[tor-bugs] #5131 [Obfsproxy]: auditing obfsproxy

[Tor Bug Tracker & Wiki]

#5131: auditing obfsproxy
-------------------------+--------------------------------------------------
 Reporter:  ioerror      |          Owner:  asn         
     Type:  enhancement  |         Status:  needs_review
 Priority:  normal       |      Milestone:              
Component:  Obfsproxy    |        Version:              
 Keywords:  security     |         Parent:              
   Points:               |   Actualpoints:              
-------------------------+--------------------------------------------------

Comment(by ioerror):

 Replying to [comment:13 Sebastian]:
 > Replying to [comment:12 ioerror]:
 > > Replying to [comment:11 Sebastian]:
 > > > What about the other comments here? Are we cool with not adding a
 gpl, and with creating a lot of warnings for non-gcc compilers? If I use
 another compiler, how can I get rid of the warnings?
 > >
 > > I don't think we need to add a GPL but I've added it:
 > > [compile_hardening 86272e0] add GPL license for m4 macros
 >
 > My apologies, that wasn't pushed when I fetched the branch the first
 time. I see it's there now.

 Heh. I just added it. No need to say you're sorry. :)

 >
 > > Also - what warnings? If a non-gcc compiler is used and it does not
 support the flag, autoconf will catch that and not add the unsupported
 flag to the CFLAGS variable. Do you have a case where this is not true?
 Are you seeing warnings right now?
 >
 > No, autoconf will only catch errors and disable flags for errors. It
 keeps flags enabled that produce warnings. Here's an example from clang:
 >
 > {{{
 >
 > clang -DHAVE_CONFIG_H   -I. -I./src  -Wall -Wwrite-strings -Werror
 -I/opt/local//include -I/opt/local//include -g -O2 -D_FORTIFY_SOURCE=2
 -fstack-protector-all -fwrapv -fPIE --param ssp-buffer-size=1 -fno-strict-
 aliasing -fno-strict-overflow -MT unittest_obfs2.o -MD -MP -MF
 .deps/unittest_obfs2.Tpo -c -o unittest_obfs2.o `test -f
 'src/test/unittest_obfs2.c' || echo './'`src/test/unittest_obfs2.c
 > clang: warning: argument unused during compilation: '--param ssp-buffer-
 size=1'
 > clang: warning: argument unused during compilation: '-fno-strict-
 overflow'
 > mv -f .deps/unittest_obfs2.Tpo .deps/unittest_obfs2.Po
 > clang -Wall -Wwrite-strings -Werror -I/opt/local//include
 -I/opt/local//include -g -O2 -D_FORTIFY_SOURCE=2 -fstack-protector-all
 -fwrapv -fPIE --param ssp-buffer-size=1 -fno-strict-aliasing -fno-strict-
 overflow  -pie -o unittests tinytest.o unittest.o unittest_container.o
 unittest_crypt.o unittest_socks.o unittest_dummy.o unittest_managed.o
 unittest_obfs2.o libobfsproxy.a -L/usr/local/lib -levent
 -L/opt/local//lib/ -lcrypto
 > clang: warning: argument unused during compilation: '--param ssp-buffer-
 size=1'
 > clang: warning: argument unused during compilation: '-pie'
 > }}}


 Wait wait - what does configure do when you use clang? With gcc I have:
 {{{
 checking whether C compiler accepts -D_FORTIFY_SOURCE=2... yes
 checking whether C compiler accepts -fstack-protector-all... yes
 checking whether C compiler accepts -fwrapv... yes
 checking whether C compiler accepts -fPIE... yes
 checking whether C compiler accepts --param ssp-buffer-size=1... yes
 checking whether C compiler accepts -fno-strict-aliasing... yes
 checking whether C compiler accepts -fno-strict-overflow... yes
 checking whether the linker accepts -pie... yes
 checking whether the linker accepts -z relro... yes
 checking whether the linker accepts -z now... yes
 }}}

 Do you have the same output with clang? If so, I think that's a bug in the
 macro that checks if the compiler/linker actually accepts a given flag.

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/5131#comment:15>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online