[Bug runtime/6481] New: timers: stp_time_timer_callback calls mod_timer with interrupts disabled

[Bug runtime/6481] New: timers: stp_time_timer_callback calls mod_timer with interrupts disabled

[fche at redhat dot com]

preempt-rt complains thusly:

May  5 11:45:23 rhelrt-3 kernel: BUG: scheduling with irqs disabled:
sirq-timer/1/0x00000000/19
May  5 11:45:23 rhelrt-3 kernel: caller is rt_spin_lock_slowlock+0x103/0x1a5
May  5 11:45:23 rhelrt-3 kernel: Pid: 19, comm: sirq-timer/1 Not tainted
2.6.24.4-48.el5rt #1
May  5 11:45:23 rhelrt-3 kernel: 
May  5 11:45:23 rhelrt-3 kernel: Call Trace:
May  5 11:45:23 rhelrt-3 kernel:  [<ffffffff81287530>] schedule+0x82/0xff
May  5 11:45:23 rhelrt-3 kernel:  [<ffffffff81287faf>]
rt_spin_lock_slowlock+0x103/0x1a5
May  5 11:45:23 rhelrt-3 kernel:  [<ffffffff81031aa5>] ?
update_rt_migration+0x18/0x99
May  5 11:45:23 rhelrt-3 kernel:  [<ffffffff81288db0>] __rt_spin_lock+0x4d/0x4f
May  5 11:45:23 rhelrt-3 kernel:  [<ffffffff81288dbb>] rt_spin_lock+0x9/0xb
May  5 11:45:23 rhelrt-3 kernel:  [<ffffffff81045e20>] lock_timer_base+0x26/0x4b
May  5 11:45:23 rhelrt-3 kernel:  [<ffffffff8104675e>] __mod_timer+0x3a/0xdf
May  5 11:45:23 rhelrt-3 kernel:  [<ffffffff810469e1>] mod_timer+0x3d/0x42
May  5 11:45:24 rhelrt-3 kernel:  [<ffffffff882e1a91>]
:stap_13c52a625a2bd38d37d461331df0cf23_2382:__stp_time_timer_callback+0x98/0

-- 
           Summary: timers: stp_time_timer_callback calls mod_timer with
                    interrupts disabled
           Product: systemtap
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: runtime
        AssignedTo: systemtap at sources dot redhat dot com
        ReportedBy: fche at redhat dot com


http://sourceware.org/bugzilla/show_bug.cgi?id=6481

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

[Bug runtime/6481] timers: stp_time_timer_callback calls mod_timer with interrupts disabled

[fche at redhat dot com]

------- Additional Comments From fche at redhat dot com  2008-05-05 17:09 -------
commit c5f8fe8; Josh, would you mind giving it an eye and a neuron to double-check?

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |joshua dot i dot stone at
                   |                            |intel dot com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED


http://sourceware.org/bugzilla/show_bug.cgi?id=6481

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

[Bug runtime/6481] timers: stp_time_timer_callback calls mod_timer with interrupts disabled

[joshua dot i dot stone at intel dot com]

------- Additional Comments From joshua dot i dot stone at intel dot com  2008-05-05 18:48 -------
RT should still keeps the soft-IRQ handler on the same CPU, so I don't see any
issue with sleeping there.  In fact, your worst-case comment won't even come to
pass, since the timer won't ever fire again until mod_timer has been called.


There's a similar possible bug in this callstack though:
  __stp_init_time->add_timer->__mod_timer->...

That's called from an IPI to setup a timer on every CPU, so interrupts are
disabled.  We might have to use an uglier mechanism to accomplish this, like a
set-affinity call to each CPU. :(

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=6481

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