diff options
author | Chris Cleeland <chris.cleeland@gmail.com> | 2003-01-22 17:23:07 +0000 |
---|---|---|
committer | Chris Cleeland <chris.cleeland@gmail.com> | 2003-01-22 17:23:07 +0000 |
commit | aa9ee633099dd7a4fe358a36f37ba338ed997589 (patch) | |
tree | 696110768d7ada931eb5b7c024e038165fda7138 | |
parent | 57a291ce50e05aa840f8b038745cdca464043d2a (diff) | |
download | ATCD-aa9ee633099dd7a4fe358a36f37ba338ed997589.tar.gz |
Wed Jan 22 11:08:57 2003 Chris Cleeland <cleeland_c@ociweb.com>
-rw-r--r-- | TAO/ChangeLog | 534 | ||||
-rw-r--r-- | TAO/tao/Messaging/Messaging_Policy_i.cpp | 4 |
2 files changed, 271 insertions, 267 deletions
diff --git a/TAO/ChangeLog b/TAO/ChangeLog index ac9ecf5d046..481c27c09ce 100644 --- a/TAO/ChangeLog +++ b/TAO/ChangeLog @@ -1,3 +1,9 @@ +Wed Jan 22 11:08:57 2003 Chris Cleeland <cleeland_c@ociweb.com> + + * tao/Messaging/Messaging_Policy_i.cpp (hook): Fixed debug message + so that it actually printed out milliseconds rather than the + microsecond component expressed as milliseconds. + Wed Jan 22 17:03:12 UTC 2003 Johnny Willemsen <jwillemsen@remedy.nl> * orbsvcs/orbsvcs/PortableGroup/UIPMC_Transport.cpp: @@ -7,8 +13,8 @@ Wed Jan 22 17:03:12 UTC 2003 Johnny Willemsen <jwillemsen@remedy.nl> Wed Jan 22 16:34:32 UTC 2003 Johnny Willemsen <jwillemsen@remedy.nl> - * Makefile.RTCORBAEvent: - * Makefile.RTSchedEvent: + * orbsvcs/orbsvcs/Makefile.RTCORBAEvent: + * orbsvcs/orbsvcs/Makefile.RTSchedEvent: Set the correct x_BUILD_DLL defines. These are needed for the Cygwin/MingW and Kylix compiler. @@ -16,16 +22,16 @@ Tue Jan 21 15:31:46 2003 Jeff Parsons <j.parsons@vanderbilt.edu> * TAO_IDL/fe/lex.yy.cpp: - Modified code handling #pragma version directive to handle - the case where one or more of the spaces between the words - in the directive contains a tab. Thanks to Dima Scub - <dima_scub@sniip.ru> for reporting the bug. This closes - [BUGID:1425]. + Modified code handling #pragma version directive to handle + the case where one or more of the spaces between the words + in the directive contains a tab. Thanks to Dima Scub + <dima_scub@sniip.ru> for reporting the bug. This closes + [BUGID:1425]. - * TAO_IDL/fe/idl.ll: + * TAO_IDL/fe/idl.ll: - Changed the file that generates lex.yy.cpp to reflect the - change above. + Changed the file that generates lex.yy.cpp to reflect the + change above. Tue Jan 21 19:28:12 UTC 2003 Johnny Willemsen <jwillemsen@remedy.nl> @@ -36,72 +42,72 @@ Tue Jan 21 19:28:12 UTC 2003 Johnny Willemsen <jwillemsen@remedy.nl> Tue Jan 21 10:38:48 2003 Boris S Kolpackov <boris@isis-server.isis.vanderbilt.edu> - * orbsvcs/tests/Security/Crash_Test/run_test.pl: fixed few - race conditions. + * orbsvcs/tests/Security/Crash_Test/run_test.pl: fixed few + race conditions. Mon Jan 20 18:46:17 2003 Nanbor Wang <nanbor@cs.wustl.edu> - * orbsvcs/ImplRepo_Service/README.txt (ImplRepoService): Fixed - incorrect usages of InitRef ID ImplRepo_Service which should - really be ImplRepoService. + * orbsvcs/ImplRepo_Service/README.txt (ImplRepoService): Fixed + incorrect usages of InitRef ID ImplRepo_Service which should + really be ImplRepoService. - * orbsvcs/ImplRepo_Service/tao_imr_i.cpp (run): - * orbsvcs/ImplRepo_Service/ImR_Activator_i.cpp (init): On Win32, - changed to use ACE_OS::hostname to get the "machine name" and - then use the name to acquire the full DNS name of the machine. - Without this extra step, ImpoRepo uses the all-caps, short form - name of the machine, which causes a mismatch with the incoming - corbaloc endpoint name. Perhaps ACE_OS::hostname needs to be - fixed on Win32 to return. These changes allow ImplRepo to - startup correctly on Win32. + * orbsvcs/ImplRepo_Service/tao_imr_i.cpp (run): + * orbsvcs/ImplRepo_Service/ImR_Activator_i.cpp (init): On Win32, + changed to use ACE_OS::hostname to get the "machine name" and + then use the name to acquire the full