are not validated but passed verbatim to the library which will
then validate them.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugectl.c | 139 +++--
man/hugectl.8 | 21 +-
On Thu, Oct 30, 2008 at 11:33:11AM +, Mel Gorman wrote:
> On Wed, Oct 29, 2008 at 04:02:53PM +0000, Andy Whitcroft wrote:
> > Add support for multiple page sizes to hugectl. Now that the library
> > supports specification of the backing page size to use for the various
> &g
It seems we are passing the LIB32/LIB64 defines to the link phase and not
the compile phase for hugectl. This means that it will not set the library
path correctly.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
Makefile |4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
On Wed, Oct 29, 2008 at 04:43:02PM +, Eric B Munson wrote:
> On Wed, 29 Oct 2008, Andy Whitcroft wrote:
>
> > Add support for multiple page sizes to hugectl. Now that the library
> > supports specification of the backing page size to use for the various
> > mappin
possible result in warnings as normal.
The sizes are not validated but passed verbatim to the library which will
then validate them.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugectl.c | 108 -
1 files changed, 71 inse
On Tue, Oct 28, 2008 at 09:44:10AM -0500, Adam Litke wrote:
> On Tue, 2008-10-28 at 13:54 +0000, Andy Whitcroft wrote:
> > When testing the library we sometimes want to influence data which becomes
> > cached within the library. To allow this we need to reset that cached data
>
e cache in the library through a new test probe interface.
Comments?
-apw
Andy Whitcroft (3):
privutils: cache and expose the kernel default page size
add a new export type for main library probe points for tests
hugetlbfs default page size is cached so expose reset for testing
h
kernel_default_hugepage_size_reset() as a test probe,
and use this within the gethugepagesizes test.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugeutils.c |4
libhugetlbfs_testprobes.h |4
tests/gethugepagesizes.c |4
3 files changed, 12 inse
routine kernel_default_hugepage_size() which looks this
up and caches the result.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Eric B Munson <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
---
hugeutils.c
.
These are external symbols but not exposed in the standard header, but
in libhugetlbfs_testprobe.h. They are also name mangled.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
libhugetlbfs_internal.h |1 +
libhugetlbfs_testprobes.h | 35 +++
On Fri, Oct 24, 2008 at 02:01:58PM +0100, Mel Gorman wrote:
> On Thu, Oct 23, 2008 at 03:54:59PM +0100, Andy Whitcroft wrote:
> > gethugepagesizes has two modes: a counting mode which tells us exactly
> > how many entries we would need; and a lookup mode which fills in a
>
implementation of read_meminfo.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
tests/gethugepagesizes.c | 187 +-
1 files changed, 151 insertions(+), 36 deletions(-)
diff --git a/tests/gethugepagesizes.c b/tests/gethugepagesizes.c
index 0
should be noted that this is a corner case where the caller passes
a buffer but requests zero entries be filled in, this should not affect
any sensible use case.
-apw
Andy Whitcroft (2):
tests: gethugepagesizes -- switch meminfo override to intercepting
open
gethugepagesizes: fix selection
when
the element count is zero which causes it to trigger at the edge case where
the caller (pointlessly) passes in a buffer with an element count of zero.
Fix behaviour to match the documentation.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugeutils.c | 19 ++--
On Wed, Oct 22, 2008 at 08:55:13PM +, Adam Litke wrote:
> The libhugetlbfs test suite has various tests for which a failure (under
> certain circumstances) is expected and not indicative of an error. It would
> be desirable to specially annotate these cases to eliminate the need for the
> t
On Wed, Oct 22, 2008 at 10:08:45AM +0100, Eric B Munson wrote:
> This patch Adjusts the HOWTO section on sizing huge page pools to reference
> the hugeadm utlity.
>
> Also it changes the number of pages from 20 to 25 for most test cases and
> calls out PowerPC 64kb pages as a specific instance whe
times. The forth makes hugeadm behave more sensibly when run on
a kernel which does not support hugepages at all. The last is a proposed
cleanup to use a common cached value for the kernel default page size as
we look it up very often and it is not cheap.
