Patch series "kasan: switch tag-based modes to stack ring from per-object
metadata", v3.
This series makes the tag-based KASAN modes use a ring buffer for storing
stack depot handles for alloc/free stack traces for slab objects instead
of per-object metadata. This ring buffer is referred to as the stack
ring.
On each alloc/free of a slab object, the tagged address of the object and
the current stack trace are recorded in the stack ring.
On each bug report, if the accessed address belongs to a slab object, the
stack ring is scanned for matching entries. The newest entries are used
to print the alloc/free stack traces in the report: one entry for alloc
and one for free.
The advantages of this approach over storing stack trace handles in
per-object metadata with the tag-based KASAN modes:
- Allows to find relevant stack traces for use-after-free bugs without
using quarantine for freed memory. (Currently, if the object was
reallocated multiple times, the report contains the latest alloc/free
stack traces, not necessarily the ones relevant to the buggy allocation.)
- Allows to better identify and mark use-after-free bugs, effectively
making the CONFIG_KASAN_TAGS_IDENTIFY functionality always-on.
- Has fixed memory overhead.
The disadvantage:
- If the affected object was allocated/freed long before the bug happened
and the stack trace events were purged from the stack ring, the report
will have no stack traces.
Discussion
==========
The proposed implementation of the stack ring uses a single ring buffer
for the whole kernel. This might lead to contention due to atomic
accesses to the ring buffer index on multicore systems.
At this point, it is unknown whether the performance impact from this
contention would be significant compared to the slowdown introduced by
collecting stack traces due to the planned changes to the latter part, see
the section below.
For now, the proposed implementation is deemed to be good enough, but this
might need to be revisited once the stack collection becomes faster.
A considered alternative is to keep a separate ring buffer for each CPU
and then iterate over all of them when printing a bug report. This
approach requires somehow figuring out which of the stack rings has the
freshest stack traces for an object if multiple stack rings have them.
Further plans
=============
This series is a part of an effort to make KASAN stack trace collection
suitable for production. This requires stack trace collection to be fast
and memory-bounded.
The planned steps are:
1. Speed up stack trace collection (potentially, by using SCS;
patches on-hold until steps #2 and #3 are completed).
2. Keep stack trace handles in the stack ring (this series).
3. Add a memory-bounded mode to stack depot or provide an alternative
memory-bounded stack storage.
4. Potentially, implement stack trace collection sampling to minimize
the performance impact.
This patch (of 34):
__kasan_metadata_size() calculates the size of the redzone for objects in
a slab cache.
When accounting for presence of kasan_free_meta in the redzone, this
function only compares free_meta_offset with 0. But free_meta_offset
could also be equal to KASAN_NO_FREE_META, which indicates that
kasan_free_meta is not present at all.
Add a comparison with KASAN_NO_FREE_META into __kasan_metadata_size().
Link: https://lkml.kernel.org/r/cover.1662411799.git.andreyknvl@google.com
Link: https://lkml.kernel.org/r/c7b316d30d90e5947eb8280f4dc78856a49298cf.1662411799.git.andreyknvl@google.com
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
Reviewed-by: Marco Elver <elver@google.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Evgenii Stepanov <eugenis@google.com>
Cc: Peter Collingbourne <pcc@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Bug: 254721825
(cherry picked from commit ca77f290cf)
Change-Id: I1feae36ac8435c0ffab4e72bcb03f03b689a0677
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
(Backport: fix conflicts with 608d4c5f9f in arch/x86/um/Makefile.)
Make KASAN run on User Mode Linux on x86_64.
The UML-specific KASAN initializer uses mmap to map the ~16TB of shadow
memory to the location defined by KASAN_SHADOW_OFFSET. kasan_init()
utilizes constructors to initialize KASAN before main().
The location of the KASAN shadow memory, starting at
KASAN_SHADOW_OFFSET, can be configured using the KASAN_SHADOW_OFFSET
option. The default location of this offset is 0x100000000000, which
keeps it out-of-the-way even on UML setups with more "physical" memory.
For low-memory setups, 0x7fff8000 can be used instead, which fits in an
immediate and is therefore faster, as suggested by Dmitry Vyukov. There
is usually enough free space at this location; however, it is a config
option so that it can be easily changed if needed.
Note that, unlike KASAN on other architectures, vmalloc allocations
still use the shadow memory allocated upfront, rather than allocating
and free-ing it per-vmalloc allocation.
If another architecture chooses to go down the same path, we should
replace the checks for CONFIG_UML with something more generic, such
as:
- A CONFIG_KASAN_NO_SHADOW_ALLOC option, which architectures could set
- or, a way of having architecture-specific versions of these vmalloc
and module shadow memory allocation options.
Also note that, while UML supports both KASAN in inline mode
(CONFIG_KASAN_INLINE) and static linking (CONFIG_STATIC_LINK), it does
not support both at the same time.
