charm.git
8 years agomake hugepage as default for uGNI
Yanhua Sun [Thu, 24 Oct 2013 01:49:36 +0000 (18:49 -0700)]
make hugepage as default for uGNI

8 years agoCharmrun: Fix autobuild failure on multicore-win32 related to cleanup
Abhishek Gupta [Wed, 23 Oct 2013 22:23:02 +0000 (17:23 -0500)]
Charmrun: Fix autobuild failure on multicore-win32 related to cleanup

8 years agoCharmrun: req_handle_abort does not retuen anything, so make it void
Abhishek Gupta [Wed, 23 Oct 2013 20:57:10 +0000 (15:57 -0500)]
Charmrun: req_handle_abort does not retuen anything, so make it void

8 years agoRemove commented out code from cklocation to cleanup.
Harshitha [Wed, 23 Oct 2013 20:26:39 +0000 (15:26 -0500)]
Remove commented out code from cklocation to cleanup.

8 years agoSuppressing two false positive cppcheck errors in tm_tree.c
Harshitha [Wed, 23 Oct 2013 20:23:21 +0000 (15:23 -0500)]
Suppressing two false positive cppcheck errors in tm_tree.c

8 years agoconvert !CMK_OPTIMIZE to CMK_ERROR_CHECKING
Eric Bohm [Wed, 23 Oct 2013 16:54:11 +0000 (11:54 -0500)]
convert !CMK_OPTIMIZE to CMK_ERROR_CHECKING

8 years agoconvert !CMK_OPTIMIZE to CMK_ERROR_CHECKING
Eric Bohm [Wed, 23 Oct 2013 16:44:03 +0000 (11:44 -0500)]
convert !CMK_OPTIMIZE to CMK_ERROR_CHECKING

8 years agoswitch !CMK_OPTIMIZE to CMK_ERROR_CHECKING
Eric Bohm [Wed, 23 Oct 2013 16:40:39 +0000 (11:40 -0500)]
switch !CMK_OPTIMIZE to CMK_ERROR_CHECKING

8 years agoconvert to use of CMK_ERROR_CHECKING
Eric Bohm [Wed, 23 Oct 2013 16:21:23 +0000 (11:21 -0500)]
convert to use of CMK_ERROR_CHECKING

8 years agotopomanager: add int specified for init_done
Nikhil Jain [Wed, 23 Oct 2013 15:25:37 +0000 (15:25 +0000)]
topomanager: add int specified for init_done

8 years agonetlrts: fix a bug in converse header, which may be the cause for all
Nikhil Jain [Wed, 23 Oct 2013 07:23:10 +0000 (02:23 -0500)]
netlrts: fix a bug in converse header, which may be the cause for all
the hangs noticed in SMP mode

8 years agonetlrts: bug fixes related to handleOneReceive function
Bilge Acun [Tue, 22 Oct 2013 23:11:26 +0000 (18:11 -0500)]
netlrts: bug fixes related to handleOneReceive function

8 years agotopomanager: give correct coordinate on BGQ when using partitions
Nikhil Jain [Mon, 14 Oct 2013 07:11:44 +0000 (07:11 +0000)]
topomanager: give correct coordinate on BGQ when using partitions

TopoManager: Makefile updates

8 years ago#7 charmxi: Suppress cppcheck memory leak warning, since NamedType keeps parentClassName
Phil Miller [Wed, 23 Oct 2013 04:02:36 +0000 (21:02 -0700)]
#7 charmxi: Suppress cppcheck memory leak warning, since NamedType keeps parentClassName

8 years agoIssue #7: Suppress cppcheck warning about leaking memory
Phil Miller [Wed, 23 Oct 2013 03:54:51 +0000 (20:54 -0700)]
Issue #7: Suppress cppcheck warning about leaking memory

We're returning an offset from the allocated memory, and can recover
the original address at deallocation. Cppcheck is just confused.

8 years agoIssue #7: Suppress cppcheck warning about uninitialized variables
Phil Miller [Wed, 23 Oct 2013 03:52:55 +0000 (20:52 -0700)]
Issue #7: Suppress cppcheck warning about uninitialized variables

By inspection, all allowed paths to the affected code actually do
initialize these fields. Some combination of two different functions
and various conditionals leading to them seems to have misled it.

