RE: os-solaris/7159: Solaris bug that causes httpd to hang in sle eping state

["Meyer, Richard" <rm...@befree.com>]

I just noticed that I can append more information to an existing bug w/o
having to submit a new one. I already submitted a new one though. Sorry
about that.


An addendum to this bug,

Learning a bit more about gdb, I've attached to a hung process and entered
gdb> info threads
(gdb) info threads
  8 Thread 3          0xef5b9790 in __lwp_sema_wait ()
  7 Thread 2 (LWP 2)  0xef5b98d8 in __signotifywait ()
  6 Thread 1 (LWP 1)  0xef5b8688 in _read ()
  5 LWP    6          0xef5b9744 in ___lwp_cond_wait ()
  4 LWP    6          0xef5b9744 in ___lwp_cond_wait ()
  3 LWP    4          0xef5b9790 in __lwp_sema_wait ()
  2 LWP    2          0xef5b98d8 in __signotifywait ()
* 1 LWP    1          0xef5b8688 in _read ()

Connecting to the unique threads and issueing >info stack calls I get :
(gdb) thread 1
[Switching to LWP    1        ]
#0  0xef5b8688 in _read ()
(gdb) info stack
#0  0xef5b8688 in _read ()
#1  0xef365b8c in _ti_read ()
#2  0x1fb58 in buff_read (fb=0xb1a48, buf=0xb1a88, nbyte=4096) at buff.c:299
#3  0x1fac8 in saferead_guts (fb=0xb1a48, buf=0xb1a88, nbyte=4096) at
buff.c:662
#4  0x1db30 in read_with_errors (fb=0xb1a48, buf=0xb1a88, nbyte=4096) at
buff.c:713
#5  0x1dffc in ap_bgets (buff=0xefffd800 "", n=8192, fb=0xb1a48) at
buff.c:866
#6  0x340a0 in getline (s=0xefffd800 "", n=8192, in=0xb1a48, fold=0) at
http_protocol.c:757
#7  0x34538 in read_request_line (r=0x644c68) at http_protocol.c:880
#8  0x34f18 in ap_read_request (conn=0x643c28) at http_protocol.c:1038
#9  0x308f8 in child_main (child_num_arg=3) at http_main.c:4166
#10 0x30ca4 in make_child (s=0xa3b78, slot=3, now=980869018) at
http_main.c:4336
#11 0x30db4 in startup_children (number_to_start=2) at http_main.c:4363
#12 0x31694 in standalone_main (argc=1, argv=0xeffffbcc) at http_main.c:4651
#13 0x32240 in main (argc=1, argv=0xeffffbcc) at http_main.c:4978

(gdb) thread 2
[Switching to LWP    2        ]
#0  0xef5b98d8 in __signotifywait ()
(gdb) info stack
#0  0xef5b98d8 in __signotifywait ()
#1  0xef35bdec in _dynamiclwps ()

(gdb) thread 3
[Switching to LWP    4        ]
#0  0xef5b9790 in __lwp_sema_wait ()
(gdb) info stack
#0  0xef5b9790 in __lwp_sema_wait ()
#1  0xef357ea0 in _park ()
#2  0xef357b84 in _swtch ()
#3  0xef35b000 in _reap_wait ()
#4  0xef35ad8c in _reaper ()

These are in addition to the stack trace I provided for the _lwp_cond_wait
thread yesterday.

BTW, we just heard from Sun and the bugfix which may address this problem
won't be available publicly until March 9,2001.



> -----Original Message-----
> From: submit@bugz.apache.org [mailto:submit@bugz.apache.org]
> Sent: Tuesday, January 30, 2001 2:30 PM
> To: rmeyer@befree.com
> Subject: Re: os-solaris/7159: Solaris bug that causes httpd to hang in
> sleeping state
> 
> 
> Thank you very much for your problem report.
> It has the internal identification `os-solaris/7159'.
> The individual assigned to look at your
> report is: apache. 
> 
> >Category:       os-solaris
> >Responsible:    apache
> >Synopsis:       Solaris bug that causes httpd to hang in 
> sleeping state
> >Arrival-Date:   Tue Jan 30 11:30:03 PST 2001
>