Development/Pahole - Revision history

S Richling: S Richling moved page Pahole to Development/Pahole: Umzug in Developmentbereich

2022-12-08T22:51:46Z

S Richling moved page Pahole to Development/Pahole: Umzug in Developmentbereich

← Older revision	Revision as of 00:51, 9 December 2022
(No difference)

S Richling at 22:35, 8 December 2022

2022-12-08T22:35:47Z

← Older revision		Revision as of 00:35, 9 December 2022
Line 142:		Line 142:
	Rearranging the above data structure would reduce the amount of holes (e.g. move above <kbd>any_source_offset</kbd> to after <kbd>agreement_specific</kbd>),		Rearranging the above data structure would reduce the amount of holes (e.g. move above <kbd>any_source_offset</kbd> to after <kbd>agreement_specific</kbd>),
	taking into account data access in the most-common code paths would also optimize cache line usage. Low hanging fruits such as eliminating (or reducing or moving to the end) <kbd>c_name</kbd>, which is solely used for application/ompi debugging, might help as well...		taking into account data access in the most-common code paths would also optimize cache line usage. Low hanging fruits such as eliminating (or reducing or moving to the end) <kbd>c_name</kbd>, which is solely used for application/ompi debugging, might help as well...


	[[Category:debugger software]][[Category:performance analysis software]][[Category:bwUniCluster]]

R Keller at 14:04, 21 February 2022

2022-02-21T14:04:42Z

← Older revision		Revision as of 16:04, 21 February 2022
Line 74:		Line 74:
	== Usage in own application ==		== Usage in own application ==
	To employ this in your own application, recompile with compiler option <kbd>-g</kbd>.		To employ this in your own application, recompile with compiler option <kbd>-g</kbd>.
	For the following code (part of the Open MPI implementation) <kbd>pahole $~~OMPI_DIR/lib~~/libmpi.so</kbd>:		For the following code (part of the Open MPI implementation) <kbd>pahole $MPI_LIB_DIR/libmpi.so</kbd>:
	<source>		<source>
	struct ompi_communicator_t {		struct ompi_communicator_t {

R Keller: Update part 2

2022-02-21T13:58:27Z

Update part 2

@@ Line 32: / Line 32: @@
 There currently is no web documentation, or tutorial. After loading the module, documentation is provided in the '''man page'''.
 <pre>$ man pahole</pre>
-= Basic commands =
+= Usage =
 == Dissecting data structures ==
 You may receive information on the padding and alignment of data structures using <kbd>pahole</kbd>.
@@ Line 75: / Line 74: @@
 == Usage in own application ==
 To employ this in your own application, recompile with compiler option <kbd>-g</kbd>.
-For the following code (part of the Open MPI implementation):
+For the following code (part of the Open MPI implementation) <kbd>pahole $OMPI_DIR/lib/libmpi.so</kbd>:

R Keller: Update part 1

2022-02-21T13:08:27Z

Update part 1

← Older revision		Revision as of 15:08, 21 February 2022
Line 1:		Line 1:
			{{Softwarepage\|devel/pahole}}

	{\| width=600px class="wikitable"		{\| width=600px class="wikitable"
	\|-		\|-
Line 20:		Line 22:
	\| Graphical Interface		\| Graphical Interface
	\| No		\| No
	\|-
	\| Included modules
	\|}		\|}
	<br>		<br>
	= Introduction =		= Introduction =
	The '''~~Dwarves Tools~~''' is a tool-set to ~~dissect~~ ~~user-~~ ~~and~~ ~~kernel~~ ~~level~~ data ~~structures~~ and performance analysis ~~using~~ ~~tracers~~.		The poke-a-hole, or short '''pahole''' tool is part of the dwarves tool-set. It dissects data structures in binary object files, showing (otherwise useless) padding and data crossing cachelines, allowing optimization and performance analysis of data structures in user- or kernel-code.
			This tool is worthy to be known by any C and Fortran developer -- the article [http://www.catb.org/esr/structure-packing/ The lost Art of Structure Packingg by Eric S. Raymond] provides an in-detail answer, why.
	<br>		<br>

	= Versions and Availability =
	A list of versions currently available on all bwHPC Clusters can be obtained from the
	<br>
	<big>

	[https://cis-hpc.uni-konstanz.de/prod.cis/ Cluster Information System CIS]

	</big>
	{{#widget:Iframe
	\|url=https://cis-hpc.uni-konstanz.de/prod.cis/bwUniCluster/devel/pahole
	\|width=99%
	\|height=120
	}}


	= Documentation =		= Documentation =
	There currently is no web documentation, or tutorial. ~~For~~ ~~'''local~~ documentation~~'''~~ ~~consult~~ the man page.		There currently is no web documentation, or tutorial. After loading the module, documentation is provided in the '''man page'''.
	<pre>$ man pahole</pre>		<pre>$ man pahole</pre>
	<br>		<br>
Line 50:		Line 36:

