* [Resend v1 1/5] linux/bitqueue.h: add the bit queue implementation
2023-07-11 14:42 [Resend v1 0/5] Implement MTE tag compression for swapped pages Alexander Potapenko
@ 2023-07-11 14:42 ` Alexander Potapenko
2023-07-11 19:20 ` Yury Norov
2023-07-11 14:42 ` [Resend v1 2/5] linux/bitqueue.h: add a KUnit test for bitqueue.h Alexander Potapenko
` (3 subsequent siblings)
4 siblings, 1 reply; 8+ messages in thread
From: Alexander Potapenko @ 2023-07-11 14:42 UTC (permalink / raw)
To: glider, catalin.marinas, will, pcc, andreyknvl
Cc: linux-kernel, linux-arm-kernel, eugenis, yury.norov
struct bitq represents a bit queue with external storage.
Its purpose is to easily pack sub-byte values, which can be used, for
example, to implement RLE algorithms.
Signed-off-by: Alexander Potapenko <glider@google.com>
---
include/linux/bitqueue.h | 144 +++++++++++++++++++++++++++++++++++++++
1 file changed, 144 insertions(+)
create mode 100644 include/linux/bitqueue.h
diff --git a/include/linux/bitqueue.h b/include/linux/bitqueue.h
new file mode 100644
index 0000000000000..c4393f703c697
--- /dev/null
+++ b/include/linux/bitqueue.h
@@ -0,0 +1,144 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * A simple bit queue which supports enqueueing/dequeueing of sub-byte values.
+ *
+ * This can be used to pack complex bitfields into byte arrays.
+ */
+#ifndef _LINUX_BITQUEUE_H
+#define _LINUX_BITQUEUE_H
+
+#include <linux/string.h>
+#include <linux/types.h>
+
+/**
+ * struct bitq - represents a bit queue with external storage.
+ * @data: data buffer used by the queue.
+ * @size: size of @data in bytes.
+ * @bit_pos: current bit position.
+ */
+struct bitq {
+ u8 *data;
+ int size, bit_pos;
+};
+
+/**
+ * bitq_init - initialize an empty bit queue.
+ * @q: struct bitq to be initialized.
+ * @data: external data buffer to use.
+ * @size: capacity in bytes.
+ *
+ * Return: 0 in the case of success, -1 if either of the pointers is NULL.
+ */
+static inline int bitq_init(struct bitq *q, u8 *data, int size)
+{
+ if (!q || !data)
+ return -1;
+ q->data = data;
+ q->size = size;
+ memset(data, 0, size);
+ q->bit_pos = 0;
+ return 0;
+}
+
+/**
+ * bitq_init_full - make a bit queue from an initialized byte array.
+ * @q: struct bitq to be initialized.
+ * @data: external data buffer to use.
+ * @size: capacity in bytes.
+ *
+ * Return: 0 in the case of success, -1 if either of the pointers is NULL.
+ */
+static inline int bitq_init_full(struct bitq *q, u8 *data, int size)
+{
+ if (!q || !data)
+ return -1;
+ q->data = data;
+ q->size = size;
+ q->bit_pos = q->size * 8;
+ return 0;
+}
+
+/**
+ * bitq_enqueue - push up to 8 bits to the end of the queue.
+ * @q: struct bitq.
+ * @value: byte containing the value to be pushed.
+ * @bits: number of bits (1 to 8) to push.
+ *
+ * Return: number of bits pushed, or -1 in the case of an error.
+ */
+static inline int bitq_enqueue(struct bitq *q, u8 value, int bits)
+{
+ int byte_pos, left_in_byte, max_pos;
+ u8 hi, lo;
+
+ if (!q || (bits < 1) || (bits > 8))
+ return -1;
+
+ max_pos = q->size * 8;
+ if ((max_pos - q->bit_pos) < bits)
+ return -1;
+
+ left_in_byte = 8 - (q->bit_pos % 8);
+ byte_pos = q->bit_pos / 8;
+ /* Clamp @value. */
+ value %= (1 << bits);
+ if (left_in_byte >= bits) {
+ /* @value fits into the current byte. */
+ value <<= (left_in_byte - bits);
+ q->data[byte_pos] |= value;
+ } else {
+ /*
+ * @value needs to be split between the current and the
+ * following bytes.
+ */
+ hi = value >> (bits - left_in_byte);
+ q->data[byte_pos] |= hi;
+ byte_pos++;
+ lo = value << (8 - (bits - left_in_byte));
+ q->data[byte_pos] |= lo;
+ }
+ q->bit_pos += bits;
+ return bits;
+}
+
+/**
+ * bitq_dequeue - pop up to 8 bits from the beginning of the queue.
+ * @q: struct bitq.
+ * @value: u8* to store the popped value (can be NULL).
+ * @bits: number of bits (1 to 8) to pop.
+ *
+ * Return: number of bits popped, or -1 in the case of an error.
+ */
+
+#include <linux/printk.h>
+static inline int bitq_dequeue(struct bitq *q, u8 *value, int bits)
+{
+ int rem_bits = 8 - bits, i;
+ u8 output;
+
+ /* Invalid arguments. */
+ if (!q || (bits < 1) || (bits > 8))
+ return -1;
+ /* Not enough space to insert @bits. */
+ if (q->bit_pos < bits)
+ return -1;
+ /* Take the first @bits bits from the first byte. */
+ output = q->data[0];
+ output >>= rem_bits;
+ if (value)
+ *value = output;
+
+ /*
+ * Shift every byte in the queue to the left by @bits, carrying over to
+ * the previous byte.
+ */
+ for (i = 0; i < q->size - 1; i++) {
+ q->data[i] = (q->data[i] << bits) |
+ (q->data[i + 1] >> rem_bits);
+ }
+ q->data[q->size - 1] <<= bits;
+ q->bit_pos -= bits;
+ return bits;
+}
+
+#endif // _LINUX_BITQUEUE_H
--
2.41.0.255.g8b1d071c50-goog
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 8+ messages in thread* Re: [Resend v1 1/5] linux/bitqueue.h: add the bit queue implementation
2023-07-11 14:42 ` [Resend v1 1/5] linux/bitqueue.h: add the bit queue implementation Alexander Potapenko
@ 2023-07-11 19:20 ` Yury Norov
2023-07-12 10:24 ` Alexander Potapenko
0 siblings, 1 reply; 8+ messages in thread
From: Yury Norov @ 2023-07-11 19:20 UTC (permalink / raw)
To: Alexander Potapenko
Cc: catalin.marinas, will, pcc, andreyknvl, linux-kernel,
linux-arm-kernel, eugenis, Andy Shevchenko, Rasmus Villemoes
+ Andy and Rasmus
On Tue, Jul 11, 2023 at 04:42:29PM +0200, Alexander Potapenko wrote:
> struct bitq represents a bit queue with external storage.
>
> Its purpose is to easily pack sub-byte values, which can be used, for
> example, to implement RLE algorithms.
Whatever it is, it's not a queue. The queue implies O(1) for insertion
and deletion, but your 'dequeue' is clearly an O(n) procedure.
I'm not sure if I completely understand the purpose of the series, but
from this description:
enqueueing/dequeueing of sub-byte values
I think, the simplest solution would be a circular queue (ringbuffer)
based on bitmaps:
Init an empty 10-bit bitmap:
Head = 0
v
..... .....
^
Tail = 1
Enqueue 6-bit value 0b101010 at the end:
Head = 0
v
10101 0....
^
Tail = 1
Dequeue 3 bits from the beginning:
Head = 0
v
...01 0....
^
Tail = 1
And so on...
See some other comments inline.
Thanks,
Yury
> Signed-off-by: Alexander Potapenko <glider@google.com>
> ---
> include/linux/bitqueue.h | 144 +++++++++++++++++++++++++++++++++++++++
> 1 file changed, 144 insertions(+)
> create mode 100644 include/linux/bitqueue.h
>
> diff --git a/include/linux/bitqueue.h b/include/linux/bitqueue.h
> new file mode 100644
> index 0000000000000..c4393f703c697
> --- /dev/null
> +++ b/include/linux/bitqueue.h
> @@ -0,0 +1,144 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * A simple bit queue which supports enqueueing/dequeueing of sub-byte values.
