/[path-sandbox]/trunk/ChangeLog
Gentoo

Contents of /trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 40 - (hide annotations) (download)
Mon Mar 21 23:30:13 2005 UTC (9 years, 9 months ago) by azarah
File size: 16566 byte(s)
Allow lchown a symlink in write-allowed path pointing to write-denied target.

1 ferringb 2 # ChangeLog for Path Sandbox
2     # Copyright 1999-2004 Gentoo Foundation; Distributed under the GPL v2
3     # $Header$
4    
5 azarah 40 22 Mar 2005; Martin Schlemmer <azarah@gentoo.org> libsandbox.c:
6     Allow lchown a symlink in write-allowed path pointing to write-denied
7     target.
8    
9 genone 39 21 Mar 2005; Marius Mauch <genone@gentoo.org> libsandbox.c:
10     Also show resolved symlink names in the log.
11    
12 azarah 37 14 Mar 2005; Martin Schlemmer <azarah@gentoo.org> Makefile.am, libsandbox.c:
13     Seems -nostdlib was the problem with the constructor/destructor - remove it
14     from Makefile.am, and change the constructor/destructor names again.
15    
16 azarah 36 14 Mar 2005; Martin Schlemmer <azarah@gentoo.org> libsandbox.c:
17     Also rename the _init() and _fini() declarations.
18    
19 azarah 35 14 Mar 2005; Martin Schlemmer <azarah@gentoo.org> libsandbox.c, sandbox.c:
20     Fixup the constructor/destructor function names again (they should be _init()
21     and _fini() it seems, and not being called caused sandbox_lib_path to be
22     unset, and thus breaking the execve() wrapper's LD_PRELOAD protection).
23     Add both the path in given SANDBOX_x variable, as well as its symlink
24     resolved path in init_env_entries(). Modify filter_path() to be able to
25     resolve paths without resolving symlinks, as well as to be able to resolve
26     symlinks. Fix a possible segfault in check_access(). Add symlink resolving
27     to check_access() resolving bug #31019. Add 'hack' for unlink, as the fix
28     for bug #31019 cause access violations if we try to remove a symlink that is
29     not in protected path, but points to a protected path. Fix a memory leak in
30     sandbox.c (sandbox_pids_file in main()). Fix the realpath() calls in main()
31     (sandbox.c) being unchecked. Fix the debug logname not having the pid in it
32     (pid_string was uninitialized). General syntax cleanups.
33    
34 ferringb 34 09 Mar 2005; Brian Harring <ferringb@gentoo.org> sandbox.c: Fixed the
35     infamous "pids file is not a regular file" w/out newline bug.
36    
37 ferringb 33 09 Mar 2005; Brian Harring <ferringb@gentoo.org> Makefile.am, configure.in:
38     Correct libc_version path detection, since it was screwing up if libdir !=
39     "/lib/".
40    
41     02 Mar 2005; Martin Schlemmer <azarah@gentoo.org> libsandbox.c:
42 azarah 32 Hack to make sure sandboxed process cannot remove a device node, bug #79836.
43    
44 ferringb 33 02 Mar 2005; Martin Schlemmer <azarah@gentoo.org> Makefile.am:
45 azarah 31 Fix symbols.in not added to dist.
46    
47 ferringb 33 02 Mar 2005; Martin Schlemmer <azarah@gentoo.org> Makefile.am, canonicalize.c,
48 azarah 30 getcwd.c, libsandbox.c, sandbox.c, sandbox.h, sandbox_futils.c:
49     White space fixes.
50    
51 ferringb 33 02 Mar 2005; Martin Schlemmer <azarah@gentoo.org> Makefile.am, canonicalize.c,
52 azarah 30 configure.in, getcwd.c, libsandbox.c, symbols.in:
53 azarah 29 Fix inverse test logic in canonicalize.c, use a strncpy. Fix gcc warning in
54     getcwd.c. Add symbols.in and logic to Makefile.am to generate symbol versions
55     for glibc and other libc's that use this. Update libsandbox.c to use these
56     symbol versions if available. Fix exec wrapper to re-export LD_PRELOAD if the
57     process unset it.
