diff mbox series

[v4,14/36] lib: add allocation tagging support for memory allocation profiling

Message ID 20240221194052.927623-15-surenb@google.com (mailing list archive)
State New
Headers show
Series Memory allocation profiling | expand

Commit Message

Suren Baghdasaryan Feb. 21, 2024, 7:40 p.m. UTC 
Introduce CONFIG_MEM_ALLOC_PROFILING which provides definitions to easily
instrument memory allocators. It registers an "alloc_tags" codetag type
with /proc/allocinfo interface to output allocation tag information when
the feature is enabled.
CONFIG_MEM_ALLOC_PROFILING_DEBUG is provided for debugging the memory
allocation profiling instrumentation.
Memory allocation profiling can be enabled or disabled at runtime using
/proc/sys/vm/mem_profiling sysctl when CONFIG_MEM_ALLOC_PROFILING_DEBUG=n.
CONFIG_MEM_ALLOC_PROFILING_ENABLED_BY_DEFAULT enables memory allocation
profiling by default.

Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Co-developed-by: Kent Overstreet <kent.overstreet@linux.dev>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
---
 Documentation/admin-guide/sysctl/vm.rst |  16 +++
 Documentation/filesystems/proc.rst      |  29 +++++
 include/asm-generic/codetag.lds.h       |  14 +++
 include/asm-generic/vmlinux.lds.h       |   3 +
 include/linux/alloc_tag.h               | 144 +++++++++++++++++++++++
 include/linux/sched.h                   |  24 ++++
 lib/Kconfig.debug                       |  25 ++++
 lib/Makefile                            |   2 +
 lib/alloc_tag.c                         | 149 ++++++++++++++++++++++++
 scripts/module.lds.S                    |   7 ++
 10 files changed, 413 insertions(+)
 create mode 100644 include/asm-generic/codetag.lds.h
 create mode 100644 include/linux/alloc_tag.h
 create mode 100644 lib/alloc_tag.c

Comments

Kees Cook Feb. 21, 2024, 11:05 p.m. UTC | #1 
On Wed, Feb 21, 2024 at 11:40:27AM -0800, Suren Baghdasaryan wrote:
> [...]
> +struct alloc_tag {
> +	struct codetag			ct;
> +	struct alloc_tag_counters __percpu	*counters;
> +} __aligned(8);
> [...]
> +#define DEFINE_ALLOC_TAG(_alloc_tag)						\
> +	static DEFINE_PER_CPU(struct alloc_tag_counters, _alloc_tag_cntr);	\
> +	static struct alloc_tag _alloc_tag __used __aligned(8)			\
> +	__section("alloc_tags") = {						\
> +		.ct = CODE_TAG_INIT,						\
> +		.counters = &_alloc_tag_cntr };
> [...]
> +static inline struct alloc_tag *alloc_tag_save(struct alloc_tag *tag)
> +{
> +	swap(current->alloc_tag, tag);
> +	return tag;
> +}

Future security hardening improvement idea based on this infrastructure:
it should be possible to implement per-allocation-site kmem caches. For
example, we could create:

struct alloc_details {
	u32 flags;
	union {
		u32 size; /* not valid after __init completes */
		struct kmem_cache *cache;
	};
};

- add struct alloc_details to struct alloc_tag
- move the tags section into .ro_after_init
- extend alloc_hooks() to populate flags and size:
	.flags = __builtin_constant_p(size) ? KMALLOC_ALLOCATE_FIXED
					    : KMALLOC_ALLOCATE_BUCKETS;
	.size = __builtin_constant_p(size) ? size : SIZE_MAX;
- during kernel start or module init, walk the alloc_tag list
  and create either a fixed-size kmem_cache or to allocate a
  full set of kmalloc-buckets, and update the "cache" member.
- adjust kmalloc core routines to use current->alloc_tag->cache instead
  of using the global buckets.

This would get us fully separated allocations, producing better than
type-based levels of granularity, exceeding what we have currently with
CONFIG_RANDOM_KMALLOC_CACHES.

Does this look possible, or am I misunderstanding something in the
infrastructure being created here?
Kent Overstreet Feb. 21, 2024, 11:29 p.m. UTC | #2 
On Wed, Feb 21, 2024 at 03:05:32PM -0800, Kees Cook wrote:
> On Wed, Feb 21, 2024 at 11:40:27AM -0800, Suren Baghdasaryan wrote:
> > [...]
> > +struct alloc_tag {
> > +	struct codetag			ct;
> > +	struct alloc_tag_counters __percpu	*counters;
> > +} __aligned(8);
> > [...]
> > +#define DEFINE_ALLOC_TAG(_alloc_tag)						\
> > +	static DEFINE_PER_CPU(struct alloc_tag_counters, _alloc_tag_cntr);	\
> > +	static struct alloc_tag _alloc_tag __used __aligned(8)			\
> > +	__section("alloc_tags") = {						\
> > +		.ct = CODE_TAG_INIT,						\
> > +		.counters = &_alloc_tag_cntr };
> > [...]
> > +static inline struct alloc_tag *alloc_tag_save(struct alloc_tag *tag)
> > +{
> > +	swap(current->alloc_tag, tag);
> > +	return tag;
> > +}
> 
> Future security hardening improvement idea based on this infrastructure:
> it should be possible to implement per-allocation-site kmem caches. For
> example, we could create:
> 
> struct alloc_details {
> 	u32 flags;
> 	union {
> 		u32 size; /* not valid after __init completes */
> 		struct kmem_cache *cache;
> 	};
> };
> 
> - add struct alloc_details to struct alloc_tag
> - move the tags section into .ro_after_init
> - extend alloc_hooks() to populate flags and size:
> 	.flags = __builtin_constant_p(size) ? KMALLOC_ALLOCATE_FIXED
> 					    : KMALLOC_ALLOCATE_BUCKETS;
> 	.size = __builtin_constant_p(size) ? size : SIZE_MAX;
> - during kernel start or module init, walk the alloc_tag list
>   and create either a fixed-size kmem_cache or to allocate a
>   full set of kmalloc-buckets, and update the "cache" member.
> - adjust kmalloc core routines to use current->alloc_tag->cache instead
>   of using the global buckets.
> 
> This would get us fully separated allocations, producing better than
> type-based levels of granularity, exceeding what we have currently with
> CONFIG_RANDOM_KMALLOC_CACHES.
> 
> Does this look possible, or am I misunderstanding something in the
> infrastructure being created here?

