Hmm.. memcpy here does not help. valgrind reports that tries to read a memory, which had previously been freed here:
> ==00:01:38:06.509 20943== Address 0x61A3430 is 0 bytes inside a block of > size 32 free'd > ==00:01:38:06.509 20943== at 0x4A0541E: free (vg_replace_malloc.c:233) > ==00:01:38:06.509 20943== by 0x30D8ED02B6: tdelete (in /lib64/libc-2.5.so) > ==00:01:38:06.509 20943== by 0x3C52A4D763: LockRelease (ow_locks.c:209) ... Here's an example of a similar guide to valgrind: http://www.seas.upenn.edu/~cis1xx/projects/CandC%2B%2B/Valgrind/GettingStarted/ Perhaps this error in the synchronization of threads... Serg. > Paul, I have tried to find any possible memory leaks in the flow below, but > I can't find any reason to the memory leak. > > ==00:01:38:06.536 20943== 360,480 (278,560 direct, 81,920 indirect) bytes in > 1,741 blocks are definitely lost in loss record 5 of 5 > ==00:01:38:06.536 20943== at 0x4A05809: malloc (vg_replace_malloc.c:149) > ==00:01:38:06.536 20943== by 0x3C52A52FBE: FS_OWQ_from_pn > (ow_parseobject.c:87) > ==00:01:38:06.536 20943== by 0x3C52A530AC: FS_OWQ_create_sibling > (ow_parseobject.c:53) > ==00:01:38:06.536 20943== by 0x3C52A58AFB: FS_r_sibling_F > (ow_sibling.c:49) > ==00:01:38:06.536 20943== by 0x3C52A27C7B: FS_slowtemp (ow_1820.c:282) > ==00:01:38:06.536 20943== by 0x3C52A53FAC: FS_read_lump (ow_read.c:472) > ==00:01:38:06.536 20943== by 0x3C52A542F4: FS_r_local (ow_read.c:427) > ==00:01:38:06.536 20943== by 0x3C52A54967: FS_r_given_bus (ow_read.c:232) > ==00:01:38:06.536 20943== by 0x3C52A54B01: FS_read_distribute > (ow_read.c:193) > ==00:01:38:06.536 20943== by 0x3C52A55223: FS_read_postparse > (ow_read.c:109) > ==00:01:38:06.536 20943== by 0x402AAD: ReadHandler (read.c:86) > ==00:01:38:06.536 20943== by 0x40339A: DataHandler (data.c:125) > > > ==00:01:38:06.508 20943== ERROR SUMMARY: 21 errors from 1 contexts > (suppressed: 4 from 1) > ==00:01:38:06.508 20943== > ==00:01:38:06.508 20943== 21 errors in context 1 of 1: > ==00:01:38:06.508 20943== Invalid read of size 8 > ==00:01:38:06.508 20943== at 0x3C52A4D8B7: LockGet (ow_locks.c:193) > ==00:01:38:06.508 20943== by 0x3C52A548A1: FS_r_given_bus (ow_read.c:231) > ==00:01:38:06.508 20943== by 0x3C52A54B01: FS_read_distribute > (ow_read.c:193) > ==00:01:38:06.508 20943== by 0x3C52A55223: FS_read_postparse > (ow_read.c:109) > ==00:01:38:06.508 20943== by 0x402AAD: ReadHandler (read.c:86) > ==00:01:38:06.508 20943== by 0x40339A: DataHandler (data.c:125) > ==00:01:38:06.508 20943== by 0x30DA6062F6: start_thread (in > /lib64/libpthread-2.5.so) > ==00:01:38:06.509 20943== by 0x30D8ED1E3C: clone (in /lib64/libc-2.5.so) > ==00:01:38:06.509 20943== Address 0x61A3430 is 0 bytes inside a block of > size 32 free'd > ==00:01:38:06.509 20943== at 0x4A0541E: free (vg_replace_malloc.c:233) > ==00:01:38:06.509 20943== by 0x30D8ED02B6: tdelete (in > /lib64/libc-2.5.so) > ==00:01:38:06.509 20943== by 0x3C52A4D763: LockRelease (ow_locks.c:209) > ==00:01:38:06.509 20943== by 0x3C52A5498B: FS_r_given_bus (ow_read.c:238) > ==00:01:38:06.509 20943== by 0x3C52A54B01: FS_read_distribute > (ow_read.c:193) > ==00:01:38:06.509 20943== by 0x3C52A55223: FS_read_postparse > (ow_read.c:109) > ==00:01:38:06.509 20943== by 0x402AAD: ReadHandler (read.c:86) > ==00:01:38:06.509 20943== by 0x40339A: DataHandler (data.c:125) > ==00:01:38:06.509 20943== by 0x30DA6062F6: start_thread (in > /lib64/libpthread-2.5.so) > ==00:01:38:06.509 20943== by 0x30D8ED1E3C: clone (in /lib64/libc-2.5.so) > > The result seems to be that LockGet() and LockRelease() fails as well.... > It might be generated by a un-successful temp-reading, and therefore the > memory isn't allocated and free'd correctly at some points. > > I have to go home now, and I won't be able to look more into this until next > year. > > /Christian > > > -----Original Message----- > From: Serg Oskin [mailto:s...