The Kernel Address Sanitizer (KASAN)

Overview

KernelAddressSANitizer (KASAN) is a dynamic memory error detector designed tofind out-of-bound and use-after-free bugs. KASAN has two modes: generic KASAN(similar to userspace ASan) and software tag-based KASAN (similar to userspaceHWASan).

KASAN uses compile-time instrumentation to insert validity checks before everymemory access, and therefore requires a compiler version that supports that.

Generic KASAN is supported in both GCC and Clang. With GCC it requires version4.9.2 or later for basic support and version 5.0 or later for detection ofout-of-bounds accesses for stack and global variables and for inlineinstrumentation mode (see the Usage section). With Clang it requires version7.0.0 or later and it doesn’t support detection of out-of-bounds accesses forglobal variables yet.

Tag-based KASAN is only supported in Clang and requires version 7.0.0 or later.

Currently generic KASAN is supported for the x86_64, arm64, xtensa, s390 andriscv architectures, and tag-based KASAN is supported only for arm64.

Usage

To enable KASAN configure kernel with:

CONFIG_KASAN = y

and choose between CONFIG_KASAN_GENERIC (to enable generic KASAN) andCONFIG_KASAN_SW_TAGS (to enable software tag-based KASAN).

You also need to choose between CONFIG_KASAN_OUTLINE and CONFIG_KASAN_INLINE.Outline and inline are compiler instrumentation types. The former producessmaller binary while the latter is 1.1 - 2 times faster.

Both KASAN modes work with both SLUB and SLAB memory allocators.For better bug detection and nicer reporting, enable CONFIG_STACKTRACE.

To augment reports with last allocation and freeing stack of the physical page,it is recommended to enable also CONFIG_PAGE_OWNER and boot with page_owner=on.

To disable instrumentation for specific files or directories, add a linesimilar to the following to the respective kernel Makefile:

  • For a single file (e.g. main.o):

    KASAN_SANITIZE_main.o := n

  • For all files in one directory:

    KASAN_SANITIZE := n

Error reports

A typical out-of-bounds access generic KASAN report looks like this:

==================================================================BUG: KASAN: slab-out-of-bounds in kmalloc_oob_right+0xa8/0xbc [test_kasan]Write of size 1 at addr ffff8801f44ec37b by task insmod/2760CPU: 1 PID: 2760 Comm: insmod Not tainted 4.19.0-rc3+ #698Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1 04/01/2014Call Trace: dump_stack+0x94/0xd8 print_address_description+0x73/0x280 kasan_report+0x144/0x187 __asan_report_store1_noabort+0x17/0x20 kmalloc_oob_right+0xa8/0xbc [test_kasan] kmalloc_tests_init+0x16/0x700 [test_kasan] do_one_initcall+0xa5/0x3ae do_init_module+0x1b6/0x547 load_module+0x75df/0x8070 __do_sys_init_module+0x1c6/0x200 __x64_sys_init_module+0x6e/0xb0 do_syscall_64+0x9f/0x2c0 entry_SYSCALL_64_after_hwframe+0x44/0xa9RIP: 0033:0x7f96443109daRSP: 002b:00007ffcf0b51b08 EFLAGS: 00000202 ORIG_RAX: 00000000000000afRAX: ffffffffffffffda RBX: 000055dc3ee521a0 RCX: 00007f96443109daRDX: 00007f96445cff88 RSI: 0000000000057a50 RDI: 00007f9644992000RBP: 000055dc3ee510b0 R08: 0000000000000003 R09: 0000000000000000R10: 00007f964430cd0a R11: 0000000000000202 R12: 00007f96445cff88R13: 000055dc3ee51090 R14: 0000000000000000 R15: 0000000000000000Allocated by task 2760: save_stack+0x43/0xd0 kasan_kmalloc+0xa7/0xd0 kmem_cache_alloc_trace+0xe1/0x1b0 kmalloc_oob_right+0x56/0xbc [test_kasan] kmalloc_tests_init+0x16/0x700 [test_kasan] do_one_initcall+0xa5/0x3ae do_init_module+0x1b6/0x547 load_module+0x75df/0x8070 __do_sys_init_module+0x1c6/0x200 __x64_sys_init_module+0x6e/0xb0 do_syscall_64+0x9f/0x2c0 entry_SYSCALL_64_after_hwframe+0x44/0xa9Freed by task 815: save_stack+0x43/0xd0 __kasan_slab_free+0x135/0x190 kasan_slab_free+0xe/0x10 kfree+0x93/0x1a0 umh_complete+0x6a/0xa0 call_usermodehelper_exec_async+0x4c3/0x640 ret_from_fork+0x35/0x40The buggy address belongs to the object at ffff8801f44ec300 which belongs to the cache kmalloc-128 of size 128The buggy address is located 123 bytes inside of 128-byte region [ffff8801f44ec300, ffff8801f44ec380)The buggy address belongs to the page:page:ffffea0007d13b00 count:1 mapcount:0 mapping:ffff8801f7001640 index:0x0flags: 0x200000000000100(slab)raw: 0200000000000100 ffffea0007d11dc0 0000001a0000001a ffff8801f7001640raw: 0000000000000000 0000000080150015 00000001ffffffff 0000000000000000page dumped because: kasan: bad access detectedMemory state around the buggy address: ffff8801f44ec200: fc fc fc fc fc fc fc fc fb fb fb fb fb fb fb fb ffff8801f44ec280: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc>ffff8801f44ec300: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 03                                                                ^ ffff8801f44ec380: fc fc fc fc fc fc fc fc fb fb fb fb fb fb fb fb ffff8801f44ec400: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc==================================================================

The header of the report provides a short summary of what kind of bug happenedand what kind of access caused it. It’s followed by a stack trace of the badaccess, a stack trace of where the accessed memory was allocated (in case badaccess happens on a slab object), and a stack trace of where the object wasfreed (in case of a use-after-free bug report). Next comes a description ofthe accessed slab object and information about the accessed memory page.