> + *
> + * This can be used to pack complex bitfields into byte arrays.
> + */
> +#ifndef _LINUX_BITQUEUE_H
> +#define _LINUX_BITQUEUE_H
> +
> +#include <linux/string.h>
> +#include <linux/types.h>
> +
> +/**
> + * struct bitq - represents a bit queue with external storage.
> + * @data: data buffer used by the queue.
> + * @size: size of @data in bytes.
> + * @bit_pos: current bit position.
> + */
> +struct bitq {
> + u8 *data;
> + int size, bit_pos;
> +};
> +
> +/**
> + * bitq_init - initialize an empty bit queue.
> + * @q: struct bitq to be initialized.
> + * @data: external data buffer to use.
> + * @size: capacity in bytes.
> + *
> + * Return: 0 in the case of success, -1 if either of the pointers is NULL.
ENIVAL?
> + */
> +static inline int bitq_init(struct bitq *q, u8 *data, int size)
> +{
> + if (!q || !data)
> + return -1;
This is a useless check. Erroneous code may (and often does) pass a
broken pointer other than NULL.
And to make it worse, this error handling (instead of run-time
segfault which can be easily caught at debugging) may make users think
that passing NULL is the right thing. Check bit/bitmap and other kernel
functions: they don't check against NULL as a general rule, except for
well-justified cases, like 'free(NULL)'.
> + q->data = data;
> + q->size = size;
> + memset(data, 0, size);
Useless memset?
> + q->bit_pos = 0;
> + return 0;
> +}
> +
> +/**
> + * bitq_init_full - make a bit queue from an initialized byte array.
> + * @q: struct bitq to be initialized.
> + * @data: external data buffer to use.
> + * @size: capacity in bytes.
> + *
> + * Return: 0 in the case of success, -1 if either of the pointers is NULL.
> + */
> +static inline int bitq_init_full(struct bitq *q, u8 *data, int size)
> +{
> + if (!q || !data)
> + return -1;
> + q->data = data;
> + q->size = size;
> + q->bit_pos = q->size * 8;
> + return 0;
> +}
This all should not reside in a header.
> +
> +/**
> + * bitq_enqueue - push up to 8 bits to the end of the queue.
> + * @q: struct bitq.
> + * @value: byte containing the value to be pushed.
> + * @bits: number of bits (1 to 8) to push.
> + *
> + * Return: number of bits pushed, or -1 in the case of an error.
> + */
> +static inline int bitq_enqueue(struct bitq *q, u8 value, int bits)
> +{
> + int byte_pos, left_in_byte, max_pos;
> + u8 hi, lo;
> +
> + if (!q || (bits < 1) || (bits > 8))
> + return -1;
Pushing 0 elements in queue is usually not an error. Implementations
usually return and do nothing. From the malloc() man page:
If size is 0, then malloc() returns a unique pointer value that
can later be successfully passed to free().
> + max_pos = q->size * 8;
> + if ((max_pos - q->bit_pos) < bits)
> + return -1;
ENOMEM? Or probably better to resize the queue.
> +
> + left_in_byte = 8 - (q->bit_pos % 8);
> + byte_pos = q->bit_pos / 8;
> + /* Clamp @value. */
> + value %= (1 << bits);
> + if (left_in_byte >= bits) {
> + /* @value fits into the current byte. */
> + value <<= (left_in_byte - bits);
> + q->data[byte_pos] |= value;
> + } else {
> + /*
> + * @value needs to be split between the current and the
> + * following bytes.
> + */
> + hi = value >> (bits - left_in_byte);
> + q->data[byte_pos] |= hi;
> + byte_pos++;
> + lo = value << (8 - (bits - left_in_byte));
> + q->data[byte_pos] |= lo;
> + }
This piece should be a bitmap_append() function, like:
bitmap_append(addr, 3, 2, 0b11) would append 0b11 to the bitmap at
offset 3. We already have bitmap_{set,get}_value8, so I suggest
to extend the interface for unaligned offsets and lengths up to
BITS_PER_LONG.
> + q->bit_pos += bits;
> + return bits;
> +}
> +
> +/**
> + * bitq_dequeue - pop up to 8 bits from the beginning of the queue.
> + * @q: struct bitq.
> + * @value: u8* to store the popped value (can be NULL).
> + * @bits: number of bits (1 to 8) to pop.
> + *
> + * Return: number of bits popped, or -1 in the case of an error.
> + */
> +
> +#include <linux/printk.h>
> +static inline int bitq_dequeue(struct bitq *q, u8 *value, int bits)
> +{
> + int rem_bits = 8 - bits, i;
> + u8 output;
> +
> + /* Invalid arguments. */
> + if (!q || (bits < 1) || (bits > 8))
> + return -1;
> + /* Not enough space to insert @bits. */
> + if (q->bit_pos < bits)
> + return -1;
> + /* Take the first @bits bits from the first byte. */
> + output = q->data[0];
> + output >>= rem_bits;
> + if (value)
> + *value = output;
> +
> + /*
> + * Shift every byte in the queue to the left by @bits, carrying over to
> + * the previous byte.
> + */
> + for (i = 0; i < q->size - 1; i++) {
> + q->data[i] = (q->data[i] << bits) |
> + (q->data[i + 1] >> rem_bits);
> + }
As I already mentioned, this is O(N), which is wrong for queues. Add a
pointer to the head in the bitq structure to avoid shifting every
byte.
BTW, we've got bitmap_shift_{left,right} for this.
> + q->data[q->size - 1] <<= bits;
> + q->bit_pos -= bits;
> + return bits;
> +}
> +
> +#endif // _LINUX_BITQUEUE_H
> --
> 2.41.0.255.g8b1d071c50-goog
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 8+ messages in thread* Re: [Resend v1 1/5] linux/bitqueue.h: add the bit queue implementation
2023-07-11 19:20 ` Yury Norov
@ 2023-07-12 10:24 ` Alexander Potapenko
0 siblings, 0 replies; 8+ messages in thread
From: Alexander Potapenko @ 2023-07-12 10:24 UTC (permalink / raw)
To: Yury Norov
Cc: catalin.marinas, will, pcc, andreyknvl, linux-kernel,
linux-arm-kernel, eugenis, Andy Shevchenko, Rasmus Villemoes
On Tue, Jul 11, 2023 at 9:20 PM Yury Norov <yury.norov@gmail.com> wrote:
>
> + Andy and Rasmus
>
> On Tue, Jul 11, 2023 at 04:42:29PM +0200, Alexander Potapenko wrote:
> > struct bitq represents a bit queue with external storage.
> >
> > Its purpose is to easily pack sub-byte values, which can be used, for
> > example, to implement RLE algorithms.
>
> Whatever it is, it's not a queue. The queue implies O(1) for insertion
> and deletion, but your 'dequeue' is clearly an O(n) procedure.
Thanks for spotting this!
I have indeed done a poor job implementing the dequeue method.
> I'm not sure if I completely understand the purpose of the series,
To implement tag compression, we need to serialize/deserialize "bit
fields" looking e.g. like this:
int largest_idx : 6;
unsigned char tags[N] : 4*N;
unsigned char sizes[N-1] : 7*(N-1)
to/from a byte array. This actually needs to be done only once, and
enqueue()/dequeue() operations do not interleave, so there is no need
for an actual queue.
I'll try to come up with something simple - maybe reimplement it as a
ring buffer, or even skip the "ring" part, because it is not needed
for my purpose.
(The struct may end up being less generic - in that case I'll move it
from include/linux arch/arm64/mm/)
> but
> from this description:
> enqueueing/dequeueing of sub-byte values
>
> I think, the simplest solution would be a circular queue (ringbuffer)
> based on bitmaps:
> > +/**
> > + * bitq_init - initialize an empty bit queue.
> > + * @q: struct bitq to be initialized.
> > + * @data: external data buffer to use.
> > + * @size: capacity in bytes.
> > + *
> > + * Return: 0 in the case of success, -1 if either of the pointers is NULL.
>
> ENIVAL?
Ack, better use the common error values.
>
> > + */
> > +static inline int bitq_init(struct bitq *q, u8 *data, int size)
> > +{
> > + if (!q || !data)
> > + return -1;
>
> This is a useless check. Erroneous code may (and often does) pass a
> broken pointer other than NULL.
