From joel.sherrill at oarcorp.com Fri May 11 20:28:19 2007 From: joel.sherrill at oarcorp.com (Joel Sherrill) Date: Fri, 11 May 2007 15:28:19 -0500 Subject: Commit Alert on CVS Message-ID: <4644D1E3.8080703@oarcorp.com> Hi, I just finished up a test sweep on sis to address issues from a series of changes I had made over the past week. All tests pass now on sis and have been spot checked on psim but you should be aware of the following changes in CVS. These WILL be in the 4.8 release series. + optional manager support now really reduces the size of your executable + Region manager is optional now that the malloc implementation does not use it. + Extension manager is optional now that the newlib extension is installed via confdefs.h as a static initialization time extension. + Semaphore manager is now optional. Any application using termios or libio needs it so be warned. I changed most of the test Makefiles to account for this. + Stack checker API cleaned up and documented in User's Guide. + CPU Usage API cleaned and documented in User's Guide. Not part of this set of commits but something to be aware of, RTEMS now supports nanosecond granularity on time of day. If the BSP Clock Driver provides a nanoseconds since last tick handler, any call that asks for TOD will be at nanosecond rather than tick granularity. At this point, only the erc32/sis BSP and the mpc6xx clock driver support this. [Can someone check the mpc6xx on real hardware? I did it on a simulator and don't completely trust it.] For those tracking things, sis's minimum is now almost 40% smaller than it was for 4.6.6. hello is showing about 15%. This does have an impact on larger applications. psx01 dropped from 144.5K to 116.9K. BIG WARNING: If you have unsubmitted work that you would like to be considered for inclusion in 4.8, you need to speak up. --joel From Thomas.Doerfler at embedded-brains.de Fri May 11 20:36:37 2007 From: Thomas.Doerfler at embedded-brains.de (Thomas Doerfler) Date: Fri, 11 May 2007 22:36:37 +0200 Subject: Commit Alert on CVS In-Reply-To: <4644D1E3.8080703@oarcorp.com> References: <4644D1E3.8080703@oarcorp.com> Message-ID: <4644D3D5.7050204@embedded-brains.de> Joel, > > BIG WARNING: If you have unsubmitted work that you would > like to be considered for inclusion in 4.8, you need to > speak up. UP. I have another BIG bunch of changes on the PowerPC exception handling code. I have gathered the code that has been copied in various CPU derivates and/or BSPs, moved it to libcpu/powerpc/new-exception and libbsp/powerpc/shared. It is not yet in a state I want to commit this, and although I am quite sure it is a BIG improvement concerning maintainence, I am sure these changes will possibly break some of the existing BSPs. I have developed them based on a CVS version about 6 weeks ago, so my next steps will be to: - update them against the current CVS HEAD - test them on several PPC BSPs: gen5200, mbx8xx, mpc750 (on a MVME2432 board). - I would like to get some feedback for other boards soon after I have commited them to CVS. (volunteers?) wkr, Thomas. > > --joel > _______________________________________________ > rtems-users mailing list > rtems-users at rtems.com > http://rtems.rtems.org/mailman/listinfo/rtems-users -- -------------------------------------------- Embedded Brains GmbH Thomas Doerfler Obere Lagerstr. 30 D-82178 Puchheim Germany email: Thomas.Doerfler at embedded-brains.de Phone: +49-89-18908079-2 Fax: +49-89-18908079-9 From joel.sherrill at oarcorp.com Tue May 15 20:54:45 2007 From: joel.sherrill at oarcorp.com (Joel Sherrill) Date: Tue, 15 May 2007 15:54:45 -0500 Subject: Rate Monotonic Manager Changes Message-ID: <464A1E15.7090605@oarcorp.com> Hi, I just committed a changes to the CVS head that integrates the libmisc/rtmonuse functionality into the Rate Monotonic Manager. The functionality is equivalent except it is more transparent to the user and documented in the User's Guide. This commit added the