[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAPDLWs_82o5cQ24=pSO6hT8KtQHkNGz5S-gUf5e647QUX+t2QA@mail.gmail.com>
Date: Tue, 17 Jan 2017 13:08:14 +0530
From: Kaiwan N Billimoria <kaiwan@...wantech.com>
To: kernel-hardening@...ts.openwall.com
Cc: Kees Cook <keescook@...omium.org>
Subject: Re: New to KSPP, would like to contribute
Hi Kees,
I've made a start. Hopefully, in the right direction too!
So, basically what I've done so far is:
- pulled in the linux-next tree, and setup my own branch
- taken the grsecurity patch (for 4.8.17) and merged those portions of
the code encompassing the CONFIG_PAX_MEMORY_SANITIZE directive
- done small fixups where required
[More on this:
1. fs/dcache.c
kmem_cache_create_usercopy() unavailable (for now at least).
Probably part of other grsec infrastructure..
So am just adding the SLAB_NO_SANITIZE flag to the usual
kmem_cache_create() call.
2. mm/slab_common.c
Compile failure:
enum pax_sanitize_mode pax_sanitize_slab __read_only = PAX_SANITIZE_SLAB_FAST;
?? -above orig statement from the grsec-4.8.17 patch fails compile with:
mm/slab_common.c:34:37: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or
‘__attribute__’ before ‘__read_only’
So I've just removed the "__read_only" attribute for now.
What's the actual approach?
3. mm/slub.c
Compile failure:
#ifdef CONFIG_PAX_MEMORY_SANITIZE
&sanitize_attr.attr,
* ?? -above statement from the grsec-4.8.17 patch fails compile with:
mm/slub.c:5337:3: error: ‘sanitize_attr’ undeclared here (not in a function)
&sanitize_attr.attr,
Just commented this out for now.
]
Below pl find the initial patch reg the work done so far:
---
diff --git a/fs/buffer.c b/fs/buffer.c
index 28484b3..b524eda 100644
--- a/fs/buffer.c
+++ b/fs/buffer.c
@@ -3511,7 +3511,7 @@ void __init buffer_init(void)
bh_cachep = kmem_cache_create("buffer_head",
sizeof(struct buffer_head), 0,
(SLAB_RECLAIM_ACCOUNT|SLAB_PANIC|
- SLAB_MEM_SPREAD),
+ SLAB_MEM_SPREAD|SLAB_NO_SANITIZE),
NULL);
/*
diff --git a/fs/dcache.c b/fs/dcache.c
index 95d71ed..77abd2f 100644
--- a/fs/dcache.c
+++ b/fs/dcache.c
@@ -3616,8 +3616,16 @@ void __init vfs_caches_init_early(void)
void __init vfs_caches_init(void)
{
+/**
+ names_cachep = kmem_cache_create_usercopy("names_cache", PATH_MAX, 0,
+ SLAB_HWCACHE_ALIGN|SLAB_PANIC| SLAB_NO_SANITIZE,
+ 0, PATH_MAX, NULL);
+ kmem_cache_create_usercopy() unavailable for now at least.
+ So just adding the SLAB_NO_SANITIZE flag to the usual call below..
