Date: Sun, 2 Jun 2013 15:54:38 -0400 From: <jfoug@....net> To: john-dev@...ts.openwall.com Subject: Re: Link failure (bleeding win32-cygwin-x86-any ) A 32 bit generic 'might' trigger this. NOTE, all of the warnings / errors you list will be going away, once I can get my changes to dyna checked in. But I really do not want to do that 'untested'. There are some very large changes. I 100% re-wrote the SHA1 code. It is as fast, or faster for ALL sha1 formats, with the exception of sha1($p) But at this time, I have only been able to build sse. The -any (or generic) do properly build all of the dynamic stuff, BUT if I can not at least do want to get a working build ,that I can do a -test=0 before pushing the commit. I 'think' it is done and working, but really need to see the results of -test=0; Jim. ---- magnum <john.magnum@...hmail.com> wrote: > On 2 Jun, 2013, at 15:52 , jfoug@....net wrote: > > I think the recent change: f873e40d51 by Sayantan is causing this problem. Here is the linkage errors: > > > > DES_fmt.o:DES_fmt.c:(.text+0x117): undefined reference to `_DES_bs_all' > > DES_fmt.o:DES_fmt.c:(.text+0x25e): undefined reference to `_DES_bs_crypt_25' > > > > So it looks like something changed, so that these functions are being called, in builds which do not have them, or possibly it was code being removed that should have been left. I really do not know. The change is complex, simply looking at the git patch. > > I tried a generic build in case it would reproduce the above, but I got this: > > dynamic_fmt.c: In function 'FreeBSDMD5Crypt_setup_nonMMX': > dynamic_fmt.c:4546:92: warning: assignment makes integer from pointer without a cast [enabled by default] > md5_items.off[i] = md5_items.off[i] = md5_items.off[i] = md5_items.off[i] = &cp1[cp-cp1]; .... lots of compile errors
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.