From: John L. <mov...@us...> - 2002-11-09 21:37:49
|
Update of /cvsroot/oprofile/oprofile In directory usw-pr-cvs1:/tmp/cvs-serv19987 Modified Files: ChangeLog TODO configure.in Log Message: Make separate objdir work. module build fails but all is OK with --with-kernel-support Index: ChangeLog =================================================================== RCS file: /cvsroot/oprofile/oprofile/ChangeLog,v retrieving revision 1.671 retrieving revision 1.672 diff -u -d -r1.671 -r1.672 --- ChangeLog 9 Nov 2002 20:52:07 -0000 1.671 +++ ChangeLog 9 Nov 2002 21:37:44 -0000 1.672 @@ -1,5 +1,13 @@ 2002-11-09 John Levon <le...@mo...> + * TODO: update + + * configure.in: + * most Makefile.am: make separate objdir work, + except for module (impossible I suppose) + +2002-11-09 John Levon <le...@mo...> + * autogen.sh: use --foreign 2002-11-09 John Levon <le...@mo...> Index: TODO =================================================================== RCS file: /cvsroot/oprofile/oprofile/TODO,v retrieving revision 1.137 retrieving revision 1.138 diff -u -d -r1.137 -r1.138 --- TODO 13 Oct 2002 02:35:53 -0000 1.137 +++ TODO 9 Nov 2002 21:37:44 -0000 1.138 @@ -1,13 +1,6 @@ -0.3 release ------------ - Before 1.0 big stuff -------------------- - o can we get pre-emption up and running ?? After a first try I think it should - be fixed by calling preempt_enable/disable before/after each op_int_operations - allowing arch depedant stuff to not worry about preemption or at least to minmize - things needed in arch dependant file o implement new pp spec o difference between two profile session. Difference must be made at source level. (This is a part of post-prof). Difference can be made also at symbols level. Important. @@ -61,14 +54,8 @@ Later ----- - o The halting problem: get two make cleans in a row to not compile anything - o multiplexing for real - o check if we can use our own RTC irq handler rather to pass through the - the kernel service. The overhead for RTC is very big actually. Take care - first from where come the overhead (read/write RTC CMOS is costly) - oprof_report: IMHO this app isn't near what we really want. There's no point doing this @@ -83,14 +70,8 @@ o merge sample files into one big report (like vtune can do repeated runs) -o move from racy sysctl to procfs, if it can be done nicely. Back compatibility - probably prevents us from making a mini-filesystem. - o turn (or only compile) daemon in C++ to allow more code sharing with pp/gui is it a good idea ? - -o negative process entries in the daemon - a fork() into the pid would always succeed, - in the meantime count the missed samples, possibly try to get info via /proc again ... o maybe we should support a CPU mask for enabling counters (useful with processor pinning ?) Index: configure.in =================================================================== RCS file: /cvsroot/oprofile/oprofile/configure.in,v retrieving revision 1.116 retrieving revision 1.117 diff -u -d -r1.116 -r1.117 --- configure.in 9 Nov 2002 20:21:04 -0000 1.116 +++ configure.in 9 Nov 2002 21:37:44 -0000 1.117 @@ -150,7 +150,7 @@ fi test "$arch" = "unknown" && AC_MSG_ERROR(Unsupported architecture) - rm -f module/arch && ln -s "$arch" module/arch + (cd ${srcdir}/module/ && rm -f arch/ && ln -s $arch arch) ARCH_MAKEFILE=module/arch/Makefile dnl check to see if kernel verion appropriate for arch |