headers
fche at redhat dot com | 1 Jun 2011 03:35
Favicon

[Bug translator/12829] New: "Too many interrupts received." under sudo

http://sourceware.org/bugzilla/show_bug.cgi?id=12829

           Summary: "Too many interrupts received." under sudo
           Product: systemtap
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: translator
        AssignedTo: systemtap <at> sourceware.org
        ReportedBy: fche <at> redhat.com

When stap is run under sudo, it sometimes receives redundant SIGfoo
for ordinary console interrupts.

--

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.
Permalink | Reply | 1 comment |
headers
fche at redhat dot com | 1 Jun 2011 03:39
Favicon

[Bug translator/12829] "Too many interrupts received." under sudo

http://sourceware.org/bugzilla/show_bug.cgi?id=12829

Frank Ch. Eigler <fche at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED

--- Comment #1 from Frank Ch. Eigler <fche at redhat dot com> 2011-06-01 01:39:17 UTC ---
commit 536a1b6

--

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.
Permalink | Reply | 0 comments |
headers
Daniel Migault | 1 Jun 2011 09:03
Picon

Re: Monitoring Modules

Hi Frank,

Thank you for your answer. We ran it on fedora with version 1.4 with the 
addr() function specified in  our script and comment it version 1.6.

Actually we could get the names of the function only by running it on 
fedora, with Ubuntu 11.04, we could only get the address. I suppose 
standard kernel miss some debug info option to get bindings between 
names and addresses?

I will post my different comments and questions in different threads on 
the mailing list.

Thanks again,
Daniel

On 05/19/11 19:54, Frank Ch. Eigler wrote:
> Hi, Daniel -
>
>> To measure how IPsec influences the performances of our system, we
>> would like to record the module of the current task. We use the
>> probemod() function but it only returns "<unknown>  ".
>> Do you know how could we get the module's name performing the task?
>> [probe timer.profile  ... ]
>> if (!user_mode()) {
>>       module=probemod()
>>       kticks[e,tid]<<<  1
> probemod(), like probefunc(), work by parsing probe point strings,
> which is not that helpful - as you've discovered.  Sorry about that.
>
(Continue reading)

Permalink | Reply | 0 comments |
headers
Daniel Migault | 1 Jun 2011 09:30
Picon

Installation from git

Hi,

The last stap version I could properly installed on my fedora  
2.6.38.6-27.fc15.x86_64 is
Systemtap translator/driver (version 1.6/0.152 commit 
release-1.5-11-ge075598 + changes)

Trying to install systemtap from a newly cloned git directory I have the 
following message :

Writing appe-Publican-Revision_History.html for 
appendix(appe-Publican-Revision_History)
Writing ix01.html for index
Writing index.html for book
     Finished html
mv: cannot stat `build/en-US/pdf/*SystemTap_Beginners_Guide*.pdf': No 
such file or directory
make[3]: *** [SystemTap_Beginners_Guide.pdf] Error 1

 From an updated git directory, I have :

Processing file build/en-US/xml_tmp/Using_SystemTap.xml -> 
build/en-US/xml/Using_SystemTap.xml
Beginning work on en-US
FATAL ERROR: XInclude:1604 in Useful_Scripts-futexes.xml on line 70: 
could not load ../testsuite/systemtap.examples/process/futexes.stp, and 
no fallback was found
  at /usr/bin/publican line 672
make[3]: *** [SystemTap_Beginners_Guide.pdf] Error 2
(Continue reading)

Permalink | Reply | 1 comment |
headers
Daniel Migault | 1 Jun 2011 09:33
Picon

Comments on Installation Guide

Hi,

I went through the documentation, and maybe I would add a few 
indications for the installation. It looks that some of them have been 
removed from previous installation guides.

Regards,
Daniel

I Install systemtap

If you want to install systemtap from rpm :
   yum install systemtap systemtap-client systemtap-grapher 
systemtap-runtime systemtap-sdt-devel systemtap-server systemtap-testsuite

If you want the latest version, then you may built it from the git 
repository :
   I.1. Download the git directory
     If you don't already have a local copy :
     git clone http://sources.redhat.com/git/systemtap.git 
<http://sources.redhat.com/git/systemtap.git> systemtap.git
     If you have already a local copy then you just need to upgrade it :
     git pull

   I.2. Get pre-requisit rpms :
     yum install xmlto latex2html gcc gcc-c++ publican nspr-devel 
nss-devel avahi-devel elfutils-devel libdwarf-devel sqlite-devel 
boost-devel xmltool rpm-libs rpm-devel rpm-build pcp-libs-devel 
perl-ExtUtils-MakeMaker bison perl-PCP-PMDA  gtkmm24-devel 
libglademm24-devel
(Continue reading)

Permalink | Reply | 2 comments |
headers
Daniel Migault | 1 Jun 2011 09:48
Picon

registration error (rc -84) and -DSTP_OVERLOAD_THRESHOLD

Hi,

While running my tap script, I receive WARNINGs and would like to knwo 
how to consider them. Does it means that matching between source code 
and memory address presents dome differences? If so, is there any way to 
correct this mismatch ? How does it really impact the probe? For 
exmaple, does a probe on tcp_new_space for example don't consider the 
function ?