Definitely possible, but... would we want this? That would produce a
_lot_ of kmem caches, and don't we already try to collapse those where
possible to reduce internal fragmentation?
Kees Cook Feb. 22, 2024, 12:25 a.m. UTC | #3 
On Wed, Feb 21, 2024 at 06:29:17PM -0500, Kent Overstreet wrote:
> On Wed, Feb 21, 2024 at 03:05:32PM -0800, Kees Cook wrote:
> > On Wed, Feb 21, 2024 at 11:40:27AM -0800, Suren Baghdasaryan wrote:
> > > [...]
> > > +struct alloc_tag {
> > > +	struct codetag			ct;
> > > +	struct alloc_tag_counters __percpu	*counters;
> > > +} __aligned(8);
> > > [...]
> > > +#define DEFINE_ALLOC_TAG(_alloc_tag)						\
> > > +	static DEFINE_PER_CPU(struct alloc_tag_counters, _alloc_tag_cntr);	\
> > > +	static struct alloc_tag _alloc_tag __used __aligned(8)			\
> > > +	__section("alloc_tags") = {						\
> > > +		.ct = CODE_TAG_INIT,						\
> > > +		.counters = &_alloc_tag_cntr };
> > > [...]
> > > +static inline struct alloc_tag *alloc_tag_save(struct alloc_tag *tag)
> > > +{
> > > +	swap(current->alloc_tag, tag);
> > > +	return tag;
> > > +}
> > 
> > Future security hardening improvement idea based on this infrastructure:
> > it should be possible to implement per-allocation-site kmem caches. For
> > example, we could create:
> > 
> > struct alloc_details {
> > 	u32 flags;
> > 	union {
> > 		u32 size; /* not valid after __init completes */
> > 		struct kmem_cache *cache;
> > 	};
> > };
> > 
> > - add struct alloc_details to struct alloc_tag
> > - move the tags section into .ro_after_init
> > - extend alloc_hooks() to populate flags and size:
> > 	.flags = __builtin_constant_p(size) ? KMALLOC_ALLOCATE_FIXED
> > 					    : KMALLOC_ALLOCATE_BUCKETS;
> > 	.size = __builtin_constant_p(size) ? size : SIZE_MAX;
> > - during kernel start or module init, walk the alloc_tag list
> >   and create either a fixed-size kmem_cache or to allocate a
> >   full set of kmalloc-buckets, and update the "cache" member.
> > - adjust kmalloc core routines to use current->alloc_tag->cache instead
> >   of using the global buckets.
> > 
> > This would get us fully separated allocations, producing better than
> > type-based levels of granularity, exceeding what we have currently with
> > CONFIG_RANDOM_KMALLOC_CACHES.
> > 
> > Does this look possible, or am I misunderstanding something in the
> > infrastructure being created here?
> 
> Definitely possible, but... would we want this?

Yes, very very much. One of the worst and mostly unaddressed weaknesses
with the kernel right now is use-after-free based type confusion[0], which
depends on merged caches (or cache reuse).

This doesn't solve cross-allocator (kmalloc/page_alloc) type confusion
(as terrifyingly demonstrated[1] by Jann Horn), but it does help with
what has been a very common case of "use msg_msg to impersonate your
target object"[2] exploitation.

> That would produce a _lot_ of kmem caches

Fewer than you'd expect, but yes, there is some overhead. However,
out-of-tree forks of Linux have successfully experimented with this
already and seen good results[3].

> and don't we already try to collapse those where possible to reduce
> internal fragmentation?

In the past, yes, but the desire for security has tended to have more
people building with SLAB_MERGE_DEFAULT=n and/or CONFIG_RANDOM_KMALLOC_CACHES=y
(or booting with "slab_nomerge").

Just doing the type safety isn't sufficient without the cross-allocator
safety, but we've also had solutions for that proposed[4].