58    
59 ferringb 33 01 Mar 2005; Brian Harring <ferringb@gentoo.org> libsandbox.c:
60 ferringb 28 killed off _init and _fini in favor of
61     void __attribute__ ((constructor)) init_func and
62     void __attribute__ ((destructor)) closing_func. _(init|func) were deprecated.
63    
64 ferringb 26 06 Dec 2004; Brian Harring <ferringb@gentoo.org> Makefile.am, libsandbox.c,
65     canonicalize.c, getcwd.c: Fixed compilation *again*. Hopefully cvs is done
66     having the hick-ups.
67    
68 ferringb 24 04 Dec 2004; Brian Harring <ferringb@gentoo.org> libsandbox.c, getcwd.c,
69     Makefile.am: Fixed compilation.
70    
71 ferringb 18 01 Dec 2004; Brian Harring <ferringb@gentoo.org> aclocal.m4:
72     Gutted the bugger so it stops checking for a c++ and fortran compiler.
73     Do *not* regenerate aclocal.m4 for making a release until a better
74     solution is created.
75    
76 ferringb 16 20 Nov 2004; Brian Harring <ferringb@gentoo.org> Makefile.am, sandbox_futils.c:
77     Removal of more hardcoded paths.
78    
79 ferringb 13 20 Nov 2004; Brian Harring <ferringb@gentoo.org> Makefile.am, configure.in,
80     sandbox_futils.c: tweaks to install sandbox.bashrc, and use it.
81    
82     19 Nov 2004; Brian Harring <ferringb@gentoo.org>:
83 ferringb 6 Sandbox is now autotooled, create-localdecls needs to be killed and the code
84     shifted into configure.in. Currently builds *one* libsandbox.so- if multiple
85     are desired (-m64 and -m32 for amd64), the ebuild should do it (imo).
86     To get to a point of testing, automake && autoconf; created requisite files w/
87     a(utomake|clocal)-1.8, and autoconf 2.59. Installs to /usr/, instead of
88     /lib and /usr/lib/portage/bin.
89    
90 ferringb 2 14 Nov 2004; Brian Harring <ferringb@gentoo.org> libsandbox.c, sandbox.c:
91     closing out bug #70225, potential overflow of the sandbox_pids_file var.
92    
93     07 Nov 2004; Brian Harring <ferringb@gentoo.org> libsandbox.c: c99 standard,
94     (think it was at least) allows intermixing of code and data segments. bug #70351
95     should be fixed by this.
96    
97     03 Nov 2004; Brian Harring <ferringb@gentoo.org> libsandbox.c, sandbox_futils.c:
98     futils fix from bug #65201 via solar, and libsandbox log path checks via #69137
99    
100     02 Aug 2004; Nicholas Jones <carpaski@gentoo.org> libsandbox.c: Code from
101     Seth Robertson that tracked down all adjuct flags for read operations that
102     do not invoke a write operation.
103    
104     04 Apr 2004; Nicholas Jones <carpaski@gentoo.org> libsandbox.c, sandbox.c:
105     Another fix from jstubbs regarding a free() on a stack variable for the
106     environment -- tracking now prevents extraneous free()'s segfault.
107    
108     04 Apr 2004; Nicholas Jones <carpaski@gentoo.org> libsandbox.c, sandbox.c:
109     J. Stubbs tracked down a new bug where mkdir was failing to the patch on
110     the lstat in mkdir... it now only returns 0 or -1 as documented for mkdir.
111     Also remove the errno = ESUCCESS settings as documentation points out that
112     a library isn't allowed to do that.
113    
114     04 Apr 2004; Nicholas Jones <carpaski@gentoo.org> libsandbox.c: Added a
115     file_security_check() function to check random potential exploits on files
116     that sandbox is to load and read -- Normally sandboxpids.tmp. This fixes
117     the 'system-crippling' exploits (bug 21923) and catches a few other
118     potential problems.
119    
120     20 Mar 2004; Nicholas Jones <carpaski@gentoo.org> Makefile: Updates for
121     32/64 bit sandbox. Made CC and LD '?=' values to allow passed in CC to work.