following directives: rtems_rate_monotonic_get_statistics rtems_rate_monotonic_reset_statistics rtems_rate_montonic_reset_all_statistics rtems_rate_montonic_report_statistics rtems_object_get_name rtems_object_get_name is a generalization of code that is in a lot of the support libraries. It does its best to take an object id and return the name for that object in string form. It is reentrant as the user must provide a buffer and length. This obsoleted the following: + File: rtems/rtmonuse.h + Method: Period_usage_Initialize + Method: Period_usage_Reset + Method: Period_usage_Update + Method: Period_usage_Dump Users of the Rate Monotonic Usage support code have some API changes between 4.7 and 4.8. Applications changes should be: + Do not include rtems/rtmonuse.h + Delete calls to Period_usage_Initialize and Period_usage_Update. + Call rtems_rate_montonic_reset_all_statistics intead of Period_usage_Reset + Call rtems_rate_montonic_report_statistics instead of Period_usage_Dump --joel From joel.sherrill at oarcorp.com Fri May 11 20:28:19 2007 From: joel.sherrill at oarcorp.com (Joel Sherrill) Date: Fri, 11 May 2007 15:28:19 -0500 Subject: Commit Alert on CVS Message-ID: <4644D1E3.8080703@oarcorp.com> Hi, I just finished up a test sweep on sis to address issues from a series of changes I had made over the past week. All tests pass now on sis and have been spot checked on psim but you should be aware of the following changes in CVS. These WILL be in the 4.8 release series. + optional manager support now really reduces the size of your executable + Region manager is optional now that the malloc implementation does not use it. + Extension manager is optional now that the newlib extension is installed via confdefs.h as a static initialization time extension. + Semaphore manager is now optional. Any application using termios or libio needs it so be warned. I changed most of the test Makefiles to account for this. + Stack checker API cleaned up and documented in User's Guide. + CPU Usage API cleaned and documented in User's Guide. Not part of this set of commits but something to be aware of, RTEMS now supports nanosecond granularity on time of day. If the BSP Clock Driver provides a nanoseconds since last tick handler, any call that asks for TOD will be at nanosecond rather than tick granularity. At this point, only the erc32/sis BSP and the mpc6xx clock driver support this. [Can someone check the mpc6xx on real hardware? I did it on a simulator and don't completely trust it.] For those tracking things, sis's minimum is now almost 40% smaller than it was for 4.6.6. hello is showing about 15%. This does have an impact on larger applications. psx01 dropped from 144.5K to 116.9K. BIG WARNING: If you have unsubmitted work that you would like to be considered for inclusion in 4.8, you need to speak up. --joel From Thomas.Doerfler at embedded-brains.de Fri May 11 20:36:37 2007 From: Thomas.Doerfler at embedded-brains.de (Thomas Doerfler) Date: Fri, 11 May 2007 22:36:37 +0200 Subject: Commit Alert on CVS In-Reply-To: <4644D1E3.8080703@oarcorp.com> References: <4644D1E3.8080703@oarcorp.com> Message-ID: <4644D3D5.7050204@embedded-brains.de> Joel, > > BIG WARNING: If you have unsubmitted work that you would > like to be considered for inclusion in 4.8, you need to > speak up. UP. I have another BIG bunch of changes on the PowerPC exception handling code. I have gathered the code that has been copied in various CPU derivates and/or BSPs, moved it to libcpu/powerpc/new-exception and libbsp/powerpc/shared. It is not yet in a state I want to commit this, and although I am quite sure it is a BIG improvement concerning maintainence, I am sure these changes will possibly break some of the existing BSPs. I have developed them based on a CVS version about 6 weeks ago, so my next steps will be to: - update them against the current CVS HEAD - test them on several PPC BSPs: gen5200, mbx8xx, mpc750 (on a MVME2432 board). - I would like to get some feedback for other boards soon after I have commited them to CVS. (volunteers?) wkr, Thomas. > > --joel > _______________________________________________ > rtems-users mailing list > rtems-users at rtems.com > http://rtems.rtems.org/mailman/listinfo/rtems-users -- -------------------------------------------- Embedded Brains GmbH Thomas Doerfler Obere Lagerstr. 