8 years ago#135: Delete old (and done) TODO comment that referenced CMK_OPTIMIZE
Phil Miller [Wed, 23 Oct 2013 02:58:47 +0000 (19:58 -0700)]
#135: Delete old (and done) TODO comment that referenced CMK_OPTIMIZE

8 years ago#135: Drop really old dead error check, that may not have been applicable for a decade
Phil Miller [Wed, 23 Oct 2013 02:54:08 +0000 (19:54 -0700)]
#135: Drop really old dead error check, that may not have been applicable for a decade

8 years agopartition: handle cpu affinity with partitions
Nikhil Jain [Tue, 22 Oct 2013 21:16:00 +0000 (16:16 -0500)]
partition: handle cpu affinity with partitions

8 years agoDelete archaic 'CMK_USE_HP_MAIN_FIX' code from non-defunct builds
Phil Miller [Wed, 23 Oct 2013 02:21:18 +0000 (19:21 -0700)]
Delete archaic 'CMK_USE_HP_MAIN_FIX' code from non-defunct builds

8 years ago#135: Drop defunct CMK_OPTIMIZE-derived configuration
Phil Miller [Wed, 23 Oct 2013 02:09:40 +0000 (19:09 -0700)]
#135: Drop defunct CMK_OPTIMIZE-derived configuration

8 years agono need to make ephemeral NormalSlabArray given that it no
Eric Bohm [Tue, 22 Oct 2013 23:39:14 +0000 (18:39 -0500)]
no need to make ephemeral NormalSlabArray given that it no
longer has the side effect of making a comlib instance

8 years agocppcheck cleanup
Eric Bohm [Tue, 22 Oct 2013 23:38:27 +0000 (18:38 -0500)]
cppcheck cleanup

8 years agocppcheck cleanup, ignoring the false positives from UNPACK operations.
Eric Bohm [Tue, 22 Oct 2013 23:11:55 +0000 (18:11 -0500)]
cppcheck cleanup, ignoring the false positives from UNPACK operations.

8 years agoMerge branch 'charm' of charmgit:charm into charm
YanhuaSun [Tue, 22 Oct 2013 20:32:32 +0000 (15:32 -0500)]
Merge branch 'charm' of charmgit:charm into charm

8 years agominor
YanhuaSun [Tue, 22 Oct 2013 20:32:26 +0000 (15:32 -0500)]
minor

8 years agocell/cell_lib/rawToProl.cpp: Fix resource leak of fopen'd file by closing before...
Michael Robson [Tue, 22 Oct 2013 20:28:33 +0000 (15:28 -0500)]
cell/cell_lib/rawToProl.cpp: Fix resource leak of fopen'd file by closing before error return

8 years agoFix cppcheck errors in seed load balancer conv-ldb folder
Harshitha [Tue, 22 Oct 2013 19:59:57 +0000 (14:59 -0500)]
Fix cppcheck errors in seed load balancer conv-ldb folder

8 years agocppcheck fix in tracing
YanhuaSun [Tue, 22 Oct 2013 19:57:43 +0000 (14:57 -0500)]
cppcheck fix in tracing

8 years agoMerge branch 'charm' of charmgit:charm into charm
YanhuaSun [Tue, 22 Oct 2013 19:40:32 +0000 (14:40 -0500)]
Merge branch 'charm' of charmgit:charm into charm

8 years agocomment out unused function in gni
YanhuaSun [Tue, 22 Oct 2013 19:40:22 +0000 (14:40 -0500)]
comment out unused function in gni

8 years agocppcheck cleanup
Eric Bohm [Tue, 22 Oct 2013 19:26:20 +0000 (14:26 -0500)]
cppcheck cleanup

8 years agoRemoving a memory leak from tm_tree.c
Harshitha [Tue, 22 Oct 2013 17:13:26 +0000 (12:13 -0500)]
Removing a memory leak from tm_tree.c

8 years agopartition: bug fix to allow progress if there is one one PE in a partition, but
Nikhil Jain [Tue, 22 Oct 2013 15:59:52 +0000 (10:59 -0500)]
partition: bug fix to allow progress if there is one one PE in a partition, but
there are multiple partitions in the job

