David Gibson wrote:
> On Wed, Feb 20, 2008 at 03:43:08PM -0600, Andrew Hastings wrote:
>> David Gibson wrote:
>>> On Thu, Feb 14, 2008 at 11:48:42AM -0600, Andrew Hastings wrote:
Disable heap shrinking by default unless HUGETLB_MORECORE_SHRINK=yes is
set in the environment.
If m
For new users of libhugetlbfs, the requirements of configuring
the hugepage pool and mounting hugetlbfs should be made explicit.
Signed-off-by: Nishanth Aravamudan <[EMAIL PROTECTED]>
diff --git a/HOWTO b/HOWTO
index a10461b..3998ef1 100644
--- a/HOWTO
+++ b/HOWTO
@@ -99,6 +99,42 @@ to use both g
Nishanth Aravamudan wrote:
> +To use libhugetlbfs features, as well as to run the testsuite, hugetlbfs
> +must be mounted:
> +
> + mkdir -p /mnt/hugetlbfs
> + mount -t hugetlbfs none /mnt/hugetlbfs
> +
> +If hugepages should be available to non-root users, the permissions on
> +the mountpo
On 22.02.2008 [18:35:53 +0530], Subrata Modak wrote:
> > We maintain a set of detailed functionality tests for hugepages in
> > libhugetlbfs. Any time you are looking for new tests to add to LTP,
> > feel free to check the source code.
>
> Adam/Nishant,
>
> I am not sure if i found the test case
On 22.02.2008 [14:20:13 -0600], Andrew Hastings wrote:
> Nish,
>
> Thanks for the review!
> Glad you suggested this! It turns out there's a dependency on the order in
> which the three libhugetlbfs constructors run, which in turns depends on
> the order the objects appear in the link line. I
On 27.02.2008 [13:29:24 -0800], Nishanth Aravamudan wrote:
> On 22.02.2008 [14:20:13 -0600], Andrew Hastings wrote:
> > Nish,
> >
> > Thanks for the review!
>
>
>
> > Glad you suggested this! It turns out there's a dependency on the order in
> > which the three libhugetlbfs constructors run, w
Nish,
Thanks! You beat me to it!
Nishanth Aravamudan wrote:
> --- /dev/null
> +++ b/init.c
> @@ -0,0 +1,8 @@
> +#include "libhugetlbfs_internal.h"
> +
> +static void __attribute__ ((constructor)) setup_libhugetlbfs(void)
> +{
> + setup_morecore();
> + setup_elflink();
> + setup_debug
Nish,
Nice! Just one comment:
Nishanth Aravamudan wrote:
> --- a/libhugetlbfs_internal.h
> +++ b/libhugetlbfs_internal.h
> @@ -29,6 +29,7 @@
> #define ALIGN(x, a) (((x) + (a) - 1) & ~((a) - 1))
>
> extern int __hugetlbfs_verbose;
> +extern int __debug;
This should be __hugetlbfs_debug to a
hugetlbfs_morecore() is currently calling mlock() presumably to
instantiate newly-mapped hugepages before returning to malloc().
However, this is ineffective for two reasons:
1. The return value of mlock is ignored. If there aren't enough
hugepages available, the process may be killed later,
morecore: dump /proc/pid/maps when the heap is forced to move and DEBUG
If HUGETLB_DEBUG is set in the environment, use a new helper function,
dump_proc_pid_maps() to indicate the address space layout. This is very
useful in finding out *why* the hugepage heap cannot be placed where
we'd like it.
I found that when using OpenMPI with libhugetlbfs there is a gigantic symbol in
the BSS. From readelf -a:
81: 120048b0 6caf2d0 OBJECT GLOBAL DEFAULT 26fields
This object is almost 109M in size! But because it is an OBJECT type and
GLOBAL, we copy it in the extracopy lo
To ease with debugging, output the address and name of extra symbols to
copy from within the BSS.
Signed-off-by: Nishanth Aravamudan <[EMAIL PROTECTED]>
diff --git a/elflink.c b/elflink.c
index aaec4a2..427f8c3 100644
--- a/elflink.c
+++ b/elflink.c
@@ -421,6 +421,10 @@ static inline int keep_sym
For debugging purposes, setting HUGETLB_DEBUG should be sufficient. It
should also imply maximum verbosity from the library, so spit out all
types of messages (DEBUG, WARNING or ERROR).
Signed-off-by: Nishanth Aravamudan <[EMAIL PROTECTED]>
diff --git a/libhugetlbfs_internal.h b/libhugetlbfs_inte
On 27.02.2008 [15:46:47 -0600], Andrew Hastings wrote:
> Nish,
>
> Thanks! You beat me to it!