I am actually a fan of defensive programming, but it's a good point
that it does not defend against non-NULL pointers, and NULL is anyway
an unexpected input value.
>
> > + q->data = data;
> > + q->size = size;
> > + memset(data, 0, size);
>
> Useless memset?
An overly cautious one, that lets us fetch values from partially
initialized bytes. This code will be removed anyway.
> > +static inline int bitq_init_full(struct bitq *q, u8 *data, int size)
> > +{
> > + if (!q || !data)
> > + return -1;
> > + q->data = data;
> > + q->size = size;
> > + q->bit_pos = q->size * 8;
> > + return 0;
> > +}
>
> This all should not reside in a header.
There's a handful of examples in include/linux where meaningful code
is written in the headers, but I agree that in this particular case it
is probably not justified by performance reasons.
> > + if (!q || (bits < 1) || (bits > 8))
> > + return -1;
>
> Pushing 0 elements in queue is usually not an error. Implementations
> usually return and do nothing. From the malloc() man page:
Agreed.
> If size is 0, then malloc() returns a unique pointer value that
> can later be successfully passed to free().
>
> > + max_pos = q->size * 8;
> > + if ((max_pos - q->bit_pos) < bits)
> > + return -1;
>
> ENOMEM? Or probably better to resize the queue.
This "queue" relies on the external storage that may not be easily
resizeable (e.g. when we are using a local u64 as a storage).
ENOMEM sounds better (should we stick to this interface).
> > + /*
> > + * @value needs to be split between the current and the
> > + * following bytes.
> > + */
> > + hi = value >> (bits - left_in_byte);
> > + q->data[byte_pos] |= hi;
> > + byte_pos++;
> > + lo = value << (8 - (bits - left_in_byte));
> > + q->data[byte_pos] |= lo;
> > + }
>
> This piece should be a bitmap_append() function, like:
> bitmap_append(addr, 3, 2, 0b11) would append 0b11 to the bitmap at
> offset 3. We already have bitmap_{set,get}_value8, so I suggest
> to extend the interface for unaligned offsets and lengths up to
> BITS_PER_LONG.
Interesting. Yeah, this could be part of bitmap.h instead.
> > + /*
> > + * Shift every byte in the queue to the left by @bits, carrying over to
> > + * the previous byte.
> > + */
> > + for (i = 0; i < q->size - 1; i++) {
> > + q->data[i] = (q->data[i] << bits) |
> > + (q->data[i + 1] >> rem_bits);
> > + }
>
> As I already mentioned, this is O(N), which is wrong for queues. Add a
> pointer to the head in the bitq structure to avoid shifting every
> byte.
>
> BTW, we've got bitmap_shift_{left,right} for this.
Ack.
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 8+ messages in thread
* [Resend v1 2/5] linux/bitqueue.h: add a KUnit test for bitqueue.h
2023-07-11 14:42 [Resend v1 0/5] Implement MTE tag compression for swapped pages Alexander Potapenko
2023-07-11 14:42 ` [Resend v1 1/5] linux/bitqueue.h: add the bit queue implementation Alexander Potapenko
@ 2023-07-11 14:42 ` Alexander Potapenko
2023-07-11 14:42 ` [Resend v1 3/5] arm64: mte: implement CONFIG_ARM64_MTE_COMP Alexander Potapenko
` (2 subsequent siblings)
4 siblings, 0 replies; 8+ messages in thread
From: Alexander Potapenko @ 2023-07-11 14:42 UTC (permalink / raw)
To: glider, catalin.marinas, will, pcc, andreyknvl
Cc: linux-kernel, linux-arm-kernel, eugenis, yury.norov
Add tests checking that struct bitq correctly handles sub-byte values.
Signed-off-by: Alexander Potapenko <glider@google.com>
---
lib/Kconfig.debug | 8 ++
lib/Makefile | 1 +
lib/test_bitqueue.c | 244 ++++++++++++++++++++++++++++++++++++++++++++
3 files changed, 253 insertions(+)
create mode 100644 lib/test_bitqueue.c
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index ce51d4dc6803e..a6598b2c250d5 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -2678,6 +2678,14 @@ config SIPHASH_KUNIT_TEST
This is intended to help people writing architecture-specific
optimized versions. If unsure, say N.
+config BITQUEUE_KUNIT_TEST
+ tristate "Test <linux/bitqueue.h>" if !KUNIT_ALL_TESTS
+ depends on KUNIT
+ default KUNIT_ALL_TESTS
+ help
+ Enable this option to test the kernel's bit queue implementation
+ (<linux/bitqueue.h>).
+
config TEST_UDELAY
tristate "udelay test driver"
help
diff --git a/lib/Makefile b/lib/Makefile
index 876fcdeae34ec..7efb6aba31cf9 100644
--- a/lib/Makefile
+++ b/lib/Makefile
@@ -394,6 +394,7 @@ CFLAGS_fortify_kunit.o += $(DISABLE_STRUCTLEAK_PLUGIN)
obj-$(CONFIG_FORTIFY_KUNIT_TEST) += fortify_kunit.o
obj-$(CONFIG_STRSCPY_KUNIT_TEST) += strscpy_kunit.o
obj-$(CONFIG_SIPHASH_KUNIT_TEST) += siphash_kunit.o
+obj-$(CONFIG_BITQUEUE_KUNIT_TEST) += test_bitqueue.o
obj-$(CONFIG_GENERIC_LIB_DEVMEM_IS_ALLOWED) += devmem_is_allowed.o
diff --git a/lib/test_bitqueue.c b/lib/test_bitqueue.c
new file mode 100644
index 0000000000000..aec04b3a5f068
--- /dev/null
+++ b/lib/test_bitqueue.c
@@ -0,0 +1,244 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Test cases for struct bitq, a simple bit queue.
+ */
+
+#include <kunit/test.h>
+#include <linux/bitqueue.h>
+#include <linux/slab.h>
+
+/* Set up a bit queue containing @size bytes. */
+static void bitq_setup(struct bitq *it, size_t size)
+{
+ u8 *data = kmalloc(size, GFP_KERNEL);
+
+ bitq_init(it, data, size);
+}
+
+/* Tear down the bit queue. */
+static void bitq_teardown(struct bitq *it)
+{
+ kfree(it->data);
+ memset(it, 0, sizeof(*it));
+}
+
+/* Test that nothing can be popped from an empty queue. */
+static void test_empty(struct kunit *test)
+{
+ struct bitq it;
+ u8 val = 0;
+
+ /* Allocate a two-byte queue. */
+ bitq_setup(&it, 2);
+
+ /* Queue is empty. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 8), -1);
+ bitq_teardown(&it);
+}
+
+/* Test that simple byte-granular enqueue/dequeue operations work. */
+static void test_basic_enqueue_dequeue(struct kunit *test)
+{
+ struct bitq it;
+ u8 val = 0;
+
+ /* Allocate a two-byte queue. */
+ bitq_setup(&it, 2);
+ /* Enqueue two 8-bit values. */
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 0xaa, 8), 8);
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 0xbb, 8), 8);
+ /* Cannot enqueue the third byte. */
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 1, 8), -1);
+ /* Dequeue two bytes. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 8), 8);
+ KUNIT_EXPECT_EQ(test, val, 0xaa);
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 8), 8);
+ KUNIT_EXPECT_EQ(test, val, 0xbb);
+
+ /* Queue is empty. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 8), -1);
+ bitq_teardown(&it);
+}
+
+/* Test that values shorter than 8 bits can be enqueued and dequeued. */
+static void test_shorter_than_byte(struct kunit *test)
+{
+ struct bitq it;
+ u8 val = 0;
+
+ /* Allocate a two-byte queue. */
+ bitq_setup(&it, 2);
+ /* Enqueue two 0b101 values. */
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 0b101, 3), 3);
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 0b101, 3), 3);
+ /* The first byte of the queue is now 0b10110100. */
+
+ /* Now dequeue three 2-bit values: 0b10, 0b11, 0b01. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 2), 2);
+ KUNIT_EXPECT_EQ(test, val, 0b10);
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 2), 2);
+ KUNIT_EXPECT_EQ(test, val, 0b11);
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 2), 2);
+ KUNIT_EXPECT_EQ(test, val, 0b01);
+
+ /* Queue is empty. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 1), -1);
+ bitq_teardown(&it);
+}
+
+/* Test that bits are carried over correctly if they do not fit. */
+static void test_carryover(struct kunit *test)
+{
+ struct bitq it;
+ u8 val = 0;
+ int i;
+
+ /* Allocate a three-byte queue. */
+ bitq_setup(&it, 3);
+ /* Enqueue 0b100 seven times. */
+ for (i = 0; i < 7; i++)
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 0b100, 3), 3);
+ /* Now dequeue three 7-bit values: 0b1001001, 0b0010010, 0b0100100. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 7), 7);
+ KUNIT_EXPECT_EQ(test, val, 0b1001001);
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 7), 7);
+ KUNIT_EXPECT_EQ(test, val, 0b0010010);
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 7), 7);
+ KUNIT_EXPECT_EQ(test, val, 0b0100100);
+
+ /* Queue is empty. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 1), -1);
+ bitq_teardown(&it);
+}
+
+/*
+ * Test case extracted from the EA0 tag compression algorithm, where
+ * carried over bits were accidentally written into the previous byte.