8 years agoPOSE: Include a human-authored header with proper include guards
Phil Miller [Tue, 22 Oct 2013 00:59:49 +0000 (17:59 -0700)]
POSE: Include a human-authored header with proper include guards

8 years agoIssue #7 - msa: Initialize variable to quiet a cppcheck warning
Phil Miller [Tue, 22 Oct 2013 00:05:28 +0000 (17:05 -0700)]
Issue #7 - msa: Initialize variable to quiet a cppcheck warning

8 years agocharmxi: Add a virtual destructor for SdagConstruct, since it has virtual methods
Phil Miller [Mon, 21 Oct 2013 22:27:10 +0000 (15:27 -0700)]
charmxi: Add a virtual destructor for SdagConstruct, since it has virtual methods

8 years agoIssue #293 - charmrun win32: const correctness in platform-specific code
Phil Miller [Mon, 21 Oct 2013 16:34:12 +0000 (09:34 -0700)]
Issue #293 - charmrun win32: const correctness in platform-specific code

8 years agoIssue #293 - charmrun: const-correctness on syncft and causalft builds
Phil Miller [Sun, 20 Oct 2013 18:21:22 +0000 (13:21 -0500)]
Issue #293 - charmrun: const-correctness on syncft and causalft builds

8 years agocharmrun daemon: const-correctness to quiet warnings
Phil Miller [Fri, 18 Oct 2013 20:56:19 +0000 (13:56 -0700)]
charmrun daemon: const-correctness to quiet warnings

8 years agoIssue #293 - charmrun: Improve const-correctness to only cast it away at exec* calls
Phil Miller [Fri, 18 Oct 2013 20:48:39 +0000 (13:48 -0700)]
Issue #293 - charmrun: Improve const-correctness to only cast it away at exec* calls

8 years agoMore const-correctness fixes for charmrun.
Nicolas Bock [Fri, 18 Oct 2013 16:34:00 +0000 (10:34 -0600)]
More const-correctness fixes for charmrun.

8 years ago Issue 135: CMK_OPTIMIZE cleanup in net, netlrts. verbs
Abhishek Gupta [Fri, 18 Oct 2013 16:25:33 +0000 (11:25 -0500)]
 Issue 135: CMK_OPTIMIZE cleanup in net, netlrts. verbs

8 years agos/#ifndef CMK_TRACE_DISABLED/#if !CMK_TRACE_DISABLED/
Eric Bohm [Fri, 18 Oct 2013 00:22:38 +0000 (19:22 -0500)]
s/#ifndef CMK_TRACE_DISABLED/#if !CMK_TRACE_DISABLED/
to protect the tracing code blocks

8 years agofix poor usage of CMK_ERROR_CHECKING
Eric Bohm [Thu, 17 Oct 2013 23:30:25 +0000 (18:30 -0500)]
fix poor usage of CMK_ERROR_CHECKING

8 years agoswap the comments so they match the code again
Eric Bohm [Thu, 17 Oct 2013 23:17:53 +0000 (18:17 -0500)]
swap the comments so they match the code again

8 years agofix reversal of meaning for CMK_ERROR_CHECKING
Eric Bohm [Thu, 17 Oct 2013 23:14:31 +0000 (18:14 -0500)]
fix reversal of meaning for CMK_ERROR_CHECKING

8 years agosweep of libs/ck-libs for CMK_OPTIMIZE issues
Eric Bohm [Thu, 17 Oct 2013 22:53:58 +0000 (17:53 -0500)]
sweep of libs/ck-libs for CMK_OPTIMIZE issues

8 years agocharmrun #293: improve const-correctness
Phil Miller [Thu, 17 Oct 2013 21:25:05 +0000 (14:25 -0700)]
charmrun #293: improve const-correctness

8 years agoFT: fix the delete issues in the previous commit
xiangni [Thu, 17 Oct 2013 21:00:11 +0000 (16:00 -0500)]
FT: fix the delete issues in the previous commit

8 years agoFT: fix cppcheck issues
xiangni [Thu, 17 Oct 2013 20:56:46 +0000 (15:56 -0500)]
FT: fix cppcheck issues

