On Wed, Jul 23, 2008 at 07:34:27PM +, Adam Litke wrote:
> The recently added relinking method makes it possible to choose from multiple
> ways to remap segments at run-time via the HUGTLB_ELFMAP environment variable.
> If nothing is specified in that variable, then no remapping will occur.
> So
On Fri, Jul 18, 2008 at 06:39:58PM +, Adam Litke wrote:
> To properly support building and installing tools (hugectl, hugeedit, hugecfg,
> etc), some changes to the initial method of building hugectl are required.
> This patch creates wildcard rules that can be used for building either 32 or
>
On Thu, Jul 24, 2008 at 09:50:30AM +0100, Andy Whitcroft wrote:
> > + switch (ret) {
> > + case '?':
> > + case 'h':
> > + print_usage();
> > + exit(EXIT_SUCCESS);
> > +
Having made the opposite error in hugectl I think that this i
On Thu, Jul 24, 2008 at 10:42:11AM +1000, David Gibson wrote:
> On Wed, Jul 23, 2008 at 04:33:40PM +0100, Andy Whitcroft wrote:
> > Add control over remap via the --text, --data, --bss, and --disable
> > options. The first three request mapping of those segments, the last
> > disables all remap.
get_huge_pages() is an API for the allocation of hugepage-backed regions of
memory. In the event it returns NULL, the application is expected to use
mmap() or malloc() if base pages would be suitable. This is boiler-plate
code that is an unnecessary burden on the application developer. This patch
a
On Wed, Jul 23, 2008 at 04:51:06PM +0100, Mel Gorman wrote:
> On (23/07/08 16:33), Andy Whitcroft didst pronounce:
> > Add support for requesting a specific library set for preload. This adds
> > the --library option which takes an optional argument. Without an argument
> > it requests use of the
Following this email are three further cleanups for the basic hugectl
infrastructure, and its interface to the library debugging. These
have come out of the ongoing work on hugectl, but stand alone as
cleanups.
Following this email are three patches:
debug: expose default and maximum verbosity l
On Thu, Jul 24, 2008 at 11:37:17AM +0100, Mel Gorman wrote:
> get_huge_pages() is an API for the allocation of hugepage-backed regions of
> memory. In the event it returns NULL, the application is expected to use
> mmap() or malloc() if base pages would be suitable. This is boiler-plate
> code that
Simplify specification of per-option help text.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugectl.c |6 +-
1 files changed, 5 insertions(+), 1 deletions(-)
diff --git a/hugectl.c b/hugectl.c
index d9b7040..8ba794c 100644
--- a/hugectl.c
+++ b/hugectl.c
@@ -40,11 +40,15 @@ ext
Move to using the POSIX standard codes EXIT_FAILURE/EXIT_SUCCESS instead
of explicit numbers.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugectl.c | 10 +++---
1 files changed, 7 insertions(+), 3 deletions(-)
diff --git a/hugectl.c b/hugectl.c
index 10e8806..d9b7040 100644
--- a
On (24/07/08 15:34), Andy Whitcroft didst pronounce:
> On Thu, Jul 24, 2008 at 11:37:17AM +0100, Mel Gorman wrote:
> > get_huge_pages() is an API for the allocation of hugepage-backed regions of
> > memory. In the event it returns NULL, the application is expected to use
> > mmap() or malloc() if b
On Thu, Jul 24, 2008 at 04:27:07PM +0100, Mel Gorman wrote:
> On (24/07/08 15:34), Andy Whitcroft didst pronounce:
> > On Thu, Jul 24, 2008 at 11:37:17AM +0100, Mel Gorman wrote:
> > > get_huge_pages() is an API for the allocation of hugepage-backed regions
> > > of
> > > memory. In the event it r
On (24/07/08 12:26), Andy Whitcroft didst pronounce:
> On Wed, Jul 23, 2008 at 04:51:06PM +0100, Mel Gorman wrote:
> > On (23/07/08 16:33), Andy Whitcroft didst pronounce:
> > > Add support for requesting a specific library set for preload. This adds
> > > the --library option which takes an optio
Expose the default and maximum verbosity levels via with the debugging
level macros so that applications can use it to enable 'all' tracing.
