httpd-bugs mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
Subject [Bug 59646] New:`proxy_handler: coredump before ap_log_rerror related to aplog_module_index use
Date Tue, 31 May 2016 14:05:51 GMT

            Bug ID: 59646
           Summary:`proxy_handler: coredump before
                    ap_log_rerror related to aplog_module_index use
           Product: Apache httpd-2
           Version: 2.4.12
          Hardware: Sun
                OS: Solaris
            Status: NEW
          Severity: normal
          Priority: P2
         Component: mod_proxy

I have several core files from our customers. Both customers are on sparc.

0007fc30df5fb0a1`proxy_handler+0x744(4c4f5f3e90, 4c4f5f3e90,
4c4f5f964e, 7fc30e501c000, 
4c4f5a37b8, 7fc30e4f073a0)
0007fc30df5fb1f1 ap_run_handler+0x4c(4c4f5f3e90, c, 4, 100200a18, 4c4f5a14b0,
0007fc30df5fb2a1 ap_invoke_handler+0x14c(4c4f5f3e90, 0, 1c, 0, 0, 0)
0007fc30df5fb371 httpd`ap_process_async_request+0x2c4(4c4f5f3e90, 4, 100033,
100193, 1001932b8, 2)
0007fc30df5fb441 ap_process_http_connection+0x8c(4c4f5f0210, 0, 1, 100193000,
100193, 100000)
0007fc30df5fb4f1 ap_run_process_connection+0x4c(4c4f5f0210, 2, 4c4f5a1a28,
100201280, 4c4f5a1a48, 100201)
0007fc30df5fb5a1`process_socket+0x2b0(4c4f5a9908, 1,
4c4f5efe80, 4c4f5f0188, 80000000, 
0007fc30df5fb671`worker_thread+0x2e4(4c4f5a9908, 4c4f5efe80,
7fc30e5e00220, 18, 0, 
0007fc30df5fb751`_lwp_start(0, 0, 0, 0, 0, 0)

>`proxy_handler+0x744::dis`proxy_handler+0x71c:       st        %o0, [%fp + 0x7db]`proxy_handler+0x720:       ldx       [%sp + 0x8c7], %o5`proxy_handler+0x724:       ld        [%o5 + 0xc], %i0`proxy_handler+0x728:       add       %l2, 0x1, %l2`proxy_handler+0x72c:       cmp       %l2, %i0`proxy_handler+0x730:       bl,pt     %icc, -0x1dc 
<`proxy_handler+0x554>`proxy_handler+0x734:       sra       %l2, 0x0, %g4`proxy_handler+0x738:       mov       -0x1, %l7`proxy_handler+0x73c:       brz,pn    %l0, +0xc    
<`proxy_handler+0x748>`proxy_handler+0x740:       ld        [%sp + 0x8e3], %l2`proxy_handler+0x744:       ld        [%l0], %l7`proxy_handler+0x748:       ldx       [%sp + 0x8d7], %o7`proxy_handler+0x74c:       sra       %l2, 0x0, %l3`proxy_handler+0x750:       add       %i5, 0x1e8, %l6`proxy_handler+0x754:       mov       %i1, %o5`proxy_handler+0x758:       clr       %o4`proxy_handler+0x75c:       stx       %l3, [%sp + 0x8bf]`proxy_handler+0x760:       mov       0x7, %o3`proxy_handler+0x764:       mov       0x489, %o1`proxy_handler+0x768:       stx       %o7, [%sp + 0x8b7]`proxy_handler+0x76c:       sra       %l7, 0x0, %o2

> $r ! grep l0 
%g0 = 0x0000000000000000                 %l0 = 0x0000000000114c00 

I cannot reproduce the issue myself. But by fiddling with the code and compiler
I was able to localize the place where it cores:

1161             ap_log_rerror(APLOG_MARK, APLOG_DEBUG, 0, r, APLOGNO(01143)
1162                           "Running scheme %s handler (attempt %d)",
1163                           scheme, attempts);

Particularly APLOG_MARK consist from:



(aplog_module_index ? *aplog_module_index : APLOG_NO_MODULE)`proxy_handler+0x738:       mov       -0x1, %l7`proxy_handler+0x73c:       brz,pn    %l0, +0xc    
<`proxy_handler+0x748>`proxy_handler+0x740:       ld        [%sp + 0x8e3], %l2`proxy_handler+0x744:       ld        [%l0], %l7

Customers also reported that Apache 2.2 works in the same scenario.
Where 2.2 seems to have:

#define APLOG_MARK     __FILE__,__LINE__

Any idea?

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

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message