-Kees

[0] https://github.com/KSPP/linux/issues/189
[1] https://googleprojectzero.blogspot.com/2021/10/how-simple-linux-kernel-memory.html
[2] https://www.willsroot.io/2021/08/corctf-2021-fire-of-salvation-writeup.html
    https://google.github.io/security-research/pocs/linux/cve-2021-22555/writeup.html#exploring-struct-msg_msg
[3] https://grsecurity.net/how_autoslab_changes_the_memory_unsafety_game
[4] https://lore.kernel.org/linux-hardening/20230915105933.495735-1-matteorizzo@google.com/
Kent Overstreet Feb. 22, 2024, 12:34 a.m. UTC | #4 
On Wed, Feb 21, 2024 at 04:25:02PM -0800, Kees Cook wrote:
> On Wed, Feb 21, 2024 at 06:29:17PM -0500, Kent Overstreet wrote:
> > On Wed, Feb 21, 2024 at 03:05:32PM -0800, Kees Cook wrote:
> > > On Wed, Feb 21, 2024 at 11:40:27AM -0800, Suren Baghdasaryan wrote:
> > > > [...]
> > > > +struct alloc_tag {
> > > > +	struct codetag			ct;
> > > > +	struct alloc_tag_counters __percpu	*counters;
> > > > +} __aligned(8);
> > > > [...]
> > > > +#define DEFINE_ALLOC_TAG(_alloc_tag)						\
> > > > +	static DEFINE_PER_CPU(struct alloc_tag_counters, _alloc_tag_cntr);	\
> > > > +	static struct alloc_tag _alloc_tag __used __aligned(8)			\
> > > > +	__section("alloc_tags") = {						\
> > > > +		.ct = CODE_TAG_INIT,						\
> > > > +		.counters = &_alloc_tag_cntr };
> > > > [...]
> > > > +static inline struct alloc_tag *alloc_tag_save(struct alloc_tag *tag)
> > > > +{
> > > > +	swap(current->alloc_tag, tag);
> > > > +	return tag;
> > > > +}
> > > 
> > > Future security hardening improvement idea based on this infrastructure:
> > > it should be possible to implement per-allocation-site kmem caches. For
> > > example, we could create:
> > > 
> > > struct alloc_details {
> > > 	u32 flags;
> > > 	union {
> > > 		u32 size; /* not valid after __init completes */
> > > 		struct kmem_cache *cache;
> > > 	};
> > > };
> > > 
> > > - add struct alloc_details to struct alloc_tag
> > > - move the tags section into .ro_after_init
> > > - extend alloc_hooks() to populate flags and size:
> > > 	.flags = __builtin_constant_p(size) ? KMALLOC_ALLOCATE_FIXED
> > > 					    : KMALLOC_ALLOCATE_BUCKETS;
> > > 	.size = __builtin_constant_p(size) ? size : SIZE_MAX;
> > > - during kernel start or module init, walk the alloc_tag list
> > >   and create either a fixed-size kmem_cache or to allocate a
> > >   full set of kmalloc-buckets, and update the "cache" member.
> > > - adjust kmalloc core routines to use current->alloc_tag->cache instead
> > >   of using the global buckets.
> > > 
> > > This would get us fully separated allocations, producing better than
> > > type-based levels of granularity, exceeding what we have currently with
> > > CONFIG_RANDOM_KMALLOC_CACHES.
> > > 
> > > Does this look possible, or am I misunderstanding something in the
> > > infrastructure being created here?
> > 
> > Definitely possible, but... would we want this?
> 
> Yes, very very much. One of the worst and mostly unaddressed weaknesses
> with the kernel right now is use-after-free based type confusion[0], which
> depends on merged caches (or cache reuse).
> 
> This doesn't solve cross-allocator (kmalloc/page_alloc) type confusion
> (as terrifyingly demonstrated[1] by Jann Horn), but it does help with
> what has been a very common case of "use msg_msg to impersonate your
> target object"[2] exploitation.

We have a ton of code that references PAGE_SIZE and uses the page
allocator completely unnecessarily - that's something worth harping
about at conferences; if we could motivate people to clean that stuff up
it'd have a lot of positive effects.

> > That would produce a _lot_ of kmem caches
> 
> Fewer than you'd expect, but yes, there is some overhead. However,
> out-of-tree forks of Linux have successfully experimented with this
> already and seen good results[3].

So in that case - I don't think there's any need for a separate
alloc_details; we'd just add a kmem_cache * to alloc_tag and then hook
into the codetag init/unload path to create and destroy the kmem caches.

No need to adjust the slab code either; alloc_hooks() itself could
dispatch to kmem_cache_alloc() instead of kmalloc() if this is in use.
Kees Cook Feb. 22, 2024, 12:57 a.m. UTC | #5 
On Wed, Feb 21, 2024 at 07:34:44PM -0500, Kent Overstreet wrote:
> On Wed, Feb 21, 2024 at 04:25:02PM -0800, Kees Cook wrote:
> > On Wed, Feb 21, 2024 at 06:29:17PM -0500, Kent Overstreet wrote:
> > > On Wed, Feb 21, 2024 at 03:05:32PM -0800, Kees Cook wrote:
> > > > On Wed, Feb 21, 2024 at 11:40:27AM -0800, Suren Baghdasaryan wrote:
> > > > > [...]
> > > > > +struct alloc_tag {
> > > > > +	struct codetag			ct;
> > > > > +	struct alloc_tag_counters __percpu	*counters;
> > > > > +} __aligned(8);
> > > > > [...]
> > > > > +#define DEFINE_ALLOC_TAG(_alloc_tag)						\
> > > > > +	static DEFINE_PER_CPU(struct alloc_tag_counters, _alloc_tag_cntr);	\
> > > > > +	static struct alloc_tag _alloc_tag __used __aligned(8)			\
> > > > > +	__section("alloc_tags") = {						\
> > > > > +		.ct = CODE_TAG_INIT,						\
> > > > > +		.counters = &_alloc_tag_cntr };
> > > > > [...]
> > > > > +static inline struct alloc_tag *alloc_tag_save(struct alloc_tag *tag)
> > > > > +{
> > > > > +	swap(current->alloc_tag, tag);
> > > > > +	return tag;
> > > > > +}
> > > > 
> > > > Future security hardening improvement idea based on this infrastructure:
> > > > it should be possible to implement per-allocation-site kmem caches. For
> > > > example, we could create:
> > > > 
> > > > struct alloc_details {
> > > > 	u32 flags;
> > > > 	union {
> > > > 		u32 size; /* not valid after __init completes */
> > > > 		struct kmem_cache *cache;
> > > > 	};
> > > > };
> > > > 
> > > > - add struct alloc_details to struct alloc_tag
> > > > - move the tags section into .ro_after_init
> > > > - extend alloc_hooks() to populate flags and size:
> > > > 	.flags = __builtin_constant_p(size) ? KMALLOC_ALLOCATE_FIXED
> > > > 					    : KMALLOC_ALLOCATE_BUCKETS;
> > > > 	.size = __builtin_constant_p(size) ? size : SIZE_MAX;
> > > > - during kernel start or module init, walk the alloc_tag list
> > > >   and create either a fixed-size kmem_cache or to allocate a
> > > >   full set of kmalloc-buckets, and update the "cache" member.
> > > > - adjust kmalloc core routines to use current->alloc_tag->cache instead
> > > >   of using the global buckets.
> > > > 
> > > > This would get us fully separated allocations, producing better than
> > > > type-based levels of granularity, exceeding what we have currently with
> > > > CONFIG_RANDOM_KMALLOC_CACHES.
> > > > 
> > > > Does this look possible, or am I misunderstanding something in the
> > > > infrastructure being created here?
> > > 
> > > Definitely possible, but... would we want this?
> > 
> > Yes, very very much. One of the worst and mostly unaddressed weaknesses
> > with the kernel right now is use-after-free based type confusion[0], which
> > depends on merged caches (or cache reuse).
> > 
> > This doesn't solve cross-allocator (kmalloc/page_alloc) type confusion
> > (as terrifyingly demonstrated[1] by Jann Horn), but it does help with
> > what has been a very common case of "use msg_msg to impersonate your
> > target object"[2] exploitation.
> 
> We have a ton of code that references PAGE_SIZE and uses the page
> allocator completely unnecessarily - that's something worth harping
> about at conferences; if we could motivate people to clean that stuff up
> it'd have a lot of positive effects.
> 
> > > That would produce a _lot_ of kmem caches
> > 
> > Fewer than you'd expect, but yes, there is some overhead. However,
> > out-of-tree forks of Linux have successfully experimented with this
> > already and seen good results[3].
> 
> So in that case - I don't think there's any need for a separate
> alloc_details; we'd just add a kmem_cache * to alloc_tag and then hook
> into the codetag init/unload path to create and destroy the kmem caches.