>
> Nishanth Aravamudan wrote:
>> --- /dev/null
>> +++ b/init.c
>> @@ -0,0 +1,8 @@
>> +#include "libhugetlbfs_internal.h"
>> +
>> +static void __attribute__ ((constructor)) setup_libhugetlbfs(void)
>> +{
On 27.02.2008 [15:48:24 -0600], Andrew Hastings wrote:
> Nish,
>
> Nice! Just one comment:
>
> Nishanth Aravamudan wrote:
>> --- a/libhugetlbfs_internal.h
>> +++ b/libhugetlbfs_internal.h
>> @@ -29,6 +29,7 @@
>> #define ALIGN(x, a) (((x) + (a) - 1) & ~((a) - 1))
>> extern int __hugetlbfs_verbos
On 27.02.2008 [14:23:18 -0800], Nishanth Aravamudan wrote:
> morecore: dump /proc/pid/maps when the heap is forced to move and DEBUG
>
> If HUGETLB_DEBUG is set in the environment, use a new helper function,
> dump_proc_pid_maps() to indicate the address space layout. This is very
> useful in find
On 27.02.2008 [14:43:38 -0800], Nishanth Aravamudan wrote:
> For debugging purposes, setting HUGETLB_DEBUG should be sufficient. It
> should also imply maximum verbosity from the library, so spit out all
> types of messages (DEBUG, WARNING or ERROR).
Update to new name for __debug:
DEBUG override
On 27.02.2008 [14:43:04 -0800], Nishanth Aravamudan wrote:
> To ease with debugging, output the address and name of extra symbols to
> copy from within the BSS.
Update to new name for __debug:
elflink: emit symbol name when extracopy symbols are found
To ease with debugging, output the address a
On 27.02.2008 [15:23:36 -0800], Nishanth Aravamudan wrote:
> On 27.02.2008 [14:43:04 -0800], Nishanth Aravamudan wrote:
> > To ease with debugging, output the address and name of extra symbols to
> > copy from within the BSS.
>
> Update to new name for __debug:
This introduced a warning actually,
Nishanth Aravamudan wrote:
> On 27.02.2008 [15:48:24 -0600], Andrew Hastings wrote:
>> Nish,
>>
>> Nice! Just one comment:
>>
>> Nishanth Aravamudan wrote:
>>> --- a/libhugetlbfs_internal.h
>>> +++ b/libhugetlbfs_internal.h
>>> @@ -29,6 +29,7 @@
>>> #define ALIGN(x, a)(((x) + (a) - 1) & ~
Nishanth Aravamudan wrote:
> On 27.02.2008 [15:46:47 -0600], Andrew Hastings wrote:
>> Nishanth Aravamudan wrote:
>>> --- /dev/null
>>> +++ b/init.c
>>> @@ -0,0 +1,8 @@
>>> +#include "libhugetlbfs_internal.h"
>>> +
>>> +static void __attribute__ ((constructor)) setup_libhugetlbfs(void)
>>> +{
>>> +
On 27.02.2008 [16:12:14 -0600], Andrew Hastings wrote:
> hugetlbfs_morecore() is currently calling mlock() presumably to
> instantiate newly-mapped hugepages before returning to malloc().
>
> However, this is ineffective for two reasons:
> 1. The return value of mlock is ignored. If there aren't
On 27.02.2008 [17:02:15 -0800], Nishanth Aravamudan wrote:
> On 27.02.2008 [16:12:14 -0600], Andrew Hastings wrote:
> > hugetlbfs_morecore() is currently calling mlock() presumably to
> > instantiate newly-mapped hugepages before returning to malloc().
> >
> > However, this is ineffective for two
So, with the 6-patch drop I did today, we have between 23 and 29 patches
post 1.2. How are folks feeling about that? I am starting to lean
towards cutting a 1.3-rc0 release to start that cycle going.
There are a few threatened features, though:
* Andrew's HUGETLB_STACK set
pros
-
On Wed, Feb 27, 2008 at 03:15:14PM -0800, Nishanth Aravamudan wrote:
> On 27.02.2008 [15:48:24 -0600], Andrew Hastings wrote:
> > Nish,
> >
> > Nice! Just one comment:
> >
> > Nishanth Aravamudan wrote:
> >> --- a/libhugetlbfs_internal.h
> >> +++ b/libhugetlbfs_internal.h
> >> @@ -29,6 +29,7 @@
>
On Tue, Feb 26, 2008 at 12:11:56AM -0800, Nishanth Aravamudan wrote:
> On 26.02.2008 [16:16:01 +1100], David Gibson wrote:
> > On Mon, Feb 25, 2008 at 09:05:42PM -0800, Nishanth Aravamudan wrote:
> > > On 26.02.2008 [15:33:55 +1100], David Gibson wrote:
> > > > On Mon, Feb 25, 2008 at 08:23:48PM -0
On Wed, Feb 27, 2008 at 03:13:20PM -0800, Nishanth Aravamudan wrote:
> On 27.02.2008 [15:46:47 -0600], Andrew Hastings wrote:
> > Nish,
> >
> > Thanks! You beat me to it!