Signed-off-by: Andy Whitcroft <[EM
where there
is no overcommit support we simply always update the maximum to match.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugeadm.c| 16 ++--
hugeutils.c | 30 ++
libhugetlbfs_privutils.h |3 +++
3
A lot of commands simply will go horribly wrong if we do not have hugepage
support in the kernel. Check for and error out if we do not have hugepage
support. Note that we still allow usage checks and the like regardless.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hug
When moving from KB to B for the default page size we must always
use size_to_smaller_unit in case we overflow.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugeutils.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/hugeutils.c b/hugeutils.c
index a
routine kernel_default_hugepage_size() which looks this
up and caches the result.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugeutils.c | 32
libhugetlbfs_privutils.h |3 +++
shm.c|2 +-
3 files chang
If no operations were specified then ensure we emit the usage.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugeadm.c |7 ++-
1 files changed, 6 insertions(+), 1 deletions(-)
diff --git a/hugeadm.c b/hugeadm.c
index 8434eef..9c7ae1c 100644
--- a/hugeadm.c
+++ b/hug
On Tue, Oct 21, 2008 at 09:06:29AM +0100, Eric B Munson wrote:
> This patch Adjusts the HOWTO section on sizing huge page pools to reference
> the hugeadm utlity.
>
> Also it changes the number of pages from 20 to 25 for most test cases and
> calls out PowerPC 64kb pages as a specific instance whe
On Mon, Oct 20, 2008 at 10:14:54PM +, Adam Litke wrote:
> Andy Whitcroft had some good suggestions on my check_linkhuge_tests patch.
> - Comment the format of LINKHUGE_WORDSIZES
> - Use a case statement instead of an if/else
>
> Signed-off-by: Adam Litke <[EMAIL PROTECTED
Document which headers should be used when declaring functions static to
the library, and functions destined for the private utility library.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugetlbfs.h |7 +++
libhugetlbfs_internal.h
We have developed a far amount of whitespace damage over time. Do a one off
cleanup of all of the breakage now.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
HOWTO|2 +-
README |2 +-
elf64ppc.c |2 +-
elf
Following this email are two patches. The first attempts to document
the various header files as suggested by Adam Litke. The second cleans
up all the whitespace issues I could detect in the source.
-apw
Andy Whitcroft (2):
document which headers to use for library static and private utility
On Mon, Oct 20, 2008 at 11:50:39AM +0100, Mel Gorman wrote:
> On (17/10/08 17:14), Andy Whitcroft didst pronounce:
>
> > Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
> > ---
> > Makefile |6 +++---
> > hugetlbfs.h |
On Mon, Oct 20, 2008 at 11:36:30AM +0100, Mel Gorman wrote:
> On (17/10/08 17:14), Andy Whitcroft didst pronounce:
> > Currently we have three types of function:
> >
> > file local -- marked static in the normal way,
> > library local -- external but prefixed with _
On Mon, Oct 20, 2008 at 11:47:10AM +0100, Mel Gorman wrote:
> On (17/10/08 17:14), Andy Whitcroft didst pronounce:
> > Add a new private utilities library consisting of various useful helpers
> > normally hidden within libhugetlbfs. Extend the library local marker idiom
> >
On Mon, Oct 20, 2008 at 12:13:38PM +0100, Mel Gorman wrote:
> On Fri, Oct 17, 2008 at 05:14:13PM +0100, Andy Whitcroft wrote:
> > Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
> > ---
> > hugetlbfs.h |1 -
> > kernel-features.c |4
On Mon, Oct 20, 2008 at 12:15:13PM +0100, Mel Gorman wrote:
> I think the subject is misleading. It says we are making it static, but
> it's really moving headers. If it's meant to read "move to private
> header", then it looks ok.
Yes that is what its mean to be.
-apw
--
On Fri, Oct 17, 2008 at 06:36:52PM +, Adam Litke wrote:
> In run_tests.sh, the function check_linkhuge_tests() checks if it is safe to
> run the legacy linkhuge tests by examining the default linker script. But
> when
> binaries are compiled with a different word size, they use a different li
On Fri, Oct 17, 2008 at 03:11:01PM -0500, Adam Litke wrote:
> On Fri, 2008-10-17 at 17:14 +0100, Andy Whitcroft wrote:
> > diff --git a/tests/Makefile b/tests/Makefile
> > index e4e1ce2..f03ec68 100644
> > --- a/tests/Makefile
> > +++ b/tests/Makefile
> > @@
Note that as this is directly referenced in assembly we cannot trivially
rename it to __lh_, nor would this make it easy to change back and forth.
Take this as an exception and put it directly in the version.lds.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
version.lds |1 +
1
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
version.lds |6 ++
1 files changed, 6 insertions(+), 0 deletions(-)
diff --git a/version.lds b/version.lds
index 0517ad3..1429230 100644
--- a/version.lds
+++ b/version.lds
@@ -15,3 +15,9 @@ HTLBFS_2.0 {
get_huge
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
elf32ppclinux.c |2 ++
elf64ppc.c |2 ++
libhugetlbfs_internal.h |3 +++
3 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/elf32ppclinux.c b/elf32ppclinux.c
index eb54411..24adaf1 100644
We have historically exported the default size versions of these functions.