+ */
names_cachep = kmem_cache_create("names_cache", PATH_MAX, 0,
- SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
+ SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_NO_SANITIZE, NULL);
+
dcache_init();
inode_init();
diff --git a/include/linux/highmem.h b/include/linux/highmem.h
index bb3f329..9daed55 100644
--- a/include/linux/highmem.h
+++ b/include/linux/highmem.h
@@ -190,6 +190,18 @@ static inline void clear_highpage(struct page *page)
kunmap_atomic(kaddr);
}
+static inline void sanitize_highpage(struct page *page)
+{
+ void *kaddr;
+ unsigned long flags;
+
+ local_irq_save(flags);
+ kaddr = kmap_atomic(page);
+ clear_page(kaddr);
+ kunmap_atomic(kaddr);
+ local_irq_restore(flags);
+}
+
static inline void zero_user_segments(struct page *page,
unsigned start1, unsigned end1,
unsigned start2, unsigned end2)
diff --git a/include/linux/slab.h b/include/linux/slab.h
index 4c53635..334bd89 100644
--- a/include/linux/slab.h
+++ b/include/linux/slab.h
@@ -23,6 +23,13 @@
#define SLAB_CONSISTENCY_CHECKS 0x00000100UL /* DEBUG: Perform
(expensive) checks on alloc/free */
#define SLAB_RED_ZONE 0x00000400UL /* DEBUG: Red zone objs in a cache */
#define SLAB_POISON 0x00000800UL /* DEBUG: Poison objects */
+
+#ifdef CONFIG_PAX_MEMORY_SANITIZE
+#define SLAB_NO_SANITIZE 0x00001000UL /* PaX: Do not sanitize objs on free */
+#else
+#define SLAB_NO_SANITIZE 0x00000000UL
+#endif
+
#define SLAB_HWCACHE_ALIGN 0x00002000UL /* Align objs on cache lines */
#define SLAB_CACHE_DMA 0x00004000UL /* Use GFP_DMA memory */
#define SLAB_STORE_USER 0x00010000UL /* DEBUG: Store the last owner
for bug hunting */
diff --git a/include/linux/slab_def.h b/include/linux/slab_def.h
index 4ad2c5a..a30bbd2 100644
--- a/include/linux/slab_def.h
+++ b/include/linux/slab_def.h
@@ -60,6 +60,10 @@ struct kmem_cache {
atomic_t allocmiss;
atomic_t freehit;
atomic_t freemiss;
+#ifdef CONFIG_PAX_MEMORY_SANITIZE
+ atomic_unchecked_t sanitized;
+ atomic_unchecked_t not_sanitized;
+#endif
#ifdef CONFIG_DEBUG_SLAB_LEAK
atomic_t store_user_clean;
#endif
diff --git a/kernel/fork.c b/kernel/fork.c
index 7da33cb..24404a6 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -2098,7 +2098,7 @@ void __init proc_caches_init(void)
sizeof(struct mm_struct), ARCH_MIN_MMSTRUCT_ALIGN,
SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_NOTRACK|SLAB_ACCOUNT,
NULL);
- vm_area_cachep = KMEM_CACHE(vm_area_struct, SLAB_PANIC|SLAB_ACCOUNT);
+ vm_area_cachep = KMEM_CACHE(vm_area_struct,
SLAB_PANIC|SLAB_ACCOUNT|SLAB_NO_SANITIZE);
mmap_init();
nsproxy_cache_init();
}
diff --git a/mm/Kconfig.debug b/mm/Kconfig.debug
index afcc550..ed3f097 100644
--- a/mm/Kconfig.debug
+++ b/mm/Kconfig.debug
@@ -10,6 +10,7 @@ config PAGE_EXTENSION
config DEBUG_PAGEALLOC
bool "Debug page memory allocations"
depends on DEBUG_KERNEL
+ depends on !PAX_MEMORY_SANITIZE
depends on !HIBERNATION || ARCH_SUPPORTS_DEBUG_PAGEALLOC && !PPC && !SPARC
depends on !KMEMCHECK
select PAGE_EXTENSION