Okay, sounds good. There needs to be a place to track "is this a fixed
size or a run-time size" choice.

> No need to adjust the slab code either; alloc_hooks() itself could
> dispatch to kmem_cache_alloc() instead of kmalloc() if this is in use.

Right, it'd go to either kmem_cache_alloc() directly, or to a modified
kmalloc() that used the passed-in cache is the base for an array of sized
buckets, rather than the global (or 16-way global) buckets.

Yay for the future!
Vlastimil Babka Feb. 28, 2024, 8:29 a.m. UTC | #6 
On 2/21/24 20:40, Suren Baghdasaryan wrote:
> 
> +static inline void alloc_tag_sub(union codetag_ref *ref, size_t bytes)
> +{
> + __alloc_tag_sub(ref, bytes);
> +}
> +
> +static inline void alloc_tag_sub_noalloc(union codetag_ref *ref, size_t bytes)
> +{
> + __alloc_tag_sub(ref, bytes);
> +}
> +

Nit: just notice these are now the same and maybe you could just drop both
wrappers and rename __alloc_tag_sub to alloc_tag_sub?
Vlastimil Babka Feb. 28, 2024, 8:41 a.m. UTC | #7 
Another thing I noticed, dunno how critical

On 2/21/24 20:40, Suren Baghdasaryan wrote:
> +static inline void __alloc_tag_sub(union codetag_ref *ref, size_t bytes)
> +{
> +	struct alloc_tag *tag;
> +
> +#ifdef CONFIG_MEM_ALLOC_PROFILING_DEBUG
> +	WARN_ONCE(ref && !ref->ct, "alloc_tag was not set\n");
> +#endif
> +	if (!ref || !ref->ct)
> +		return;

This is quite careful.

> +
> +	tag = ct_to_alloc_tag(ref->ct);
> +
> +	this_cpu_sub(tag->counters->bytes, bytes);
> +	this_cpu_dec(tag->counters->calls);
> +
> +	ref->ct = NULL;
> +}
> +
> +static inline void alloc_tag_sub(union codetag_ref *ref, size_t bytes)
> +{
> +	__alloc_tag_sub(ref, bytes);
> +}
> +
> +static inline void alloc_tag_sub_noalloc(union codetag_ref *ref, size_t bytes)
> +{
> +	__alloc_tag_sub(ref, bytes);
> +}
> +
> +static inline void alloc_tag_ref_set(union codetag_ref *ref, struct alloc_tag *tag)
> +{
> +#ifdef CONFIG_MEM_ALLOC_PROFILING_DEBUG
> +	WARN_ONCE(ref && ref->ct,
> +		  "alloc_tag was not cleared (got tag for %s:%u)\n",\
> +		  ref->ct->filename, ref->ct->lineno);
> +
> +	WARN_ONCE(!tag, "current->alloc_tag not set");
> +#endif
> +	if (!ref || !tag)
> +		return;

This too.

> +
> +	ref->ct = &tag->ct;
> +	/*
> +	 * We need in increment the call counter every time we have a new
> +	 * allocation or when we split a large allocation into smaller ones.
> +	 * Each new reference for every sub-allocation needs to increment call
> +	 * counter because when we free each part the counter will be decremented.
> +	 */
> +	this_cpu_inc(tag->counters->calls);
> +}
> +
> +static inline void alloc_tag_add(union codetag_ref *ref, struct alloc_tag *tag, size_t bytes)
> +{
> +	alloc_tag_ref_set(ref, tag);

We might have returned from alloc_tag_ref_set() due to !tag

> +	this_cpu_add(tag->counters->bytes, bytes);

But here we still assume it's valid.

> +}
> +
Suren Baghdasaryan Feb. 28, 2024, 6:05 p.m. UTC | #8 
On Wed, Feb 28, 2024 at 8:29 AM Vlastimil Babka <vbabka@suse.cz> wrote:
>
> On 2/21/24 20:40, Suren Baghdasaryan wrote:
> >
> > +static inline void alloc_tag_sub(union codetag_ref *ref, size_t bytes)
> > +{
> > + __alloc_tag_sub(ref, bytes);
> > +}
> > +
> > +static inline void alloc_tag_sub_noalloc(union codetag_ref *ref, size_t bytes)
> > +{
> > + __alloc_tag_sub(ref, bytes);
> > +}
> > +
>
> Nit: just notice these are now the same and maybe you could just drop both
> wrappers and rename __alloc_tag_sub to alloc_tag_sub?

Ack.