Pass 5: starting run.
WARNING: probe 
kernel.function("tcp_new_space <at> net/ipv4/tcp_input.c:4885") (address 
0xffffffff814064f0) registration error (rc -84)
WARNING: probe 
kernel.function("tcp_update_reordering <at> net/ipv4/tcp_input.c:935") 
(address 0xffffffff81404578) registration error (rc -84)
WARNING: probe 
kernel.function("tcp_should_expand_sndbuf <at> net/ipv4/tcp_input.c:4854") 
(address 0xffffffff814064f0) registration error (rc -84)
WARNING: probe 
kernel.function("ablkcipher_get_spot <at> crypto/ablkcipher.c:71") (address 
0xffffffff812059bc) registration error (rc -84)
[...]

By the way, I also noticed some problems with the 
-DSTP_OVERLOAD_THRESHOLD. The script does not run and sometime it may 
result by frezzing the OS. We don't have any problems with the 
-DSTP_NO_OVERLOAD option for example. I am using gcc version 4.6.0 
20110509 (Red Hat 4.6.0-7) (GCC).
(Continue reading)

Permalink | Reply | 8 comments |
headers
Frank Ch. Eigler | 1 Jun 2011 15:37
Picon
Favicon
Gravatar

Re: registration error (rc -84) and -DSTP_OVERLOAD_THRESHOLD


mglt.biz wrote:

> [...]  While running my tap script, I receive WARNINGs and would
> like to knwo how to consider them. [...]

> WARNING: probe
> kernel.function("tcp_new_space <at> net/ipv4/tcp_input.c:4885") (address
> 0xffffffff814064f0) registration error (rc -84)

This is an error (-EILSEQ) that comes from the kernel, most likely
arch/x86/kernel/kprobes.c.  It may be a somehow blacklisted function,
or perhaps a problem with the 'optimized kprobes' widget.  It may help
to get a vmlinux disassembly around that 0xffffffff814064f0 address.

> How does it really impact the probe? For exmaple, does a probe on
> tcp_new_space for example don't consider the function ?

That's correct, such a probe will just not be activated, and its
handler never run.

> By the way, I also noticed some problems with the
> -DSTP_OVERLOAD_THRESHOLD. The script does not run and sometime it
> may result by frezzing the OS. We don't have any problems with the
> -DSTP_NO_OVERLOAD option for example. [...]

We'd probably need some more details.  Mis-set THRESHOLD type values
should do no worse than abort the script prematurely.  Try 
   # stap -t -v
Share the script and its invocation.
(Continue reading)

Permalink | Reply | 7 comments |
headers
dsmith at redhat dot com | 1 Jun 2011 21:30
Favicon

[Bug runtime/12834] New: itrace not handling utrace engines properly

http://sourceware.org/bugzilla/show_bug.cgi?id=12834

           Summary: itrace not handling utrace engines properly
           Product: systemtap
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: runtime
        AssignedTo: systemtap <at> sourceware.org
        ReportedBy: dsmith <at> redhat.com

On the current version of utrace (> RHEL5), the itrace runtime code isn't
calling utrace_engine_put() after attaching to a new task.  This will break
utrace's reference count on the utrace_engine structure, and cause a slow
memory leak.

--

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.
Permalink | Reply | 1 comment |
headers
Frank Ch. Eigler | 2 Jun 2011 00:23
Picon
Favicon
Gravatar

Re: Installation from git

Daniel Migault <mglt.biz <at> gmail.com> writes:

> [...]
> The last stap version I could properly installed on my fedora
> 2.6.38.6-27.fc15.x86_64 is
> Systemtap translator/driver (version 1.6/0.152 commit
> release-1.5-11-ge075598 + changes)

Great.

> Trying to install systemtap from a newly cloned git directory I have
> the following message :
>
> [...]
> mv: cannot stat `build/en-US/pdf/*SystemTap_Beginners_Guide*.pdf': No
> such file or directory

This sounds like https://bugzilla.redhat.com/show_bug.cgi?id=704298
Consider configuring your build tree with --disable-docs for now.

- FChE
Permalink | Reply | 0 comments |
headers
Atsushi Tsuji | 2 Jun 2011 02:38
Picon

[PATCH] Fix error checking in utrace_attach_task()

utrace_attach_task() returns not ENOENT but ESRCH in case of
UTRACE_ATTACH_CREATE if an error occurs. So this patch fixes
error checking to correctly ignore processes in "mortally
wounded" state.

Signed-off-by: Atsushi Tsuji <a-tsuji <at> bk.jp.nec.com>
---
 runtime/task_finder.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/runtime/task_finder.c b/runtime/task_finder.c
index 2c52175..552972b 100644
--- a/runtime/task_finder.c
+++ b/runtime/task_finder.c
 <at>  <at>  -506,7 +506,7  <at>  <at>  __stp_utrace_attach(struct task_struct *tsk,
 	engine = utrace_attach_task(tsk, UTRACE_ATTACH_CREATE, ops, data);
 	if (IS_ERR(engine)) {
 		int error = -PTR_ERR(engine);
-		if (error != ENOENT) {
+		if (error != ESRCH) {
 			_stp_error("utrace_attach returned error %d on pid %d",
 				   error, (int)tsk->pid);
 			rc = error;
-- 1.7.2.1
Permalink | Reply | 2 comments |

Gmane