public inbox for systemtap@sourceware.org
 help / color / mirror / Atom feed
From: "agentzh at gmail dot com" <sourceware-bugzilla@sourceware.org>
To: systemtap@sourceware.org
Subject: [Bug tapsets/25290] New: process(EXE).begin may occasionally miss already-running target processes from EXE
Date: Wed, 18 Dec 2019 07:32:00 -0000	[thread overview]
Message-ID: <bug-25290-6586@http.sourceware.org/bugzilla/> (raw)

https://sourceware.org/bugzilla/show_bug.cgi?id=25290

            Bug ID: 25290
           Summary: process(EXE).begin may occasionally miss
                    already-running target processes from EXE
           Product: systemtap
           Version: unspecified
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: tapsets
          Assignee: systemtap at sourceware dot org
          Reporter: agentzh at gmail dot com
  Target Milestone: ---

I've noted that the process(EXE).begin probes may sometimes miss target
processes which are already running when staprun is started. To reproduce this:

1. Prepare a minimal a.c C program file:

```
#include <unistd.h>

int main (int argc, char** argv) {
    while (1) {
        usleep(1);
    }
    return 0;
}
```

Compile it like this:

    gcc -g a.c

which results in ./a.out generated. Run this program in a separate terminal
window like this:


    ./a.out

Keep this process running (it has an infinite loop so it will never exit by
itself).

2. Prepare a minimal a.stp script file:

```
global max_cnt = 5, cnt = 0;
global pids

probe process("/home/agentzh/a.out").begin {
    pids[pid()] = 1
}

probe process("/home/agentzh/a.out").end {
    delete pids[pid()];
}

probe process("/lib64/libc.so.6").function("usleep") {
    if (!pids[pid()])
        next;

    if (++cnt > max_cnt)
        exit();

    println("usleep(", @var("useconds"), ")");
}

probe timer.s(5) {
    warn("timer expired");
    exit();
}

probe begin {
    warn("Start tracing...")
}
```

And then compile this file to a kernel module file, usleep.ko:


    stap -p4 -m usleep a.stp


3. Run this usleep.ko module in a shell loop:


    ( while true; do echo ======; sudo staprun usleep.ko; sleep 0.1; done ) |&
tee a.txt

after a few minutes (be patient!), we should see "timer expired" messages from
the output file a.txt:

    $ grep -a timer a.txt
    WARNING: timer expired
    WARNING: timer expired
    WARNING: timer expired
    WARNING: timer expired
    WARNING: timer expired
    WARNING: timer expired
    WARNING: timer expired
    WARNING: timer expired

You may get fewer lines here though.

The timer expired message should never be printed when the process(EXE).begin
probe is fired properly.

I tried the latest master branch of stap on kernel 5.0.16-100.fc28.x86_64 from
Fedora 28, kernel 4.15.0-72-generic x86_64 from Ubuntu 18.04, and kernels
3.10.0-957.27.2.el7.x86_64 and 3.10.0-1062.9.1.el7.x86_64 from CentOS 7, all
show the same problem. I just noted that this is much harder to reproduce on
kernel 5.0.16 than earlier versions of the kernel.

The stap version:

```
$ /opt/stap/bin/stap -V
Systemtap translator/driver (version 4.3/0.174/0.177, commit
release-4.2-10-g1427836ac118)
Copyright (C) 2005-2019 Red Hat, Inc. and others
This is free software; see the source for copying conditions.
tested kernel versions: 2.6.32 ... 5.4-rc6
enabled features: BPF LIBSQLITE3 NLS
```

Any hints on debugging this further? Thanks!

-- 
You are receiving this mail because:
You are the assignee for the bug.

             reply	other threads:[~2019-12-18  7:32 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-12-18  7:32 agentzh at gmail dot com [this message]
2019-12-19 21:25 ` [Bug tapsets/25290] " agentzh at gmail dot com
2019-12-20  6:47 ` agentzh at gmail dot com
2019-12-20 23:24 ` agentzh at gmail dot com
2019-12-20 23:25 ` agentzh at gmail dot com

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=bug-25290-6586@http.sourceware.org/bugzilla/ \
    --to=sourceware-bugzilla@sourceware.org \
    --cc=systemtap@sourceware.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).