In the last section the report shows memory state around the accessed address.Reading this part requires some understanding of how KASAN works.

The state of each 8 aligned bytes of memory is encoded in one shadow byte.Those 8 bytes can be accessible, partially accessible, freed or be a redzone.We use the following encoding for each shadow byte: 0 means that all 8 bytesof the corresponding memory region are accessible; number N (1 <= N <= 7) meansthat the first N bytes are accessible, and other (8 - N) bytes are not;any negative value indicates that the entire 8-byte word is inaccessible.We use different negative values to distinguish between different kinds ofinaccessible memory like redzones or freed memory (see mm/kasan/kasan.h).

In the report above the arrows point to the shadow byte 03, which means thatthe accessed address is partially accessible.

For tag-based KASAN this last report section shows the memory tags around theaccessed address (see Implementation details section).

Implementation details

Generic KASAN

From a high level, our approach to memory error detection is similar to thatof kmemcheck: use shadow memory to record whether each byte of memory is safeto access, and use compile-time instrumentation to insert checks of shadowmemory on each memory access.

Generic KASAN dedicates 1/8th of kernel memory to its shadow memory (e.g. 16TBto cover 128TB on x86_64) and uses direct mapping with a scale and offset totranslate a memory address to its corresponding shadow address.

Here is the function which translates an address to its corresponding shadowaddress:

static inline void *kasan_mem_to_shadow(const void *addr){    return ((unsigned long)addr >> KASAN_SHADOW_SCALE_SHIFT)            + KASAN_SHADOW_OFFSET;}

whereKASAN_SHADOW_SCALE_SHIFT=3.

Compile-time instrumentation is used to insert memory access checks. Compilerinserts function calls (__asan_load*(addr), __asan_store*(addr)) before eachmemory access of size 1, 2, 4, 8 or 16. These functions check whether memoryaccess is valid or not by checking corresponding shadow memory.

GCC 5.0 has possibility to perform inline instrumentation. Instead of makingfunction calls GCC directly inserts the code to check the shadow memory.This option significantly enlarges kernel but it gives x1.1-x2 performanceboost over outline instrumented kernel.

Software tag-based KASAN

Tag-based KASAN uses the Top Byte Ignore (TBI) feature of modern arm64 CPUs tostore a pointer tag in the top byte of kernel pointers. Like generic KASAN ituses shadow memory to store memory tags associated with each 16-byte memorycell (therefore it dedicates 1/16th of the kernel memory for shadow memory).

On each memory allocation tag-based KASAN generates a random tag, tags theallocated memory with this tag, and embeds this tag into the returned pointer.Software tag-based KASAN uses compile-time instrumentation to insert checksbefore each memory access. These checks make sure that tag of the memory thatis being accessed is equal to tag of the pointer that is used to access thismemory. In case of a tag mismatch tag-based KASAN prints a bug report.

Software tag-based KASAN also has two instrumentation modes (outline, thatemits callbacks to check memory accesses; and inline, that performs the shadowmemory checks inline). With outline instrumentation mode, a bug report issimply printed from the function that performs the access check. With inlineinstrumentation a brk instruction is emitted by the compiler, and a dedicatedbrk handler is used to print bug reports.

A potential expansion of this mode is a hardware tag-based mode, which woulduse hardware memory tagging support instead of compiler instrumentation andmanual shadow memory manipulation.

What memory accesses are sanitised by KASAN?

The kernel maps memory in a number of different parts of the addressspace. This poses something of a problem for KASAN, which requiresthat all addresses accessed by instrumented code have a valid shadowregion.

The range of kernel virtual addresses is large: there is not enoughreal memory to support a real shadow region for every address thatcould be accessed by the kernel.

By default

By default, architectures only map real memory over the shadow regionfor the linear mapping (and potentially other small areas). For allother areas - such as vmalloc and vmemmap space - a single read-onlypage is mapped over the shadow area. This read-only shadow pagedeclares all memory accesses as permitted.

This presents a problem for modules: they do not live in the linearmapping, but in a dedicated module space. By hooking in to the moduleallocator, KASAN can temporarily map real shadow memory to coverthem. This allows detection of invalid accesses to module globals, forexample.

This also creates an incompatibility withVMAP_STACK: if the stacklives in vmalloc space, it will be shadowed by the read-only page, andthe kernel will fault when trying to set up the shadow data for stackvariables.

CONFIG_KASAN_VMALLOC

WithCONFIG_KASAN_VMALLOC, KASAN can cover vmalloc space at thecost of greater memory usage. Currently this is only supported on x86.

This works by hooking into vmalloc and vmap, and dynamicallyallocating real shadow memory to back the mappings.

Most mappings in vmalloc space are small, requiring less than a fullpage of shadow space. Allocating a full shadow page per mapping wouldtherefore be wasteful. Furthermore, to ensure that different mappingsuse different shadow pages, mappings would have to be aligned toKASAN_SHADOW_SCALE_SIZE*PAGE_SIZE.

Instead, we share backing space across multiple mappings. We allocatea backing page when a mapping in vmalloc space uses a particular pageof the shadow region. This page can be shared by other vmallocmappings later on.

We hook in to the vmap infrastructure to lazily clean up unused shadowmemory.

To avoid the difficulties around swapping mappings around, we expectthat the part of the shadow region that covers the vmalloc space willnot be covered by the early shadow page, but will be leftunmapped. This will require changes in arch-specific code.

This allowsVMAP_STACK support on x86, and can simplify support ofarchitectures that do not have a fixed module region.