122    
123     20 Mar 2004; Nicholas Jones <carpaski@gentoo.org> libsandbox.c:
124     bug 42048 -- Fixed the lstat/errno conditions for mkdir <caleb@g.o>.
125     Added the 64/32 bit sandbox patch for AMD64 bug 32963 <brad/azarah>.
126    
127     29 Feb 2004; Martin Schlemmer <azarah@gentoo.org> sandbox.c, sandbox_futils.c :
128     Fix permissions and group of pids file and logs. Permissions should be 0664
129     and group should be 'portage'. Bug #34260.
130    
131     28 Feb 2004; Martin Schlemmer <azarah@gentoo.org> libsandbox.c :
132     Besides a small cleanup, redo how we replace LD_PRELOAD in the environ passed
133     to the real execve (in our execve wrapper). Seems that on some arches (sparc
134     among others) do not allow us to tamper with the readonly copy passed to
135     execve, so pass our own copy of the environment. Bug #42290.
136    
137     11 Jan 2004; Nicholas Jones <carpaski@gentoo.org> create-decls:
138     Changed tail to head and added a notice about duration of glibc check.
139    
140     21 Dec 2003; Nicholas Jones <carpaski@gentoo.org> create-decls:
141     Changed the glibc subversion check to use /usr/bin/* instead of /bin/sh
142     as there isn't a guarentee that it is dynamic.
143    
144     02 Nov 2003; Martin Schlemmer <azarah@gentoo.org> libsandbox.c :
145     If 'file' passed to before_syscall(const char *func, const char *file) is
146     invalid, we should set errno to ENOENT, and not EINVAL. This should
147     close bug #32238.
148    
149     14 Oct 2003; Martin Schlemmer <azarah@gentoo.org> libsandbox.c :
150     Fix a bug that occurs mainly on 64bit arch, where the file passed to
151     the functions we wrap, is invalid, and then cause canonicalize to pass
152     garbage to before_syscall(), thanks to great detective work from
153     Andrea Luzzardi <al@sig11.org> (bug #29846).
154    
155     13 Oct 2003; Martin Schlemmer <azarah@gentoo.org> create-localdecls :
156     Add a uClibc detection patch from Peter S. Mazinger <ps.m@gmx.net>.
157    
158     13 Oct 2003; Martin Schlemmer <azarah@gentoo.org> libsandbox.c :
159     Fix a bug in libsandbox.c 's checking in the rename wrapper - it basically
160     only checked the destination patch, and not the source, so we could move
161     a protected file to a unprotected directory, and then delete/modify it.
162     Thanks to Andrea Luzzardi (scox) <al@sig11.org>, bug #30992, for this fix.
163    
164     12 Oct 2003; Nicholas Jones <carpaski@gentoo.org> sandbox.c :
165     Added python2.3 to the predict section/variable.
166    
167     28 Sep 2003; Martin Schlemmer <azarah@gentoo.org> libsandbox.c, sandbox.c,
168     sandbox.h, sandbox_futils.c :
169     Add support to set the pids file via SANDBOX_PIDS_FILE at startup. If
170     it is not set, it will revert to its old value.
171    
172     27 Sep 2003; Martin Schlemmer <azarah@gentoo.org> libsandbox.c :
173     Fix our mkdir wrapper to check if the dir exist, and return EEXIST if so,
174     rather than failing with a violation, bug #29748.
175    
176     27 Jul 2003; Martin Schlemmer <azarah@gentoo.org> libsandbox.c :
177     Fix canonicalize() to ignore calls with path = "".
178    
179     27 Jul 2003; Martin Schlemmer <azarah@gentoo.org> getcwd.c, libsandbox.c,
180     sandbox_futils.c, canonicalize.c :
181     Once again coreutils fails, as my systems had 2.5 kernel, the getcwd system
182     call handled strings larger than PATH_MAX (bug #21766). It however does not
183     work the same on 2.4 kernels.