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 21ea508..d7b3d95 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -994,6 +994,10 @@ static __always_inline bool
free_pages_prepare(struct page *page,
{
int bad = 0;
+#ifdef CONFIG_PAX_MEMORY_SANITIZE
+ unsigned long index = 1UL << order;
+#endif
+
VM_BUG_ON_PAGE(PageTail(page), page);
trace_mm_page_free(page, order);
@@ -1040,6 +1044,12 @@ static __always_inline bool
free_pages_prepare(struct page *page,
debug_check_no_obj_freed(page_address(page),
PAGE_SIZE << order);
}
+
+#ifdef CONFIG_PAX_MEMORY_SANITIZE
+ for (; index; --index)
+ sanitize_highpage(page + index - 1);
+#endif
+
arch_free_page(page, order);
kernel_poison_pages(page, 1 << order, 0);
kernel_map_pages(page, 1 << order, 0);
@@ -1696,8 +1706,9 @@ static inline int check_new_page(struct page *page)
static inline bool free_pages_prezeroed(bool poisoned)
{
- return IS_ENABLED(CONFIG_PAGE_POISONING_ZERO) &&
- page_poisoning_enabled() && poisoned;
+ return IS_ENABLED(CONFIG_PAX_MEMORY_SANITIZE) ||
+ (IS_ENABLED(CONFIG_PAGE_POISONING_ZERO) && page_poisoning_enabled()
&& poisoned);
+
}
#ifdef CONFIG_DEBUG_VM
@@ -1753,11 +1764,13 @@ static void prep_new_page(struct page *page,
unsigned int order, gfp_t gfp_flags
int i;
bool poisoned = true;
+#ifndef CONFIG_PAX_MEMORY_SANITIZE
for (i = 0; i < (1 << order); i++) {
struct page *p = page + i;
if (poisoned)
poisoned &= page_is_poisoned(p);
}
+#endif
post_alloc_hook(page, order, gfp_flags);
diff --git a/mm/rmap.c b/mm/rmap.c
index 91619fd..c97da9e 100644
--- a/mm/rmap.c
+++ b/mm/rmap.c
@@ -428,10 +428,10 @@ static void anon_vma_ctor(void *data)
void __init anon_vma_init(void)
{
anon_vma_cachep = kmem_cache_create("anon_vma", sizeof(struct anon_vma),
- 0, SLAB_DESTROY_BY_RCU|SLAB_PANIC|SLAB_ACCOUNT,
+ 0, SLAB_DESTROY_BY_RCU|SLAB_PANIC|SLAB_ACCOUNT|SLAB_NO_SANITIZE,
anon_vma_ctor);
anon_vma_chain_cachep = KMEM_CACHE(anon_vma_chain,
- SLAB_PANIC|SLAB_ACCOUNT);
+ SLAB_PANIC|SLAB_ACCOUNT|SLAB_NO_SANITIZE);
}
/*
diff --git a/mm/slab.c b/mm/slab.c
index 4f2ec6b..6dc9f4a 100644
--- a/mm/slab.c
+++ b/mm/slab.c
@@ -3511,6 +3511,20 @@ void ___cache_free(struct kmem_cache *cachep, void *objp,
struct array_cache *ac = cpu_cache_get(cachep);
check_irq_off();
+
+#ifdef CONFIG_PAX_MEMORY_SANITIZE
+ if (cachep->flags & (SLAB_POISON | SLAB_NO_SANITIZE))
+ STATS_INC_NOT_SANITIZED(cachep);
+ else {
+ memset(objp, PAX_MEMORY_SANITIZE_VALUE, cachep->object_size);
+
+ if (cachep->ctor)
+ cachep->ctor(objp);
+
+ STATS_INC_SANITIZED(cachep);
+ }
+#endif
+
kmemleak_free_recursive(objp, cachep->flags);
objp = cache_free_debugcheck(cachep, objp, caller);
@@ -4157,6 +4171,14 @@ void slabinfo_show_stats(struct seq_file *m,
struct kmem_cache *cachep)
seq_printf(m, " : cpustat %6lu %6lu %6lu %6lu",
allochit, allocmiss, freehit, freemiss);
}
+#ifdef CONFIG_PAX_MEMORY_SANITIZE
+ {
+ unsigned long sanitized = atomic_read_unchecked(&cachep->sanitized);
+ unsigned long not_sanitized = atomic_read_unchecked(&cachep->not_sanitized);
+
+ seq_printf(m, " : pax %6lu %6lu", sanitized, not_sanitized);
+ }
+#endif
#endif
}
diff --git a/mm/slab.h b/mm/slab.h