+ */
+static void test_carryover_ea0(struct kunit *test)
+{
+ struct bitq it;
+ u8 val = 0;
+
+ /* Allocate a three-byte queue. */
+ bitq_setup(&it, 3);
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 0b100, 3), 3);
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 0b1010, 4), 4);
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 0b0000, 4), 4);
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 0b1010, 4), 4);
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 0b1011, 4), 4);
+
+ /* Now dequeue two byte values: 0b10010100, 0b00010101. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 8), 8);
+ KUNIT_EXPECT_EQ(test, val, 0b10010100);
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 8), 8);
+ KUNIT_EXPECT_EQ(test, val, 0b00010101);
+ /* And the remaining 0b011. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 3), 3);
+ KUNIT_EXPECT_EQ(test, val, 0b011);
+
+ /* Queue is empty. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 1), -1);
+ bitq_teardown(&it);
+}
+
+/* Test that upper bits of the pushed value are discarded. */
+static void test_trim_upper_bits(struct kunit *test)
+{
+ struct bitq it;
+ u8 val = 0;
+
+ /* Allocate a two-byte queue. */
+ bitq_setup(&it, 2);
+ /* Enqueue two values that do not fit into 4 bits. */
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 0xab, 4), 4);
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 0xab, 4), 4);
+ /* The first byte of the queue is now 0xbb. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 8), 8);
+ KUNIT_EXPECT_EQ(test, val, 0xbb);
+
+ /* Queue is empty. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 1), -1);
+ bitq_teardown(&it);
+}
+
+/* Another test for discarding the upper bits. */
+static void test_trim_upper_bits2(struct kunit *test)
+{
+ struct bitq it;
+ u8 val = 0;
+
+ /* Allocate a two-byte queue. */
+ bitq_setup(&it, 2);
+ /* Push seven zero bits. */
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 0, 7), 7);
+ /* Push a single 1 bit, but pass a bigger value to bitq_enqueue(). */
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 0xff, 1), 1);
+ /* The first byte of the queue is now 0x01. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 8), 8);
+ KUNIT_EXPECT_EQ(test, val, 0x01);
+
+ /* Queue is empty. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 1), -1);
+ bitq_teardown(&it);
+}
+
+/* Test that a NULL value can be used as output of bitq_dequeue() */
+static void test_dequeue_to_null(struct kunit *test)
+{
+ struct bitq it;
+
+ /* Allocate a two-byte queue. */
+ bitq_setup(&it, 2);
+ /* Enqueue a byte value. */
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 0xab, 8), 8);
+ /* Dequeue the byte, but discard its value. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, NULL, 8), 8);
+
+ /* Queue is empty. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, NULL, 1), -1);
+ bitq_teardown(&it);
+}
+
+/* Test that bitq_init_full works. */
+static void test_init_full(struct kunit *test)
+{
+ struct bitq it;
+ u8 data[2] = { 0xaa, 0xbb };
+ u8 val = 0;
+
+ /* Initialize a queue with the contents of @data */
+ bitq_init_full(&it, data, 2);
+ /* Cannot enqueue anything else. */
+ KUNIT_EXPECT_EQ(test, bitq_enqueue(&it, 1, 8), -1);
+ /* Dequeue two bytes. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 8), 8);
+ KUNIT_EXPECT_EQ(test, val, 0xaa);
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, &val, 8), 8);
+ KUNIT_EXPECT_EQ(test, val, 0xbb);
+
+ /* Queue is empty. */
+ KUNIT_EXPECT_EQ(test, bitq_dequeue(&it, NULL, 1), -1);
+}
+
+static struct kunit_case bitq_test_cases[] = {
+ KUNIT_CASE(test_empty),
+ KUNIT_CASE(test_basic_enqueue_dequeue),
+ KUNIT_CASE(test_shorter_than_byte),
+ KUNIT_CASE(test_carryover),
+ KUNIT_CASE(test_carryover_ea0),
+ KUNIT_CASE(test_trim_upper_bits),
+ KUNIT_CASE(test_trim_upper_bits2),
+ KUNIT_CASE(test_dequeue_to_null),
+ KUNIT_CASE(test_init_full),
+ {}
+};
+
+static struct kunit_suite bitq_test_suite = {
+ .name = "bitq",
+ .test_cases = bitq_test_cases,
+};
+kunit_test_suites(&bitq_test_suite);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Alexander Potapenko <glider@google.com>");
--
2.41.0.255.g8b1d071c50-goog
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 8+ messages in thread* [Resend v1 3/5] arm64: mte: implement CONFIG_ARM64_MTE_COMP
2023-07-11 14:42 [Resend v1 0/5] Implement MTE tag compression for swapped pages Alexander Potapenko
2023-07-11 14:42 ` [Resend v1 1/5] linux/bitqueue.h: add the bit queue implementation Alexander Potapenko
2023-07-11 14:42 ` [Resend v1 2/5] linux/bitqueue.h: add a KUnit test for bitqueue.h Alexander Potapenko
@ 2023-07-11 14:42 ` Alexander Potapenko
2023-07-11 14:42 ` [Resend v1 4/5] arm64: mte: add a test for MTE tags compression Alexander Potapenko
2023-07-11 14:42 ` [Resend v1 5/5] arm64: mte: add compression support to mteswap.c Alexander Potapenko
4 siblings, 0 replies; 8+ messages in thread
From: Alexander Potapenko @ 2023-07-11 14:42 UTC (permalink / raw)
To: glider, catalin.marinas, will, pcc, andreyknvl
Cc: linux-kernel, linux-arm-kernel, eugenis, yury.norov
The config implements the EA0 algorithm suggested by Evgenii Stepanov
to compress the memory tags for ARM MTE during swapping.
The algorithm is based on RLE and specifically targets 128-byte buffers
of tags corresponding to a single page. In the common case a buffer
can be compressed into 63 bits, making it possible to store it without
additional memory allocation.
Suggested-by: Evgenii Stepanov <eugenis@google.com>
Signed-off-by: Alexander Potapenko <glider@google.com>
---
arch/arm64/Kconfig | 10 +
arch/arm64/include/asm/mtecomp.h | 60 +++++
arch/arm64/mm/Makefile | 1 +
arch/arm64/mm/mtecomp.c | 398 +++++++++++++++++++++++++++++++
4 files changed, 469 insertions(+)
create mode 100644 arch/arm64/include/asm/mtecomp.h
create mode 100644 arch/arm64/mm/mtecomp.c
diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
index 343e1e1cae10a..b25b584a0a9cb 100644
--- a/arch/arm64/Kconfig
+++ b/arch/arm64/Kconfig
@@ -2065,6 +2065,16 @@ config ARM64_EPAN
if the cpu does not implement the feature.
endmenu # "ARMv8.7 architectural features"
+config ARM64_MTE_COMP
+ bool "Tag compression for ARM64 MTE"
+ default y
+ depends on ARM64_MTE
+ help
+ Enable tag compression support for ARM64 MTE.
+
+ 128-byte tag buffers corresponding to 4K pages can be compressed using
+ the EA0 algorithm to save heap memory.