Use these in the main library.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
---
debug.c |2 +-
On (24/07/08 16:35), Andy Whitcroft didst pronounce:
> On Thu, Jul 24, 2008 at 04:27:07PM +0100, Mel Gorman wrote:
> > On (24/07/08 15:34), Andy Whitcroft didst pronounce:
> > > On Thu, Jul 24, 2008 at 11:37:17AM +0100, Mel Gorman wrote:
> > > > get_huge_pages() is an API for the allocation of huge
To properly support building and installing tools (hugectl, hugeedit, hugecfg,
etc), some changes to the initial method of building hugectl are required.
This patch creates wildcard rules that can be used for building either 32 or 64
bit versions of all tools. Additionally, a native wordsize is se
get_huge_pages() is an API for the allocation of hugepage-backed regions of
memory. In the event it returns NULL, the application is expected to recover,
possibly by calling mmap() or malloc() to use base pages. This is boiler-plate
code that is an unnecessary burden on the application developer. T
On Thu, 24 Jul 2008, Andy Whitcroft wrote:
> Move to using the POSIX standard codes EXIT_FAILURE/EXIT_SUCCESS instead
> of explicit numbers.
>
> Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Eric B Munson <[EMAIL PROTECTED]>
signature.asc
Description: Digital signature
---
On Thu, Jul 24, 2008 at 03:45:01PM +, Adam Litke wrote:
> To properly support building and installing tools (hugectl, hugeedit, hugecfg,
> etc), some changes to the initial method of building hugectl are required.
> This patch creates wildcard rules that can be used for building either 32 or
>
On Thu, 2008-07-24 at 10:24 +0100, Andy Whitcroft wrote:
> I am a little confused by why we consider bit size at all for the 'host'
> applications. We seem to build one or other only for host applications
> even where the host supports both sizes. But the makefile has to specify
> which they are.
On Tue, 2008-07-22 at 18:42 +0100, Mel Gorman wrote:
> +int shmget(key_t key, size_t size, int shmflg)
> +{
> + static int (*real_shmget)(key_t key, size_t size, int shmflg) = NULL;
> + void *handle;
> + char *error;
> + int retval;
> + char *hugetlbshm_env;
> + size_t align
On Thu, Jul 24, 2008 at 03:45:08PM +, Adam Litke wrote:
> The recently added relinking method makes it possible to choose from multiple
> ways to remap segments at run-time via the HUGTLB_ELFMAP environment variable.
> If nothing is specified in that variable, then no remapping will occur.
> So
On (24/07/08 11:35), Adam Litke didst pronounce:
> On Tue, 2008-07-22 at 18:42 +0100, Mel Gorman wrote:
> > +int shmget(key_t key, size_t size, int shmflg)
> > +{
> > + static int (*real_shmget)(key_t key, size_t size, int shmflg) = NULL;
> > + void *handle;
> > + char *error;
> > + int ret
Add control over remap via the --text, --data, --bss, and --disable
options. The first three request mapping of those segments, the last
disables all remap. Where the combinations requested cannot be exactly
handled then the request is "widened" to get that segment remapped,
for example if you re
This is a further revision of the basic hugectl support incorporating
further feedback, and a number of other cleanups which came out of the work
to address them. This is based on my cleanup stack 'hugectl cleanups II
(V1)'. Things seem to be stabalising here.
Needing resolution:
1) are we happ
Allow the user to request the specific hugetlb configuration which will
be used for the selected options, exposed via the --dry-run option.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
---
hugectl.c | 18 +-
1 files changed, 17 insertions(+), 1 deletions(-)
diff --git a/hu
Handle the HUGETLB_VERBOSE and HUGETLB_DEBUG flags such that we will emit
messages at the same levels as defined for the library. Use the common
header to define the levels. Convert all existing errors to use the new
macros.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel Gorman
Add support for mapping the heap via morecore through the --heap option.
Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Mel Gorman <[EMAIL PROTECTED]>
---
hugectl.c |7 +++
1 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/hugectl.c b/hugectl.c
index 0387c5b..79e
Preload is only useful when linking against an non-relinked binary
and only then when we are only attempting to back the heap. No other
combination can usefully use preload. Using preload with a relinked
binary is safe as long as you are only requesting backing of the heap.