30 D-82178 Puchheim Germany email: Thomas.Doerfler at embedded-brains.de Phone: +49-89-18908079-2 Fax: +49-89-18908079-9 From joel.sherrill at oarcorp.com Tue May 15 20:54:45 2007 From: joel.sherrill at oarcorp.com (Joel Sherrill) Date: Tue, 15 May 2007 15:54:45 -0500 Subject: Rate Monotonic Manager Changes Message-ID: <464A1E15.7090605@oarcorp.com> Hi, I just committed a changes to the CVS head that integrates the libmisc/rtmonuse functionality into the Rate Monotonic Manager. The functionality is equivalent except it is more transparent to the user and documented in the User's Guide. This commit added the following directives: rtems_rate_monotonic_get_statistics rtems_rate_monotonic_reset_statistics rtems_rate_montonic_reset_all_statistics rtems_rate_montonic_report_statistics rtems_object_get_name rtems_object_get_name is a generalization of code that is in a lot of the support libraries. It does its best to take an object id and return the name for that object in string form. It is reentrant as the user must provide a buffer and length. This obsoleted the following: + File: rtems/rtmonuse.h + Method: Period_usage_Initialize + Method: Period_usage_Reset + Method: Period_usage_Update + Method: Period_usage_Dump Users of the Rate Monotonic Usage support code have some API changes between 4.7 and 4.8. Applications changes should be: + Do not include rtems/rtmonuse.h + Delete calls to Period_usage_Initialize and Period_usage_Update. + Call rtems_rate_montonic_reset_all_statistics intead of Period_usage_Reset + Call rtems_rate_montonic_report_statistics instead of Period_usage_Dump --joel From joel.sherrill at oarcorp.com Fri May 11 20:28:19 2007 From: joel.sherrill at oarcorp.com (Joel Sherrill) Date: Fri, 11 May 2007 15:28:19 -0500 Subject: Commit Alert on CVS Message-ID: <4644D1E3.8080703@oarcorp.com> Hi, I just finished up a test sweep on sis to address issues from a series of changes I had made over the past week. All tests pass now on sis and have been spot checked on psim but you should be aware of the following changes in CVS. These WILL be in the 4.8 release series. + optional manager support now really reduces the size of your executable + Region manager is optional now that the malloc implementation does not use it. + Extension manager is optional now that the newlib extension is installed via confdefs.h as a static initialization time extension. + Semaphore manager is now optional. Any application using termios or libio needs it so be warned. I changed most of the test Makefiles to account for this. + Stack checker API cleaned up and documented in User's Guide. + CPU Usage API cleaned and documented in User's Guide. Not part of this set of commits but something to be aware of, RTEMS now supports nanosecond granularity on time of day. If the BSP Clock Driver provides a nanoseconds since last tick handler, any call that asks for TOD will be at nanosecond rather than tick granularity. At this point, only the erc32/sis BSP and the mpc6xx clock driver support this. [Can someone check the mpc6xx on real hardware? I did it on a simulator and don't completely trust it.] For those tracking things, sis's minimum is now almost 40% smaller than it was for 4.6.6. hello is showing about 15%. This does have an impact on larger applications. psx01 dropped from 144.5K to 116.9K. BIG WARNING: If you have unsubmitted work that you would like to be considered for inclusion in 4.8, you need to speak up. --joel From Thomas.Doerfler at embedded-brains.de Fri May 11 20:36:37 2007 From: Thomas.Doerfler at embedded-brains.de (Thomas Doerfler) Date: Fri, 11 May 2007 22:36:37 +0200 Subject: Commit Alert on CVS In-Reply-To: <4644D1E3.8080703@oarcorp.com> References: <4644D1E3.8080703@oarcorp.com> Message-ID: <4644D3D5.7050204@embedded-brains.de> Joel, > > BIG WARNING: If you have unsubmitted work that you would > like to be considered for inclusion in 4.8, you need to > speak up. UP. I have another BIG bunch of changes on the PowerPC exception handling code. I have gathered the code that has been copied in various CPU derivates and/or BSPs, moved it to libcpu/powerpc/new-exception and libbsp/powerpc/shared. It is not yet in a state I want to commit this, and although I am quite sure it is a BIG improvement concerning maintainence, I am sure these changes will possibly break some of the existing BSPs. I have developed them based on a CVS version about 6 weeks ago, so my next steps will be to: - update them against the current CVS HEAD - test them on several PPC BSPs: gen5200, mbx8xx, mpc750 (on a MVME2432 board). - I would like to get some feedback for other boards soon after I have commited them to CVS. (volunteers?) wkr, Thomas. > > --joel > _______________________________________________ > rtems-users mailing list > rtems-users at rtems.com > http://rtems.rtems.org/mailman/listinfo/rtems-users -- -------------------------------------------- Embedded Brains GmbH Thomas Doerfler Obere Lagerstr. 30 D-82178 Puchheim Germany email: Thomas.Doerfler at embedded-brains.de Phone: +49-89-18908079-2 Fax: +49-89-18908079-9 From joel.sherrill at oarcorp.com Tue May 15 20:54:45 2007 From: joel.sherrill at oarcorp.com (Joel Sherrill) Date: Tue, 15 May 2007 15:54:45 -0500 Subject: Rate Monotonic Manager Changes Message-ID: <464A1E15.7090605@oarcorp.com> Hi, I just committed a changes to the CVS head that integrates the libmisc/rtmonuse functionality into the Rate Monotonic Manager. The functionality is equivalent except it is more transparent to the user and documented in the User's Guide. This commit added the following directives: rtems_rate_monotonic_get_statistics rtems_rate_monotonic_reset_statistics rtems_rate_montonic_reset_all_statistics rtems_rate_montonic_report_statistics rtems_object_get_name rtems_object_get_name is a generalization of code that is in a lot of the support libraries. It does its best to take an object id and return the name for that object in string form. It is reentrant as the user must provide a buffer and length. This obsoleted the following: + File: rtems/rtmonuse.h + Method: Period_usage_Initialize + Method: Period_usage_Reset + Method: Period_usage_Update + Method: Period_usage_Dump Users of the Rate Monotonic Usage support code have some API changes between 4.7 and 4.8. Applications changes should be: + Do not include rtems/rtmonuse.h + Delete calls to Period_usage_Initialize and Period_usage_Update. + Call rtems_rate_montonic_reset_all_statistics intead of Period_usage_Reset + Call rtems_rate_montonic_report_statistics instead of Period_usage_Dump --joel From joel.sherrill at oarcorp.com Fri May 11 20:28:19 2007 From: joel.sherrill at oarcorp.com (Joel Sherrill) Date: Fri, 11 May 2007 15:28:19 -0500 Subject: Commit Alert on CVS Message-ID: <4644D1E3.8080703@oarcorp.com> Hi, I just finished up a test sweep on sis to address issues from a series of changes I had made over the past week. All tests pass now on sis and have been spot checked on psim but you should be aware of the following changes in CVS. These WILL be in the 4.8 release series. + optional manager support now really reduces the size of your executable + Region manager is optional now that the malloc implementation does not use it. + Extension manager is optional now that the newlib extension is installed via confdefs.h as a static initialization time extension. + Semaphore manager is now optional. Any application using termios or libio needs it so be warned. I changed most of the test Makefiles to account for this. + Stack checker API cleaned up and documented in