Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 7 Oct 2018 00:29:08 +0300
From:      Konstantin Belousov <kostikbel@gmail.com>
To:        freebsd-security@freebsd.org
Subject:   Re: [FreeBSD-Announce] FreeBSD Security Advisory FreeBSD-SA-18:12.elf
Message-ID:  <20181006212908.GU5335@kib.kiev.ua>
In-Reply-To: <20181006184636.GT5335@kib.kiev.ua>
References:  <20180912054309.61C6B13269@freefall.freebsd.org> <20181006173525.GC813@lena.kiev> <20181006182104.GS5335@kib.kiev.ua> <20181006184636.GT5335@kib.kiev.ua>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Oct 06, 2018 at 09:46:36PM +0300, Konstantin Belousov wrote:
> On Sat, Oct 06, 2018 at 09:21:04PM +0300, Konstantin Belousov wrote:
> > On Sat, Oct 06, 2018 at 08:35:26PM +0300, Lena@lena.kiev.ua wrote:
> > > > Insufficient validation was performed in the ELF header parser, and malformed
> > > > or otherwise invalid ELF binaries were not rejected as they should be.
> > > 
> > > What is invalid in the /usr/local/share/google-earth/googleearth-bin
> > > binary of the port google-earth-7.1.5.1557,3 ?
> > > 
> > > FreeBSD 11.2-RELEASE-p4 Sep 27 GENERIC i386, the binary:
> > > https://drive.google.com/file/d/1SgHk8ijSp2F9UcQGlx44psT832TdIEL0/view
> > > 
> > > ~ $ googleearth
> > > Invalid PT_INTERP
> > > exec: ./googleearth-bin: Exec format error
> > > ~ $ readelf --program-headers /usr/local/share/google-earth/googleearth-bin
> > > 
> > > Elf file type is EXEC (Executable file)
> > > Entry point 0x8048650
> > > There are 8 program headers, starting at offset 52
> > > 
> > > Program Headers:
> > >   Type           Offset   VirtAddr   PhysAddr   FileSiz MemSiz  Flg Align
> > >   PHDR           0x000034 0x08048034 0x08048034 0x00100 0x00100 R E 0x4
> > >   INTERP         0x000134 0x08048134 0x08048134 0x00011 0x00011 R   0x1
> > >       [Requesting program interpreter: /lib/ld-linux.so.2]
> > As you see, the file delcares that file/memory length of the interpreter
> > name' segment is 0x11 == 16 decimal. But the string does not end on
> > byte 16, which is not NUL.  We tighten the checks and do require that
> > PT_INTERP string is valid by checking that it is NUL-terminated at the
> > offset declared by the size.
> As emaste pointed out, I am off by one, i.e. replace 16 by 17 in the text
> above.

But we might be somewhat nicer in this case.  Try the following.

diff --git a/sys/kern/imgact_elf.c b/sys/kern/imgact_elf.c
index f4302d46665..88f8a1ed2fa 100644
--- a/sys/kern/imgact_elf.c
+++ b/sys/kern/imgact_elf.c
@@ -872,9 +872,23 @@ __CONCAT(exec_, __elfN(imgact))(struct image_params *imgp)
 				interp = __DECONST(char *, imgp->image_header) +
 				    phdr[i].p_offset;
 				if (interp[interp_name_len - 1] != '\0') {
-					uprintf("Invalid PT_INTERP\n");
-					error = ENOEXEC;
-					goto ret;
+					/*
+					 * ELF specification requires
+					 * that PT_INTERP contained
+					 * NUL-terminated string.  If
+					 * it is not, try to fix the
+					 * path and still execute the
+					 * binary.
+					 */
+					VOP_UNLOCK(imgp->vp, 0);
+					interp_buf = malloc(interp_name_len + 1,
+					    M_TEMP, M_WAITOK);
+					vn_lock(imgp->vp, LK_EXCLUSIVE |
+					    LK_RETRY);
+					memcpy(interp_buf, interp,
+					    interp_name_len);
+					interp_buf[interp_name_len] = '\0';
+					interp = interp_buf;
 				}
 			}
 			break;



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20181006212908.GU5335>