It makes sense therefore to export their new multi-pagesize versions as
part of the V2.1 API.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
version.lds |2 ++
1 files changed, 2 insertions(+), 0 del
None of these flags or debugging flags need to be exported. Move them all
library static:
__hugetlbfs_debug
__hugetlbfs_hostname
__hugetlbfs_prefault
__hugetlbfs_verbose
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
libhugetlbfs_internal.h |4 +
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugetlbfs.h |1 -
kernel-features.c |4 ++--
libhugetlbfs_internal.h |3 +++
3 files changed, 5 insertions(+), 3 deletions(-)
diff --git a/hugetlbfs.h b/hugetlbfs.h
index 06c427f..fc23f9f 100644
--- a/huget
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
Makefile |6 +++---
hugetlbfs.h | 16
hugeutils.c |6 +-
init_privutils.c |2 ++
libhugetlbfs_internal.h | 12
libhugetlbfs_privutils.h
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugeutils.c |4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/hugeutils.c b/hugeutils.c
index 81a834a..e053691 100644
--- a/hugeutils.c
+++ b/hugeutils.c
@@ -122,7 +122,7 @@ struct hugetlb_pool_counter_
converts all current library local functions to this new idiom.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
alloc.c |2 +-
debug.c |2 +-
elflink.c |6 +++---
hugeadm.c | 14 +++---
huget
only routines exported. This makes it very hard for the two libraries to
interfere with each other when both are linked to the same binary, which is
particularly important when testing the library; we must test the real one.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
Ma
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
Makefile |2 +-
hugetlbfs.h |9 -
kernel-features.c|1 +
libhugetlbfs_privutils.h | 10 ++
4 files changed, 12 insertions(+), 10 deletions(-)
diff --git a/Makefile b/Ma
exported. The final two patches export the four routines we
have added to the offical API in this release.
-apw
Andy Whitcroft (14):
move to a new library local idiom
add a new private utilities library
stop exposing pool configuration in the main library
make hugetlb_counter_info
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugeutils.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/hugeutils.c b/hugeutils.c
index e053691..a319e89 100644
--- a/hugeutils.c
+++ b/hugeutils.c
@@ -51,7 +51,7 @@
static int hugepagesize_errno;
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugetlbfs.h |4
libhugetlbfs_internal.h |4
2 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/hugetlbfs.h b/hugetlbfs.h
index fc23f9f..6b64ee7 100644
--- a/hugetlbfs.h
+++ b/hugetlbfs.h
@@
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
version.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/version.c b/version.c
index 4070fe0..0ab886a 100644
--- a/version.c
+++ b/version.c
@@ -1,3 +1,3 @@
#include "version.h"
-const char libh
Add a new utility to list page sizes supported on the system.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
---
Makefile |8 +++-
pagesize.c | 143
2 files changed, 15
When calling gethugepagesizes in a 32 bit environment it is possible that
some page sizeds will not be representable in the result. Ensure we detect
and elide these.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
Acked-by: Adam Litke <[
Rename and export the pool counter update interface.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Adam Litke <[EMAIL PROTECTED]>
---
hugetlbfs.h |2 ++
hugeutils.c | 26 +-
2 files changed, 15 insertions(+), 13 deletions(-)
diff --git a/h
EMAIL PROTECTED]>
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugetlbfs.h |1 +
hugeutils.c | 65 +
tests/Makefile |3 +-
tests/gethugepagesizes.c | 226 ++
tests/run_tests.sh
, which have pages assigned to the pool and which are exposed
via a mount point.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
---
hugeadm.c | 40
1 files changed, 40 insertions(+), 0 deletions(-)
Sort the output of the pool configuration listing --pool-list, and now
that the default size will not be first mark it in a new Default column.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Adam Litke <[EMAIL PROTECTED]>
---
hugeadm.c
requested should memory not be available. In this case we must
recalculate the overcommit.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
---
hugeadm.c | 138 ++-
hugeutils.c
Utilise the new gethugepagesizes call to implement getpagesizes() call
returning all page sizes on the system, both base page sizes and all
hugepage sizes.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugetlbfs.h |1 +
hugeutils.c
we have all of the negative range available so it seems
logical to use a negative value to indicate error.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Adam Litke <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
---
hugeutils.c | 10 --
From: Adam Litke <[EMAIL PROTECTED]>
This patch fixes two intertwined issues:
When the SHM_HUGETLB flag is passed to shmget(), the system default huge page
size is used to back the shared memory segment. Unlike with mmap() there is
not yet a way to use an alternate huge page size for shared memo
Add a very basic framework for hugeadm, the administrator pool configuration
helper.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
Acked-by: Adam Litke <[EMAIL PROTECTED]>
---
Makefile |8 -
cream about
fundamentals. For instance we have just change the name to hugeadm.