DNS name of the machine. + Without this extra step, ImpoRepo uses the all-caps, short form + name of the machine, which causes a mismatch with the incoming + corbaloc endpoint name. Perhaps ACE_OS::hostname needs to be + fixed on Win32 to return. These changes allow ImplRepo to + startup correctly on Win32. - * orbsvcs/ImplRepo_Service/ImR_Locator_i.cpp (server_is_running): - The following change seemed to be missing for some reason. - Figured out what to do and fixed it. + * orbsvcs/ImplRepo_Service/ImR_Locator_i.cpp (server_is_running): + The following change seemed to be missing for some reason. + Figured out what to do and fixed it. - Mon Nov 11 09:28:43 2002 Priyanka Gontla <pgontla@ociweb.com> + Mon Nov 11 09:28:43 2002 Priyanka Gontla <pgontla@ociweb.com> - * orbsvcs/ImplRepo_Service/ImR_Locator_i.cpp: - Get the whole hostname name including the domain instead - of just the hostname. - Thanks to Kobi Cohen-Arazi <kobi@mivzak.com> for pointing - out a exception thats raised. + * orbsvcs/ImplRepo_Service/ImR_Locator_i.cpp: + Get the whole hostname name including the domain instead + of just the hostname. + Thanks to Kobi Cohen-Arazi <kobi@mivzak.com> for pointing + out a exception thats raised. Mon Jan 20 18:05:26 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> - * tao/LF_Connect_Strategy.cpp: Removed debugging statements that - needed access to the transport object. The transport object - could dissappear even before the thread gets to this point and - it is only chaos after that. + * tao/LF_Connect_Strategy.cpp: Removed debugging statements that + needed access to the transport object. The transport object + could dissappear even before the thread gets to this point and + it is only chaos after that. - * tao/IIOP_Connector.cpp: - * tao/Strategies/UIOP_Connector.cpp: - * orbsvcs/orbsvcs/SSLIOP/IIOP_SSL_Connector.cpp (make_connection): - * orbsvcs/orbsvcs/SSLIOP/SSLIOP_Connector.cpp: Added debugging - statements to make up for the loss from the above change. + * tao/IIOP_Connector.cpp: + * tao/Strategies/UIOP_Connector.cpp: + * orbsvcs/orbsvcs/SSLIOP/IIOP_SSL_Connector.cpp (make_connection): + * orbsvcs/orbsvcs/SSLIOP/SSLIOP_Connector.cpp: Added debugging + statements to make up for the loss from the above change. - * tao/Leader_Follower.cpp: Added comments and also made sure that - we access the transport object only when debugging is turned on. + * tao/Leader_Follower.cpp: Added comments and also made sure that + we access the transport object only when debugging is turned on. - Thanks to Duane Binder for sharing his purify logs from his runs - that he ran for a long time (which wouldnt have been possible at - all for us) + Thanks to Duane Binder for sharing his purify logs from his runs + that he ran for a long time (which wouldnt have been possible at + all for us) Mon Jan 20 16:42:34 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> - * tao/Muxed_TMS.cpp (bind_dispatcher): Added a lock to synchronize - acces to the map. It *was* initially thought that - synchronization will not be required since we always believed - that only one thread could be active in the map. This is *not* - true since a thread could be accessing the transport from the - invocation side whereas another thread could be accessing the - transport and the map underneath from the LF. This is really - really bad. + * tao/Muxed_TMS.cpp (bind_dispatcher): Added a lock to synchronize + acces to the map. It *was* initially thought that + synchronization will not be required since we always believed + that only one thread could be active in the map. This is *not* + true since a thread could be accessing the transport from the + invocation side whereas another thread could be accessing the + transport and the map underneath from the LF. This is really + really bad. Mon Jan 20 13:32:30 2003 Nanbor Wang <nanbor@cs.wustl.edu> - * orbsvcs/ImplRepo_Service/README.txt: Fixed a broken HTML tag. + * orbsvcs/ImplRepo_Service/README.txt: Fixed a broken HTML tag. Mon Jan 20 19:20:32 UTC 2003 Johnny Willemsen <jwillemsen@remedy.nl> @@ -121,13 +127,13 @@ Mon Jan 20 12:03:03 2003 Jeff Parsons <j.parsons@vanderbilt.edu> * tao/DynamicInterface/Request.cpp: - Fixed a bug in poll_response(). Thanks to Jaroslaw Nozderko - <jarek@eko.net.pl> for reporting it. + Fixed a bug in poll_response(). Thanks to Jaroslaw Nozderko + <jarek@eko.net.pl> for reporting it. Mon Jan 20 08:00:44 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> - * tao/Connection_Handler.cpp: Removed multiple inclusions of an - header. + * tao/Connection_Handler.cpp: Removed multiple inclusions of an + header. Mon Jan 20 10:00:22 UTC 2003 Johnny Willemsen <jwillemsen@remedy.nl> @@ -149,19 +155,19 @@ Sun Jan 19 16:56:12 UTC 2003 Johnny Willemsen <jwillemsen@remedy.nl> Sat Jan 18 12:07:14 2003 Chris Cleeland <cleeland_c@ociweb.com> - * orbsvcs/orbsvcs/Notify/Properties.cpp: Fixed my fix by removing - a #define I used to test the fix. + * orbsvcs/orbsvcs/Notify/Properties.cpp: Fixed my fix by removing + a #define I used to test the fix. Sat Jan 18 10:40:58 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> - * tao/orbconf.h: Added TAO_RD_TABLE_SIZE and set the value to be - 16. + * tao/orbconf.h: Added TAO_RD_TABLE_SIZE and set the value to be + 16. - * tao/Muxed_TMS.cpp: - * tao/Muxed_TMS.h: Used the TAO_RD_TABLE_SIZE to configure the - internal hash map. This is now hard coded. We should have a way - for the user to configure the size of these hash maps and it will - be integrated after 1.3.2. + * tao/Muxed_TMS.cpp: + * tao/Muxed_TMS.h: Used the TAO_RD_TABLE_SIZE to configure the + internal hash map. This is now hard coded. We should have a way + for the user to configure the size of these hash maps and it will + be integrated after 1.3.2. Sat Jan 18 09:08:12 UTC 2003 Johnny Willemsen <jwillemsen@remedy.nl> @@ -176,92 +182,92 @@ Sat Jan 18 09:08:12 UTC 2003 Johnny Willemsen <jwillemsen@remedy.nl> Fri Jan 17 14:22:48 2003 Chris Cleeland <cleeland_c@ociweb.com> - * orbsvcs/orbsvcs/Notify/Properties.cpp: Fixed explicit - instantiation to instantiate TAO_Singleton rather than - ACE_Singleton. This fixes the RH71_Explicit_Templates build. + * orbsvcs/orbsvcs/Notify/Properties.cpp: Fixed explicit + instantiation to instantiate TAO_Singleton rather than + ACE_Singleton. This fixes the RH71_Explicit_Templates build. Fri Jan 17 08:34:57 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> - * orbsvcs/orbsvcs/Event/EC_Gateway_UDP.h: Removed an extra - ';'. This may nor may not help the Borland builds. + * orbsvcs/orbsvcs/Event/EC_Gateway_UDP.h: Removed an extra + ';'. This may nor may not help the Borland builds. Thu Jan 16 17:24:30 2003 Pradeep Gore <pradeep@oomworks.com> - * orbsvcs/orbsvcs/Notify/ETCL_FilterFactory.cpp: - * orbsvcs/orbsvcs/Notify/ETCL_FilterFactory.h: - Make the FilterFactory a RefCounted servant. - This allows us to cleanup this object when the ChannelFactory - dies. - - * orbsvcs/orbsvcs/NotifyExt.idl: - * orbsvcs/orbsvcs/Notify/EventChannelFactory.cpp: - * orbsvcs/orbsvcs/Notify/EventChannelFactory.h: - Extend the ChannelFactory interface to include a destroy method. - This allows us to check for memory leaks using a tool like - purify when the channel factory is destroyed. - - * orbsvcs/orbsvcs/Notify/EventType.cpp: - * orbsvcs/orbsvcs/Notify/EventType.h: - * orbsvcs/orbsvcs/Notify/EventTypeSeq.cpp: - * orbsvcs/orbsvcs/Notify/EventTypeSeq.h: - Made the <dump> methods const. - - * orbsvcs/orbsvcs/Notify/Object.cpp: - * orbsvcs/orbsvcs/Notify/Object_T.cpp: - * orbsvcs/orbsvcs/Notify/Object_T.h: - The <shutdown_> flag is now set in the Object_T::destroy method. - Note that if multiple threads that encounter dispatch failure - try to destroy the proxy object, only the