	= Basic commands =		= Basic commands =
			== Dissecting data structures ==
	You may receive information on the padding and alignment of data structures using <kbd>pahole</kbd>.		You may receive information on the padding and alignment of data structures using <kbd>pahole</kbd>.
	If interested in the data layout pass the option to expand data structures, e.g. the kernel's		If interested in the data layout pass the option to expand data structures, e.g. the kernel's
Line 83:		Line 70:
	</source>		</source>

	This nicely shows, where the compiler needed to insert padding to adhere to the architecture's alignment requirements specified by the ABI.		This nicely shows, where the compiler needed to insert padding to adhere to the architecture's alignment requirements specified by the ABI. Additionally it layouts the crossing of past cacheline boundaries, which might be problematic in false sharing of cachelines in multi-threaded programming.
	Additionally it layouts the crossing of past cacheline boundaries, which might be problematic in false sharing of cachelines in multi-threaded programming.
	Both of this information may be used by You to re-layout your data-structures to minimize them and limit cache-thrashing.		Both of this information may be used by You to re-layout your data-structures to minimize them and limit cache-thrashing.

			== Usage in own application ==
			To employ this in your own application, recompile with compiler option <kbd>-g</kbd>.
			For the following code (part of the Open MPI implementation):



	[[Category:debugger software]][[Category:performance analysis software]][[Category:bwUniCluster]]		[[Category:debugger software]][[Category:performance analysis software]][[Category:bwUniCluster]]

R Keller at 16:10, 4 November 2020

2020-11-04T16:10:14Z

← Older revision		Revision as of 18:10, 4 November 2020
Line 87:		Line 87:
	Both of this information may be used by You to re-layout your data-structures to minimize them and limit cache-thrashing.		Both of this information may be used by You to re-layout your data-structures to minimize them and limit cache-thrashing.

	[[Category:debugger software]][[Category:bwUniCluster]]		[[Category:debugger software]][[Category:performance analysis software]][[Category:bwUniCluster]]

R Keller: Created page with "{| width=600px class="wikitable" |- ! Description !! Content |- | module load | devel/pahole |- | Availability | bwUniCluster |- | License | [http://www.gnu.org/licenses/..."

2020-11-04T15:05:14Z

Created page with "{| width=600px class="wikitable" |- ! Description !! Content |- | module load | devel/pahole |- | Availability | bwUniCluster |- | License | [http://www.gnu.org/licenses/..."

New page

{| width=600px class="wikitable"
|-
! Description !! Content
|-
| module load
| devel/pahole
|-
| Availability
| [[bwUniCluster]]
|-
| License
| [http://www.gnu.org/licenses/ GPL]
|-
|Citing
| n/a
|-
| Links
| [https://github.com/acmel/dwarves Homepage] | [https://fedorapeople.org/~acme/dwarves/ Releases]
|-
| Graphical Interface
| No
|-
| Included modules
|}
<br>
= Introduction =
The '''Dwarves Tools''' is a tool-set to dissect user- and kernel level data structures and performance analysis using tracers.
<br>

= Versions and Availability =
A list of versions currently available on all bwHPC Clusters can be obtained from the
<br>
<big>

[https://cis-hpc.uni-konstanz.de/prod.cis/ Cluster Information System CIS]

</big>
{{#widget:Iframe
|url=https://cis-hpc.uni-konstanz.de/prod.cis/bwUniCluster/devel/pahole
|width=99%
|height=120
}}

= Documentation =
There currently is no web documentation, or tutorial. For '''local documentation''' consult the man page.
<pre>$ man pahole</pre>
<br>

= Basic commands =
You may receive information on the padding and alignment of data structures using <kbd>pahole</kbd>.
If interested in the data layout pass the option to expand data structures, e.g. the kernel's
data structure for every single task, use <kbd>pahole -E task_struct</kbd>
<source lang="c">
struct task_struct {
struct thread_info {
long unsigned int flags; /* 0 8 */
unsigned int status; /* 8 4 */
} thread_info; /* 0 16 */

/* XXX last struct has 4 bytes of padding */

volatile long int state; /* 16 8 */
void * stack; /* 24 8 */
struct {
int counter; /* 32 4 */
} usage; /* 32 4 */
unsigned int flags; /* 36 4 */
unsigned int ptrace; /* 40 4 */

/* XXX 4 bytes hole, try to pack */

struct llist_node {
struct llist_node * next; /* 48 8 */
} wake_entry; /* 48 8 */
int on_cpu; /* 56 4 */
unsigned int cpu; /* 60 4 */
/* --- cacheline 1 boundary (64 bytes) --- */
unsigned int wakee_flips; /* 64 4 */
...
}
</source>

This nicely shows, where the compiler needed to insert padding to adhere to the architecture's alignment requirements specified by the ABI.
Additionally it layouts the crossing of past cacheline boundaries, which might be problematic in false sharing of cachelines in multi-threaded programming.
Both of this information may be used by You to re-layout your data-structures to minimize them and limit cache-thrashing.

[[Category:debugger software]][[Category:bwUniCluster]]