index de6579d..e982e3b 100644
--- a/mm/slab.h
+++ b/mm/slab.h
@@ -71,6 +71,35 @@ extern struct list_head slab_caches;
/* The slab cache that manages slab cache information */
extern struct kmem_cache *kmem_cache;
+#ifdef CONFIG_PAX_MEMORY_SANITIZE
+#ifdef CONFIG_X86_64
+#define PAX_MEMORY_SANITIZE_VALUE '\xfe'
+#else
+#define PAX_MEMORY_SANITIZE_VALUE '\xff'
+#endif
+enum pax_sanitize_mode {
+ PAX_SANITIZE_SLAB_OFF = 0,
+ PAX_SANITIZE_SLAB_FAST,
+ PAX_SANITIZE_SLAB_FULL,
+};
+
+extern enum pax_sanitize_mode pax_sanitize_slab;
+
+static inline unsigned long pax_sanitize_slab_flags(unsigned long flags)
+{
+ if (pax_sanitize_slab == PAX_SANITIZE_SLAB_OFF || (flags &
SLAB_DESTROY_BY_RCU))
+ flags |= SLAB_NO_SANITIZE;
+ else if (pax_sanitize_slab == PAX_SANITIZE_SLAB_FULL)
+ flags &= ~SLAB_NO_SANITIZE;
+ return flags;
+}
+#else
+static inline unsigned long pax_sanitize_slab_flags(unsigned long flags)
+{
+ return flags;
+}
+#endif
+
unsigned long calculate_alignment(unsigned long flags,
unsigned long align, unsigned long size);
@@ -120,7 +149,7 @@ static inline unsigned long
kmem_cache_flags(unsigned long object_size,
/* Legal flag mask for kmem_cache_create(), for various configurations */
#define SLAB_CORE_FLAGS (SLAB_HWCACHE_ALIGN | SLAB_CACHE_DMA | SLAB_PANIC | \
- SLAB_DESTROY_BY_RCU | SLAB_DEBUG_OBJECTS )
+ SLAB_DESTROY_BY_RCU | SLAB_DEBUG_OBJECTS | SLAB_NO_SANITIZE)
#if defined(CONFIG_DEBUG_SLAB)
#define SLAB_DEBUG_FLAGS (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER)
diff --git a/mm/slab_common.c b/mm/slab_common.c
index 1dfc209..80bf59e 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -30,6 +30,37 @@ LIST_HEAD(slab_caches);
DEFINE_MUTEX(slab_mutex);
struct kmem_cache *kmem_cache;
+#ifdef CONFIG_PAX_MEMORY_SANITIZE
+/**
+enum pax_sanitize_mode pax_sanitize_slab __read_only = PAX_SANITIZE_SLAB_FAST;
+
+?? -above orig statement from the grsec-4.8.17 patch fails compile with:
+mm/slab_common.c:34:37: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or
‘__attribute__’ before ‘__read_only’
+ pax_sanitize_mode pax_sanitize_slab __read_only = PAX_SANITIZE_SLAB_FAST;
+*/
+enum pax_sanitize_mode pax_sanitize_slab = PAX_SANITIZE_SLAB_FAST;
+static int __init pax_sanitize_slab_setup(char *str)
+{
+ if (!str)
+ return 0;
+
+ if (!strcmp(str, "0") || !strcmp(str, "off")) {
+ pr_info("PaX slab sanitization: %s\n", "disabled");
+ pax_sanitize_slab = PAX_SANITIZE_SLAB_OFF;
+ } else if (!strcmp(str, "1") || !strcmp(str, "fast")) {
+ pr_info("PaX slab sanitization: %s\n", "fast");
+ pax_sanitize_slab = PAX_SANITIZE_SLAB_FAST;
+ } else if (!strcmp(str, "full")) {
+ pr_info("PaX slab sanitization: %s\n", "full");
+ pax_sanitize_slab = PAX_SANITIZE_SLAB_FULL;
+ } else
+ pr_err("PaX slab sanitization: unsupported option '%s'\n", str);
+
+ return 0;
+}
+early_param("pax_sanitize_slab", pax_sanitize_slab_setup);
+#endif
+
/*
* Set of flags that will prevent slab merging
*/
@@ -1136,6 +1167,9 @@ static void print_slabinfo_header(struct seq_file *m)
#ifdef CONFIG_DEBUG_SLAB