+
config ARM64_SVE
bool "ARM Scalable Vector Extension support"
default y
diff --git a/arch/arm64/include/asm/mtecomp.h b/arch/arm64/include/asm/mtecomp.h
new file mode 100644
index 0000000000000..65a3730cc50d9
--- /dev/null
+++ b/arch/arm64/include/asm/mtecomp.h
@@ -0,0 +1,60 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifndef __ASM_MTECOMP_H
+#define __ASM_MTECOMP_H
+
+#include <linux/types.h>
+
+/*
+ * ea0_compress() - compress the given tag array.
+ * @tags: 128-byte array to read the tags from.
+ *
+ * Compresses the tags and returns a 64-bit opaque handle pointing to the
+ * tag storage. May allocate memory, which is freed by @ea0_release_handle().
+ */
+u64 ea0_compress(u8 *tags);
+
+/*
+ * ea0_decompress() - decompress the tag array addressed by the handle.
+ * @handle: handle returned by @ea0_decompress()
+ * @tags: 128-byte array to write the tags to.
+ *
+ * Reads the compressed data and writes it into the user-supplied tag array.
+ * Returns true on success, false on error.
+ */
+bool ea0_decompress(u64 handle, u8 *tags);
+
+/*
+ * ea0_release_handle() - release the handle returned by ea0_compress().
+ * @handle: handle returned by ea0_compress().
+ */
+void ea0_release_handle(u64 handle);
+
+/* Functions below are exported for testing purposes. */
+
+/*
+ * ea0_storage_size() - calculate the memory occupied by compressed tags.
+ * @handle: storage handle returned by ea0_compress.
+ */
+int ea0_storage_size(u64 handle);
+
+/*
+ * ea0_tags_to_ranges() - break @tags into arrays of tag ranges.
+ * @tags: 128-byte array containing 256 MTE tags.
+ * @out_tags: u8 array to store the tag of every range.
+ * @out_sizes: u16 array to store the size of every range.
+ * @out_len: length of @out_tags and @out_sizes (output parameter, initially
+ * equal to lengths of out_tags[] and out_sizes[]).
+ */
+void ea0_tags_to_ranges(u8 *tags, u8 *out_tags, short *out_sizes, int *out_len);
+
+/*
+ * ea0_ranges_to_tags() - fill @tags using given tag ranges.
+ * @r_tags: u8[256] containing the tag of every range.
+ * @r_sizes: u16[256] containing the size of every range.
+ * @r_len: length of @r_tags and @r_sizes.
+ * @tags: 128-byte array to write the tags to.
+ */
+void ea0_ranges_to_tags(u8 *r_tags, short *r_sizes, int r_len, u8 *tags);
+
+#endif // __ASM_MTECOMP_H
diff --git a/arch/arm64/mm/Makefile b/arch/arm64/mm/Makefile
index dbd1bc95967d0..46778f6dd83c2 100644
--- a/arch/arm64/mm/Makefile
+++ b/arch/arm64/mm/Makefile
@@ -10,6 +10,7 @@ obj-$(CONFIG_TRANS_TABLE) += trans_pgd.o
obj-$(CONFIG_TRANS_TABLE) += trans_pgd-asm.o
obj-$(CONFIG_DEBUG_VIRTUAL) += physaddr.o
obj-$(CONFIG_ARM64_MTE) += mteswap.o
+obj-$(CONFIG_ARM64_MTE_COMP) += mtecomp.o
KASAN_SANITIZE_physaddr.o += n
obj-$(CONFIG_KASAN) += kasan_init.o
diff --git a/arch/arm64/mm/mtecomp.c b/arch/arm64/mm/mtecomp.c
new file mode 100644
index 0000000000000..01f7d22665b49
--- /dev/null
+++ b/arch/arm64/mm/mtecomp.c
@@ -0,0 +1,398 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+/*
+ * MTE tag compression algorithm.
+ * Proposed by Evgenii Stepanov <eugenis@google.com>
+ */
+
+/*
+ * EA0 stands for "Evgenii's Algorithm 0", as the initial proposal contained two
+ * compression algorithms.
+ *
+ * The algorithm attempts to compress a 128-byte (MTE_GRANULES_PER_PAGE / 2)
+ * array of tags into a smaller byte sequence that can be stored in a
+ * 16-, 32-, or 64-byte buffer. A special case is storing the tags inline in
+ * an 8-byte pointer.
+ *
+ * We encapsulate tag storage memory management in this module, because it is
+ * tightly coupled with the pointer representation.
+ * ea0_compress(*tags) takes a 128-byte buffer and returns an opaque value
+ * that can be stored in Xarray
+ * ea_decompress(*ptr, *tags) takes the opaque value and loads the tags into
+ * the provided 128-byte buffer.
+ *
+ *
+ *
+ * The compression algorithm works as follows.
+ *
+ * 1. The input array of 128 bytes is transformed into tag ranges (two arrays:
+ * @r_tags containing tag values and @r_sizes containing range lengths) by
+ * ea0_tags_to_ranges(). Note that @r_sizes sums up to 256.
+ *
+ * 2. Depending on the number N of ranges, the following storage class is picked:
+ * N <= 6: 8 bytes (inline case, no allocation required);
+ * 6 < N <= 11: 16 bytes
+ * 11 < N <= 23: 32 bytes
+ * 23 < N <= 46: 64 bytes
+ * 46 < N: 128 bytes (no compression will be performed)
+ *
+ * 3. The number of the largest element of @r_sizes is stored in @largest_idx.
+ * The element itself is thrown away from @r_sizes, because it can be
+ * reconstructed from the sum of the remaining elements. Note that now none
+ * of the remaining @r_sizes elements is greater than 127.
+ *
+ * 4. For the inline case, the following values are stored in the 8-byte handle:
+ * largest_idx : i4
+ * r_tags[0..5] : i4 x 6
+ * r_sizes[0..4] : i7 x 5
+ * (if N is less than 6, @r_tags and @r_sizes are padded up with zero values)
+ *
+ * Because @largest_idx is <= 5, bit 63 of the handle is always 0 (so it can
+ * be stored in the Xarray), and bits 62..60 cannot all be 1, so it can be
+ * distinguished from a kernel pointer.
+ *
+ * 5. For the out-of-line case, the storage is allocated from one of the
+ * "mte-tags-{16,32,64,128}" kmem caches. The resulting pointer is aligned
+ * on 8 bytes, so its bits 2..0 can be used to store the size class:
+ * - 0 for 128 bytes
+ * - 1 for 16
+ * - 2 for 32
+ * - 4 for 64.
+ * Bit 63 of the pointer is zeroed out, so that it can be stored in Xarray.
+ *
+ * 6. The data layout in the allocated storage is as follows:
+ * largest_idx : i6
+ * r_tags[0..N] : i4 x N
+ * r_sizes[0..N-1] : i7 x (N-1)
+ *
+ *
+ *
+ * The decompression algorithm performs the steps below.
+ *
+ * 1. Decide if data is stored inline (bits 62..60 of the handle != 0b111) or
+ * out-of line.
+ *
+ * 2. For the inline case, treat the handle itself as the input buffer.
+ *
+ * 3. For the out-of-line case, look at bits 2..0 of the handle to understand
+ * the input buffer length. To obtain the pointer to the input buffer, unset
+ * bits 2..0 of the handle and set bit 63.
+ *
+ * 4. If the input buffer is 128 byte long, copy its contents to the output
+ * buffer.
+ *
+ * 5. Otherwise, read @largest_idx, @r_tags and @r_sizes from the input buffer.
+ * Calculate the removed largest element of @r_sizes:
+ * largest = 256 - sum(r_sizes)
+ * and insert it into @r_sizes at position @largest_idx.
+ *
+ * 6. While @r_sizes[i] > 0, add a 4-bit value @r_tags[i] to the output buffer
+ * @r_sizes[i] times.
