Hi, I don't know all the details, but it failed on x86-32 and it did not on x86-64. You can have a look at this change, which fixed the problem:
http://hg.youterm.com/radare/rev/700fb4cb81ca -- Sebastian On Wed, Nov 04, 2009 at 01:41:58PM +0100, Lluís Batlle wrote: > The fault, although triggered by -O2, may not be compiler's. > > Sebastian, do you remember more details? I'm interested in that. gcc > version, platform ... > Did gmp build well with that compiler, if you tried? > > Regards, > Lluís. > > 2009/11/4 pancake <[email protected]>: > > Heheh, ok, it was a gcc fault :) stupid buggy optimizations ;) > > > > Sebastian Reichel wrote: > >> > >> Hi, > >> > >> I'm very sorry, I guess you are using my package from the Debian > >> repository. You should update to a more recent radare version, > >> because your bug has been fixed upstream. It seems the DD, which > >> uploads my packages forgot my last update :( > >> I will prepare a new package once I've got enough time to do so. > >> As alternative to upgrading to a more recent radare version you > >> can also disable code analysis by putting > >> e file.analyzing=false > >> in your .radarerc, but I guess upgrading is better :P > >> > >> @pancake: The bug was 32bit only when compiling with -O2 ;) > >> > >> -- Sebastian > >> > >> On Wed, Nov 04, 2009 at 08:55:11AM +0100, pancake wrote: > >> > >>> > >>> Looks like a problem related in cons.c > >>> > >>> Are you using the debian package? Or a snapshot compiled by you? I > >>> would recommend you to get it from mercurial to ensure that the bug > >>> is not yet fixed in mainstream. > >>> > >>> If it is compiled by you, can you provide the gcc version and cflags > >>> used? > >>> > >>> To debug the segfault I will use valgrind and gdb. If the binary has > >>> been built with debug info you will get a backtrace using the bt > >>> command in gdb. > >>> > >>> If the bug cannot be reproduced by gdb, try by running it in > >>> valgrind. If you can't get it in any of these ways just enable core > >>> files and do some postmortem analysis. > >>> > >>> Send those backtraces to the mailing list, so we can point to the > >>> problem. > >>> > >>> Btw is this happening on 32 or 64 bits? > >>> > >>> Let me know :) > >>> > >>> Thanks for reporting > >>> > >>> On Nov 4, 2009, at 5:55 AM, "jcyang" <[email protected]> wrote: > >>> > >>> > >>>> > >>>> Hi, > >>>> I'm new to linux revese engineering,and even newer to radare.So I > >>>> tried to follow the radare book step by step,but I failed on the > >>>> first step. > >>>> > >>>> When I run 'radare -d /bin/ls',the radare failed in the analyze > >>>> code section with segment failed.Unexpectedly,when I try to ues > >>>> 'typescript' to log the error message,radare runs successfully,but > >>>> if I exit the typescript mode,then the error message turns out > >>>> agian. > >>>> > >>>> I am running debian lenny ,radare 2009.07.22. > >>>> > >>>> Whether it is a bug?If it is,what I need to provide to help fix > >>>> the bug? > >>>> > >>>> thanks. > >>>> > >>>> ------------------------------------------------------------------------ > >>>> > >>>> _______________________________________________ > >>>> radare mailing list > >>>> [email protected] > >>>> http://lists.nopcode.org/listinfo.cgi/radare-nopcode.org > >>>> > > > > _______________________________________________ > > radare mailing list > > [email protected] > > http://lists.nopcode.org/listinfo.cgi/radare-nopcode.org > > > _______________________________________________ > radare mailing list > [email protected] > http://lists.nopcode.org/listinfo.cgi/radare-nopcode.org
signature.asc
Description: Digital signature
_______________________________________________ radare mailing list [email protected] http://lists.nopcode.org/listinfo.cgi/radare-nopcode.org
