[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <005f01cc1240$02166210$06432630$@net>
Date: Sat, 14 May 2011 10:05:40 -0400
From: "Robert Harris" <rs904c@...scape.net>
To: <john-users@...ts.openwall.com>
Subject: RE: ITIMER_REAL' error in bench.c
John Users and Alex,
That worked!
Problem setting in bench.c:
#define _XOPEN_SOURCE /* for setitimer(2) */
Setting that fixed bench.c:
#define _XOPEN_SOURCE 500 /* for setitimer(2) */
Alex,
Why does the number 500 work? I tried other numbers like 2200, 200, and
400 and they all failed.
-----Original Message-----
From: Solar Designer [mailto:solar@...nwall.com]
Sent: Friday, May 13, 2011 12:14 PM
To: john-users@...ts.openwall.com
Subject: Re: [john-users] ITIMER_REAL' error in bench.c
Robert, Jim, all -
On Wed, May 11, 2011 at 05:47:08PM -0400, Robert Harris wrote:
> I get the following error trying to compile
>
> JtR 1.7.7 and JtR 1.7.7 jumbo 1 on Solaris x86.
>
> bench.c: In function `benchmark_format':
> bench.c:96: error: storage size of 'it' isn't known
> bench.c:140: warning: implicit declaration of function `setitimer'
> bench.c:140: error: `ITIMER_REAL' undeclared (first use in this function)
This must be a regression introduced with this change:
http://cvsweb.openwall.com/cgi/cvsweb.cgi/Owl/packages/john/john/src/bench.c
.diff?r1=1.8;r2=1.9
documented as:
"Define _XOPEN_SOURCE for setitimer(2). Problem seen when compiling
with musl."
musl is an alternative libc for Linux:
http://www.etalabs.net/musl/
Apparently, Solaris thinks that having _XOPEN_SOURCE merely defined
without requesting a specific version of the X/Open specification means
that we want a fairly old version.
Thus, a workaround is to remove "#define _XOPEN_SOURCE" (this breaks
compiles with musl, but should allow the code to compile on Solaris).
A proper fix is to figure out what to replace the "#define _XOPEN_SOURCE"
with. Maybe replace _XOPEN_SOURCE with _POSIX_SOURCE. Maybe specify a
version, like:
#define _XOPEN_SOURCE 500
This is what signals.c uses because it also needs siginterrupt(3).
Robert - can you please give "#define _XOPEN_SOURCE 500" a try? If it
works, this is likely the fix I'll commit.
To give another example, c3_fmt.c uses this weird combination:
#define _XOPEN_SOURCE 4 /* for crypt(3) */
#define _XOPEN_SOURCE_EXTENDED
#define _XOPEN_VERSION 4
#define _XPG4_2
which is needed (based on my experience with popa3d) to get crypt(3)
defined on a variety of systems.
We need to come up with similarly portable yet preferably minimal and
clean settings for getting setitimer(2) defined. Hopefully, a simple
"#define _XOPEN_SOURCE 500" will be it.
Thanks,
Alexander
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
