Segfault in mprime v25.11 build 2

Graff · 2009-12-02, 05:19

This has now happened a couple of times: an mprime job running under
64-bit Linux has segfaulted. The problem seems to always occur immediately
after a factor is found via P-1 factoring.

[Worker #7 Dec 1 14:36] M51409019 stage 2 complete. 633806 transforms. Time: 33.
[Worker #7 Dec 1 14:36] Starting stage 2 GCD - please be patient.
[Worker #7 Dec 1 14:39] Stage 2 GCD complete. Time: 150.078 sec.
[Worker #7 Dec 1 14:39] P-1 found a factor in stage #2, B1=610000, B2=17842500.
[Worker #7 Dec 1 14:39] M51409019 has a factor: 374449412920910477112607
Segmentation fault

The system is a quad-core Xeon E5520. The version of mprime is v25.11
build 2. I don't recall seeing other reports of segfaults.

Gareth

ckdo · 2009-12-02, 12:13

There was one here (v25.6 back then). Dunno if anything was done about it since...

lycorn · 2009-12-02, 17:22

That also happened with one of my machines a couple of weeks ago, immediately after finding a factor in P-1 Stage 1.
It was a Windows system, though. Prime95 version 9.

harlee · 2009-12-02, 22:01

Also runnng P-1 factor to low limits (4M range) on an XP 32-bit system and the program crashed the last three times a factor was found. This happened with both 25.11 and 25.9.

Graff · 2009-12-06, 05:41

Another segfault, another factor found by P-1.

Gareth

Graff · 2009-12-24, 03:47

Quote:

Originally Posted by Graff

Another segfault, another factor found by P-1.

Gareth

Several more factors > 2^64 found by P-1 and every time mprime crashes.

RichD · 2009-12-24, 16:33

Code:

[Worker #1 Dec 21 20:08] Starting stage 1 GCD - please be patient.
[Worker #1 Dec 21 20:11] Stage 1 GCD complete. Time: 193.497 sec.
[Worker #1 Dec 21 20:11] P-1 found a factor in stage #1, B1=610000.
[Worker #1 Dec 21 20:11] M51476981 has a factor: 1318460442877998352872047
mprime(4342,0x100787000) malloc: *** error for object 0x100516110: incorrect checksum for freed object - object was probably modified after being freed.
*** set a breakpoint in malloc_error_break to debug
Segmentation fault

I've seen it too on a 64-bit Mac system. At least it reports the factor before barfing. Then again, it shuts down all workers until to look at the progress of mprime which may be days.

sdbardwick · 2009-12-24, 17:11

Prime95 Win64 v25.9b4 exhibited the same problem (find factor via P-1, report factor, crash) at least once on my i5-750, Win7 64 bit system.

TheJudger · 2009-12-26, 22:51

I have noticed these segfaults asweel.

Linux, mprime x86-64 25.9b2, one worker thread, one CPU-core:
mprime segfaulted after P-1 factor was found and BEFORE the result was sent.

Found 146 P-1 factors on these machines (some were found with earlier versions of mprime (v25.7/v25.8)), at least two segfaults occured after a factor was found.

harlee · 2010-01-04, 21:51

Had two more crashes today on my XP 32-bit system. First time, it crashed after (or during) reporting the factor and the second crashed before reporting the factor.Both times, the results.txt file was update and the system still had the prime.spl. Verified the v5 server did get the first factor but it doesn't show the second one. Here are the entries for the prime log.

First crash:

PrimeNet success code with additional info:
CPU credit is 0.0291 GHz-days.
[Mon Jan 04 06:06:25 2010 - ver 25.11]

Code:

Sending result to server: UID: Squeeky_Squirrel/P4_2600, M5421019 has a factor: 1215446653308321070783, AID: 7A270AF99B44BC1BC4644B855805F220

URL: http://v5.mersenne.org/v5server/?v=0.95&px=GIMPS&t=ar&g=A25D8726F0C63524FDFC4CFFA33B40D5&k=7A270AF99B44BC1BC4644B855805F220&m=UID:+Squeeky_Squirrel/P4_2600,+M5421019+has+a+factor:+1215446653308321070783,+AID:+7A270AF99B44BC1BC4644B855805F220%0A&r=2&d=1&A=1&b=2&n=5421019&c=-1&B1=60000&B2=1035000&f=1215446653308321070783&fftlen=327680&ss=41&sh=6219FD505B4B07EA416E37E813A92E70

There is nothing in the prime.log file about sending info for the second factor.