first one will run the - destroy method, others will check for the shutdown flag and - return. This fixes the random crashes in some of the Linux - builds on the scoreboard (orbsvcs/tests/Notify/MT_Dispatching). - - * orbsvcs/orbsvcs/Notify/Proxy.cpp: - * orbsvcs/orbsvcs/Notify/Proxy.h: - Changed the <subscribed_types> method to return a copy of the - curent subscriptions. This method has been corrected to be - thread safe. - - * orbsvcs/orbsvcs/Notify/Peer.cpp: - Changed to new usage of Proxy::subscribed_types. - - * orbsvcs/orbsvcs/Notify/Properties.h: - Use TAO_Singleton instead of ACE_Singleton. + * orbsvcs/orbsvcs/Notify/ETCL_FilterFactory.cpp: + * orbsvcs/orbsvcs/Notify/ETCL_FilterFactory.h: + Make the FilterFactory a RefCounted servant. + This allows us to cleanup this object when the ChannelFactory + dies. + + * orbsvcs/orbsvcs/NotifyExt.idl: + * orbsvcs/orbsvcs/Notify/EventChannelFactory.cpp: + * orbsvcs/orbsvcs/Notify/EventChannelFactory.h: + Extend the ChannelFactory interface to include a destroy method. + This allows us to check for memory leaks using a tool like + purify when the channel factory is destroyed. + + * orbsvcs/orbsvcs/Notify/EventType.cpp: + * orbsvcs/orbsvcs/Notify/EventType.h: + * orbsvcs/orbsvcs/Notify/EventTypeSeq.cpp: + * orbsvcs/orbsvcs/Notify/EventTypeSeq.h: + Made the <dump> methods const. + + * orbsvcs/orbsvcs/Notify/Object.cpp: + * orbsvcs/orbsvcs/Notify/Object_T.cpp: + * orbsvcs/orbsvcs/Notify/Object_T.h: + The <shutdown_> flag is now set in the Object_T::destroy method. + Note that if multiple threads that encounter dispatch failure + try to destroy the proxy object, only the first one will run the + destroy method, others will check for the shutdown flag and + return. This fixes the random crashes in some of the Linux + builds on the scoreboard (orbsvcs/tests/Notify/MT_Dispatching). + + * orbsvcs/orbsvcs/Notify/Proxy.cpp: + * orbsvcs/orbsvcs/Notify/Proxy.h: + Changed the <subscribed_types> method to return a copy of the + curent subscriptions. This method has been corrected to be + thread safe. + + * orbsvcs/orbsvcs/Notify/Peer.cpp: + Changed to new usage of Proxy::subscribed_types. + + * orbsvcs/orbsvcs/Notify/Properties.h: + Use TAO_Singleton instead of ACE_Singleton. Thu Jan 16 16:27:59 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> * orbsvcs/orbsvcs/Event/EC_Gateway_UDP.{h,cpp}: Added patches from - this checkin "Mon Oct 28 18:21:18 CET 2002 Oliver Kellogg + this checkin "Mon Oct 28 18:21:18 CET 2002 Oliver Kellogg <oliver.kellogg@sysde.eads.net>". The patches mysteriously - dissappeared. Thanks to Oliver Kellogg for alerting me. + dissappeared. Thanks to Oliver Kellogg for alerting me. Thu Jan 16 13:25:10 2003 Nanbor Wang <nanbor@cs.wustl.edu> - * docs/rtcorba/features.html: Fixed obsolete references to the - location of various priority mapping source files. + * docs/rtcorba/features.html: Fixed obsolete references to the + location of various priority mapping source files. Thu Jan 16 12:29:00 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> - * COPYING: Extended the validity of the copyright notice from 2002 - to 2003. Should have been done yesterday.., anyway, better late - than never. + * COPYING: Extended the validity of the copyright notice from 2002 + to 2003. Should have been done yesterday.., anyway, better late + than never. Thu Jan 16 10:52:07 2003 Jeff Parsons <j.parsons@vanderbilt.edu> * TAO_IDL/be/be_visitor_interface/smart_proxy_ch.cpp: - * TAO_IDL/be/be_visitor_interface/smart_proxy_cs.cpp: - - In the generated TAO_*_Proxy_Factory_Adapter class, - changed signatures of register_proxy_factory() and - unregister_proxy_factory() to return void. Error - reporting through return type is not necessary, since - native exceptions or an environment variable are - already used, and there is no code in either of these - methods that sets the