Comments?
-apw
Adam Litke (4):
utils: Make pool counter manipulation code part of the library API
shm: Shared memory always uses the meminfo page size
Do not allow zero as an explicit page size
lib: Add get
Allow a utility control which source files it is made up of. This allows
us to reuse some of the utility functions from within libhugetlbfs directly
without linking against the library.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
Ac
: Adam Litke <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugeutils.c | 22 +++---
1 files changed, 11 insertions(+), 11 deletions(-)
diff --git a/hugeutils.c b/hugeutils.c
index 22ea4
Expose pool allocations for each of the available page sizes via the
--pool-list command. Exposes the miniumum and maximimum values indicating
the dedicated pages and the size of the overcommit for all supported page
sizes.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel
Pull out all the common filenames and prefixes to a common set of defines.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
Acked-by: Adam Litke <[EMAIL PROTECTED]>
---
hugeutils.c | 16 ++--
1 files changed, 10 insertio
fixed up by the next patch.
[EMAIL PROTECTED]: consolidate makefile updates in one place]
Signed-off-by: Adam Litke <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugetlbfs.h
Where utilities are going to use routines out of the library directly it
is helpful to share the debugging routines whilst maintaining the identity
of the message producer. Allow the utilities to override the name shown.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Adam
On Fri, Oct 10, 2008 at 02:10:21PM -0500, Adam Litke wrote:
> On Fri, 2008-10-10 at 16:58 +0100, Andy Whitcroft wrote:
> > diff --git a/libhugetlbfs_internal.h b/libhugetlbfs_internal.h
> > index 197fa05..2912c5e 100644
> > --- a/libhugetlbfs_internal.h
> > ++
On Fri, Oct 10, 2008 at 02:30:43PM -0500, Adam Litke wrote:
> On Fri, 2008-10-10 at 16:58 +0100, Andy Whitcroft wrote:
> > Expose possible page sizes. Expose the page sizes which are available
> > (have pages assigned to them) via --page-sizes. Expose all possible page
> >
Utilise the new gethugepagesizes call to implement getpagesizes() call
returning all page sizes on the system, both base page sizes and all
hugepage sizes.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugetlbfs.h |1 +
hugeutils.c
Add a new utility to list page sizes supported on the system.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
---
Makefile |8 +++-
pagesize.c | 143
2 files changed, 15
When calling gethugepagesizes in a 32 bit environment it is possible that
some page sizeds will not be representable in the result. Ensure we detect
and elide these.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
---
hugeutils.c |4 +
EMAIL PROTECTED]>
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugetlbfs.h |1 +
hugeutils.c | 65 +
tests/Makefile |3 +-
tests/gethugepagesizes.c | 226 ++
tests/run_tests.sh
in --pool-list. Adjustments are either absolute,
or relative with a + or - prefix. For example the following incantion
increases the 2MB page pool minimum by 10 pages.
--pool-minimum-adjust 2048kb:+10
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL
Rename and export the pool counter update interface.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugetlbfs.h |2 ++
hugeutils.c | 26 +-
2 files changed, 15 insertions(+), 13 deletions(-)
diff --git a/hugetlbfs.h b/hugetlbfs.h
index 2103515..f
Sort the output of the pool configuration listing --pool-list, and now
that the default size will not be first mark it in a new Default column.
[should this new column be Flags and us be putting 'D' in there for instance?]
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Expose possible page sizes. Expose the page sizes which are available
(have pages assigned to them) via --page-sizes. Expose all possible page
sizes (supported by the processor) via --page-sizes-all. Page sizes are
listed in numerical order.
Signed-off-by: Andy Whitcroft <[EMAIL PROTEC
: Adam Litke <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugeutils.c | 22 +++---
1 files changed, 11 insertions(+), 11 deletions(-)
diff --git a/hugeutils.c b/hugeutils.c
index 22ea4
Pull out all the common filenames and prefixes to a common set of defines.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
---
hugeutils.c | 16 ++--
1 files changed, 10 insertions(+), 6 deletions(-)
diff --git a/hugeutils.c b
fixed up by the next patch.