> >
> > Nishanth Aravamudan wrote:
> >> --- /dev/null
> >> +++ b/init.c
> >> @@ -0,0 +1,8 @@
> >> +#include "libhugetlbfs_inter
On 28.02.2008 [13:00:34 +1100], David Gibson wrote:
> On Tue, Feb 26, 2008 at 12:11:56AM -0800, Nishanth Aravamudan wrote:
> > On 26.02.2008 [16:16:01 +1100], David Gibson wrote:
> > > On Mon, Feb 25, 2008 at 09:05:42PM -0800, Nishanth Aravamudan wrote:
> > > > On 26.02.2008 [15:33:55 +1100], David
On 28.02.2008 [12:47:08 +1100], David Gibson wrote:
> On Wed, Feb 27, 2008 at 03:15:14PM -0800, Nishanth Aravamudan wrote:
> > On 27.02.2008 [15:48:24 -0600], Andrew Hastings wrote:
> > > Nish,
> > >
> > > Nice! Just one comment:
> > >
> > > Nishanth Aravamudan wrote:
> > >> --- a/libhugetlbfs_int
On 28.02.2008 [12:43:13 +1100], David Gibson wrote:
> On Wed, Feb 27, 2008 at 03:13:20PM -0800, Nishanth Aravamudan wrote:
> > On 27.02.2008 [15:46:47 -0600], Andrew Hastings wrote:
> > > Nish,
> > >
> > > Thanks! You beat me to it!
> > >
> > > Nishanth Aravamudan wrote:
> > >> --- /dev/null
> > >
On Wed, Feb 27, 2008 at 11:29:03AM -0600, Andrew Hastings wrote:
> David Gibson wrote:
> > On Wed, Feb 20, 2008 at 03:43:08PM -0600, Andrew Hastings wrote:
> >> David Gibson wrote:
> >>> On Thu, Feb 14, 2008 at 11:48:42AM -0600, Andrew Hastings wrote:
> Disable heap shrinking by default unless
On Wed, Feb 27, 2008 at 06:17:02PM -0800, Nishanth Aravamudan wrote:
> On 28.02.2008 [13:00:34 +1100], David Gibson wrote:
> > On Tue, Feb 26, 2008 at 12:11:56AM -0800, Nishanth Aravamudan wrote:
> > > On 26.02.2008 [16:16:01 +1100], David Gibson wrote:
> > > > On Mon, Feb 25, 2008 at 09:05:42PM -0
On 28.02.2008 [13:28:47 +1100], David Gibson wrote:
> On Wed, Feb 27, 2008 at 06:17:02PM -0800, Nishanth Aravamudan wrote:
> > On 28.02.2008 [13:00:34 +1100], David Gibson wrote:
> > > On Tue, Feb 26, 2008 at 12:11:56AM -0800, Nishanth Aravamudan wrote:
> > > > On 26.02.2008 [16:16:01 +1100], David
On Wed, Feb 27, 2008 at 06:34:37PM -0800, Nishanth Aravamudan wrote:
> On 28.02.2008 [13:28:47 +1100], David Gibson wrote:
> > On Wed, Feb 27, 2008 at 06:17:02PM -0800, Nishanth Aravamudan wrote:
[snip]
> > > Hrm, I guess I hadn't thought about it that way. I really am curious
> > > about these sem
On 28.02.2008 [13:55:58 +1100], David Gibson wrote:
> On Wed, Feb 27, 2008 at 06:34:37PM -0800, Nishanth Aravamudan wrote:
> > On 28.02.2008 [13:28:47 +1100], David Gibson wrote:
> > > On Wed, Feb 27, 2008 at 06:17:02PM -0800, Nishanth Aravamudan wrote:
> [snip]
> > > > Hrm, I guess I hadn't though
On 27.02.2008 [23:20:35 -0800], Nishanth Aravamudan wrote:
> On 28.02.2008 [13:55:58 +1100], David Gibson wrote:
> > On Wed, Feb 27, 2008 at 06:34:37PM -0800, Nishanth Aravamudan wrote:
> > > On 28.02.2008 [13:28:47 +1100], David Gibson wrote:
> > > > On Wed, Feb 27, 2008 at 06:17:02PM -0800, Nisha
36 matches
Mail list logo