@oskin.ru] > Sent: Tuesday, December 23, 2008 4:10 PM > To: owfs-developers@lists.sourceforge.net > Subject: Re: [Owfs-developers] general protection > > Running: > /usr/bin/valgrind --time-stamp=yes --leak-check=full --leak-resolution=high > --log-file=/tmp/owserver.log --trace-children=yes --undef-value-errors=yes > --verbose /usr/sbin/owserver -p 30003 -d /dev/ttyS0 -t 30 --foreground > --fatal_debug --fatal_debug_file=/tmp/owserver_fatal > > Not after you see the "Invalid read of size 8", or after Ctrl-C file > /tmp/owserver_fatal* is not created. > > > Serg. > >>> I have tried to stress-test owserver as much as I can here on >>> different platforms, and I can't reproduce the errors. >> In my case, the owserver used (by owread) 2xNagios every minute and > 2xCacti (2 threads each) every 5 minutes. >> As shown in the logs, this error does not occur immediately, but after a > while, in the latter case through 11 hours. >> In order to avoid a hardware error, I tried to completely change the > server - it did not help. >>> Can you start with the debug-output from all pthread-calls as well.. >>> /usr/sbin/owserver -p 30003 -d /dev/ttyS0 -t 30 --foreground >>> --fatal_debug --fatal_debug_file=/tmp/owserver_fatal >>> >>> If /tmp/owserver_fatal.pid is created and filled with debug-messages, >>> then some pthread-calls are failing on your server. >> --fatal_debug I added to /etc/sysconfig/owserver, but forgot to add when > run by valgrind. Now add ... >>> BTW: I will be on vacation between December 24th to 31'th, so I will >>> not be able to do much more on this after tonight. >> When owserver runs through valgrind, the system performs its >> functions... :) >> >> Merry Christmas! :) >> >> Serg. >> >>> /Christian >>> >>> >>> -----Original Message----- >>> From: Serg Oskin [mailto:s...@oskin.ru] >>> Sent: Tuesday, December 23, 2008 12:11 PM >>> To: owfs-developers@lists.sourceforge.net >>> Subject: Re: [Owfs-developers] general protection >>> >>> If I run owserver as daemon, he was in such cases crashed. >>> If I run well >>> valgrind ... /usr/sbin/owserver -p 30003 -d /dev/ttyS0 -t 30 >>> --foreground he continues to work. According to my experience this is >>> also a sign of memory leaks. >>> >>> This is beginning to occur when owserver started simultaneously use >>> several >>> (4-6) clients, rather than 1-2 as before. >>> >>> Serg. >>> >>>> I'm pretty clue-less why you get this error-message when >>>> "memcpy(&(pn->lock), &(opaque->key), sizeof(struct devlock *));" is >>> called. >>>> ==00:01:31:14.563 2670== Invalid read of size 8 >>>> ==00:01:31:14.563 2670== at 0x4C568B7: LockGet (ow_locks.c:193) >>>> ==00:01:31:14.563 2670== by 0x4C5D8A1: FS_r_given_bus (ow_read.c:231) >>>> ==00:01:31:14.563 2670== by 0x4C5DB01: FS_read_distribute >>> (ow_read.c:193) >>>> But... your owserver didn't seem to crash now? Is the major problem >>>> fixed now? >>>> I have checked in some various fixed some minutes ago, but nothing >>>> special that should affect this issue. >>>> >>>> /Christian >>>> >>>> >>>> -----Original Message----- >>>> From: Serg Oskin [mailto:s...