>
> --
> To unsubscribe from this group and stop receiving emails from it, send an email to kernel-team+unsubscribe@android.com.
>
Suren Baghdasaryan Feb. 28, 2024, 6:07 p.m. UTC | #9 
On Wed, Feb 28, 2024 at 8:41 AM Vlastimil Babka <vbabka@suse.cz> wrote:
>
> Another thing I noticed, dunno how critical
>
> On 2/21/24 20:40, Suren Baghdasaryan wrote:
> > +static inline void __alloc_tag_sub(union codetag_ref *ref, size_t bytes)
> > +{
> > +     struct alloc_tag *tag;
> > +
> > +#ifdef CONFIG_MEM_ALLOC_PROFILING_DEBUG
> > +     WARN_ONCE(ref && !ref->ct, "alloc_tag was not set\n");
> > +#endif
> > +     if (!ref || !ref->ct)
> > +             return;
>
> This is quite careful.
>
> > +
> > +     tag = ct_to_alloc_tag(ref->ct);
> > +
> > +     this_cpu_sub(tag->counters->bytes, bytes);
> > +     this_cpu_dec(tag->counters->calls);
> > +
> > +     ref->ct = NULL;
> > +}
> > +
> > +static inline void alloc_tag_sub(union codetag_ref *ref, size_t bytes)
> > +{
> > +     __alloc_tag_sub(ref, bytes);
> > +}
> > +
> > +static inline void alloc_tag_sub_noalloc(union codetag_ref *ref, size_t bytes)
> > +{
> > +     __alloc_tag_sub(ref, bytes);
> > +}
> > +
> > +static inline void alloc_tag_ref_set(union codetag_ref *ref, struct alloc_tag *tag)
> > +{
> > +#ifdef CONFIG_MEM_ALLOC_PROFILING_DEBUG
> > +     WARN_ONCE(ref && ref->ct,
> > +               "alloc_tag was not cleared (got tag for %s:%u)\n",\
> > +               ref->ct->filename, ref->ct->lineno);
> > +
> > +     WARN_ONCE(!tag, "current->alloc_tag not set");
> > +#endif
> > +     if (!ref || !tag)
> > +             return;
>
> This too.
>
> > +
> > +     ref->ct = &tag->ct;
> > +     /*
> > +      * We need in increment the call counter every time we have a new
> > +      * allocation or when we split a large allocation into smaller ones.
> > +      * Each new reference for every sub-allocation needs to increment call
> > +      * counter because when we free each part the counter will be decremented.
> > +      */
> > +     this_cpu_inc(tag->counters->calls);
> > +}
> > +
> > +static inline void alloc_tag_add(union codetag_ref *ref, struct alloc_tag *tag, size_t bytes)
> > +{
> > +     alloc_tag_ref_set(ref, tag);
>
> We might have returned from alloc_tag_ref_set() due to !tag
>
> > +     this_cpu_add(tag->counters->bytes, bytes);
>
> But here we still assume it's valid.

Yes, this is a blunder on my side after splitting alloc_tag_ref_set()
into a separate function. I'll fix this in the next version. Thanks!

>
> > +}
> > +
>
> --
> To unsubscribe from this group and stop receiving emails from it, send an email to kernel-team+unsubscribe@android.com.
>
diff mbox series

Patch

diff --git a/Documentation/admin-guide/sysctl/vm.rst b/Documentation/admin-guide/sysctl/vm.rst
index c59889de122b..e86c968a7a0e 100644
--- a/Documentation/admin-guide/sysctl/vm.rst
+++ b/Documentation/admin-guide/sysctl/vm.rst
@@ -43,6 +43,7 @@  Currently, these files are in /proc/sys/vm:
 - legacy_va_layout
 - lowmem_reserve_ratio
 - max_map_count
+- mem_profiling         (only if CONFIG_MEM_ALLOC_PROFILING=y)
 - memory_failure_early_kill
 - memory_failure_recovery
 - min_free_kbytes
@@ -425,6 +426,21 @@  e.g., up to one or two maps per allocation.
 The default value is 65530.
 
 
+mem_profiling
+==============
+
+Enable memory profiling (when CONFIG_MEM_ALLOC_PROFILING=y)
+
+1: Enable memory profiling.
+
+0: Disable memory profiling.
+
+Enabling memory profiling introduces a small performance overhead for all
+memory allocations.
+
+The default value depends on CONFIG_MEM_ALLOC_PROFILING_ENABLED_BY_DEFAULT.
+
+
 memory_failure_early_kill:
 ==========================
 
diff --git a/Documentation/filesystems/proc.rst b/Documentation/filesystems/proc.rst
index 104c6d047d9b..8150dc3d689c 100644
--- a/Documentation/filesystems/proc.rst
+++ b/Documentation/filesystems/proc.rst
@@ -688,6 +688,7 @@  files are there, and which are missing.
  ============ ===============================================================
  File         Content
  ============ ===============================================================
+ allocinfo    Memory allocations profiling information
  apm          Advanced power management info
  bootconfig   Kernel command line obtained from boot config,
  	      and, if there were kernel parameters from the
@@ -953,6 +954,34 @@  also be allocatable although a lot of filesystem metadata may have to be
 reclaimed to achieve this.
 
