This is the mail archive of the
systemtap-cvs@sourceware.org
mailing list for the systemtap project.
[SCM] systemtap: system-wide probe/trace tool branch, jistone/remote, created. release-1.3-389-gebff2ed
- From: jistone at sourceware dot org
- To: systemtap-cvs at sourceware dot org
- Date: 15 Dec 2010 01:38:03 -0000
- Subject: [SCM] systemtap: system-wide probe/trace tool branch, jistone/remote, created. release-1.3-389-gebff2ed
- Reply-to: systemtap at sourceware dot org
This is an automated email from the git hooks/post-receive script. It was
generated because a ref change was pushed to the repository containing
the project "systemtap: system-wide probe/trace tool".
The branch, jistone/remote has been created
at ebff2ed0ae97ced371a0b24a4f74d433513d9574 (commit)
- Log -----------------------------------------------------------------
commit ebff2ed0ae97ced371a0b24a4f74d433513d9574
Author: Josh Stone <jistone@redhat.com>
Date: Tue Dec 14 16:15:45 2010 -0800
PR12115: New support for multiple remote targets
* session.cxx (systemtap_session::clone): Returns a systemtap_session
for a requested arch/release. If one such was already created, go
ahead and use it.
* remote.h (run -> start, finish): Separate the launch from the wait.
* remote.cxx (direct, ssh_remote): Implement separated start and finish.
(ssh_remote::ssh_remote): Use systemtap_session::clone to get a
session matching the remote target.
* main.cxx (pass_5): Take a vector of targets to run in parallel.
(cleanup): Perform cleanup for subsessions too.
(main): Loop over all specified remote_uris.
commit 01cc94dc3b1217127f2ae838cd01c73b0aed130d
Author: Josh Stone <jistone@redhat.com>
Date: Mon Dec 13 16:46:34 2010 -0800
PR12115: Prepare utils for multiple inferiors
We'll need to juggle multiple child process if we are to run multiple
remote targets in parallel.
* util.cxx (spawned_pid -> spawned_pids): No longer an only child,
hopefully it won't act so spoiled anymore.
(wait_for_spawn -> stap_waitpid): Now wait for a specific pid. Clear
it from the set when dead (but not forgotten).
(stap_spawn): Like stap_system, but return the pid and don't wait up.
(stap_system): Call stap_spawn, then wait for it.
(kill_stap_spawn): Must be fair, kill them all.
commit b2cbfd402694fb449f2b5be999015820ecd86732
Author: Josh Stone <jistone@redhat.com>
Date: Mon Dec 13 16:45:52 2010 -0800
PR12115: Give remote-ssh a tty, and quiet status messages
Having a tty lets ssh pass signals through to the remote process.
* remote.cxx (ssh_remote): Use -t with ssh commands, and -q with ssh and
scp commands.
commit daa752066a495311f792788df878194f7ff2d67d
Author: Josh Stone <jistone@redhat.com>
Date: Tue Nov 30 18:37:15 2010 -0800
PR12115: Early support for remote execution
This corresponds to the second draft posted on bugzilla.
* Makefile.am: Add remote.cxx to the build.
* Makefile.in: Regenerate.
* buildrun.cxx (run_pass -> make_run_command): Rename, and don't
actually run the thing yet, so we can control *where* we run it.
* main.cxx (pass_5): Take a target where we want to run.
(main): Check if we have a remote uri, and go with it.
* remote.cxx: New, define classes for local and ssh execution.
* remote.h: New, define the generic remote class.
* session.cxx (systemtap_session::systemtap_session): Move the logic to
normalize machine architecture names to util.cxx.
(systemtap_session::parse_cmdline): Add --remote.
* session.h (systemtap_session::remote_uris): New.
* util.cxx (normalize_machine): Moved from session.cxx.
(wait_for_spawn): Refactor common code.
(stap_system): Use wait_for_spawn.
(stap_system_read): New, reads stdout from subprocess.
Like popen, but with c++ streams and a signal-managed pid.
-----------------------------------------------------------------------
hooks/post-receive
--
systemtap: system-wide probe/trace tool