184    
185     To fix, I added the posix implementation of getcwd() (from glibc cvs) that
186     do not need the system call. We use the default getcwd() function via a
187     wrapper (egetcwd), and then lstat the returned path. If lstat fails, it
188     means the current directory was removed, OR that the the system call for
189     getcwd failed (curious is that it do not fail and return NULL or set
190     errno, but rather just truncate the retured directory - usually from the
191     start), and if so, we use the generic getcwd() function (__egetcwd). Note
192     that we do not use the generic version all the time, as it calls lstat()
193     a great number of times, and performance degrade much.
194    
195     29 Jun 2003; Martin Schlemmer <azarah@gentoo.org> create-localdecls,
196     libsandbox.c :
197     Make sure SB_PATH_MAX will not wrap. Fix two possible memory leaks.
198    
199     22 Jun 2003; Martin Schlemmer <azarah@gentoo.org> libsandbox.c, canonicalize.c
200     create-localdecls :
201     When checking path names of files accessed, we need to canonicalize it, else
202     it may be a symlink in a 'write allowed' directory pointing to a file in a
203     directory we should not have write access to.
204    
205     With something like coreutils-5.0, we have two problems:
206     1) One of the tests checks if getcwd() can return a path longer than
207     PATH_MAX. This test then tries to create a dir which even while
208     created local (mkdir("conftest2")), it ends up being resolved with
209     a name that is much larger than PATH_MAX. The problem now is that
210     canonicalize() have undefined behaviour when the path was too long
211     (returned wrongly truncated paths, etc), and pass the wrong path to
212     before_syscall() (causing the bogus sandbox violations).
213     2) The ecanonicalize() function we used, along with the canonicalize()
214     function did not support longer than PATH_MAX. This is partly a
215     cause for 1), but the error checking (rather lack of it) of calls
216     to erealpath() in canonicalize() was the prime reason for 1).
217    
218     As we do not use this canonicalized name to call the function, we resolve this
219     by fixing canonicalize() to do better error checking, and ecanonicalize() as
220     well as all functions in libsandbox.c to use a PATH_MAX of 'PATH_MAX * 2'.
221     While they will resolve paths properly now, and can check if a write/read is
222     allowed, the functions called from the sandboxed environment will still work
223     as expected.
224    
225     This should resolve bug #21766.
226    
227     06 Apr 2003; Martin Schlemmer <azarah@gentoo.org> libsandbox.c :
228     For some reason sandbox fails with a 'open_wr' if you run 'locale -a' under
229     it (bug #16298).
230    
231     Problem is that for some reason locale fopen's locale.alias with mode "rm".
232    
233     -------------------------------------------------------
234     nosferatu root # grep fopen locale.log
235     fopen("/usr/share/locale/locale.alias", "rm"ACCESS DENIED open_wr: /usr/share/locale/locale.alias
236     nosferatu root #
237     --------------------------------------------------------
238    
239     I checked the source of locale, but it have fopen with mode 'r', so
240     not sure where the "rm" mode comes from. Anyhow, changed the check in
241     before_syscall_open_char() to also see mode "rm" as readonly.
242    
243     23 Feb 2003; Martin Schlemmer <azarah@gentoo.org> create-localdecls :
244    
245     Add glibc-2.3 support.
246    
247     22 Feb 2003; Martin Schlemmer <azarah@gentoo.org> sandbox.c :
248    
249     Some /etc/ld.so.preload fixes. Just changed the #if defines to cover all
250     operations releated to preload, as well as only try to modify ld.so.preload
251     if we can. Also modify to write the pid to /tmp/sandboxpids.tmp even when
252     not using ld.so.preload. Fix to not write this instance of sandbox's pid
253     to /tmp/sandboxpids.tmp on exit if this is not the last sandbox running.
254    
255     22 Feb 2003; Nicholas Jones <carpaski@gentoo.org> Makefile :
256    
257     Changed the LD to CC for hppa.
258    
259     22 Feb 2003; Nicholas Jones <carpaski@gentoo.org> create-localdecls :
260    
261     Killed the previous changes I made.
262    
263     17 Feb 2003; Nicholas Jones <carpaski@gentoo.org> create-localdecls :
264    
265     Added parisc to BROKEN_RTLD_ARCHLIST to see if it we can fix the relocation probs.
266    
267     09 Jan 2003; J Robert Ray <jrray@gentoo.org> sandbox.c :
268    
269     Don't segfault if $HOME isn't set, set $HOME to "/" instead. Fixes bug 10868.
270    
271     16 Dec 2002; Martin Schlemmer <azarah@gentoo.org> create-localdecls :
272    
273     Fix memory leak for mips, bug #12236. Thanks to Torgeir Hansen <torgeir@trenger.ro>
274     for this fix.