[EMAIL PROTECTED]: consolidate makefile updates in one place]
Signed-off-by: Adam Litke <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugetlbfs.h
Allow a utility control which source files it is made up of. This allows
us to reuse some of the utility functions from within libhugetlbfs directly
without linking against the library.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
-
we have all of the negative range available so it seems
logical to use a negative value to indicate error.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Adam Litke <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
---
hugeutils.c | 10 --
Expose pool allocations for each of the available page sizes via the
--pool-list command. Exposes the miniumum and maximimum values indicating
the dedicated pages and the size of the overcommit for all supported page
sizes.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel
From: Adam Litke <[EMAIL PROTECTED]>
This patch fixes two intertwined issues:
When the SHM_HUGETLB flag is passed to shmget(), the system default huge page
size is used to back the shared memory segment. Unlike with mmap() there is
not yet a way to use an alternate huge page size for shared memo
Where utilities are going to use routines out of the library directly it
is helpful to share the debugging routines whilst maintaining the identity
of the message producer. Allow the utilities to override the name shown.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Adam
Add a very basic framework for hugeadm, the administrator pool configuration
helper.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
---
Makefile |8 -
hugeadm.c | 95 ++
stance we have just change the name to hugeadm.
Comments?
-apw
Adam Litke (4):
utils: Make pool counter manipulation code part of the library API
shm: Shared memory always uses the meminfo page size
Do not allow zero as an explicit page size
lib: Add gethugepagesizes() API call
An
On Fri, Oct 10, 2008 at 02:23:14PM +0100, Mel Gorman wrote:
> On (09/10/08 20:20), Andy Whitcroft didst pronounce:
> > Utilise the new gethugepagesizes call to implement getpagesizes() call
> > returning all page sizes on the system, both base page sizes and all
> > hugepage s
On Fri, Oct 10, 2008 at 12:22:40PM +0100, Mel Gorman wrote:
> On (09/10/08 20:20), Andy Whitcroft didst pronounce:
> > Allow resize of specific pools via the --pool-minimum-adjust and
> > --pool-maximum-adjust commands. The first affects the number of static
> > pages in
On Fri, Oct 10, 2008 at 12:29:47PM +0100, Mel Gorman wrote:
[...]
> > + if (adjust_str[0] == '+' || adjust_str[0] == '-') {
> > +
> > + /*
> > +* XXX: reading HUGEPAGES_TOTAL gives you the total pages
> > +* in use not what you wrote.
> > +*/
> > +
On Fri, Oct 10, 2008 at 12:17:28PM +0100, Mel Gorman wrote:
> On (09/10/08 20:20), Andy Whitcroft didst pronounce:
> > Rename and export the pool counter update interface.
> >
> > Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
> > ---
> > hugetl
On Fri, Oct 10, 2008 at 12:09:07PM +0100, Mel Gorman wrote:
> On (09/10/08 20:20), Andy Whitcroft didst pronounce:
> > Expose pool allocations for each of the available page sizes via the
> > --pool-list command. Exposes the miniumum and maximimum values indicating
> > the de
On Fri, Oct 10, 2008 at 11:00:51AM +0100, Mel Gorman wrote:
> On (09/10/08 20:20), Andy Whitcroft didst pronounce:
> > From: Adam Litke <[EMAIL PROTECTED]>
> >
> > The need to set and get pool counters for multiple huge page sizes is about
> > to
> > ex
Ensure we report any unparsed options that getopts_long recognises but which
have no implementation.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugeedit.c |4
1 files changed, 4 insertions(+), 0 deletions(-)
diff --git a/hugeedit.c b/hugeedit.c
index 2785200..17b052c
On Thu, Oct 09, 2008 at 09:05:08AM -0500, Adam Litke wrote:
> On Thu, 2008-10-09 at 11:47 +0100, Andy Whitcroft wrote:
> > When we hit the end of the arguments we will get a valid -1 return
> > from getopts_long. We should suppress this to prevent confusion.
>
> Does this af
in --pool-list. Adjustments are either absolute,
or relative with a + or - prefix. For exmaple the following incantion
increases the 2MB page pool minimum by 10 pages.
--pool-minimum-adjust 2048kb:+10
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hpoolcfg.c
Expose possible page sizes. Expose the page sizes which are available
(have pages assigned to them) via --page-sizes. Expose all possible page
sizes (supported by the processor) via --page-sizes-all.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hpoolcfg.c
Rename and export the pool counter update interface.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugetlbfs.h |2 ++
hugeutils.c | 28 ++--
2 files changed, 16 insertions(+), 14 deletions(-)
diff --git a/hugetlbfs.h b/hugetlbfs.h
index 2103515..f
1 - 100 of 267 matches
Mail list logo