 
+allocinfo
+~~~~~~~
+
+Provides information about memory allocations at all locations in the code
+base. Each allocation in the code is identified by its source file, line
+number, module (if originates from a loadable module) and the function calling
+the allocation. The number of bytes allocated and number of calls at each
+location are reported.
+
+Example output.
+
+::
+
+    > sort -rn /proc/allocinfo
+   127664128    31168 mm/page_ext.c:270 func:alloc_page_ext
+    56373248     4737 mm/slub.c:2259 func:alloc_slab_page
+    14880768     3633 mm/readahead.c:247 func:page_cache_ra_unbounded
+    14417920     3520 mm/mm_init.c:2530 func:alloc_large_system_hash
+    13377536      234 block/blk-mq.c:3421 func:blk_mq_alloc_rqs
+    11718656     2861 mm/filemap.c:1919 func:__filemap_get_folio
+     9192960     2800 kernel/fork.c:307 func:alloc_thread_stack_node
+     4206592        4 net/netfilter/nf_conntrack_core.c:2567 func:nf_ct_alloc_hashtable
+     4136960     1010 drivers/staging/ctagmod/ctagmod.c:20 [ctagmod] func:ctagmod_start
+     3940352      962 mm/memory.c:4214 func:alloc_anon_folio
+     2894464    22613 fs/kernfs/dir.c:615 func:__kernfs_new_node
+     ...
+
+
 meminfo
 ~~~~~~~
 
diff --git a/include/asm-generic/codetag.lds.h b/include/asm-generic/codetag.lds.h
new file mode 100644
index 000000000000..64f536b80380
--- /dev/null
+++ b/include/asm-generic/codetag.lds.h
@@ -0,0 +1,14 @@ 
+/* SPDX-License-Identifier: GPL-2.0-only */
+#ifndef __ASM_GENERIC_CODETAG_LDS_H
+#define __ASM_GENERIC_CODETAG_LDS_H
+
+#define SECTION_WITH_BOUNDARIES(_name)	\
+	. = ALIGN(8);			\
+	__start_##_name = .;		\
+	KEEP(*(_name))			\
+	__stop_##_name = .;
+
+#define CODETAG_SECTIONS()		\
+	SECTION_WITH_BOUNDARIES(alloc_tags)
+
+#endif /* __ASM_GENERIC_CODETAG_LDS_H */
diff --git a/include/asm-generic/vmlinux.lds.h b/include/asm-generic/vmlinux.lds.h
index 5dd3a61d673d..c9997dc50c50 100644
--- a/include/asm-generic/vmlinux.lds.h
+++ b/include/asm-generic/vmlinux.lds.h
@@ -50,6 +50,8 @@ 
  *               [__nosave_begin, __nosave_end] for the nosave data
  */
 
+#include <asm-generic/codetag.lds.h>
+
 #ifndef LOAD_OFFSET
 #define LOAD_OFFSET 0
 #endif
@@ -366,6 +368,7 @@ 
 	. = ALIGN(8);							\
 	BOUNDED_SECTION_BY(__dyndbg_classes, ___dyndbg_classes)		\
 	BOUNDED_SECTION_BY(__dyndbg, ___dyndbg)				\
+	CODETAG_SECTIONS()						\
 	LIKELY_PROFILE()		       				\
 	BRANCH_PROFILE()						\
 	TRACE_PRINTKS()							\
diff --git a/include/linux/alloc_tag.h b/include/linux/alloc_tag.h
new file mode 100644
index 000000000000..be3ba955846c
--- /dev/null
+++ b/include/linux/alloc_tag.h
@@ -0,0 +1,144 @@ 
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * allocation tagging
+ */
+#ifndef _LINUX_ALLOC_TAG_H
+#define _LINUX_ALLOC_TAG_H
+
+#include <linux/bug.h>
+#include <linux/codetag.h>
+#include <linux/container_of.h>
+#include <linux/preempt.h>
+#include <asm/percpu.h>
+#include <linux/cpumask.h>
+#include <linux/static_key.h>
+
+struct alloc_tag_counters {
+	u64 bytes;
+	u64 calls;
+};
+
+/*
+ * An instance of this structure is created in a special ELF section at every
+ * allocation callsite. At runtime, the special section is treated as
+ * an array of these. Embedded codetag utilizes codetag framework.
+ */
+struct alloc_tag {
+	struct codetag			ct;
+	struct alloc_tag_counters __percpu	*counters;
+} __aligned(8);
+
+#ifdef CONFIG_MEM_ALLOC_PROFILING
+
+static inline struct alloc_tag *ct_to_alloc_tag(struct codetag *ct)
+{
+	return container_of(ct, struct alloc_tag, ct);
+}
+
+#ifdef ARCH_NEEDS_WEAK_PER_CPU
+/*
+ * When percpu variables are required to be defined as weak, static percpu
+ * variables can't be used inside a function (see comments for DECLARE_PER_CPU_SECTION).
+ */
+#error "Memory allocation profiling is incompatible with ARCH_NEEDS_WEAK_PER_CPU"
+#endif
+
+#define DEFINE_ALLOC_TAG(_alloc_tag)						\
+	static DEFINE_PER_CPU(struct alloc_tag_counters, _alloc_tag_cntr);	\
+	static struct alloc_tag _alloc_tag __used __aligned(8)			\
+	__section("alloc_tags") = {						\
+		.ct = CODE_TAG_INIT,						\
+		.counters = &_alloc_tag_cntr };
+
+DECLARE_STATIC_KEY_MAYBE(CONFIG_MEM_ALLOC_PROFILING_ENABLED_BY_DEFAULT,
+			mem_alloc_profiling_key);
+
+static inline bool mem_alloc_profiling_enabled(void)
+{
+	return static_branch_maybe(CONFIG_MEM_ALLOC_PROFILING_ENABLED_BY_DEFAULT,
+				   &mem_alloc_profiling_key);
+}
+
+static inline struct alloc_tag_counters alloc_tag_read(struct alloc_tag *tag)
+{
+	struct alloc_tag_counters v = { 0, 0 };
+	struct alloc_tag_counters *counter;
+	int cpu;
+
+	for_each_possible_cpu(cpu) {
+		counter = per_cpu_ptr(tag->counters, cpu);