seq_puts(m, " : globalstat <listallocs> <maxobjs> <grown> <reaped>
<error> <maxfreeable> <nodeallocs> <remotefrees> <alienoverflow>");
seq_puts(m, " : cpustat <allochit> <allocmiss> <freehit> <freemiss>");
+#ifdef CONFIG_PAX_MEMORY_SANITIZE
+ seq_puts(m, " : pax <sanitized> <not_sanitized>");
+#endif
#endif
seq_putc(m, '\n');
}
diff --git a/mm/slob.c b/mm/slob.c
index eac04d43..f455845 100644
--- a/mm/slob.c
+++ b/mm/slob.c
@@ -365,6 +365,11 @@ static void slob_free(void *block, int size)
return;
}
+#ifdef CONFIG_PAX_MEMORY_SANITIZE
+ if (pax_sanitize_slab && !(c && (c->flags & SLAB_NO_SANITIZE)))
+ memset(block, PAX_MEMORY_SANITIZE_VALUE, size);
+#endif
+
if (!slob_page_free(sp)) {
/* This slob page is about to become partially free. Easy! */
sp->units = units;
diff --git a/mm/slub.c b/mm/slub.c
index 067598a..9253eb6 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -2911,6 +2911,23 @@ static __always_inline void do_slab_free(struct
kmem_cache *s,
void *tail_obj = tail ? : head;
struct kmem_cache_cpu *c;
unsigned long tid;
+
+#ifdef CONFIG_PAX_MEMORY_SANITIZE
+ if (!(s->flags & SLAB_NO_SANITIZE)) {
+ int offset = s->offset ? 0 : sizeof(void *);
+ void *x = head;
+
+ while (1) {
+ memset(x + offset, PAX_MEMORY_SANITIZE_VALUE, s->object_size - offset);
+ if (s->ctor)
+ s->ctor(x);
+ if (x == tail_obj)
+ break;
+ x = get_freepointer(s, x);
+ }
+ }
+#endif
+
redo:
/*
* Determine the currently cpus per cpu slab.
@@ -5316,6 +5333,15 @@ static struct attribute *slab_attrs[] = {
#ifdef CONFIG_ZONE_DMA
&cache_dma_attr.attr,
#endif
+/**
+#ifdef CONFIG_PAX_MEMORY_SANITIZE
+ &sanitize_attr.attr,
+ * ?? -above statement from the grsec-4.8.17 patch fails compile with:
+ mm/slub.c:5337:3: error: ‘sanitize_attr’ undeclared here (not in a function)
+ &sanitize_attr.attr,
+ ...
+#endif
+ */
#ifdef CONFIG_NUMA
&remote_node_defrag_ratio_attr.attr,
#endif
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 7ad67d7..dde2364 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -3453,12 +3453,14 @@ void __init skb_init(void)
skbuff_head_cache = kmem_cache_create("skbuff_head_cache",
sizeof(struct sk_buff),
0,
- SLAB_HWCACHE_ALIGN|SLAB_PANIC,
+ SLAB_HWCACHE_ALIGN|SLAB_PANIC|
+ SLAB_NO_SANITIZE,
NULL);
skbuff_fclone_cache = kmem_cache_create("skbuff_fclone_cache",
sizeof(struct sk_buff_fclones),
0,
- SLAB_HWCACHE_ALIGN|SLAB_PANIC,
+ SLAB_HWCACHE_ALIGN|SLAB_PANIC|
+ SLAB_NO_SANITIZE,
NULL);
}
diff --git a/security/Kconfig b/security/Kconfig
index 118f454..3ad5110 100644
--- a/security/Kconfig
+++ b/security/Kconfig
@@ -4,6 +4,46 @@
menu "Security options"
+menu "Miscellaneous hardening features"
+
+config PAX_MEMORY_SANITIZE
+ bool "Sanitize all freed memory"
+ default y if (GRKERNSEC_CONFIG_AUTO && GRKERNSEC_CONFIG_PRIORITY_SECURITY)
+ help
+ By saying Y here the kernel will erase memory pages and slab objects
+ as soon as they are freed. This in turn reduces the lifetime of data
+ stored in them, making it less likely that sensitive information such
+ as passwords, cryptographic secrets, etc stay in memory for too long.