Therefore unconditiona
Add support for requesting a specific library set for preload. This adds
the --library option which takes an optional argument. Without an argument
it requests use of the specific libraries installed with the version
of hugectl in use. An argument is treated as a prefix to the library
directorie
On (24/07/08 17:44), Andy Whitcroft didst pronounce:
> Add control over remap via the --text, --data, --bss, and --disable
> options. The first three request mapping of those segments, the last
> disables all remap. Where the combinations requested cannot be exactly
> handled then the request is
On 24.07.2008 [15:45:01 +], Adam Litke wrote:
> To properly support building and installing tools (hugectl, hugeedit, hugecfg,
> etc), some changes to the initial method of building hugectl are required.
> This patch creates wildcard rules that can be used for building either 32 or
> 64
> bit
The recently added relinking method makes it possible to choose from multiple
ways to remap segments at run-time via the HUGTLB_ELFMAP environment variable.
If nothing is specified in that variable, then no remapping will occur.
Sometimes it is desirable to set a default remapping mode so that
HUGE
On 24.07.2008 [14:46:17 +0100], Andy Whitcroft wrote:
> Following this email are three further cleanups for the basic hugectl
> infrastructure, and its interface to the library debugging. These
> have come out of the ongoing work on hugectl, but stand alone as
> cleanups.
>
> Following this email
On 24.07.2008 [09:59:37 -0700], Nishanth Aravamudan wrote:
> On 24.07.2008 [15:45:01 +], Adam Litke wrote:
> > To properly support building and installing tools (hugectl, hugeedit,
> > hugecfg,
> > etc), some changes to the initial method of building hugectl are required.
> > This patch create
The recently added relinking method makes it possible to choose from multiple
ways to remap segments at run-time via the HUGTLB_ELFMAP environment variable.
If nothing is specified in that variable, then no remapping will occur.
Sometimes it is desirable to set a default remapping mode so that
HUGE
On Thu, 2008-07-24 at 17:44 +0100, Andy Whitcroft wrote:
> Add control over remap via the --text, --data, --bss, and --disable
> options. The first three request mapping of those segments, the last
> disables all remap. Where the combinations requested cannot be exactly
> handled then the request
On Thu, 2008-07-24 at 17:44 +0100, Andy Whitcroft wrote:
> Add support for mapping the heap via morecore through the --heap option.
>
> Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
> Acked-by: Mel Gorman <[EMAIL PROTECTED]>
Acked-by: Adam Litke <[EMAIL PROTECTED]>
> ---
> hugectl.c |7 +
On Thu, 2008-07-24 at 17:44 +0100, Andy Whitcroft wrote:
> Preload is only useful when linking against an non-relinked binary
> and only then when we are only attempting to back the heap. No other
> combination can usefully use preload. Using preload with a relinked
> binary is safe as long as yo
On Thu, Jul 24, 2008 at 06:12:02PM +, Adam Litke wrote:
> The recently added relinking method makes it possible to choose from multiple
> ways to remap segments at run-time via the HUGTLB_ELFMAP environment variable.
> If nothing is specified in that variable, then no remapping will occur.
> So
On Thu, 2008-07-24 at 17:44 +0100, Andy Whitcroft wrote:
> Allow the user to request the specific hugetlb configuration which will
> be used for the selected options, exposed via the --dry-run option.
>
> Signed-off-by: Andy Whitcroft <[EMAIL PROTECTED]>
Acked-by: Adam Litke <[EMAIL PROTECTED]>
>
On Thu, 2008-07-24 at 17:44 +0100, Andy Whitcroft wrote:
> Add support for requesting a specific library set for preload. This adds
> the --library option which takes an optional argument. Without an argument
> it requests use of the specific libraries installed with the version
> of hugectl in u
On Thu, 2008-07-24 at 17:44 +0100, Andy Whitcroft wrote:
> Handle the HUGETLB_VERBOSE and HUGETLB_DEBUG flags such that we will emit
> messages at the same levels as defined for the library. Use the common
> header to define the levels. Convert all existing errors to use the new
> macros.