8 years agoCharmrun: Improved const-correctness at some places
Abhishek Gupta [Thu, 17 Oct 2013 19:49:45 +0000 (14:49 -0500)]
Charmrun: Improved const-correctness at some places

8 years agoCharmrun: fixed cppchek issues
Abhishek Gupta [Thu, 17 Oct 2013 19:11:34 +0000 (14:11 -0500)]
Charmrun: fixed cppchek issues

8 years agocppcheck cleanup sweep for unowned modules
Eric Bohm [Thu, 17 Oct 2013 18:31:42 +0000 (13:31 -0500)]
cppcheck cleanup sweep for unowned modules

8 years agoCharmrun-cleanup: bug fix - rectify makefile dependencies
Abhishek Gupta [Mon, 14 Oct 2013 18:28:27 +0000 (13:28 -0500)]
Charmrun-cleanup: bug fix - rectify makefile dependencies

8 years agoRemove duplicated charmrun copies across net, netlrts, verbs lrts to src/util/charmru...
Abhishek [Sat, 5 Oct 2013 05:32:52 +0000 (00:32 -0500)]
Remove duplicated charmrun copies across net, netlrts, verbs lrts to src/util/charmrun-src. Modified Makefile appropiately

8 years agochanges to provide clean cppcheck and quell compiler complaints
Eric Bohm [Wed, 16 Oct 2013 23:22:01 +0000 (18:22 -0500)]
changes to provide clean cppcheck and quell compiler complaints

8 years agotiming trial and parameter settings to examine runtime of synctest
Eric Bohm [Wed, 16 Oct 2013 20:37:33 +0000 (15:37 -0500)]
timing trial and parameter settings to examine runtime of synctest

8 years agoFixed bug in CkReductionMgr::reduceMessages().
Nicolas Bock [Thu, 5 Sep 2013 10:04:06 +0000 (04:04 -0600)]
Fixed bug in CkReductionMgr::reduceMessages().

The first contribution of a child node is a single message. In this case
CkReductionMgr::reduceMessages() simply returns the message without calling
the reducer. However, when using the CkReduction::set reducer this behavior is
incorrect, and instead the reducer should be called even for one single
message so that the message is wrapped into a setElement struct. In the
current implementation, the reduction becomes corrupted because the
CkReduction::set() method one tier up in the reduction sizes the remote
contribution incorrectly assuming a setElement struct and not a simple
message.

8 years agoFix the log2 not found errors in machines which does not have log2. Use CmiLog2.
Harshitha [Wed, 16 Oct 2013 17:47:14 +0000 (12:47 -0500)]
Fix the log2 not found errors in machines which does not have log2. Use CmiLog2.

8 years agocleanup: do not initialize int with a character
Nikhil Jain [Wed, 16 Oct 2013 05:56:06 +0000 (00:56 -0500)]
cleanup: do not initialize int with a character

8 years agocth: bug fix in an earlier commit (assign before use)
Nikhil Jain [Wed, 16 Oct 2013 05:47:40 +0000 (00:47 -0500)]
cth: bug fix in an earlier commit (assign before use)

8 years ago#74: Removing initcalls across codebase
Ronak Buch [Tue, 15 Oct 2013 19:31:07 +0000 (14:31 -0500)]
#74: Removing initcalls across codebase

8 years agoAdd appropriate include headers for log2 and time function
Harshitha [Tue, 15 Oct 2013 13:52:43 +0000 (08:52 -0500)]
Add appropriate include headers for log2 and time function

8 years agoAdd DistributedLB to the set of load balancers in charm.
Harshitha [Mon, 14 Oct 2013 21:41:07 +0000 (16:41 -0500)]
Add DistributedLB to the set of load balancers in charm.

8 years agocleanup: issue 7; address errors pointed to by cppcheck
Nikhil Jain [Mon, 14 Oct 2013 03:45:32 +0000 (22:45 -0500)]
cleanup: issue 7; address errors pointed to by cppcheck

8 years agoFix cppcheck error in ldb specifically in tm_mapping file.
Harshitha [Sun, 13 Oct 2013 20:35:54 +0000 (15:35 -0500)]
Fix cppcheck error in ldb specifically in tm_mapping file.