+		v.bytes += counter->bytes;
+		v.calls += counter->calls;
+	}
+
+	return v;
+}
+
+static inline void __alloc_tag_sub(union codetag_ref *ref, size_t bytes)
+{
+	struct alloc_tag *tag;
+
+#ifdef CONFIG_MEM_ALLOC_PROFILING_DEBUG
+	WARN_ONCE(ref && !ref->ct, "alloc_tag was not set\n");
+#endif
+	if (!ref || !ref->ct)
+		return;
+
+	tag = ct_to_alloc_tag(ref->ct);
+
+	this_cpu_sub(tag->counters->bytes, bytes);
+	this_cpu_dec(tag->counters->calls);
+
+	ref->ct = NULL;
+}
+
+static inline void alloc_tag_sub(union codetag_ref *ref, size_t bytes)
+{
+	__alloc_tag_sub(ref, bytes);
+}
+
+static inline void alloc_tag_sub_noalloc(union codetag_ref *ref, size_t bytes)
+{
+	__alloc_tag_sub(ref, bytes);
+}
+
+static inline void alloc_tag_ref_set(union codetag_ref *ref, struct alloc_tag *tag)
+{
+#ifdef CONFIG_MEM_ALLOC_PROFILING_DEBUG
+	WARN_ONCE(ref && ref->ct,
+		  "alloc_tag was not cleared (got tag for %s:%u)\n",\
+		  ref->ct->filename, ref->ct->lineno);
+
+	WARN_ONCE(!tag, "current->alloc_tag not set");
+#endif
+	if (!ref || !tag)
+		return;
+
+	ref->ct = &tag->ct;
+	/*
+	 * We need in increment the call counter every time we have a new
+	 * allocation or when we split a large allocation into smaller ones.
+	 * Each new reference for every sub-allocation needs to increment call
+	 * counter because when we free each part the counter will be decremented.
+	 */
+	this_cpu_inc(tag->counters->calls);
+}
+
+static inline void alloc_tag_add(union codetag_ref *ref, struct alloc_tag *tag, size_t bytes)
+{
+	alloc_tag_ref_set(ref, tag);
+	this_cpu_add(tag->counters->bytes, bytes);
+}
+
+#else /* CONFIG_MEM_ALLOC_PROFILING */
+
+#define DEFINE_ALLOC_TAG(_alloc_tag)
+static inline bool mem_alloc_profiling_enabled(void) { return false; }
+static inline void alloc_tag_sub(union codetag_ref *ref, size_t bytes) {}
+static inline void alloc_tag_sub_noalloc(union codetag_ref *ref, size_t bytes) {}
+static inline void alloc_tag_add(union codetag_ref *ref, struct alloc_tag *tag,
+				 size_t bytes) {}
+
+#endif /* CONFIG_MEM_ALLOC_PROFILING */
+
+#endif /* _LINUX_ALLOC_TAG_H */
diff --git a/include/linux/sched.h b/include/linux/sched.h
index ffe8f618ab86..eede1f92bcc6 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -770,6 +770,10 @@  struct task_struct {
 	unsigned int			flags;
 	unsigned int			ptrace;
 
+#ifdef CONFIG_MEM_ALLOC_PROFILING
+	struct alloc_tag		*alloc_tag;
+#endif
+
 #ifdef CONFIG_SMP
 	int				on_cpu;
 	struct __call_single_node	wake_entry;
@@ -810,6 +814,7 @@  struct task_struct {
 	struct task_group		*sched_task_group;
 #endif
 
+
 #ifdef CONFIG_UCLAMP_TASK
 	/*
 	 * Clamp values requested for a scheduling entity.
@@ -2183,4 +2188,23 @@  static inline int sched_core_idle_cpu(int cpu) { return idle_cpu(cpu); }
 
 extern void sched_set_stop_task(int cpu, struct task_struct *stop);
 
+#ifdef CONFIG_MEM_ALLOC_PROFILING
+static inline struct alloc_tag *alloc_tag_save(struct alloc_tag *tag)
+{
+	swap(current->alloc_tag, tag);
+	return tag;
+}
+
+static inline void alloc_tag_restore(struct alloc_tag *tag, struct alloc_tag *old)
+{
+#ifdef CONFIG_MEM_ALLOC_PROFILING_DEBUG
+	WARN(current->alloc_tag != tag, "current->alloc_tag was changed:\n");
+#endif
+	current->alloc_tag = old;
+}
+#else
+#define alloc_tag_save(_tag)			NULL
+#define alloc_tag_restore(_tag, _old)		do {} while (0)
+#endif
+
 #endif
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index 0be2d00c3696..78d258ca508f 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -972,6 +972,31 @@  config CODE_TAGGING
 	bool
 	select KALLSYMS
 
+config MEM_ALLOC_PROFILING
+	bool "Enable memory allocation profiling"
+	default n
+	depends on PROC_FS
+	depends on !DEBUG_FORCE_WEAK_PER_CPU
+	select CODE_TAGGING
+	help
+	  Track allocation source code and record total allocation size
+	  initiated at that code location. The mechanism can be used to track
+	  memory leaks with a low performance and memory impact.
+
+config MEM_ALLOC_PROFILING_ENABLED_BY_DEFAULT
+	bool "Enable memory allocation profiling by default"
+	default y
+	depends on MEM_ALLOC_PROFILING
+
+config MEM_ALLOC_PROFILING_DEBUG
+	bool "Memory allocation profiler debugging"
+	default n
+	depends on MEM_ALLOC_PROFILING
+	select MEM_ALLOC_PROFILING_ENABLED_BY_DEFAULT
+	help
+	  Adds warnings with helpful error messages for memory allocation
+	  profiling.
+
 source "lib/Kconfig.kasan"
 source "lib/Kconfig.kfence"
 source "lib/Kconfig.kmsan"
diff --git a/lib/Makefile b/lib/Makefile
index 6b48b22fdfac..859112f09bf5 100644
--- a/lib/Makefile
+++ b/lib/Makefile
@@ -236,6 +236,8 @@  obj-$(CONFIG_OF_RECONFIG_NOTIFIER_ERROR_INJECT) += \