return value to anything other - than 0. Although the methods are public, the class - itself has a protected constructor, and the methods - are intended for TAO internal use only. - This closes [BUGID:844]. + * TAO_IDL/be/be_visitor_interface/smart_proxy_cs.cpp: + + In the generated TAO_*_Proxy_Factory_Adapter class, + changed signatures of register_proxy_factory() and + unregister_proxy_factory() to return void. Error + reporting through return type is not necessary, since + native exceptions or an environment variable are + already used, and there is no code in either of these + methods that sets the return value to anything other + than 0. Although the methods are public, the class + itself has a protected constructor, and the methods + are intended for TAO internal use only. + This closes [BUGID:844]. Wed Jan 15 18:02:49 2003 Bala <bala@cs.wustl.edu> @@ -269,44 +275,44 @@ Wed Jan 15 18:02:49 2003 Bala <bala@cs.wustl.edu> Tue Jan 14 22:11:14 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> - * tests/Crashed_Callback/client.dsp: - * tests/Crashed_Callback/server.dsp: - * tests/MProfile_Connection_Timeout/client.dsp: - * tests/MT_Timeout/MT_Timeout_Client.dsp: - * tests/MT_Timeout/MT_Timeout_Server.dsp: Fixed the release - configurations. + * tests/Crashed_Callback/client.dsp: + * tests/Crashed_Callback/server.dsp: + * tests/MProfile_Connection_Timeout/client.dsp: + * tests/MT_Timeout/MT_Timeout_Client.dsp: + * tests/MT_Timeout/MT_Timeout_Server.dsp: Fixed the release + configurations. Tue Jan 14 20:46:40 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> - * tests/Crashed_Callback/client.dsp: - * tests/Crashed_Callback/server.dsp: Fixed the release build - configuration. Thanks to Nanbor for pointing this out. + * tests/Crashed_Callback/client.dsp: + * tests/Crashed_Callback/server.dsp: Fixed the release build + configuration. Thanks to Nanbor for pointing this out. Tue Jan 14 16:37:15 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> - * tests/Stack_Recursion/server.dsp: Adding this file to the repo. + * tests/Stack_Recursion/server.dsp: Adding this file to the repo. Tue Jan 14 16:37:44 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> - * orbsvcs/tests/Security/Crash_Test/run_test.pl: The timeout was - adjusted. + * orbsvcs/tests/Security/Crash_Test/run_test.pl: The timeout was + adjusted. Tue Jan 14 15:10:53 2003 Jeff Parsons <j.parsons@vanderbilt.edu> * TAO_IDL/driver/drv_args.cpp: - Added a check and graceful exit for the case where the -H command - line option is not followed by a selection (i.e., is followed by - another dash-prefixed command line option or by nothing). The case - where the selection exists but is invalid was already handled. - Thanks to Chip Jones <jones_c@ociweb.com> for reporting the bug. - This fix closes [BUGID:1422]. + Added a check and graceful exit for the case where the -H command + line option is not followed by a selection (i.e., is followed by + another dash-prefixed command line option or by nothing). The case + where the selection exists but is invalid was already handled. + Thanks to Chip Jones <jones_c@ociweb.com> for reporting the bug. + This fix closes [BUGID:1422]. Tue Jan 14 16:13:09 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> - * orbsvcs/IFR_Service/Makefile.IFR_Service: Added a check for - TAO_IFRService library before going head to actually compile - stuff. + * orbsvcs/IFR_Service/Makefile.IFR_Service: Added a check for + TAO_IFRService library before going head to actually compile + stuff. Tue Jan 14 07:05:21 UTC 2003 Johnny Willemsen <jwillemsen@remedy.nl> @@ -331,9 +337,9 @@ Mon Jan 13 10:12:20 2003 Jeff Parsons <j.parsons@vanderbilt.edu> * TAO_IDL/be/be_visitor_interface/interface_ch.cpp: - Removed generation of - 'friend class <remote/thru-POA/direct proxy class name>' for local - interfaces. + Removed generation of + 'friend