Change Summary for MPICH2 for week of May 06, 2007

Information on all weekly change summaries is available.

List of revisions

• (robl) - fixed some compile/linker errors
• (robl) - needed to define a structure for 'extra_state' information- fixed some compile/linker errors
• (robl) - needed to redo the query and free functions- needed to define a structure for 'extra_state' information- fixed some compile/linker errors
• (robl) - probably only works for UFS (if anything)- still need to fix up code for all other file system drivers
• (robl) - still need to fix up code for all other file system drivers
• (robl) -- MPI implementation does that for us- probably only works for UFS (if anything)- still need to fix up code for all other file system drivers
• (chan) Added CallStack subpackage support.
• (gropp) Added MPI_OFFSET_KIND_VAL to ensure that a valid kind parameter is used in tests of Fortran (before, in some cases an invalid Fortran program was generated with integer (kind=), that is, with an empty string for the kind value)
• (gropp) Added a mpich2version program
• (gropp) Added a send-flood test, particularly for shared memory methods
• (gropp) Added a test to help find memory leaks in type_dup
• (chan) Added extra instruction on configure/make/install with different shells.
• (gropp) Added missing hyphen in prefix
• (gropp) Added more diagnostics when check for stdio linking with Fortran fails
• (gropp) Added support for a private copy of comm work
• (gropp) Also alloc_mem in Fortran
• (gropp) Also fixed Fortran versions of Accumulate, Get, and Put
• (rbutler) Catch case of sending to pmiSock already closed.
• (gropp) Change the test for arrays in the shell to handle the case where the shell execute neither branch of the test
• (gropp) Checked for the non-portable POSIX_MUTEX_ERRORCHECK_NP before using it, we need to fix these thread files in general (it makes no sense to define macros for these low-usage, non-performance-critical operations such as create mutex)
• (gropp) cleaned up thread definitions for portability
• (gropp) Fix some formatting
• (gropp) Fix typo in test for done with finding a buffered send segement
• (gropp) Fix use of address-sized ints in RMA and use only local include files to avoid problems with directory separators in Windows
• (gropp) Fixed the handling of the second arg in Win_create
• (gropp) Fixed typo in stdio with Fortran and Fortran 90 test and added diagnostics to the Fortran 77 version
• (gropp) Fixed use of full mpiimpl.h in rlogtime.c by including only the necessary headers
• (jayesh) GenBuild(): Suppressed processing of Makefile variables
• (robl) if we call the channel's wait function, we'll wait forever if the onlyoutstanding request is a generalized request. instead, poll repeatedly. (canrevisit when generalized classes and a wait routine is in place).
• (gropp) Improved error messages when programs do not exit as expected
• (robl) initial commit of the 'MPIX_Grequest_start' infrastructure providing a hook foruser-defined requests to make progress w/o a thread.
• (chan) initial deposit of CallStack.
• (robl) Let us use MPIX_Grequest_start extensions. No longer need to manage requests-- MPI implementation does that for us- probably only works for UFS (if anything)- still need to fix up code for all other file system drivers
• (gropp) Make sure an address-sized int is used
• (gropp) Moved the version information into a separate file to simplify the build of mpich2version
• (buntinas) nemesis blocking: multithread bugfix
• (buntinas) nemesis:blocking fixed bug in init code where some nodes had only one proc and others had more than one. Those with one node werent calling the same number of PMI_Barriers
• (buntinas) nemesis:blocking updated src/pm to HEAD
• (robl) null out additional member
• (robl) outstanding request is a generalized request. instead, poll repeatedly. (canrevisit when generalized classes and a wait routine is in place).
• (robl) small cleanup helper to instantiate and complete a generalized request (usefulwhen nonblocking routine calls blocking)
• (gropp) Use USE_RDMA_READV/WRITEV to control compilation of code that uses ptrace to access memory in another process. This is non-portable (ptrace has different implementations in different versions of Unix) and may also require special support from the kernel (to avoid possible security issues)
• (robl) user-defined requests to make progress w/o a thread.
• (robl) when nonblocking routine calls blocking)
• (robl) wow, this touched a lot of files. Rip out old ROMIO request management code.Let us use MPIX_Grequest_start extensions. No longer need to manage requests-- MPI implementation does that for us- probably only works for UFS (if anything)- still need to fix up code for all other file system drivers

Files for each revision