 obj-$(CONFIG_FUNCTION_ERROR_INJECTION) += error-inject.o
 
 obj-$(CONFIG_CODE_TAGGING) += codetag.o
+obj-$(CONFIG_MEM_ALLOC_PROFILING) += alloc_tag.o
+
 lib-$(CONFIG_GENERIC_BUG) += bug.o
 
 obj-$(CONFIG_HAVE_ARCH_TRACEHOOK) += syscall.o
diff --git a/lib/alloc_tag.c b/lib/alloc_tag.c
new file mode 100644
index 000000000000..f09c8a422bc2
--- /dev/null
+++ b/lib/alloc_tag.c
@@ -0,0 +1,149 @@ 
+// SPDX-License-Identifier: GPL-2.0-only
+#include <linux/alloc_tag.h>
+#include <linux/fs.h>
+#include <linux/gfp.h>
+#include <linux/module.h>
+#include <linux/proc_fs.h>
+#include <linux/seq_buf.h>
+#include <linux/seq_file.h>
+
+static struct codetag_type *alloc_tag_cttype;
+
+DEFINE_STATIC_KEY_MAYBE(CONFIG_MEM_ALLOC_PROFILING_ENABLED_BY_DEFAULT,
+			mem_alloc_profiling_key);
+
+static void *allocinfo_start(struct seq_file *m, loff_t *pos)
+{
+	struct codetag_iterator *iter;
+	struct codetag *ct;
+	loff_t node = *pos;
+
+	iter = kzalloc(sizeof(*iter), GFP_KERNEL);
+	m->private = iter;
+	if (!iter)
+		return NULL;
+
+	codetag_lock_module_list(alloc_tag_cttype, true);
+	*iter = codetag_get_ct_iter(alloc_tag_cttype);
+	while ((ct = codetag_next_ct(iter)) != NULL && node)
+		node--;
+
+	return ct ? iter : NULL;
+}
+
+static void *allocinfo_next(struct seq_file *m, void *arg, loff_t *pos)
+{
+	struct codetag_iterator *iter = (struct codetag_iterator *)arg;
+	struct codetag *ct = codetag_next_ct(iter);
+
+	(*pos)++;
+	if (!ct)
+		return NULL;
+
+	return iter;
+}
+
+static void allocinfo_stop(struct seq_file *m, void *arg)
+{
+	struct codetag_iterator *iter = (struct codetag_iterator *)m->private;
+
+	if (iter) {
+		codetag_lock_module_list(alloc_tag_cttype, false);
+		kfree(iter);
+	}
+}
+
+static void alloc_tag_to_text(struct seq_buf *out, struct codetag *ct)
+{
+	struct alloc_tag *tag = ct_to_alloc_tag(ct);
+	struct alloc_tag_counters counter = alloc_tag_read(tag);
+	s64 bytes = counter.bytes;
+
+	seq_buf_printf(out, "%12lli %8llu ", bytes, counter.calls);
+	codetag_to_text(out, ct);
+	seq_buf_putc(out, ' ');
+	seq_buf_putc(out, '\n');
+}
+
+static int allocinfo_show(struct seq_file *m, void *arg)
+{
+	struct codetag_iterator *iter = (struct codetag_iterator *)arg;
+	char *bufp;
+	size_t n = seq_get_buf(m, &bufp);
+	struct seq_buf buf;
+
+	seq_buf_init(&buf, bufp, n);
+	alloc_tag_to_text(&buf, iter->ct);
+	seq_commit(m, seq_buf_used(&buf));
+	return 0;
+}
+
+static const struct seq_operations allocinfo_seq_op = {
+	.start	= allocinfo_start,
+	.next	= allocinfo_next,
+	.stop	= allocinfo_stop,
+	.show	= allocinfo_show,
+};
+
+static void __init procfs_init(void)
+{
+	proc_create_seq("allocinfo", 0444, NULL, &allocinfo_seq_op);
+}
+
+static bool alloc_tag_module_unload(struct codetag_type *cttype,
+				    struct codetag_module *cmod)
+{
+	struct codetag_iterator iter = codetag_get_ct_iter(cttype);
+	struct alloc_tag_counters counter;
+	bool module_unused = true;
+	struct alloc_tag *tag;
+	struct codetag *ct;
+
+	for (ct = codetag_next_ct(&iter); ct; ct = codetag_next_ct(&iter)) {
+		if (iter.cmod != cmod)
+			continue;
+
+		tag = ct_to_alloc_tag(ct);
+		counter = alloc_tag_read(tag);
+
+		if (WARN(counter.bytes,
+			 "%s:%u module %s func:%s has %llu allocated at module unload",
+			 ct->filename, ct->lineno, ct->modname, ct->function, counter.bytes))
+			module_unused = false;
+	}
+
+	return module_unused;
+}
+
+static struct ctl_table memory_allocation_profiling_sysctls[] = {
+	{
+		.procname	= "mem_profiling",
+		.data		= &mem_alloc_profiling_key,
+#ifdef CONFIG_MEM_ALLOC_PROFILING_DEBUG
+		.mode		= 0444,
+#else
+		.mode		= 0644,
+#endif
+		.proc_handler	= proc_do_static_key,
+	},
+	{ }
+};
+
+static int __init alloc_tag_init(void)
+{
+	const struct codetag_type_desc desc = {
+		.section	= "alloc_tags",
+		.tag_size	= sizeof(struct alloc_tag),
+		.module_unload	= alloc_tag_module_unload,
+	};
+
+	alloc_tag_cttype = codetag_register_type(&desc);
+	if (IS_ERR_OR_NULL(alloc_tag_cttype))
+		return PTR_ERR(alloc_tag_cttype);
+
+	register_sysctl_init("vm", memory_allocation_profiling_sysctls);
+	procfs_init();
+
+	return 0;
+}
+module_init(alloc_tag_init);
diff --git a/scripts/module.lds.S b/scripts/module.lds.S
index bf5bcf2836d8..45c67a0994f3 100644
--- a/scripts/module.lds.S
+++ b/scripts/module.lds.S
@@ -9,6 +9,8 @@ 
 #define DISCARD_EH_FRAME	*(.eh_frame)
 #endif
 
+#include <asm-generic/codetag.lds.h>
+
 SECTIONS {
 	/DISCARD/ : {
 		*(.discard)
@@ -47,12 +49,17 @@  SECTIONS {
 	.data : {
 		*(.data .data.[0-9a-zA-Z_]*)
 		*(.data..L*)
+		CODETAG_SECTIONS()
 	}
 
 	.rodata : {
 		*(.rodata .rodata.[0-9a-zA-Z_]*)
 		*(.rodata..L*)
 	}
+#else
+	.data : {
+		CODETAG_SECTIONS()
+	}
 #endif
 }