From: Jiri Kosina <jkosina@suse.cz>
i386 and x86_64: randomize brk()
This patch randomizes the location of the heap (brk) for i386 and x86_64.
The range is randomized in the range starting at current brk location up
to 0x02000000 offset for both architectures. This, together with
pie-executable-randomization.patch and
pie-executable-randomization-fix.patch, should make the address space
randomization on i386 and x86_64 complete.
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
diff --git a/arch/i386/kernel/process.c b/arch/i386/kernel/process.c
index 8466471..8e0624d 100644
--- a/arch/i386/kernel/process.c
+++ b/arch/i386/kernel/process.c
@@ -949,3 +949,17 @@ unsigned long arch_align_stack(unsigned long sp)
sp -= get_random_int() % 8192;
return sp & ~0xf;
}
+
+unsigned long arch_randomize_brk(unsigned long brk)
+{
+ unsigned long new_brk;
+ unsigned long range_end;
+
+ range_end = brk + 0x02000000;
+ new_brk = randomize_range(brk, range_end, 0);
+ if (new_brk)
+ return new_brk;
+ else
+ return brk;
+}
+
diff --git a/arch/x86_64/kernel/process.c b/arch/x86_64/kernel/process.c
index 2842f50..b20f0eb 100644
--- a/arch/x86_64/kernel/process.c
+++ b/arch/x86_64/kernel/process.c
@@ -902,3 +902,17 @@ unsigned long arch_align_stack(unsigned long sp)
sp -= get_random_int() % 8192;
return sp & ~0xf;
}
+
+unsigned long arch_randomize_brk(unsigned long brk)
+{
+ unsigned long new_brk;
+ unsigned long range_end;
+
+ range_end = brk + 0x02000000;
+ new_brk = randomize_range(brk, range_end, 0);
+ if (new_brk)
+ return new_brk;
+ else
+ return brk;
+}
+
diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c
index d65f1d9..7afec71 100644
--- a/fs/binfmt_elf.c
+++ b/fs/binfmt_elf.c
@@ -47,6 +47,9 @@ static int load_elf_binary(struct linux_binprm *bprm, struct pt_regs *regs);
static int load_elf_library(struct file *);
static unsigned long elf_map (struct file *, unsigned long, struct elf_phdr *, int, int, unsigned long);
+/* overriden by architectures ...We need a prototype of arch_randomize_brk() in scope for all callers and implementations, so that the compiler can perform the appropriate typechecking. --- a/include/linux/mm.h~i386-and-x86_64-randomize-brk-fix +++ a/include/linux/mm.h @@ -1017,6 +1017,7 @@ out: extern int do_munmap(struct mm_struct *, unsigned long, size_t); extern unsigned long do_brk(unsigned long, unsigned long); +extern unsigned long arch_randomize_brk(unsigned long brk); /* filemap.c */ extern unsigned long page_unuse(struct page *); _ -
You are right, thanks a lot, please fold your fix into my patch. -- Jiri Kosina SUSE Labs -
I seem to have a different copy of this patch from the one which you
originally sent. It's the one which implements arch_randomize_brk() in
each arch's header file. Mayeb it wasn't cc'ed to a mailing list of maybe
I just lost the email, dunno.
Anyway, it breaks on ia64:
CC arch/ia64/ia32/binfmt_elf32.o
In file included from arch/ia64/ia32/binfmt_elf32.c:49:
arch/ia64/ia32/../../../fs/binfmt_elf.c: In function `load_elf_binary':
arch/ia64/ia32/../../../fs/binfmt_elf.c:1085: error: implicit declaration of function `arch_randomize_brk'
make[1]: *** [arch/ia64/ia32/binfmt_elf32.o] Error 1
make: *** [arch/ia64/ia32/binfmt_elf32.o] Error 2
This is because ia64 takes the exceptional stupidity which is our
elf-handling build system and adds an extra layer of stupidity on top of
it. Look:
akpm2:/usr/src/25> head include/asm-ia64/elf.h
#ifndef _ASM_IA64_ELF_H
#define _ASM_IA64_ELF_H
/*
* ELF-specific definitions.
*
* Copyright (C) 1998-1999, 2002-2004 Hewlett-Packard Co
* David Mosberger-Tang <davidm@hpl.hp.com>
*/
akpm2:/usr/src/25> grep -r _ASM_IA64_ELF_H arch/ia64
arch/ia64/ia32/ia32priv.h:#define _ASM_IA64_ELF_H /* Don't include elf.h */
can you believe this stuff?