2009-12-24, 17:11	#8
sdbardwick Aug 2002 North San Diego County 2²×3×67 Posts	Prime95 Win64 v25.9b4 exhibited the same problem (find factor via P-1, report factor, crash) at least once on my i5-750, Win7 64 bit system. Last fiddled with by sdbardwick on 2009-12-24 at 17:12

2010-01-04, 21:51	#10
harlee Sep 2006 Odenton, MD, USA 191 Posts	Had two more crashes today on my XP 32-bit system. First time, it crashed after (or during) reporting the factor and the second crashed before reporting the factor.Both times, the results.txt file was update and the system still had the prime.spl. Verified the v5 server did get the first factor but it doesn't show the second one. Here are the entries for the prime log. First crash: PrimeNet success code with additional info: CPU credit is 0.0291 GHz-days. [Mon Jan 04 06:06:25 2010 - ver 25.11] Code: Sending result to server: UID: Squeeky_Squirrel/P4_2600, M5421019 has a factor: 1215446653308321070783, AID: 7A270AF99B44BC1BC4644B855805F220 URL: http://v5.mersenne.org/v5server/?v=0.95&px=GIMPS&t=ar&g=A25D8726F0C63524FDFC4CFFA33B40D5&k=7A270AF99B44BC1BC4644B855805F220&m=UID:+Squeeky_Squirrel/P4_2600,+M5421019+has+a+factor:+1215446653308321070783,+AID:+7A270AF99B44BC1BC4644B855805F220%0A&r=2&d=1&A=1&b=2&n=5421019&c=-1&B1=60000&B2=1035000&f=1215446653308321070783&fftlen=327680&ss=41&sh=6219FD505B4B07EA416E37E813A92E70 There is nothing in the prime.log file about sending info for the second factor.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[Patch] CPU affinity prompt problem in mprime Linux / OS X build	Explorer09	Software	1	2017-03-01 02:34
msieve QS segfault when given large worktodo.ini	fivemack	Msieve	2	2011-10-25 20:11
segfault	junky	NFSNET Discussion	0	2006-07-06 03:25
segfault with large input on Opteron	sean	GMP-ECM	2	2005-08-15 09:59
PRP3 segfault with big numbers.	Washuu	Software	6	2005-04-07 17:08

2009-12-02, 05:19	#1
Graff Jul 2006 USA (UT-5) via UK (UT) 2²·59 Posts	Segfault in mprime v25.11 build 2 This has now happened a couple of times: an mprime job running under 64-bit Linux has segfaulted. The problem seems to always occur immediately after a factor is found via P-1 factoring. [Worker #7 Dec 1 14:36] M51409019 stage 2 complete. 633806 transforms. Time: 33. [Worker #7 Dec 1 14:36] Starting stage 2 GCD - please be patient. [Worker #7 Dec 1 14:39] Stage 2 GCD complete. Time: 150.078 sec. [Worker #7 Dec 1 14:39] P-1 found a factor in stage #2, B1=610000, B2=17842500. [Worker #7 Dec 1 14:39] M51409019 has a factor: 374449412920910477112607 Segmentation fault The system is a quad-core Xeon E5520. The version of mprime is v25.11 build 2. I don't recall seeing other reports of segfaults. Gareth

2009-12-02, 12:13	#2
ckdo Dec 2007 Cleves, Germany 2·5·53 Posts	There was one here (v25.6 back then). Dunno if anything was done about it since...

2009-12-02, 17:22	#3
lycorn "GIMFS" Sep 2002 Oeiras, Portugal 2×5×157 Posts	That also happened with one of my machines a couple of weeks ago, immediately after finding a factor in P-1 Stage 1. It was a Windows system, though. Prime95 version 9.

2009-12-02, 22:01	#4
harlee Sep 2006 Odenton, MD, USA BF₁₆ Posts	Also runnng P-1 factor to low limits (4M range) on an XP 32-bit system and the program crashed the last three times a factor was found. This happened with both 25.11 and 25.9.

2009-12-06, 05:41	#5
Graff Jul 2006 USA (UT-5) via UK (UT) 2²·59 Posts	Another segfault, another factor found by P-1. Gareth

2009-12-26, 22:51	#9
TheJudger "Oliver" Mar 2005 Germany 2×557 Posts	I have noticed these segfaults asweel. Linux, mprime x86-64 25.9b2, one worker thread, one CPU-core: mprime segfaulted after P-1 factor was found and BEFORE the result was sent. Found 146 P-1 factors on these machines (some were found with earlier versions of mprime (v25.7/v25.8)), at least two segfaults occured after a factor was found.