diff options
author | Andy Lutomirski <luto@amacapital.net> | 2014-12-19 19:04:11 -0500 |
---|---|---|
committer | Andy Lutomirski <luto@amacapital.net> | 2014-12-20 19:56:57 -0500 |
commit | 394f56fe480140877304d342dec46d50dc823d46 (patch) | |
tree | 7a5aaa30bdeb3bbcfb6bd08003bf3d1bd4bef1bf /arch | |
parent | e589c9e13aeb0c5539bf1314b3a78442ea8fc0c2 (diff) |
x86_64, vdso: Fix the vdso address randomization algorithm
The theory behind vdso randomization is that it's mapped at a random
offset above the top of the stack. To avoid wasting a page of
memory for an extra page table, the vdso isn't supposed to extend
past the lowest PMD into which it can fit. Other than that, the
address should be a uniformly distributed address that meets all of
the alignment requirements.
The current algorithm is buggy: the vdso has about a 50% probability
of being at the very end of a PMD. The current algorithm also has a
decent chance of failing outright due to incorrect handling of the
case where the top of the stack is near the top of its PMD.
This fixes the implementation. The paxtest estimate of vdso
"randomisation" improves from 11 bits to 18 bits. (Disclaimer: I
don't know what the paxtest code is actually calculating.)
It's worth noting that this algorithm is inherently biased: the vdso
is more likely to end up near the end of its PMD than near the
beginning. Ideally we would either nix the PMD sharing requirement
or jointly randomize the vdso and the stack to reduce the bias.
In the mean time, this is a considerable improvement with basically
no risk of compatibility issues, since the allowed outputs of the
algorithm are unchanged.
As an easy test, doing this:
for i in `seq 10000`
do grep -P vdso /proc/self/maps |cut -d- -f1
done |sort |uniq -d
used to produce lots of output (1445 lines on my most recent run).
A tiny subset looks like this:
7fffdfffe000
7fffe01fe000
7fffe05fe000
7fffe07fe000
7fffe09fe000
7fffe0bfe000
7fffe0dfe000
Note the suspicious fe000 endings. With the fix, I get a much more
palatable 76 repeated addresses.
Reviewed-by: Kees Cook <keescook@chromium.org>
Cc: stable@vger.kernel.org
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Diffstat (limited to 'arch')
-rw-r--r-- | arch/x86/vdso/vma.c | 45 |
1 files changed, 29 insertions, 16 deletions
diff --git a/arch/x86/vdso/vma.c b/arch/x86/vdso/vma.c index 009495b9ab4b..1c9f750c3859 100644 --- a/arch/x86/vdso/vma.c +++ b/arch/x86/vdso/vma.c | |||
@@ -41,12 +41,17 @@ void __init init_vdso_image(const struct vdso_image *image) | |||
41 | 41 | ||
42 | struct linux_binprm; | 42 | struct linux_binprm; |
43 | 43 | ||
44 | /* Put the vdso above the (randomized) stack with another randomized offset. | 44 | /* |
45 | This way there is no hole in the middle of address space. | 45 | * Put the vdso above the (randomized) stack with another randomized |
46 | To save memory make sure it is still in the same PTE as the stack top. | 46 | * offset. This way there is no hole in the middle of address space. |
47 | This doesn't give that many random bits. | 47 | * To save memory make sure it is still in the same PTE as the stack |
48 | 48 | * top. This doesn't give that many random bits. | |
49 | Only used for the 64-bit and x32 vdsos. */ | 49 | * |
50 | * Note that this algorithm is imperfect: the distribution of the vdso | ||
51 | * start address within a PMD is biased toward the end. | ||
52 | * | ||
53 | * Only used for the 64-bit and x32 vdsos. | ||
54 | */ | ||
50 | static unsigned long vdso_addr(unsigned long start, unsigned len) | 55 | static unsigned long vdso_addr(unsigned long start, unsigned len) |
51 | { | 56 | { |
52 | #ifdef CONFIG_X86_32 | 57 | #ifdef CONFIG_X86_32 |
@@ -54,22 +59,30 @@ static unsigned long vdso_addr(unsigned long start, unsigned len) | |||
54 | #else | 59 | #else |
55 | unsigned long addr, end; | 60 | unsigned long addr, end; |
56 | unsigned offset; | 61 | unsigned offset; |
57 | end = (start + PMD_SIZE - 1) & PMD_MASK; | 62 | |
63 | /* | ||
64 | * Round up the start address. It can start out unaligned as a result | ||
65 | * of stack start randomization. | ||
66 | */ | ||
67 | start = PAGE_ALIGN(start); | ||
68 | |||
69 | /* Round the lowest possible end address up to a PMD boundary. */ | ||
70 | end = (start + len + PMD_SIZE - 1) & PMD_MASK; | ||
58 | if (end >= TASK_SIZE_MAX) | 71 | if (end >= TASK_SIZE_MAX) |
59 | end = TASK_SIZE_MAX; | 72 | end = TASK_SIZE_MAX; |
60 | end -= len; | 73 | end -= len; |
61 | /* This loses some more bits than a modulo, but is cheaper */ | 74 | |
62 | offset = get_random_int() & (PTRS_PER_PTE - 1); | 75 | if (end > start) { |
63 | addr = start + (offset << PAGE_SHIFT); | 76 | offset = get_random_int() % (((end - start) >> PAGE_SHIFT) + 1); |
64 | if (addr >= end) | 77 | addr = start + (offset << PAGE_SHIFT); |
65 | addr = end; | 78 | } else { |
79 | addr = start; | ||
80 | } | ||
66 | 81 | ||
67 | /* | 82 | /* |
68 | * page-align it here so that get_unmapped_area doesn't | 83 | * Forcibly align the final address in case we have a hardware |
69 | * align it wrongfully again to the next page. addr can come in 4K | 84 | * issue that requires alignment for performance reasons. |
70 | * unaligned here as a result of stack start randomization. | ||
71 | */ | 85 | */ |
72 | addr = PAGE_ALIGN(addr); | ||
73 | addr = align_vdso_addr(addr); | 86 | addr = align_vdso_addr(addr); |
74 | 87 | ||
75 | return addr; | 88 | return addr; |