User's Guide. + CPU Usage API cleaned and documented in User's Guide. Not part of this set of commits but something to be aware of, RTEMS now supports nanosecond granularity on time of day. If the BSP Clock Driver provides a nanoseconds since last tick handler, any call that asks for TOD will be at nanosecond rather than tick granularity. At this point, only the erc32/sis BSP and the mpc6xx clock driver support this. [Can someone check the mpc6xx on real hardware? I did it on a simulator and don't completely trust it.] For those tracking things, sis's minimum is now almost 40% smaller than it was for 4.6.6. hello is showing about 15%. This does have an impact on larger applications. psx01 dropped from 144.5K to 116.9K. BIG WARNING: If you have unsubmitted work that you would like to be considered for inclusion in 4.8, you need to speak up. --joel From Thomas.Doerfler at embedded-brains.de Fri May 11 20:36:37 2007 From: Thomas.Doerfler at embedded-brains.de (Thomas Doerfler) Date: Fri, 11 May 2007 22:36:37 +0200 Subject: Commit Alert on CVS In-Reply-To: <4644D1E3.8080703@oarcorp.com> References: <4644D1E3.8080703@oarcorp.com> Message-ID: <4644D3D5.7050204@embedded-brains.de> Joel, > > BIG WARNING: If you have unsubmitted work that you would > like to be considered for inclusion in 4.8, you need to > speak up. UP. I have another BIG bunch of changes on the PowerPC exception handling code. I have gathered the code that has been copied in various CPU derivates and/or BSPs, moved it to libcpu/powerpc/new-exception and libbsp/powerpc/shared. It is not yet in a state I want to commit this, and although I am quite sure it is a BIG improvement concerning maintainence, I am sure these changes will possibly break some of the existing BSPs. I have developed them based on a CVS version about 6 weeks ago, so my next steps will be to: - update them against the current CVS HEAD - test them on several PPC BSPs: gen5200, mbx8xx, mpc750 (on a MVME2432 board). - I would like to get some feedback for other boards soon after I have commited them to CVS. (volunteers?) wkr, Thomas. > > --joel > _______________________________________________ > rtems-users mailing list > rtems-users at rtems.com > http://rtems.rtems.org/mailman/listinfo/rtems-users -- -------------------------------------------- Embedded Brains GmbH Thomas Doerfler Obere Lagerstr. 30 D-82178 Puchheim Germany email: Thomas.Doerfler at embedded-brains.de Phone: +49-89-18908079-2 Fax: +49-89-18908079-9 From joel.sherrill at oarcorp.com Tue May 15 20:54:45 2007 From: joel.sherrill at oarcorp.com (Joel Sherrill) Date: Tue, 15 May 2007 15:54:45 -0500 Subject: Rate Monotonic Manager Changes Message-ID: <464A1E15.7090605@oarcorp.com> Hi, I just committed a changes to the CVS head that integrates the libmisc/rtmonuse functionality into the Rate Monotonic Manager. The functionality is equivalent except it is more transparent to the user and documented in the User's Guide. This commit added the following directives: rtems_rate_monotonic_get_statistics rtems_rate_monotonic_reset_statistics rtems_rate_montonic_reset_all_statistics rtems_rate_montonic_report_statistics rtems_object_get_name rtems_object_get_name is a generalization of code that is in a lot of the support libraries. It does its best to take an object id and return the name for that object in string form. It is reentrant as the user must provide a buffer and length. This obsoleted the following: + File: rtems/rtmonuse.h + Method: Period_usage_Initialize + Method: Period_usage_Reset + Method: Period_usage_Update + Method: Period_usage_Dump Users of the Rate Monotonic Usage support code have some API changes between 4.7 and 4.8. Applications changes should be: + Do not include rtems/rtmonuse.h + Delete calls to Period_usage_Initialize and Period_usage_Update. + Call rtems_rate_montonic_reset_all_statistics intead of Period_usage_Reset + Call rtems_rate_montonic_report_statistics instead of Period_usage_Dump --joel