Bug 164251 - segfault on startup on x86_64
segfault on startup on x86_64
Status: CLOSED CURRENTRELEASE
Product: Fedora
Classification: Fedora
Component: john (Show other bugs)
4
x86_64 Linux
medium Severity high
: ---
: ---
Assigned To: Till Maas
Fedora Extras Quality Assurance
:
Depends On:
Blocks: FE5Target
  Show dependency treegraph
 
Reported: 2005-07-26 05:11 EDT by Noa Resare
Modified: 2007-11-30 17:11 EST (History)
3 users (show)

See Also:
Fixed In Version: 1.7.0.2-3
Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
Environment:
Last Closed: 2007-03-01 14:59:24 EST
Type: ---
Regression: ---
Mount Type: ---
Documentation: ---
CRM:
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---


Attachments (Terms of Use)
autodetect platforms where sizeof(void*) == sizeof(long) (1015 bytes, patch)
2005-07-26 05:11 EDT, Noa Resare
no flags Details | Diff
Fix odd chars and syntax issues (1.17 KB, patch)
2007-02-27 11:52 EST, Ville Skyttä
no flags Details | Diff

  None (edit)
Description Noa Resare 2005-07-26 05:11:30 EDT
Description of problem:

Due to an incorrect assumpution about the size of pointers on the x86_64
platform john segfaults on startup on that platform.

Version-Release number of selected component (if applicable):
john-1.6-4

How reproducible:
always

Steps to Reproduce:
1. unshadow /etc/passwd /etc/shadow > to_crack
2. john to_crack

  
Actual results:

Program received signal SIGSEGV, Segmentation fault.
0x00000000004187d5 in ldr_init_database (db=0x537660, options=0x55a810)
    at loader.c:78
78                      issep_map[(ARCH_INDEX)*pos] = 1;


Expected results:
Some password cracking

Additional info:

The attached patch has been tested on x86_64 and vanilla x86, and I have
verified that it actually cracks simple passwords on both platforms.

I haven't looked into submitting this upstream yet.
Comment 1 Noa Resare 2005-07-26 05:11:30 EDT
Created attachment 117149 [details]
autodetect platforms where sizeof(void*) == sizeof(long)
Comment 2 Gavin Henry 2005-07-26 05:14:17 EDT
Thanks for this report and the patch!!! 

I will test it later this week and get the new RPMS upstream.

Thanks again,

Gavin.
Comment 3 Ville Skyttä 2006-09-23 13:26:01 EDT
Ping?  This still happens in FC5, john-1.6-4.
Comment 4 Gavin Henry 2006-09-23 14:28:12 EDT
Haven't I done this yet? Sorry. Will get on it.
Comment 5 Hans de Goede 2006-11-13 07:45:38 EST
ping again?
Comment 6 Gavin Henry 2006-11-13 07:48:02 EST
Should get this nailed this week.

Thanks for your patience.

Gavin.
Comment 7 Gavin Henry 2006-12-05 04:40:24 EST
Sending for rebuild now.
Comment 8 Ville Skyttä 2006-12-06 07:39:26 EST
Build failures, looks like it was not even tested if the %prep stage succeeds
(let alone the patched functionality) before submitting the build job :/

http://buildsys.fedoraproject.org/build-status/job.psp?uid=22950
http://buildsys.fedoraproject.org/build-status/job.psp?uid=22951
Comment 9 Gavin Henry 2006-12-06 09:17:27 EST
Sorry about that. Will test again this week.

Gavin.
Comment 10 Till Maas 2007-01-10 09:50:15 EST
john 1.7.0.2-2 has been built for development as job 25388

Please check whether or not john works now on x86_64.
Comment 11 Till Maas 2007-02-26 18:44:31 EST
I will assume the bug is fixed when you do not respond soon.
Comment 12 Ville Skyttä 2007-02-27 02:53:39 EST
Well, the package is not available in the devel x86_64 repo.
Comment 13 Till Maas 2007-02-27 03:37:14 EST
The buildsystem told me the build succeeded, what can I do now?

http://buildsys.fedoraproject.org/build-status/job.psp?uid=25388
Comment 14 Ville Skyttä 2007-02-27 11:52:08 EST
Created attachment 148873 [details]
Fix odd chars and syntax issues

Dunno, but the binary package is missing from the ppc repository too, ditto in
the FC-6 repo.	Before doing anything else, I'd suggest taking a look at these
findings:

There seems to be an odd character of some kind on lines 64 and 83, the "%"
preceding {target_mmx} is not a regular percent sign - I suppose that will
cause some problems.

There is also syntactic problems in the specfile on line 47 (and I suppose on
line 64 and 85 too, they're essentially the same as line 47 after the percent
sign problem has been corrected) when eval'd on x86_64:

$ rpm -q --specfile john.spec
error: syntax error in expression
error: john.spec:47: parseExpressionBoolean returns -1

The attached patch should fix all these issues.  And looks like a local x86_64
FC6 build no longer crashes.
Comment 15 Till Maas 2007-02-27 14:52:52 EST
I applied your patch, the build succeeded, I hope it will be published, too:

http://buildsys.fedoraproject.org/logs/fedora-development-extras/28366-john-1.7.0.2-3.fc7/
Comment 16 Ville Skyttä 2007-02-27 16:32:49 EST
Looks good so far, at least all binary packages are in the needsign queue as
expected.  The FC6 branch needs the fix too, BTW.
Comment 17 Till Maas 2007-03-01 14:59:24 EST
(In reply to comment #16)
> Looks good so far, at least all binary packages are in the needsign queue as
> expected.  The FC6 branch needs the fix too, BTW.

FC6 has been rebuilt, too.

Note You need to log in before you can comment on or make changes to this bug.