8 years agoFor causal message logging, revert back to using StartReduction message
Harshitha [Sat, 12 Oct 2013 03:26:45 +0000 (22:26 -0500)]
For causal message logging, revert back to using StartReduction message

8 years agoHandle the case for checkpoint restart with shrink expand
Harshitha [Mon, 7 Oct 2013 17:53:04 +0000 (12:53 -0500)]
Handle the case for checkpoint restart with shrink expand

8 years agoAn implementation of reductions which does not require ReductionStarting messages...
Harshitha [Fri, 4 Oct 2013 16:01:16 +0000 (11:01 -0500)]
An implementation of reductions which does not require ReductionStarting messages in the case where there is no barren leaf

8 years agonetlrts: cleanup, remove macros that are already defined in machine-common-core
Bilge Acun [Sat, 12 Oct 2013 20:25:05 +0000 (15:25 -0500)]
netlrts: cleanup, remove macros that are already defined in machine-common-core

8 years agonetlrts: fix the bug which occurred during GM and MX removal
Bilge Acun [Fri, 11 Oct 2013 19:32:58 +0000 (14:32 -0500)]
netlrts: fix the bug which occurred during GM and MX removal

8 years agoRevert "netlrts: fix the bug which occurred during GM and MX removal"
Bilge Acun [Fri, 11 Oct 2013 19:31:36 +0000 (14:31 -0500)]
Revert "netlrts: fix the bug which occurred during GM and MX removal"

This reverts commit 5aeb63bababb3f9493301d2de0a0b65ef840a0aa.

8 years agonetlrts: fix the bug which occurred during GM and MX removal
Bilge Acun [Fri, 11 Oct 2013 19:11:18 +0000 (14:11 -0500)]
netlrts: fix the bug which occurred during GM and MX removal

8 years agofix for bitvector
YanhuaSun [Fri, 11 Oct 2013 16:01:18 +0000 (11:01 -0500)]
fix for bitvector

8 years agofix bitvector compare according to lexicographic order, length is secondary
YanhuaSun [Fri, 11 Oct 2013 14:02:13 +0000 (09:02 -0500)]
fix bitvector compare  according to lexicographic order, length is secondary

8 years agominor
YanhuaSun [Fri, 11 Oct 2013 03:53:53 +0000 (22:53 -0500)]
minor

8 years agoadd abort to avoid the SMP many-to-many hang without async
YanhuaSun [Fri, 11 Oct 2013 03:40:52 +0000 (22:40 -0500)]
add abort to avoid the SMP many-to-many hang without async

8 years ago#277: Exposing lb start trigger as CkStartLB() and adding documentation for it.
Ronak Buch [Thu, 10 Oct 2013 22:20:38 +0000 (17:20 -0500)]
#277: Exposing lb start trigger as CkStartLB() and adding documentation for it.

8 years agoisomalloc: handle cases with reallocation of NULL
Nikhil Jain [Thu, 10 Oct 2013 14:23:55 +0000 (09:23 -0500)]
isomalloc: handle cases with reallocation of NULL

8 years agoFT: fix compiler warning for unused return value
xiangni [Thu, 10 Oct 2013 05:02:20 +0000 (00:02 -0500)]
FT: fix compiler warning for unused return value

8 years agonetlrts: remove GM and MX from netlrts
Bilge Acun [Wed, 9 Oct 2013 21:55:49 +0000 (16:55 -0500)]
netlrts: remove GM and MX from netlrts
codes within CMK_USE_MX and CMK_USE_GM macro, and gm, mx specific
machine files are deleted

8 years agonetlrts: broadcasts are now handled by the common core code
Bilge Acun [Wed, 9 Oct 2013 20:57:07 +0000 (15:57 -0500)]
netlrts: broadcasts are now handled by the common core code
by handleOneReceive function. All of the Hypercube and SpanningTree
send code is removed.