+ */
+
+#include <linux/bitqueue.h>
+#include <linux/gfp.h>
+#include <linux/module.h>
+#include <asm/mtecomp.h>
+#include <linux/slab.h>
+#include <linux/swab.h>
+#include <linux/string.h>
+#include <linux/types.h>
+
+/* The handle must fit into an Xarray value. */
+#define HANDLE_MASK ~(BIT_ULL(63))
+
+/* Out-of-line handles have 0b111 in bits 62..60. */
+#define NOINLINE_MASK (BIT_ULL(62) | BIT_ULL(61) | BIT_ULL(60))
+
+/* Cache index is stored in the lowest pointer bits. */
+#define CACHE_ID_MASK (BIT_ULL(2) | BIT_ULL(1) | BIT_ULL(0))
+
+/* Four separate caches to store out-of-line data. */
+#define NUM_CACHES 4
+static struct kmem_cache *mtecomp_caches[NUM_CACHES];
+
+/* Translate allocation size into mtecomp_caches[] index. */
+static int ea0_size_to_cache_id(int len)
+{
+ switch (len) {
+ case 16:
+ return 1;
+ case 32:
+ return 2;
+ case 64:
+ return 3;
+ default:
+ return 0;
+ }
+}
+
+/* Translate mtecomp_caches[] index into allocation size. */
+static int ea0_cache_id_to_size(int id)
+{
+ switch (id) {
+ case 1:
+ return 16;
+ case 2:
+ return 32;
+ case 3:
+ return 64;
+ default:
+ return 128;
+ }
+}
+
+/* Transform tags into tag ranges. */
+void ea0_tags_to_ranges(u8 *tags, u8 *out_tags, short *out_sizes, int *out_len)
+{
+ u8 prev_tag = 0xff;
+ int cur_idx = -1;
+ u8 cur_tag;
+ int i;
+
+ memset(out_tags, 0, *out_len * sizeof(*out_tags));
+ memset(out_sizes, 0, *out_len * sizeof(*out_sizes));
+ for (i = 0; i < MTE_GRANULES_PER_PAGE; i++) {
+ cur_tag = tags[i / 2];
+ if (i % 2)
+ cur_tag = cur_tag % 16;
+ else
+ cur_tag = cur_tag / 16;
+ if (cur_tag == prev_tag) {
+ out_sizes[cur_idx]++;
+ } else {
+ cur_idx++;
+ prev_tag = cur_tag;
+ out_tags[cur_idx] = prev_tag;
+ out_sizes[cur_idx] = 1;
+ }
+ }
+ *out_len = cur_idx + 1;
+}
+
+/* Transform tag ranges back into tags. */
+void ea0_ranges_to_tags(u8 *r_tags, short *r_sizes, int r_len, u8 *tags)
+{
+ struct bitq iter;
+ int i, j;
+
+ bitq_init(&iter, tags, 128);
+ for (i = 0; i < r_len; i++) {
+ for (j = 0; j < r_sizes[i]; j++)
+ bitq_enqueue(&iter, r_tags[i], 4);
+ }
+}
+
+/* Translate @num_ranges into the allocation size needed to hold them. */
+static int ea0_alloc_size(int num_ranges)
+{
+ if (num_ranges <= 6)
+ return 8;
+ if (num_ranges <= 11)
+ return 16;
+ if (num_ranges <= 23)
+ return 32;
+ if (num_ranges <= 46)
+ return 64;
+ return 128;
+}
+
+/* Translate allocation size into maximum number of ranges that it can hold. */
+static int ea0_size_to_ranges(int size)
+{
+ switch (size) {
+ case 8:
+ return 6;
+ case 16:
+ return 11;
+ case 32:
+ return 23;
+ case 64:
+ return 46;
+ default:
+ return 0;
+ }
+}
+
+/* Is the data stored inline in the handle itself? */
+static bool ea0_is_inline(u64 handle)
+{
+ return (handle & NOINLINE_MASK) != NOINLINE_MASK;
+}
+
+/* Get the size of the buffer backing @handle. */
+int ea0_storage_size(u64 handle)
+{
+ if (ea0_is_inline(handle))
+ return 8;
+ return ea0_cache_id_to_size(handle & CACHE_ID_MASK);
+}
+EXPORT_SYMBOL(ea0_storage_size);
+
+/* Compress ranges into the buffer of the given length. */
+void ea0_compress_to_buf(int len, u8 *tags, short *sizes, u8 *buf, int buflen)
+{
+ int largest_idx = -1, i;
+ short largest = 0;
+ struct bitq iter;
+
+ bitq_init(&iter, buf, buflen);
+ for (i = 0; i < len; i++) {
+ if (i == len)
+ break;
+ if (sizes[i] > largest) {
+ largest = sizes[i];
+ largest_idx = i;
+ }
+ }
+ if (len <= 6)
+ /* Inline case, @buflen <= 8. */
+ bitq_enqueue(&iter, largest_idx, 4);
+ else
+ bitq_enqueue(&iter, largest_idx, 6);
+ for (i = 0; i < len; i++)
+ bitq_enqueue(&iter, tags[i], 4);
+ for (i = len; i < ea0_size_to_ranges(buflen); i++)
+ bitq_enqueue(&iter, 0, 4);
+ for (i = 0; i < len; i++) {
+ if (i == largest_idx)
+ continue;
+ bitq_enqueue(&iter, sizes[i], 7);
+ }
+}
+
+/* Compress the data inline. */
+static u64 ea0_compress_inline(int len, u8 *tags, short *sizes)
+{
+ u64 result;
+
+ ea0_compress_to_buf(len, tags, sizes, (u8 *)&result, sizeof(result));
+ result = be64_to_cpu(result);
+ return result;
+}
+
+/* Compress @tags and return a handle. */
+u64 ea0_compress(u8 *tags)
+{
+ int alloc_size, cache_id;
+ struct kmem_cache *cache;
+ short r_sizes[256];
+ u8 r_tags[256];
+ int r_len = ARRAY_SIZE(r_tags);
+ u8 *storage;
+
+ ea0_tags_to_ranges(tags, r_tags, r_sizes, &r_len);
+ alloc_size = ea0_alloc_size(r_len);
+ if (alloc_size == 8)
+ return ea0_compress_inline(r_len, r_tags, r_sizes);
+ cache_id = ea0_size_to_cache_id(alloc_size);
+ cache = mtecomp_caches[cache_id];
+ storage = kmem_cache_alloc(cache, GFP_KERNEL);
+ if (alloc_size < 128) {
+ ea0_compress_to_buf(r_len, r_tags, r_sizes, storage,
+ alloc_size);
+ return ((u64)storage | cache_id) & HANDLE_MASK;
+ }
+ memcpy(storage, tags, alloc_size);
+ return (u64)storage & HANDLE_MASK;
+}
+
+/* Decompress the contents of the given buffer into @tags. */
+static bool ea0_decompress_from_buf(u8 *buf, int buflen, u8 *tags)
+{
+ int bits, largest_idx, i, r_len = ea0_size_to_ranges(buflen);
+ short r_sizes[46], sum = 0;
+ u8 r_tags[46];
+ struct bitq iter;
+ u8 val;
+
+ bitq_init_full(&iter, buf, buflen);
+ bits = bitq_dequeue(&iter, &val, (buflen == 8) ? 4 : 6);
+ if (bits == -1)
+ return false;
+ largest_idx = val;
+ for (i = 0; i < r_len; i++) {
+ bits = bitq_dequeue(&iter, &val, 4);
+ if (bits == -1)
+ return false;
+ r_tags[i] = val;
+ }
+ for (i = 0; i < r_len; i++) {
+ if (i == largest_idx)
+ continue;
+ bits = bitq_dequeue(&iter, &val, 7);
+ if (bits == -1)
+ return false;
+ if (!val) {
+ r_len = i;
+ break;
+ }
+ r_sizes[i] = val;
+ sum += val;
+ }
+ if (sum >= 256)
+ return false;
+ r_sizes[largest_idx] = 256 - sum;
+ ea0_ranges_to_tags(r_tags, r_sizes, r_len, tags);
+ return true;
+}
+
+/* Get pointer to the out-of-line storage from a handle. */
+static void *ea0_storage(u64 handle)
+{
+ if (ea0_is_inline(handle))
+ return NULL;
+ return (void *)((handle & (~CACHE_ID_MASK)) | BIT_ULL(63));
+}
+
+/* Decompress tags from the buffer referenced by @handle. */
+bool ea0_decompress(u64 handle, u8 *tags)
+{
+ u8 *storage = ea0_storage(handle);
+ int size = ea0_storage_size(handle);
+
+ if (size == 128) {
+ memcpy(tags, storage, size);
+ return true;
+ }
+ if (size == 8) {
+ handle = cpu_to_be64(handle);
+ return ea0_decompress_from_buf((u8 *)&handle, sizeof(handle),
+ tags);
+ }
+ return ea0_decompress_from_buf(storage, size, tags);
+}
+EXPORT_SYMBOL(ea0_decompress);
+
+/* Release the memory referenced by @handle. */
+void ea0_release_handle(u64 handle)
+{
+ void *storage = ea0_storage(handle);
+ int size = ea0_storage_size(handle);
+ struct kmem_cache *c;
+
+ if (!handle || !storage)
+ return;
+
+ c = mtecomp_caches[ea0_size_to_cache_id(size)];
+ kmem_cache_free(c, storage);
+}
+EXPORT_SYMBOL(ea0_release_handle);
+
+/* Set up mtecomp_caches[]. */
+static int mtecomp_init(void)
+{
+ char name[16];
+ int size;
+ int i;
+
+ for (i = 0; i < NUM_CACHES; i++) {
+ size = ea0_cache_id_to_size(i);
+ snprintf(name, ARRAY_SIZE(name), "mte-tags-%d", size);
+ mtecomp_caches[i] =
+ kmem_cache_create(name, size, size, 0, NULL);
+ }
+ return 0;
+}
+
+module_init(mtecomp_init);
--
2.41.0.255.g8b1d071c50-goog
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 8+ messages in thread* [Resend v1 4/5] arm64: mte: add a test for MTE tags compression
2023-07-11 14:42 [Resend v1 0/5] Implement MTE tag compression for swapped pages Alexander Potapenko
` (2 preceding siblings ...)