class <remote/thru-POA/direct proxy class name>' for local + interfaces. Mon Jan 13 12:47:12 UTC 2003 Johnny Willemsen <jwillemsen@remedy.nl> @@ -348,146 +354,146 @@ Mon Jan 13 12:47:12 UTC 2003 Johnny Willemsen <jwillemsen@remedy.nl> Fri Jan 10 19:36:56 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> - * tao/Transport_Cache_Manager.h: - * tao/Transport_Cache_Manager.cpp: Added a debug - statement. Removed the method close_entries () which was under - #if 0 guard. + * tao/Transport_Cache_Manager.h: + * tao/Transport_Cache_Manager.cpp: Added a debug + statement. Removed the method close_entries () which was under + #if 0 guard. - The method is_entry_idle () was trying to make the transport as - busy, which is horribly horribly bad. The method is_entry_idle - () should only do one thing, ie. check whether the entry is idle - or not! + The method is_entry_idle () was trying to make the transport as + busy, which is horribly horribly bad. The method is_entry_idle + () should only do one thing, ie. check whether the entry is idle + or not! Fri Jan 10 08:28:34 2003 Frank Hunleth <frank@hunleth.com> - * orbsvcs/orbsvcs/PortableGroup/UIPMC_Connection_Handler.cpp: + * orbsvcs/orbsvcs/PortableGroup/UIPMC_Connection_Handler.cpp: - Removed call to unsubscribe from multicast socket as this happens - automatically. Thanks to Don Hinton <dhinton@ieee.org> for - pointing this out. + Removed call to unsubscribe from multicast socket as this happens + automatically. Thanks to Don Hinton <dhinton@ieee.org> for + pointing this out. - * orbsvcs/orbsvcs/PortableGroup/UIPMC_Transport.cpp: - * orbsvcs/orbsvcs/PortableGroup/UIPMC_Transport.h: + * orbsvcs/orbsvcs/PortableGroup/UIPMC_Transport.cpp: + * orbsvcs/orbsvcs/PortableGroup/UIPMC_Transport.h: - Added code to check length of MIOP UniqueID so that it can be - properly parsed. Thanks to Chau Vincent - <Vincent.Chau@fr.thalesgroup.com> for finding the problem when - performing MIOP interoperability testing. Also, added code to - set the UniqueID and packet length fields properly per update to - the MIOP spec and tightened bounds checks on MIOP packets to - prevent buffer overruns. + Added code to check length of MIOP UniqueID so that it can be + properly parsed. Thanks to Chau Vincent + <Vincent.Chau@fr.thalesgroup.com> for finding the problem when + performing MIOP interoperability testing. Also, added code to + set the UniqueID and packet length fields properly per update to + the MIOP spec and tightened bounds checks on MIOP packets to + prevent buffer overruns. - * orbsvcs/tests/Miop/McastHello/client.cpp: + * orbsvcs/tests/Miop/McastHello/client.cpp: - Fixed spelling error. + Fixed spelling error. Thu Jan 9 11:01:24 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> - * tao/try_macros.h: Removed the file from the repo. This is no - longer used and seems like a good time to remove it. Thanks to - Oliver Kellog for reminding us. + * tao/try_macros.h: Removed the file from the repo. This is no + longer used and seems like a good time to remove it. Thanks to + Oliver Kellog for reminding us. Tue Jan 7 09:02:52 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> - * performance-tests/Memory/IORsize/Foo_i.h: Fixed fuzz errors. + * performance-tests/Memory/IORsize/Foo_i.h: Fixed fuzz errors. Mon Jan 6 17:31:03 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> - * tao/Reactive_Connect_Strategy.cpp: - * tao/LF_Connect_Strategy.cpp: Used the cached pointer to access - the ORB_Core, instead of accessing the ORB_Core in the - transport. The reason why this should be done this way is too - involved to explain here, but suffice to say that in highly - multi-threaded environments the transports could dis-sappear - before the connection handlers. Accessing the ORB_Core from the - transport could then lead to vague crashes. + * tao/Reactive_Connect_Strategy.cpp: + * tao/LF_Connect_Strategy.cpp: Used the cached pointer to access + the ORB_Core, instead of accessing the ORB_Core in the + transport. The reason why this should be done this way is too + involved to explain here, but suffice to say that in highly + multi-threaded environments the transports could dis-sappear + before the connection handlers. Accessing the ORB_Core from the + transport could then lead to vague crashes. Mon Jan 6 17:12:40 2003 Balachandran Natarajan <bala@isis-server.isis.vanderbilt.edu> - * performance-tests/Memory/IORsize/client.cpp: - * performance-tests/Memory/IORsize/server.cpp: - * performance-tests/Memory/IORsize/Memory_Growth.h: - * performance-tests/Memory/IORsize/Memory_Growth.cpp: - * performance-tests/Memory/IORsize/Foo_i.h: - * performance-tests/Memory/IORsize/Foo_i.cpp: - * performance-tests/Memory/IORsize/Makefile: - * performance-tests/Memory/IORsize/Test.idl: A simple, again a - very simple test, to measure the memory consumed by the server - for unmarshalling an IOR. This test will not be compiled in all - the builds, but only in the builds where the scripts are run. + * performance-tests/Memory/IORsize/client.cpp: + * performance-tests/Memory/IORsize/server.cpp: + * performance-tests/Memory/IORsize/Memory_Growth.h: + * performance-tests/Memory/IORsize/Memory_Growth.cpp: + * performance-tests/Memory/IORsize/Foo_i.h: + * performance-tests/Memory/IORsize/Foo_i.cpp: + * performance-tests/Memory/IORsize/Makefile: + * performance-tests/Memory/IORsize/Test.idl: A simple, again a + very simple test, to measure the memory consumed by the server + for unmarshalling an IOR. This test will not be compiled in all + the builds, but only in the builds where the scripts are run. - * performance-tests/Memory/Makefile: Added the new test to the - list of directories. + * performance-tests/Memory/Makefile: Added the new test to the + list of directories. Sat Jan 04 12:15:39 2003 Pradeep Gore <pradeep@oomworks.com> - * orbsvcs/Notify_Service/Notify_Server.cpp: - Removed extra ACE_CHECK_RETURN causing build errors. + * orbsvcs/Notify_Service/Notify_Server.cpp: + Removed extra ACE_CHECK_RETURN causing build errors. Fri Jan 3 14:43:56 2003 Jeff Parsons <j.parsons@vanderbilt.edu> * TAO_IDL/util/utl_global.cpp: - * TAO_IDL/include/idl_global.h: + * TAO_IDL/include/idl_global.h: - Changes to the management of the IDL compiler's stack of prefixes, - to address a problem that appeared when there is a chain of - included IDL files. Thanks to Adee Ran <adeeran@yahoo.com> for - reporting the problem and supplying example IDL files to - reproduce it. + Changes to the management of the IDL compiler's stack of prefixes, + to address a problem that appeared when there is a chain of + included IDL files. Thanks to Adee Ran <adeeran@yahoo.com> for + reporting the problem and supplying example IDL files to + reproduce it. Fri Jan 03 14:22:14 2003 Pradeep Gore <pradeep@oomworks.com> - * orbsvcs/Notify_Service/Notify_Server.cpp: - Moved the call to <init> to catch any resulting exceptions and - exit gracefully. - - The following changes are made to fix the crashes observed on the - scoreboard in the $TAO_ROOT/orbsvcs/tests/Notify/Sequence_Multi_Test. - Thanks to Irfan and Bala for helping to fix this problem. - - * orbsvcs/orbsvcs/Notify/Container_T.cpp: - Changed the order of destruction of the Notify objects. The tree - is destroyed bottom up. This ensures that proxy objects that - refer to a valid parent during shutdown. - - * orbsvcs/orbsvcs/Notify/Method_Request.h: - * orbsvcs/orbsvcs/Notify/Method_Request.inl: - * orbsvcs/orbsvcs/Notify/Refcountable.h: - * orbsvcs/orbsvcs/Notify/Refcountable.inl: - Moved the TAO_NS_Refcountable_Guard class from Method_Request.* - files into Refcountable.