8 years ago#276 Doc: Adding StartLB() documentation
Ronak Buch [Wed, 9 Oct 2013 16:32:31 +0000 (11:32 -0500)]
#276 Doc: Adding StartLB() documentation

8 years agoNDMeshStreamer: check validity of flush period when enabling periodic flushing.
Lukasz Wesolowski [Tue, 8 Oct 2013 18:32:30 +0000 (13:32 -0500)]
NDMeshStreamer: check validity of flush period when enabling periodic flushing.

8 years agomachine layer: add a real barrier to new lrts versions
Nikhil Jain [Tue, 8 Oct 2013 06:41:54 +0000 (01:41 -0500)]
machine layer: add a real barrier to new lrts versions

8 years agomachine layer: fix exit issues in verbs layer
Nikhil Jain [Mon, 7 Oct 2013 03:10:10 +0000 (22:10 -0500)]
machine layer: fix exit issues in verbs layer

8 years agomachine layer: make netlrts exit do what is needed for a clean exit
Nikhil Jain [Mon, 7 Oct 2013 00:33:29 +0000 (19:33 -0500)]
machine layer: make netlrts exit do what is needed for a clean exit

8 years agoexamples: run jacobi for a fixed iter count
Nikhil Jain [Mon, 7 Oct 2013 00:32:49 +0000 (19:32 -0500)]
examples: run jacobi for a fixed iter count

8 years agomachine layer: explicit use of 32 bit flags for netlrts
Nikhil Jain [Sun, 6 Oct 2013 23:06:59 +0000 (18:06 -0500)]
machine layer: explicit use of 32 bit flags for netlrts

8 years agoAutomated the parameter selection for qlogic-specific ibverbs parameters if the defau...
Abhishek Gupta [Sat, 5 Oct 2013 06:05:14 +0000 (01:05 -0500)]
Automated the parameter selection for qlogic-specific ibverbs parameters if the default ones fail.

8 years agoin SMP, potentially there us a race condition between rank0 calling initDone,
Gengbin Zheng [Fri, 4 Oct 2013 17:40:24 +0000 (12:40 -0500)]
in SMP, potentially there us a race condition between rank0 calling initDone,
which sendTriggers, and PE 0 calls bdcastRO which broadcast readonlys if this
readonly message arrives later, we need to call trigger again to trigger
initDone() on all ranks. We therefore needs to make sure initDone() is exactly

8 years agoNDMeshStreamer: reworking the interface for termination via QD.
Lukasz Wesolowski [Fri, 4 Oct 2013 03:42:51 +0000 (22:42 -0500)]
NDMeshStreamer: reworking the interface for termination via QD.

8 years agomachine layer: revert changes to GNI layer, default may work
Nikhil Jain [Thu, 3 Oct 2013 18:49:35 +0000 (11:49 -0700)]
machine layer: revert changes to GNI layer, default may work

8 years agoampi: temporarily handle IGNORED status using variables on stack
Nikhil Jain [Thu, 3 Oct 2013 02:38:55 +0000 (21:38 -0500)]
ampi: temporarily handle IGNORED status using variables on stack

8 years agomachinelayer: comment out verbose print
Nikhil Jain [Thu, 3 Oct 2013 02:36:45 +0000 (21:36 -0500)]
machinelayer: comment out verbose print

8 years agopartition: add API to support parameter setup during application execution
Nikhil Jain [Wed, 2 Oct 2013 22:12:11 +0000 (17:12 -0500)]
partition: add API to support parameter setup during application execution

partition: add missing files, and remove the unnecessary .h file

partition: bug fixes in the interface

8 years agopartition: patch to add recursive bisection strategy for partitioning
Jim Phillips [Wed, 2 Oct 2013 19:22:15 +0000 (14:22 -0500)]
partition: patch to add recursive bisection strategy for partitioning

8 years agomachine layer: add a separate variable for system libraries - needed to give preferance
Nikhil Jain [Wed, 2 Oct 2013 06:02:31 +0000 (06:02 +0000)]
machine layer: add a separate variable for system libraries - needed to give preferance
to charm's inc directory

8 years agoisomalloc: correct report randomized stack pointer - anything nonzero is
Nikhil Jain [Wed, 2 Oct 2013 04:05:52 +0000 (23:05 -0500)]
isomalloc: correct report randomized stack pointer - anything nonzero is
randomized stack in modern OS