>
> Si
On Thu, Jul 24, 2008 at 01:49:24PM -0500, Adam Litke wrote:
> On Thu, 2008-07-24 at 17:44 +0100, Andy Whitcroft wrote:
> > Add support for requesting a specific library set for preload. This adds
> > the --library option which takes an optional argument. Without an argument
> > it requests use of
(Note, how GLIBC_2.0 and GLIB_2.3 have to be overidden is the big change
in this version. It was necessary to pass all tests on PPC64)
There are applications that are not hugepage-aware but use shared memory such
as the postgres database. It is possible that the SHM_HUGETLB flag could
be added to
The following two patches enable shmget() to be overridden to add the
SHM_HUGETLB flag. The first patch scopes local functions specifically
so that shmget() can be unversioned. The second patch adds the version
of shmget() to override glibc with.
Changelog since V4
o Do not use versioned shmget.
To override shmget(), it is necessary for the symbol to be
unversioned. However, all unversioned symbols are given local scope to avoid
internal functions being called accidently. This patch marks the internal-only
export functions clearly with the prefix __lh_ and then versions them to
be only of
There are applications that are not hugepage-aware but use shared memory such
as the postgres database. It is possible that the SHM_HUGETLB flag could
be added to such an application via libhugetlbfs and overriding shmget()
to add the SHM_HUGETLB to the flags. This patch enables libhugetlbfs to ad
On 24.07.2008 [19:41:38 +0100], Andy Whitcroft wrote:
> On Thu, Jul 24, 2008 at 06:12:02PM +, Adam Litke wrote:
> > The recently added relinking method makes it possible to choose from
> > multiple
> > ways to remap segments at run-time via the HUGTLB_ELFMAP environment
> > variable.
> > If n
Hi all,
I am happy to announce the 2.0-pre1 release of libhugetlbfs. We've had a
flurry of development recently and I'm trying to balance regular
releases (which means regular distribution releases as well) with new
features. In that vein, I've decided to postpone the following patchsets
until 2.1
On 24.07.2008 [23:36:29 +0100], Mel Gorman wrote:
>
> There are applications that are not hugepage-aware but use shared memory such
> as the postgres database. It is possible that the SHM_HUGETLB flag could
> be added to such an application via libhugetlbfs and overriding shmget()
> to add the SHM
On 24.07.2008 [23:36:09 +0100], Mel Gorman wrote:
>
> To override shmget(), it is necessary for the symbol to be
> unversioned. However, all unversioned symbols are given local scope to
> avoid internal functions being called accidently. This patch marks the
> internal-only export functions clearl
On (24/07/08 17:13), Nishanth Aravamudan didst pronounce:
> On 24.07.2008 [23:36:09 +0100], Mel Gorman wrote:
> >
> > To override shmget(), it is necessary for the symbol to be
> > unversioned. However, all unversioned symbols are given local scope to
> > avoid internal functions being called acci
On 25.07.2008 [01:16:30 +0100], Mel Gorman wrote:
> On (24/07/08 17:13), Nishanth Aravamudan didst pronounce:
> > On 24.07.2008 [23:36:09 +0100], Mel Gorman wrote:
> > >
> > > To override shmget(), it is necessary for the symbol to be
> > > unversioned. However, all unversioned symbols are given l
On 24.07.2008 [21:48:19 +0100], Andy Whitcroft wrote:
> On Thu, Jul 24, 2008 at 01:49:24PM -0500, Adam Litke wrote:
> > On Thu, 2008-07-24 at 17:44 +0100, Andy Whitcroft wrote:
> > > Add support for requesting a specific library set for preload. This adds
> > > the --library option which takes an
On 24.07.2008 [17:44:56 +0100], Andy Whitcroft wrote:
> Preload is only useful when linking against an non-relinked binary
> and only then when we are only attempting to back the heap. No other
> combination can usefully use preload. Using preload with a relinked
> binary is safe as long as you a
1、代理注册外商投资企业、外商驻华商务机构、注册香港及海外公司、离岸公司、注册国内公司、
注册外地驻穗办事处、注册国际及香港书刊号；
2、香港公司周年申报、年度报税，公司股份转股、公司转手、增加或减少股东服务；
3、代办香港律师公证书（用于外商投资企业之用）；
4、代理广东省内及广州企业会计报税、记帐服务；
5、代客撰写各类投标书、商业计划书、项目建议书及可行性分析报告、标准文书、法律诉讼书。
公司：广州工商注册代理公司
地址：广州市环市东路中央商务区
电话:15918644701
E-mail:[EMAIL PROTECTED]
如此邮件对您造成不便，敬请原谅。
―――
57 matches
Mail list logo