2023-07-11 14:42 ` [Resend v1 3/5] arm64: mte: implement CONFIG_ARM64_MTE_COMP Alexander Potapenko
@ 2023-07-11 14:42 ` Alexander Potapenko
2023-07-11 14:42 ` [Resend v1 5/5] arm64: mte: add compression support to mteswap.c Alexander Potapenko
4 siblings, 0 replies; 8+ messages in thread
From: Alexander Potapenko @ 2023-07-11 14:42 UTC (permalink / raw)
To: glider, catalin.marinas, will, pcc, andreyknvl
Cc: linux-kernel, linux-arm-kernel, eugenis, yury.norov
Ensure that tag sequences containing alternating values are compressed
to buffers of expected size and correctly decompressed afterwards.
Signed-off-by: Alexander Potapenko <glider@google.com>
---
arch/arm64/Kconfig | 10 ++
arch/arm64/mm/Makefile | 1 +
arch/arm64/mm/test_mtecomp.c | 175 +++++++++++++++++++++++++++++++++++
3 files changed, 186 insertions(+)
create mode 100644 arch/arm64/mm/test_mtecomp.c
diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
index b25b584a0a9cb..31fc50208b383 100644
--- a/arch/arm64/Kconfig
+++ b/arch/arm64/Kconfig
@@ -2075,6 +2075,16 @@ config ARM64_MTE_COMP
128-byte tag buffers corresponding to 4K pages can be compressed using
the EA0 algorithm to save heap memory.
+config ARM64_MTE_COMP_KUNIT_TEST
+ tristate "Test tag compression for ARM64 MTE" if !KUNIT_ALL_TESTS
+ default KUNIT_ALL_TESTS
+ depends on KUNIT && ARM64_MTE_COMP
+ help
+ Test EA0 compression algorithm enabled by CONFIG_ARM64_MTE_COMP.
+
+ Ensure that tag sequences containing alternating values are compressed
+ to buffers of expected size and correctly decompressed afterwards.
+
config ARM64_SVE
bool "ARM Scalable Vector Extension support"
default y
diff --git a/arch/arm64/mm/Makefile b/arch/arm64/mm/Makefile
index 46778f6dd83c2..170dc62b010b9 100644
--- a/arch/arm64/mm/Makefile
+++ b/arch/arm64/mm/Makefile
@@ -11,6 +11,7 @@ obj-$(CONFIG_TRANS_TABLE) += trans_pgd-asm.o
obj-$(CONFIG_DEBUG_VIRTUAL) += physaddr.o
obj-$(CONFIG_ARM64_MTE) += mteswap.o
obj-$(CONFIG_ARM64_MTE_COMP) += mtecomp.o
+obj-$(CONFIG_ARM64_MTE_COMP_KUNIT_TEST) += test_mtecomp.o
KASAN_SANITIZE_physaddr.o += n
obj-$(CONFIG_KASAN) += kasan_init.o
diff --git a/arch/arm64/mm/test_mtecomp.c b/arch/arm64/mm/test_mtecomp.c
new file mode 100644
index 0000000000000..67bef6f28dac4
--- /dev/null
+++ b/arch/arm64/mm/test_mtecomp.c
@@ -0,0 +1,175 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Test cases for EA0, the compression algorithm for MTE tags.
+ */
+
+#include <asm/mtecomp.h>
+#include <kunit/test.h>
+#include <linux/slab.h>
+#include <linux/types.h>
+
+/*
+ * Test that ea0_tags_to_ranges() produces a single range for a zero-filled tag
+ * buffer.
+ */
+static void test_tags_to_ranges_zero(struct kunit *test)
+{
+ u8 tags[128], dtags[128];
+ short r_sizes[256];
+ int r_len = 256;
+ u8 r_tags[256];
+
+ memset(tags, 0, 128);
+ ea0_tags_to_ranges(tags, r_tags, r_sizes, &r_len);
+ KUNIT_EXPECT_EQ(test, r_len, 1);
+ KUNIT_EXPECT_EQ(test, r_tags[0], 0);
+ KUNIT_EXPECT_EQ(test, r_sizes[0], 256);
+ ea0_ranges_to_tags(r_tags, r_sizes, r_len, dtags);
+ KUNIT_EXPECT_EQ(test, memcmp(tags, dtags, 128), 0);
+}
+
+/*
+ * Test that a small number of different tags is correctly transformed into
+ * ranges.
+ */
+static void test_tags_to_ranges_simple(struct kunit *test)
+{
+ u8 tags[128], dtags[128];
+ const u8 ex_tags[] = { 0xa, 0x0, 0xa, 0xb, 0x0 };
+ const short ex_sizes[] = { 1, 2, 2, 1, 250 };
+ short r_sizes[256];
+ int r_len = 256;
+ u8 r_tags[256];
+
+ memset(tags, 0, 128);
+ tags[0] = 0xa0;
+ tags[1] = 0x0a;
+ tags[2] = 0xab;
+ ea0_tags_to_ranges(tags, r_tags, r_sizes, &r_len);
+ KUNIT_EXPECT_EQ(test, r_len, 5);
+ KUNIT_EXPECT_EQ(test, memcmp(r_tags, ex_tags, sizeof(ex_tags)), 0);
+ KUNIT_EXPECT_EQ(test, memcmp(r_sizes, ex_sizes, sizeof(ex_sizes)), 0);
+ ea0_ranges_to_tags(r_tags, r_sizes, r_len, dtags);
+ KUNIT_EXPECT_EQ(test, memcmp(tags, dtags, 128), 0);
+}
+
+/* Test that repeated 0xa0 byte produces 256 ranges of length 1. */
+static void test_tags_to_ranges_repeated(struct kunit *test)
+{
+ u8 tags[128], dtags[128];
+ short r_sizes[256];
+ int r_len = 256;
+ u8 r_tags[256];
+
+ memset(tags, 0xa0, 128);
+ ea0_tags_to_ranges(tags, r_tags, r_sizes, &r_len);
+ KUNIT_EXPECT_EQ(test, r_len, 256);
+ ea0_ranges_to_tags(r_tags, r_sizes, r_len, dtags);
+ KUNIT_EXPECT_EQ(test, memcmp(tags, dtags, 128), 0);
+}
+
+/* Test that a zero-filled array is compressed into inline storage. */
+static void test_compress_zero(struct kunit *test)
+{
+ u8 tags[128], dtags[128];
+ u64 handle;
+
+ memset(tags, 0, 128);
+ handle = ea0_compress(tags);
+ KUNIT_EXPECT_EQ(test, handle & BIT_ULL(63), 0);
+ /* Tags are stored inline. */
+ KUNIT_EXPECT_EQ(test, ea0_storage_size(handle), 8);
+ KUNIT_EXPECT_TRUE(test, ea0_decompress(handle, dtags));
+ KUNIT_EXPECT_EQ(test, memcmp(tags, dtags, 128), 0);
+}
+
+/*
+ * Test that a very small number of tag ranges ends up compressed into 8 bytes.