Anyway, I'm presently running with this loveliness:
--- a/include/asm-ia64/elf.h~i386-and-x86_64-randomize-brk-2-fix
+++ a/include/asm-ia64/elf.h
@@ -249,8 +249,10 @@ do { \
#endif /* __KERNEL__ */
+#endif /* _ASM_IA64_ELF_H */
+
+#ifndef IA64_IS_WEIRD
static inline void arch_randomize_brk(void)
{
}
-
-#endif /* _ASM_IA64_ELF_H */
+#endif
_
Wanna see if there's something saner we can do please?
-
err, with the obvious `#define IA64_IS_WEIRD' in there.. -
And on s390 In file included from arch/s390/kernel/binfmt_elf32.c:202: arch/s390/kernel/../../../fs/binfmt_elf.c: In function 'load_elf_binary': arch/s390/kernel/../../../fs/binfmt_elf.c:1088: error: implicit declaration of function 'arch_randomize_brk' I'll drop the patch. Really we should fix the elf mess before we try and change it any more. -
This should be fixed trivially in the very same way as for x86_64, i.e. something like diff --git a/arch/s390/kernel/binfmt_elf32.c b/arch/s390/kernel/binfmt_elf32.c index f1e40ca..4b5432d 100644 --- a/arch/s390/kernel/binfmt_elf32.c +++ b/arch/s390/kernel/binfmt_elf32.c @@ -199,5 +199,6 @@ cputime_to_compat_timeval(const cputime_t cputime, struct compat_timeval *value) value->tv_sec = cputime / 1000000; } +extern void arch_randomize_brk(void); #include "../../../fs/binfmt_elf.c" right? Maybe this would be needed for other architectures too :( I will check this. I'd really like the patch go in, so that we have really full I agree that it is a total mess. We'd need someone brave enough to rewrite this crap :/ Thanks, -- Jiri Kosina SUSE Labs -
... back to this a week old thread about already dropped patch. I am thinking about going back to the original idea of just simply defining ARCH_HAS_RANDOMIZE_BRK and not caring for the ELF crap any more for now. This way the patch is as small as possible, and doesn't interfere with the ELF cross-arch craziness at all. Here I posted such version of the patch: http://lkml.org/lkml/2007/8/22/254 and here you asked me to put empty stubs into elf.h, which turned out to be too headache for some archs: http://lkml.org/lkml/2007/8/22/492 Does going back to the original ARCH_HAS_RANDOMIZE_BRK sound acceptable now, before the ELF stuff gets completely rewritten one day? Thanks, -- Jiri Kosina -
And here it goes, rebased on top of current Linus' -git, please consider queuing in -mm for 2.6.25. From: Jiri Kosina <jkosina@suse.cz> x86: randomize brk() Randomize the location of the heap (brk) for i386 and x86_64. The range is randomized in the range starting at current brk location up to 0x02000000 offset for both architectures. This, together with pie-executable-randomization.patch and pie-executable-randomization-fix.patch, should make the address space randomization on i386 and x86_64 complete. Arjan says: This is known to break older versions of some emacs variants, whose dumper code assumed that the last variable declared in the program is equal to the start of the dynamically allocated memory region. (The dumper is the code where emacs effectively dumps core at the end of it's compilation stage; this coredump is then loaded as the main program during normal use) iirc this was 5 years or so; we found this way back when I was at RH and we first did the security stuff there (including this brk randomization). It wasn't all variants of emacs, and it got fixed as a result (I vaguely remember that emacs already had code to deal with it for other archs/oses, just ifdeffed wrongly). It's a rare and wrong assumption as a general thing, just on x86 it mostly happened to be true (but to be honest, it'll break too if gcc does something fancy or if the linker does a non-standard order). Still its something we should at least document. Note 2: afaik it only broke the emacs *build*. I'm not 100% sure about that (it IS 5 years ago) though. Signed-off-by: Jiri Kosina <jkosina@suse.cz> diff --git a/arch/x86/kernel/process_32.c b/arch/x86/kernel/process_32.c index 7b89958..b6578c7 100644 --- a/arch/x86/kernel/process_32.c +++ b/arch/x86/kernel/process_32.c @@ -971,3 +971,10 @@ unsigned long arch_align_stack(unsigned long sp) sp -= get_random_int() % 8192; return sp & ~0xf; } + +unsigned long arch_randomize_brk(struct mm_struct ...
Yes, it should the the last version of this patch, and it should also have Oh, that's pretty crazy indeed. Including of binfmt_elf.c from all over the place is crazy as hell by itself, but this certainly adds a lot of Apart from rewriting the whole thing to make more sense, I don't see currently a way out of it better than your patch :( -- Jiri Kosina SUSE Labs -