Signed-off-by: Patricia Alfonso <trishalfonso@google.com>
Co-developed-by: Vincent Whitchurch <vincent.whitchurch@axis.com>
Signed-off-by: Vincent Whitchurch <vincent.whitchurch@axis.com>
Signed-off-by: David Gow <davidgow@google.com>
Reviewed-by: Johannes Berg <johannes@sipsolutions.net>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Reviewed-by: Andrey Konovalov <andreyknvl@gmail.com>
Signed-off-by: Richard Weinberger <richard@nod.at>
Bug: 254721825
(cherry picked from commit 5b301409e8)
Change-Id: I7b6a5dfef80dd8e4684db738c35d80e6405b2f8b
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
This is just the same as PAGE_ALIGN(), but rounds the address down, not
up.
Suggested-by: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: David Gow <davidgow@google.com>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Richard Weinberger <richard@nod.at>
Bug: 254721825
(cherry picked from commit 335e52c28c)
Change-Id: I33bebcffbe4af3937e00243228b71d531b56e89d
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
(Backport: no real conflicts, inverse commit order commit 918002bdbe
"arm64: mte: Avoid setting PG_mte_tagged if no tags cleared or restored".
This revert patch has previously made into Android through stable in
commit add4bc9281, but was then reverted in stable commit a8a007c5b1.
The reason for the revert: this revert patch was initially considered a
standalone fix, but it actually required more patches to work properly:
70c248aca9 ("mm: kasan: Skip unpoisoning of user pages")
6d05141a39 ("mm: kasan: Skip page unpoisoning only if __GFP_SKIP_KASAN_UNPOISON")
Now that these patches are backported (in 1f2cb45568 and bc6ed581f4),
apply the revert patch again.)
This reverts commit e5b8d92189.
Pages mapped in user-space with PROT_MTE have the allocation tags either
zeroed or copied/restored to some user values. In order for the kernel
to access such pages via page_address(), resetting the tag in
page->flags was necessary. This tag resetting was deferred to
set_pte_at() -> mte_sync_page_tags() but it can race with another CPU
reading the flags (via page_to_virt()):
P0 (mte_sync_page_tags): P1 (memcpy from virt_to_page):
Rflags!=0xff
Wflags=0xff
DMB (doesn't help)
Wtags=0
Rtags=0 // fault
Since now the post_alloc_hook() function resets the page->flags tag when
unpoisoning is skipped for user pages (including the __GFP_ZEROTAGS
case), revert the arm64 commit calling page_kasan_tag_reset().
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will@kernel.org>
Cc: Vincenzo Frascino <vincenzo.frascino@arm.com>
Cc: Andrey Konovalov <andreyknvl@gmail.com>
Cc: Peter Collingbourne <pcc@google.com>
Reviewed-by: Vincenzo Frascino <vincenzo.frascino@arm.com>
Acked-by: Andrey Konovalov <andreyknvl@gmail.com>
Link: https://lore.kernel.org/r/20220610152141.2148929-5-catalin.marinas@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
Bug: 254721825
(cherry picked from commit 20794545c1)
Change-Id: I2a1cae227249e386056894af40f61b86425569f2
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
__kasan_unpoison_pages() colours the memory with a random tag and stores
it in page->flags in order to re-create the tagged pointer via
page_to_virt() later. When the tag from the page->flags is read, ensure
that the in-memory tags are already visible by re-ordering the
page_kasan_tag_set() after kasan_unpoison(). The former already has
barriers in place through try_cmpxchg(). On the reader side, the order
is ensured by the address dependency between page->flags and the memory
access.
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Reviewed-by: Andrey Konovalov <andreyknvl@gmail.com>
Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Cc: Vincenzo Frascino <vincenzo.frascino@arm.com>
Reviewed-by: Vincenzo Frascino <vincenzo.frascino@arm.com>
Link: https://lore.kernel.org/r/20220610152141.2148929-2-catalin.marinas@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
Bug: 254721825
(cherry picked from commit ed0a6d1d97)
Change-Id: I21744380b630fa5c8c6174ca0d4f063ff75b6acd
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
HW_TAGS KASAN skips zeroing page_alloc allocations backing vmalloc
mappings via __GFP_SKIP_ZERO. Instead, these pages are zeroed via
kasan_unpoison_vmalloc() by passing the KASAN_VMALLOC_INIT flag.
The problem is that __kasan_unpoison_vmalloc() does not zero pages when
either kasan_vmalloc_enabled() or is_vmalloc_or_module_addr() fail.
Thus:
1. Change __vmalloc_node_range() to only set KASAN_VMALLOC_INIT when
__GFP_SKIP_ZERO is set.
2. Change __kasan_unpoison_vmalloc() to always zero pages when the
KASAN_VMALLOC_INIT flag is set.
3. Add WARN_ON() asserts to check that KASAN_VMALLOC_INIT cannot be set
in other early return paths of __kasan_unpoison_vmalloc().
Also clean up the comment in __kasan_unpoison_vmalloc.
