OProfile 0.5.3 has been released
[Posted May 28, 2003 by cook]
| From: |
| John Levon <levon@movementarian.org> |
| To: |
| oprofile-list@lists.sf.net, editor@lwn.net |
| Subject: |
| OProfile 0.5.3 has been released |
| Date: |
| Tue, 27 May 2003 20:25:38 +0100 |
OProfile is a system-wide profiler for Linux.
OProfile 0.5.3 has been released. OProfile is still in alpha,
but has been proven stable for many users.
Release notes
-------------
It is recommended that you upgrade to this version for future
2.5 kernel compatibility.
For 2.2 kernels, the module must be compiled as the same user
that owns the kernel source tree.
nosmp is not supported in kernels before 2.4.10 (bug #463087).
The pre-emptable kernel option is not supported in 2.4 (bug #478516).
Power management on laptops can be incompatible with OProfile in 2.4 (bug #554927).
The utilities op_start, op_stop, op_dump, and op_session have been removed,
as threatened in the last three releases. opcontrol is the replacement.
New features
------------
The annotated source output has changed so that each line is prefixed by the
counts, and the header is now a footer. This maintains the line numbering
in the annotated output. The same change has been done for annotated assembly
output.
The daemon now handles --no-vmlinux with 2.4 kernels.
Bug fixes
---------
Some sanity checking on the minimum event count has been re-introduced, to
help guard against typos locking up the machine (bug #715923)
Some serious problems with hyper-threaded Pentium 4s have been fixed.
Some Hammer and i386 event descriptions have been fixed.
An IA-64 compile problem has been fixed.
Make opcontrol --ctrX-event=none work properly.
sparc and sparc64 in 2.5 are theoretically supported.
In 2.4, the watermark size is now proportional to the buffer size.
We were doing a sanity check on matching CPU speeds. This makes
no sense and has been removed.
A rare race that could cause a sample file to be corrupted when
running oprofpp has been fixed.
More STL patterns are matched by --smart-demangle.
(
Log in to post comments)