* files. - - * orbsvcs/orbsvcs/Notify/Object.cpp: - Moved the <worker_task_> and <proxy_poa_> data member shutdown - from the destructor to the <shutdown> method. - - * orbsvcs/orbsvcs/Notify/Object_T.cpp: - Replaced the TAO_NS_Object_RefCount_Guard with - TAO_NS_Refcountable_Guard. - - * orbsvcs/orbsvcs/Notify/Peer.cpp: - Protect the Peer from being destroyed while it is dispatching - updates by using a TAO_NS_Refcountable_Guard. - - * orbsvcs/orbsvcs/Notify/Types.h: - Removed the TAO_NS_Object_RefCount_Guard class. We use - TAO_NS_Refcountable_Guard instead. - - * orbsvcs/orbsvcs/Notify/Sequence/Batch_Buffering_Strategy.cpp: - * orbsvcs/orbsvcs/Notify/Sequence/Batch_Buffering_Strategy.h: - Modified the <dequeue_available> method to return the number of - pending events. - - * orbsvcs/orbsvcs/Notify/Sequence/SequencePushConsumer.cpp: - * orbsvcs/orbsvcs/Notify/Sequence/SequencePushConsumer.h: - + Protect the call to the event batch <push> method with - TAO_NS_Refcountable_Guard so that the proxy (and hence this - peer) is not destroyed while the push is in progress. - + Reset the pacing error on dispatch error so that we don't - schedule a timer on an object that is about to be destroyed. - + Schedule a timer in the <handle_timeout> method only if there - are pending events. + * orbsvcs/Notify_Service/Notify_Server.cpp: + Moved the call to <init> to catch any resulting exceptions and + exit gracefully. + + The following changes are made to fix the crashes observed on the + scoreboard in the $TAO_ROOT/orbsvcs/tests/Notify/Sequence_Multi_Test. + Thanks to Irfan and Bala for helping to fix this problem. + + * orbsvcs/orbsvcs/Notify/Container_T.cpp: + Changed the order of destruction of the Notify objects. The tree + is destroyed bottom up. This ensures that proxy objects that + refer to a valid parent during shutdown. + + * orbsvcs/orbsvcs/Notify/Method_Request.h: + * orbsvcs/orbsvcs/Notify/Method_Request.inl: + * orbsvcs/orbsvcs/Notify/Refcountable.h: + * orbsvcs/orbsvcs/Notify/Refcountable.inl: + Moved the TAO_NS_Refcountable_Guard class from Method_Request.* + files into Refcountable.* files. + + * orbsvcs/orbsvcs/Notify/Object.cpp: + Moved the <worker_task_> and <proxy_poa_> data member shutdown + from the destructor to the <shutdown> method. + + * orbsvcs/orbsvcs/Notify/Object_T.cpp: + Replaced the TAO_NS_Object_RefCount_Guard with + TAO_NS_Refcountable_Guard. + + * orbsvcs/orbsvcs/Notify/Peer.cpp: + Protect the Peer from being destroyed while it is dispatching + updates by using a TAO_NS_Refcountable_Guard. + + * orbsvcs/orbsvcs/Notify/Types.h: + Removed the TAO_NS_Object_RefCount_Guard class. We use + TAO_NS_Refcountable_Guard instead. + + * orbsvcs/orbsvcs/Notify/Sequence/Batch_Buffering_Strategy.cpp: + * orbsvcs/orbsvcs/Notify/Sequence/Batch_Buffering_Strategy.h: + Modified the <dequeue_available> method to return the number of + pending events. + + * orbsvcs/orbsvcs/Notify/Sequence/SequencePushConsumer.cpp: + * orbsvcs/orbsvcs/Notify/Sequence/SequencePushConsumer.h: + + Protect the call to the event batch <push> method with + TAO_NS_Refcountable_Guard so that the proxy (and hence this + peer) is not destroyed while the push is in progress. + + Reset the pacing error on dispatch error so that we don't + schedule a timer on an object that is about to be destroyed. + + Schedule a timer in the <handle_timeout> method only if there + are pending events. Fri Jan 3 07:48:01 2003 Chad Elliott <elliott_c@ociweb.com> diff --git a/TAO/tao/Messaging/Messaging_Policy_i.cpp b/TAO/tao/Messaging/Messaging_Policy_i.cpp index 1e103072f6d..c210c779248 100644 --- a/TAO/tao/Messaging/Messaging_Policy_i.cpp +++ b/TAO/tao/Messaging/Messaging_Policy_i.cpp @@ -80,11 +80,9 @@ TAO_RelativeRoundtripTimeoutPolicy::hook (TAO_ORB_Core *orb_core, if (TAO_debug_level > 0) { - CORBA::ULong msecs = - ACE_static_cast (CORBA::ULong, microseconds / 1000); ACE_DEBUG ((LM_DEBUG, ACE_TEXT ("TAO (%P|%t) Timeout is <%u>\n"), - msecs)); + time_value.msec ())); } } ACE_CATCHANY |