275    
276     4 Dec 2002; J Robert Ray <jrray@gentoo.org> sandbox.h sandbox_futils.c :
277    
278     sandbox_futils defined a dirname() function that was masking the same
279     function in glibc and was broken (e.g.: SANDBOX_DIR was being set to
280     '/usr/lib/portage/bi/'). Fixed function to return expected results and
281     renamed it to sb_dirname() to no longer mask the glibc function. Closes bug
282     11231.
283    
284     4 Dec 2002; Martin Schlemmer <azarah@gentoo.org> :
285    
286     Fix a segfault in libsandbox.c if canonicalize() was called with
287     first parameter = NULL.
288    
289     1 Sep 2002; Martin Schlemmer <azarah@gentoo.org> :
290    
291     Fix my braindead 'return 1;' in a void function. Updated sandbox.c,
292     cleanup() for this.
293    
294     Change cleanup() in sandbox.c not to exit with fail status if
295     the pidsfile is missing. We really should still display sandbox
296     violations if they occured.
297    
298     31 Aug 2002; Martin Schlemmer <azarah@gentoo.org> :
299    
300     Update cleanup() in sandbox.c to remove the PIDSFILE if this is
301     the last sandbox running.
302    
303     25 Aug 2002; Martin Schlemmer <azarah@gentoo.org> :
304    
305     Major cleanups to mainly libsandbox.c again.
306    
307     22 Aug 2002; Martin Schlemmer <azarah@gentoo.org> :
308    
309     Add copyrights to sandbox.h and sandbox_futils.h. If wrong, the
310     parties involved should please contact me so that we can fix it.
311    
312     Add opendir wrapper to libsandbox.c.
313    
314     21 Aug 2002; Martin Schlemmer <azarah@gentoo.org> :
315    
316     Do some more cleanups to ecanonicalize(), as it dropped filenames in
317     rare cases (after my symlink cleanups), and caused glibc to bork.
318     These fixes went into canonicalize.c.
319    
320     20 Aug 2002; Martin Schlemmer <azarah@gentoo.org> :
321    
322     Fix spawn_shell() and main() in sandbox.c to properly return fail
323     status.
324    
325     19 Aug 2002; Martin Schlemmer <azarah@gentoo.org> :
326    
327     The new canonicalize() function in libsandbox.c also resolved symlinks,
328     which caused on cleaning sandbox errors if the symlink pointed to a
329     file in the live root. Ripped out canonicalize() and realpath() from
330     glibc; removed the symlink stuff, and changed them to ecanonicalize()
331     and erealpath().
332    
333     18 Aug 2002; Martin Schlemmer <azarah@gentoo.org> :
334    
335     Ripped out all the wrappers, and implemented those of InstallWatch.
336     Losts of cleanups and bugfixes. Implement a execve that forces
337     $LIBSANDBOX in $LD_PRELOAD. We can now thus do away with the feared
338     /etc/ld.so.preload (*g*) ... Made the needed changes to sandbox.c,
339     sandbox.h and sandbox_futils.c. Rewrote the Makefile for most
340     parts; it now have an install target.
341    
342     Reformat the whole thing to look somewhat like the reworked sandbox.c
343     and new sandbox.h and sandbox_futils.c from:
344    
345     Brad House <brad@mainstreetsoftworks.com>.
346    
347     Additional Copyrights now due to the InstallWatch code:
348    
349     Copyright (C) 1998-9 Pancrazio `Ezio' de Mauro <p@demauro.net>

Properties

Name Value
svn:eol-style native
svn:keywords Author Date Id Revision

  ViewVC Help
Powered by ViewVC 1.1.20