@oskin.ru] >>>> Sent: Monday, December 22, 2008 7:50 PM >>>> To: owfs-developers@lists.sourceforge.net >>>> Subject: Re: [Owfs-developers] general protection >>>> >>>> I later corrected this. :) >>>> But this did not help. :( >>>> >>>> Serg. >>>> >>>>> Sorry... I made a typo in the code... the memcpy row should look >>>>> like >>>> this: >>>>> memcpy(&(pn->lock), &(opaque->key), sizeof(struct devlock *)); >>>>> >>>>> Forgot to get the pointer's to the variables, and therefore it >>>>> ended up with a segmentation fault instead... >>>>> Can you try to change the row and recompile with memcpy again? >>>>> >>>>> /Christian >>>>> >>>>> >>>>> -----Original Message----- >>>>> From: Serg Oskin [mailto:s...@oskin.ru] >>>>> Sent: Monday, December 22, 2008 12:13 PM >>>>> To: owfs-developers@lists.sourceforge.net >>>>> Subject: Re: [Owfs-developers] general protection >>>>> >>>>>> Now running CVS-version on 9:30 UTC 2008-12-22 ... >>>>> Results of file attachments. >>>>> >>>>> Serg. >>>>> >>>>>> Serg. >>>>>> >>>>>>> Hi Serg, >>>>>>> >>>>>>> Interesting log-files... It seems that your compiler generate >>>>>>> wrong >>>>> code... >>>>>>> ==00:06:33:57.651 2275== Invalid read of size 8 >>>>>>> ==00:06:33:57.651 2275== at 0x4C56559: LockGet (ow_locks.c:195) >>>>>>> >>>>>>> ==00:06:33:57.651 2275== Address 0x5A0D750 is 0 bytes inside a >>>>>>> block of size 32 free'd >>>>>>> ==00:06:33:57.651 2275== at 0x4A0541E: free >>> (vg_replace_malloc.c:233) >>>>>>> ==00:06:33:57.651 2275== by 0x30D8ED02B6: tdelete (in >>>>> /lib64/libc-2.5.so) >>>>>>> tsearch() seem to return a pointer to opaque, but "pn->lock = >>>>> opaque->key" >>>>>>> results "Invalid read of size 8"... ? >>>>>>> I have made some changes in the code, and that might fix the problem. > >>>>>>> Size of "struct devlock" might have be unknown at some places and >>>>>>> I have moved around the definitions a bit. >>>>>>> >>>>>>> Can you checkout the latest CVS-version and try it? >>>>>>> >>>>>>> >>>>>>> >>>>>>> If this doesn't work, you can try to edit >>>>>>> module/owlib/src/c/ow_locks.c and change two rows... >>>>>>> pn->lock = (struct devlock *)opaque->key; /* Serg: >>>>>>> Invalid read of size 8 */ >>>>>>> /* Why should a pointer compare fail? Unaligned memory? >>>>>>> Perhaps try to copy the pointer with memcpy() instead. >>>>>>> Will this help? >>>>>>> */ >>>>>>> //memcpy(pn->lock, opaque->key, sizeof(struct devlock >>>>>>> *)); >>>>>>> >>>>>>> Comment out the row "pn->lock = " and uncomment memcpy instead.... >>>>>>> Will this work for better for you? >>>>>>> It should remove the warning of "Invalid read of size 8" at >>>>>>> least, and perhaps everything will work then as well. >>>>>>> >>>>>>> BTW: Which platform are you using? I have a feeling that your >>>>>>> source isn't compiled with -m64, even if it should be... >>>>>>> Could you look at the host_cpu in config.log and "uname -a" >>>>>>> # grep host_cpu cvs/owfs/config.log # uname -a >>>>>>> >>>>>>> /Christian >>>>>>> >>>>>>> >>>>>>> -----Original Message----- >>>>>>> From: Serg Oskin [mailto:s...@oskin.ru] >>>>>>> Sent: Saturday, December 20, 2008 10:33 PM >>>>>>> To: owfs-developers@lists.sourceforge.net >>>>>>> Subject: Re: [Owfs-developers] general protection >>>>>>> >>>>>>> Ctrl-C pressed. >>>>>>> >>>>>>> >>>>>>>> Tried - the result of old. >>>>>>>> Version: from CVS at Dec 20 2008 12:00 UTC. >>>>>>>> >>>>>>>> Message in /tmp/owfs_fatal I received only once during the "kill >>>>>>>> owserver_pid": >>>>>>>> ow_connect.c:322 mutex_destroy failed rc=16 [Device or resource >>>>>>>> busy] >>>>>>>> >>>>>>>> Serg. >>>>>>> >>>>>>> >>>>>>> __________ Information from ESET NOD32 Antivirus, version of >>>>>>> virus signature database 3709 (20081220) __________ >>>>>>> >>>>>>> The message was checked by ESET NOD32 Antivirus. >>>>>>> >>>>>>> http://www.eset.com >>>>>>> >>>>>>> >>>>>>> >>>>>>> __________ Information from ESET NOD32 Antivirus, version of >>>>>>> virus signature database 3709 (20081220) __________ >>>>>>> >>>>>>> The message was checked by ESET NOD32 Antivirus. >>>>>>> >>>>>>> http://www.eset.com >>>>>>> >>>>>>> >>>>>>> >>>>>>> ----------------------------------------------------------------- >>>>>>> --- >>>>>>> - >>>>>>> --------- _______________________________________________ >>>>>>> Owfs-developers mailing list >>>>>>> Owfs-developers@lists.sourceforge.net >>>>>>> https://lists.sourceforge.net/lists/listinfo/owfs-developers >>>>>>> >>>>>>> >>>>>> ------------------------------------------------------------------ >>>>>> --- >>>>>> - >>>>>> -------- _______________________________________________ >>>>>> Owfs-developers mailing list >>>>>> Owfs-developers@lists.sourceforge.net >>>>>> https://lists.sourceforge.net/lists/listinfo/owfs-developers >>>>>> >>>>>> >>>>> >>>>> >>>>> __________ Information from ESET NOD32 Antivirus, version of virus >>>>> signature database 3710 (20081222) __________ >>>>> >>>>> The message was checked by ESET NOD32 Antivirus. >>>>> >>>>> http://www.eset.com >>>>> >>>>> >>>>> >>>>> __________ Information from ESET NOD32 Antivirus, version of virus >>>>> signature database 3710 (20081222) __________ >>>>> >>>>> The message was checked by ESET NOD32 Antivirus. >>>>> >>>>> http://www.eset.com >>>>> >>>>> >>>>> >>>>> ------------------------------------------------------------------- >>>>> --- >>>>> -------- _______________________________________________ >>>>> Owfs-developers mailing list >>>>> Owfs-developers@lists.sourceforge.net >>>>> https://lists.sourceforge.net/lists/listinfo/owfs-developers >>>>> >>>>> >>>> >>>> >>>> __________ Information from ESET NOD32 Antivirus, version of virus >>> signature >>>> database 3710 (20081222) __________ >>>> >>>> The message was checked by ESET NOD32 Antivirus. >>>> >>>> http://www.eset.com >>>> >>>> >>>> >>>> >>> --------------------------------------------------------------------- >>> ------- >>> -- >>>> _______________________________________________ >>>> Owfs-developers mailing list >>>> Owfs-developers@lists.sourceforge.net >>>> https://lists.sourceforge.net/lists/listinfo/owfs-developers >>>> >>>> >>> --------------------------------------------------------------------- >>> ------- >>> -- >>> _______________________________________________ >>> Owfs-developers mailing list >>> Owfs-developers@lists.sourceforge.net >>> https://lists.sourceforge.net/lists/listinfo/owfs-developers >>> >>> >>> --------------------------------------------------------------------- >>> --------- _______________________________________________ >>> Owfs-developers mailing list >>> Owfs-developers@lists.sourceforge.net >>> https://lists.sourceforge.net/lists/listinfo/owfs-developers >>> >>> >> ---------------------------------------------------------------------- >> -------- _______________________________________________ >> Owfs-developers mailing list >> Owfs-developers@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/owfs-developers >> >> > > > __________ Information from ESET NOD32 Antivirus, version of virus signature > database 3713 (20081223) __________ > > The message was checked by ESET NOD32 Antivirus. > > http://www.eset.com > > > > ------------------------------------------------------------------------------ > _______________________________________________ > Owfs-developers mailing list > Owfs-developers@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/owfs-developers > > ------------------------------------------------------------------------------ _______________________________________________ Owfs-developers mailing list Owfs-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/owfs-developers
