Hi Stefan, no i was mistaken customer isn't using mod_proxy - but I think this is the patch causing me problems: https://github.com/apache/httpd/commit/a61a4bd02e483fb45d433343740d0130ee3d8d5d
What do you think? Greets, Stefan Am 22.01.2017 um 17:17 schrieb Stefan Eissing: > >> Am 22.01.2017 um 17:14 schrieb Stefan Priebe - Profihost AG >> <s.pri...@profihost.ag>: >> >> *arg* it's just mod_proxy - just saw thread safety and apr bucket aloc. > > ??? Can you elaborate? Is your finding the known hcheck bug or something else? > >> Stefan >> >> Am 22.01.2017 um 17:06 schrieb Stefan Priebe - Profihost AG: >>> Looks like others have the same crashes too: >>> https://bz.apache.org/bugzilla/show_bug.cgi?id=60071 >>> and >>> https://github.com/apache/httpd/commit/8e63c3c9372cd398f57357099aa941cbba695758 >>> >>> So it looks like mod_http2 is running fine now. Thanks a lot Stefan. >>> >>> Yann i think i can start testing your mpm patch again after the >>> segfaults in 2.4 branch are fixed. >>> >>> Greets, >>> Stefan >>> >>> Am 22.01.2017 um 13:16 schrieb Stefan Priebe: >>>> Hi, >>>> >>>> and a new one but also in ap_start_lingering_close: >>>> >>>> Program terminated with signal SIGSEGV, Segmentation fault. >>>> #0 apr_palloc (pool=pool@entry=0x7f455805e138, in_size=in_size@entry=32) >>>> at memory/unix/apr_pools.c:684 >>>> #0 apr_palloc (pool=pool@entry=0x7f455805e138, in_size=in_size@entry=32) >>>> at memory/unix/apr_pools.c:684 >>>> #1 0x00007f456bc5d8b4 in apr_brigade_create (p=0x7f455805e138, >>>> list=0x7f45040034e8) at buckets/apr_brigade.c:61 >>>> #2 0x000055e165efa319 in ap_shutdown_conn (c=c@entry=0x7f455805e458, >>>> flush=flush@entry=1) at connection.c:76 >>>> #3 0x000055e165efa40d in ap_flush_conn (c=0x7f455805e458) at >>>> connection.c:95 >>>> #4 ap_start_lingering_close (c=0x7f455805e458) at connection.c:145 >>>> #5 0x000055e165f942dd in start_lingering_close_blocking (cs=<optimized >>>> out>) >>>> at event.c:876 >>>> #6 process_socket (my_thread_num=<optimized out>, >>>> my_child_num=<optimized out>, cs=0x7f455805e3c8, sock=<optimized out>, >>>> p=<optimized out>, thd=<optimized out>) at event.c:1153 >>>> #7 worker_thread (thd=0x7f455805e138, dummy=0x20) at event.c:2001 >>>> #8 0x00007f456b80a0a4 in start_thread () >>>> from /lib/x86_64-linux-gnu/libpthread.so.0 >>>> #9 0x00007f456b53f62d in clone () from /lib/x86_64-linux-gnu/libc.so.6 >>>> >>>> Stefan >>>> >>>> Am 21.01.2017 um 19:31 schrieb Stefan Priebe: >>>>> All last traces come from event, proces_longering_close ap_push_pool but >>>>> end in different functions. It looks like a race somewhere and it just >>>>> races at different function in the event of close and pool clear. >>>>> >>>>> Might there be two places where the same pool gets cleared? >>>>> >>>>> Stefan >>>>> >>>>> Am 21.01.2017 um 19:07 schrieb Stefan Priebe: >>>>>> Hi Stefan, >>>>>> >>>>>> thanks. No crashes where h2 comes up. But i still have these and no idea >>>>>> how to find out who and why they're crashing. >>>>>> >>>>>> Using host libthread_db library >>>>>> "/lib/x86_64-linux-gnu/libthread_db.so.1". >>>>>> Core was generated by `/usr/local/apache2/bin/httpd -k start'. >>>>>> Program terminated with signal SIGSEGV, Segmentation fault. >>>>>> #0 allocator_free (node=0x0, allocator=0x7f6e08066540) >>>>>> at memory/unix/apr_pools.c:381 >>>>>> #0 allocator_free (node=0x0, allocator=0x7f6e08066540) >>>>>> at memory/unix/apr_pools.c:381 >>>>>> #1 apr_pool_clear (pool=0x7f6e0808d238) at memory/unix/apr_pools.c:793 >>>>>> #2 0x00000000004fe528 in ap_push_pool (queue_info=0x0, >>>>>> pool_to_recycle=0x7f6e08066548) at fdqueue.c:234 >>>>>> #3 0x00000000004fa2c8 in process_lingering_close (cs=0x7f6e0808d4c8, >>>>>> pfd=0x1d3bf98) at event.c:1439 >>>>>> #4 0x00000000004fd410 in listener_thread (thd=0x1d3cb70, >>>>>> dummy=0x7f6e0808d4c8) >>>>>> at event.c:1704 >>>>>> #5 0x00007f6e1aed20a4 in start_thread () >>>>>> from /lib/x86_64-linux-gnu/libpthread.so.0 >>>>>> #6 0x00007f6e1aa0362d in clone () from /lib/x86_64-linux-gnu/libc.so.6 >>>>>> (gdb) (gdb) quit >>>>>> >>>>>> Reading symbols from /usr/local/apache/bin/httpd...Reading symbols from >>>>>> /usr/lib/debug//usr/local/apache2/bin/httpd...done. >>>>>> done. >>>>>> [Thread debugging using libthread_db enabled] >>>>>> Using host libthread_db library >>>>>> "/lib/x86_64-linux-gnu/libthread_db.so.1". >>>>>> Core was generated by `/usr/local/apache2/bin/httpd -k start'. >>>>>> Program terminated with signal SIGSEGV, Segmentation fault. >>>>>> #0 allocator_free (node=0x0, allocator=0x7f6e08053ae0) >>>>>> at memory/unix/apr_pools.c:381 >>>>>> #0 allocator_free (node=0x0, allocator=0x7f6e08053ae0) >>>>>> at memory/unix/apr_pools.c:381 >>>>>> #1 apr_pool_clear (pool=0x7f6e08076bb8) at memory/unix/apr_pools.c:793 >>>>>> #2 0x00000000004fe528 in ap_push_pool (queue_info=0x0, >>>>>> pool_to_recycle=0x7f6e08053ae8) at fdqueue.c:234 >>>>>> #3 0x00000000004fa2c8 in process_lingering_close (cs=0x7f6e08076e48, >>>>>> pfd=0x1d3bf98) at event.c:1439 >>>>>> #4 0x00000000004fd410 in listener_thread (thd=0x1d3cb70, >>>>>> dummy=0x7f6e08076e48) >>>>>> at event.c:1704 >>>>>> #5 0x00007f6e1aed20a4 in start_thread () >>>>>> from /lib/x86_64-linux-gnu/libpthread.so.0 >>>>>> #6 0x00007f6e1aa0362d in clone () from /lib/x86_64-linux-gnu/libc.so.6 >>>>>> (gdb) (gdb) quit >>>>>> >>>>>> Stefan >>>>>> >>>>>> Am 21.01.2017 um 17:03 schrieb Stefan Eissing: >>>>>>> Stefan, >>>>>>> >>>>>>> made a release at https://github.com/icing/mod_h2/releases/tag/v1.8.9 >>>>>>> with all patches and improved (hopefully) on them a bit. If you dare >>>>>>> to drop that into your installation, that'd be great. >>>>>>> >>>>>>> Cheers, >>>>>>> >>>>>>> Stefan >>>>>>> >>>>>>>> Am 21.01.2017 um 15:25 schrieb Stefan Priebe <s.pri...@profihost.ag>: >>>>>>>> >>>>>>>> and i got another crash here: >>>>>>>> >>>>>>>> 2346 static void run_cleanups(cleanup_t **cref) >>>>>>>> 2347 { >>>>>>>> 2348 cleanup_t *c = *cref; >>>>>>>> 2349 >>>>>>>> 2350 while (c) { >>>>>>>> 2351 *cref = c->next; >>>>>>>> 2352 (*c->plain_cleanup_fn)((void *)c->data); <== here >>>>>>>> 2353 c = *cref; >>>>>>>> 2354 >>>>>>>> >>>>>>>> which looks similar to the other crash. >>>>>>>> >>>>>>>> #0 0x00007fe4bbd33e1b in run_cleanups (cref=<optimized out>) at >>>>>>>> memory/unix/apr_pools.c:2352 >>>>>>>> #1 apr_pool_clear (pool=0x7fe4a804dac8) at >>>>>>>> memory/unix/apr_pools.c:772 >>>>>>>> #2 0x00000000004feb38 in ap_push_pool >>>>>>>> (queue_info=0x6d616e79642d3733, pool_to_recycle=0x2) at fdqueue.c:234 >>>>>>>> #3 0x00000000004fa8d8 in process_lingering_close (cs=0x7fe4a804dd58, >>>>>>>> pfd=0x25d3f98) at event.c:1439 >>>>>>>> >>>>>>>> Details: >>>>>>>> (gdb) print c >>>>>>>> $1 = (cleanup_t *) 0x7fe4a804e9f0 >>>>>>>> (gdb) print *c >>>>>>>> $2 = {next = 0x7fe4a804e870, data = 0x6d616e79642d3733, >>>>>>>> plain_cleanup_fn = 0x392d3734322e6369, >>>>>>>> child_cleanup_fn = 0x617465722e722d35} >>>>>>>> (gdb) print *c->data >>>>>>>> Attempt to dereference a generic pointer. >>>>>>>> (gdb) print *c->plain_cleanup_fn >>>>>>>> Cannot access memory at address 0x392d3734322e6369 >>>>>>>> (gdb) >>>>>>>> >>>>>>>> Stefan >>>>>>>> >>>>>>>> Am 21.01.2017 um 15:18 schrieb Stefan Priebe: >>>>>>>>> Hi, >>>>>>>>> >>>>>>>>> #0 apr_pool_cleanup_kill (p=0x7fe4a8072358, >>>>>>>>> data=data@entry=0x7fe4a80723e0, >>>>>>>>> cleanup_fn=cleanup_fn@entry=0x7fe4bbd38a40 <socket_cleanup>) at >>>>>>>>> memory/unix/apr_pools.c:2276 >>>>>>>>> >>>>>>>>> it crashes here in apr: >>>>>>>>> 2276 if (c->data == data && c->plain_cleanup_fn == >>>>>>>>> cleanup_fn) { >>>>>>>>> >>>>>>>>> some lines before c becomes this >>>>>>>>> 2264 c = p->cleanups; >>>>>>>>> >>>>>>>>> p is: >>>>>>>>> (gdb) print *p >>>>>>>>> $1 = {parent = 0x256f138, child = 0x7fe46c0751c8, sibling = >>>>>>>>> 0x7fe4a8096888, ref = 0x7fe4a8069fe8, cleanups = 0x7fe478159748, >>>>>>>>> free_cleanups = 0x7fe478159788, allocator = 0x7fe4a803b490, >>>>>>>>> subprocesses = 0x0, abort_fn = 0x43da00 <abort_on_oom>, >>>>>>>>> user_data = 0x0, tag = 0x502285 "transaction", active = >>>>>>>>> 0x7fe478158d70, self = 0x7fe4a8072330, >>>>>>>>> self_first_avail = 0x7fe4a80723d0 "X#\a\250\344\177", pre_cleanups = >>>>>>>>> 0x7fe4a8072ab8} >>>>>>>>> >>>>>>>>> wouldn't the error mean that p->cleanups is NULL? >>>>>>>>> >>>>>>>>> (gdb) print *p->cleanups >>>>>>>>> $2 = {next = 0x7fe478159628, data = 0x7fe478159648, >>>>>>>>> plain_cleanup_fn = >>>>>>>>> 0x7fe4bbd2ffd0 <apr_unix_file_cleanup>, >>>>>>>>> child_cleanup_fn = 0x7fe4bbd2ff70 <apr_unix_child_file_cleanup>} >>>>>>>>> >>>>>>>>> So p->cleanups->data is 0x7fe478159648 and data is 0x7fe4a80723e0? >>>>>>>>> >>>>>>>>> I don't get why it's segfaulting >>>>>>>>> >>>>>>>>> Stefan >>>>>>>>> Am 21.01.2017 um 09:50 schrieb Yann Ylavic: >>>>>>>>>> Hi Stefan, >>>>>>>>>> >>>>>>>>>> On Sat, Jan 21, 2017 at 9:45 AM, Stefan Priebe >>>>>>>>>> <s.pri...@profihost.ag> >>>>>>>>>> wrote: >>>>>>>>>>> >>>>>>>>>>> after running the whole night. These are the only ones still >>>>>>>>>>> happening. >>>>>>>>>>> Should i revert the mpm patch to check whether it's the source? >>>>>>>>>> >>>>>>>>>> Yes please, we need to determine... >>>>>>>>>> >>>>>>>>>> Thanks, >>>>>>>>>> Yann. >>>>>>>>>> >>>>>>> >>>>>>> Stefan Eissing >>>>>>> >>>>>>> <green/>bytes GmbH >>>>>>> Hafenstrasse 16 >>>>>>> 48155 Münster >>>>>>> www.greenbytes.de >>>>>>> > > Stefan Eissing > > <green/>bytes GmbH > Hafenstrasse 16 > 48155 Münster > www.greenbytes.de >