Link: https://lkml.kernel.org/r/4bc503537efdc539ffc3f461c1b70162eea31cf6.1654798516.git.andreyknvl@google.com
Fixes: 23689e91fb ("kasan, vmalloc: add vmalloc tagging for HW_TAGS")
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
Cc: Marco Elver <elver@google.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Bug: 254721825
(cherry picked from commit 6c2f761dad)
Change-Id: I07d3f8dc3cd28f43852a04b741a1c0b5a65a4ff9
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
This function is only called from assembly, no need for a prototype
declaration in a header file. In addition, add #ifdef around the
function since it is only used when CONFIG_KASAN_HW_TAGS.
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Reported-by: kernel test robot <lkp@intel.com>
Bug: 254721825
(cherry picked from commit 78cdaf3f42)
Change-Id: I222d17ae6cf5df53c922ef8c126115efaf6d9eee
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
BUG: sleeping function called from invalid context at kernel/locking/spinlock_rt.c:46
in_atomic(): 1, irqs_disabled(): 1, non_block: 0, pid: 1, name: swapper/0
preempt_count: 1, expected: 0
...........
CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.17.1-rt16-yocto-preempt-rt #22
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009),
BIOS rel-1.15.0-0-g2dd4b9b3f840-prebuilt.qemu.org 04/01/2014
Call Trace:
<TASK>
dump_stack_lvl+0x60/0x8c
dump_stack+0x10/0x12
__might_resched.cold+0x13b/0x173
rt_spin_lock+0x5b/0xf0
___cache_free+0xa5/0x180
qlist_free_all+0x7a/0x160
per_cpu_remove_cache+0x5f/0x70
smp_call_function_many_cond+0x4c4/0x4f0
on_each_cpu_cond_mask+0x49/0xc0
kasan_quarantine_remove_cache+0x54/0xf0
kasan_cache_shrink+0x9/0x10
kmem_cache_shrink+0x13/0x20
acpi_os_purge_cache+0xe/0x20
acpi_purge_cached_objects+0x21/0x6d
acpi_initialize_objects+0x15/0x3b
acpi_init+0x130/0x5ba
do_one_initcall+0xe5/0x5b0
kernel_init_freeable+0x34f/0x3ad
kernel_init+0x1e/0x140
ret_from_fork+0x22/0x30
When the kmem_cache_shrink() was called, the IPI was triggered, the
___cache_free() is called in IPI interrupt context, the local-lock or
spin-lock will be acquired. On PREEMPT_RT kernel, these locks are
replaced with sleepbale rt-spinlock, so the above problem is triggered.
Fix it by moving the qlist_free_allfrom() from IPI interrupt context to
task context when PREEMPT_RT is enabled.
[akpm@linux-foundation.org: reduce ifdeffery]
Link: https://lkml.kernel.org/r/20220401134649.2222485-1-qiang1.zhang@intel.com
Signed-off-by: Zqiang <qiang1.zhang@intel.com>
Acked-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Andrey Konovalov <andreyknvl@gmail.com>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Bug: 254721825
(cherry picked from commit 07d067e4f2)
Change-Id: I0a0555265d2e05f31cf2cd58b318d3602ba3cc6a
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
With MTE, even if the pte allows an access, a mismatched tag somewhere
within a page can still cause a fault. Select ARCH_HAS_SUBPAGE_FAULTS if
MTE is enabled and implement the probe_subpage_writeable() function.
Note that get_user() is sufficient for the writeable MTE check since the
same tag mismatch fault would be triggered by a read. The caller of
probe_subpage_writeable() will need to check the pte permissions
(put_user, GUP).
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20220423100751.1870771-3-catalin.marinas@arm.com
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Bug: 254721825
(cherry picked from commit f3ba50a7a1)
Change-Id: Ia12bd6d2c49a742794abf5658fb2ddaa063a5b67
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
(Backport: no conflict, neighboring lines changes.)
On hardware with features like arm64 MTE or SPARC ADI, an access fault
can be triggered at sub-page granularity. Depending on how the
fault_in_writeable() function is used, the caller can get into a
live-lock by continuously retrying the fault-in on an address different
from the one where the uaccess failed.
In the majority of cases progress is ensured by the following
conditions:
1. copy_to_user_nofault() guarantees at least one byte access if the
user address is not faulting.
2. The fault_in_writeable() loop is resumed from the first address that
could not be accessed by copy_to_user_nofault().
If the loop iteration is restarted from an earlier (initial) point, the
loop is repeated with the same conditions and it would live-lock.
Introduce an arch-specific probe_subpage_writeable() and call it from
the newly added fault_in_subpage_writeable() function. The arch code
with sub-page faults will have to implement the specific probing
functionality.
Note that no other fault_in_subpage_*() functions are added since they
have no callers currently susceptible to a live-lock.
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Link: https://lore.kernel.org/r/20220423100751.1870771-2-catalin.marinas@arm.com
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Bug: 254721825
(cherry picked from commit da32b58172)
Change-Id: I8362937496a2a8709686af9f97009b00a21b1f5d
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