+
+ This is especially useful for programs whose runtime is short, long
+ lived processes and the kernel itself benefit from this as long as
+ they ensure timely freeing of memory that may hold sensitive
+ information.
+
+ A nice side effect of the sanitization of slab objects is the
+ reduction of possible info leaks caused by padding bytes within the
+ leaky structures. Use-after-free bugs for structures containing
+ pointers can also be detected as dereferencing the sanitized pointer
+ will generate an access violation.
+
+ The tradeoff is performance impact, on a single CPU system kernel
+ compilation sees a 3% slowdown, other systems and workloads may vary
+ and you are advised to test this feature on your expected workload
+ before deploying it.
+
+ The slab sanitization feature excludes a few slab caches per default
+ for performance reasons. To extend the feature to cover those as
+ well, pass "pax_sanitize_slab=full" as kernel command line parameter.
+
+ To reduce the performance penalty by sanitizing pages only, albeit
+ limiting the effectiveness of this feature at the same time, slab
+ sanitization can be disabled with the kernel command line parameter
+ "pax_sanitize_slab=off".
+
+ Note that this feature does not protect data stored in live pages,
+ e.g., process memory swapped to disk may stay there for a long time.
+endmenu
+
source security/keys/Kconfig
config SECURITY_DMESG_RESTRICT
---
Am pretty sure it's definitely imperfect, just hope it's not horribly
wrong! - pl let me know.
What should I change, etc...
TIA,
Regards,
Kaiwan.
On Thu, Jan 12, 2017 at 3:34 PM, Kaiwan N Billimoria
<kaiwan@...wantech.com> wrote:
> Thanks Greg.
> Will do..
>
>
> Regards,
> Kaiwan.
>
> Kaiwan N Billimoria
> ✉ kaiwan@...wantech.com
> ✉ kaiwan.billimoria@...il.com
>
> kaiwanTECH
> http://kaiwantech.in
> Do visit our enhanced website.
>
> 4931, 11th Floor, Highpoint IV, 45 Palace Road, Bangalore 560 001.
> +91.9845016788 (M)
> TeleFax: +91.80.22389396 | Alt. Tel: +91.80.64500257
>
> LinkedIn: https://in.linkedin.com/in/kaiwanbillimoria
>
> I blog here:
> Tech : http://kaiwantech.wordpress.com |
> Running : http://kaiwanbill.blogspot.in/
>
> "Don't be afraid that your life will end,
> be afraid that it will never begin."
>
> On Thu, Jan 12, 2017 at 1:07 PM, Greg KH <gregkh@...uxfoundation.org> wrote:
>>
>> On Thu, Jan 12, 2017 at 09:01:05AM +0530, Kaiwan N Billimoria wrote:
>> > Okay, thanks, I see it better now :)
>> >
>> > Forgive my silly qs:
>> > so as I understand it, the objective is to start merging in some of the
>> > grsec
>> > code into mainline, yes? So, which kernel do we work on? the latest rc
>> > or do I
>> > work on the one where the above patch applies (i.e. 4.8.17)?
>>
>> You can't go back in time and get stuff merged into an old patch, so
>> always work on the latest kernel tree. Use linux-next so you don't end
>> up duplicating any effort that people have already done.
>>
>> I suggest reading Documentation/development_process/ as a summary of how
>> all of this works.
>>
>> thanks,
>>
>> greg k-h
>>
>
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