+ */
+static void test_compress_simple(struct kunit *test)
+{
+ u8 tags[128], dtags[128];
+ u64 handle;
+
+ memset(tags, 0, 128);
+ tags[0] = 0xa0;
+ tags[1] = 0x0a;
+ tags[2] = 0xab;
+
+ handle = ea0_compress(tags);
+ KUNIT_EXPECT_EQ(test, handle & BIT_ULL(63), 0);
+ /* Tags are stored inline. */
+ KUNIT_EXPECT_EQ(test, ea0_storage_size(handle), 8);
+ KUNIT_EXPECT_TRUE(test, ea0_decompress(handle, dtags));
+ KUNIT_EXPECT_EQ(test, memcmp(tags, dtags, 128), 0);
+}
+
+/*
+ * Generate a buffer that will contain @nranges of tag ranges, test that it
+ * compresses into @exp_size bytes and decompresses into the original tag
+ * sequence.
+ */
+static void compress_range_helper(struct kunit *test, int nranges, int exp_size)
+{
+ u8 tags[128], dtags[128];
+ u64 handle;
+ int i;
+
+ memset(tags, 0, 128);
+
+ if (nranges > 1) {
+ nranges--;
+ for (i = 0; i < nranges / 2; i++)
+ tags[i] = 0xab;
+ if (nranges % 2)
+ tags[nranges / 2] = 0xa0;
+ }
+
+ handle = ea0_compress(tags);
+ KUNIT_EXPECT_EQ(test, handle & BIT_ULL(63), 0);
+ KUNIT_EXPECT_EQ(test, ea0_storage_size(handle), exp_size);
+ KUNIT_EXPECT_TRUE(test, ea0_decompress(handle, dtags));
+ KUNIT_EXPECT_EQ(test, memcmp(tags, dtags, 128), 0);
+}
+
+/*
+ * Test that every number of tag ranges is correctly compressed and
+ * decompressed.
+ */
+static void test_compress_ranges(struct kunit *test)
+{
+ int i, exp_size;
+
+ for (i = 1; i <= 256; i++) {
+ if (i < 7)
+ exp_size = 8;
+ else if (i < 12)
+ exp_size = 16;
+ else if (i < 24)
+ exp_size = 32;
+ else if (i < 47)
+ exp_size = 64;
+ else
+ exp_size = 128;
+ compress_range_helper(test, i, exp_size);
+ }
+}
+
+static struct kunit_case mtecomp_test_cases[] = {
+ KUNIT_CASE(test_tags_to_ranges_zero),
+ KUNIT_CASE(test_tags_to_ranges_simple),
+ KUNIT_CASE(test_tags_to_ranges_repeated),
+ KUNIT_CASE(test_compress_zero),
+ KUNIT_CASE(test_compress_simple),
+ KUNIT_CASE(test_compress_ranges),
+ {}
+};
+
+static struct kunit_suite mtecomp_test_suite = {
+ .name = "mtecomp",
+ .test_cases = mtecomp_test_cases,
+};
+kunit_test_suites(&mtecomp_test_suite);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Alexander Potapenko <glider@google.com>");
--
2.41.0.255.g8b1d071c50-goog
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 8+ messages in thread* [Resend v1 5/5] arm64: mte: add compression support to mteswap.c
2023-07-11 14:42 [Resend v1 0/5] Implement MTE tag compression for swapped pages Alexander Potapenko
` (3 preceding siblings ...)
2023-07-11 14:42 ` [Resend v1 4/5] arm64: mte: add a test for MTE tags compression Alexander Potapenko
@ 2023-07-11 14:42 ` Alexander Potapenko
4 siblings, 0 replies; 8+ messages in thread
From: Alexander Potapenko @ 2023-07-11 14:42 UTC (permalink / raw)
To: glider, catalin.marinas, will, pcc, andreyknvl
Cc: linux-kernel, linux-arm-kernel, eugenis, yury.norov
Define the internal mteswap.h interface:
- _mte_alloc_and_save_tags()
- _mte_free_saved_tags()
- _mte_restore_tags()
, that encapsulates saving tags for a struct page (together with memory
allocation), restoring tags, and deleting the storage allocated for them.
These functions accept opaque pointers, which may point to 128-byte
tag buffers, as well as smaller buffers containing compressed tags, or
have compressed tags stored directly in them.
The existing code from mteswap.c operating with uncompressed tags is split
away into mteswap_nocomp.c, and the newly introduced mteswap_comp.c
provides compression with the EA0 algorithm. The latter implementation
is picked if CONFIG_ARM64_MTE_COMP=y.
Soon after booting Android, tag compression saves ~2.5x memory previously
spent by mteswap.c on tag allocations. With the growing uptime, the
savings reach 20x and even more.
Signed-off-by: Alexander Potapenko <glider@google.com>
---
arch/arm64/mm/Makefile | 5 ++++
arch/arm64/mm/mteswap.c | 19 ++++++-------
arch/arm64/mm/mteswap.h | 12 ++++++++
arch/arm64/mm/mteswap_comp.c | 50 ++++++++++++++++++++++++++++++++++
arch/arm64/mm/mteswap_nocomp.c | 37 +++++++++++++++++++++++++
5 files changed, 112 insertions(+), 11 deletions(-)
create mode 100644 arch/arm64/mm/mteswap.h
create mode 100644 arch/arm64/mm/mteswap_comp.c
create mode 100644 arch/arm64/mm/mteswap_nocomp.c
diff --git a/arch/arm64/mm/Makefile b/arch/arm64/mm/Makefile
index 170dc62b010b9..46a798e2b67cb 100644
--- a/arch/arm64/mm/Makefile
+++ b/arch/arm64/mm/Makefile
@@ -11,6 +11,11 @@ obj-$(CONFIG_TRANS_TABLE) += trans_pgd-asm.o
obj-$(CONFIG_DEBUG_VIRTUAL) += physaddr.o
obj-$(CONFIG_ARM64_MTE) += mteswap.o
obj-$(CONFIG_ARM64_MTE_COMP) += mtecomp.o
+ifdef CONFIG_ARM64_MTE_COMP
+obj-$(CONFIG_ARM64_MTE) += mteswap_comp.o
+else
+obj-$(CONFIG_ARM64_MTE) += mteswap_nocomp.o
+endif
obj-$(CONFIG_ARM64_MTE_COMP_KUNIT_TEST) += test_mtecomp.o
KASAN_SANITIZE_physaddr.o += n
diff --git a/arch/arm64/mm/mteswap.c b/arch/arm64/mm/mteswap.c
index cd508ba80ab1b..e072c1209fbb6 100644
--- a/arch/arm64/mm/mteswap.c
+++ b/arch/arm64/mm/mteswap.c
@@ -7,6 +7,8 @@
#include <linux/swapops.h>
#include <asm/mte.h>
+#include "mteswap.h"
+
static DEFINE_XARRAY(mte_pages);
void *mte_allocate_tag_storage(void)
@@ -27,20 +29,18 @@ int mte_save_tags(struct page *page)
if (!page_mte_tagged(page))
return 0;
- tag_storage = mte_allocate_tag_storage();
+ tag_storage = _mte_alloc_and_save_tags(page);
if (!tag_storage)
return -ENOMEM;
- mte_save_page_tags(page_address(page), tag_storage);
-
/* page_private contains the swap entry.val set in do_swap_page */
ret = xa_store(&mte_pages, page_private(page), tag_storage, GFP_KERNEL);
if (WARN(xa_is_err(ret), "Failed to store MTE tags")) {
- mte_free_tag_storage(tag_storage);
+ _mte_free_saved_tags(tag_storage);
return xa_err(ret);
} else if (ret) {
/* Entry is being replaced, free the old entry */
- mte_free_tag_storage(ret);
+ _mte_free_saved_tags(ret);
}
return 0;
@@ -53,10 +53,7 @@ void mte_restore_tags(swp_entry_t entry, struct page *page)
if (!tags)
return;
- if (try_page_mte_tagging(page)) {
- mte_restore_page_tags(page_address(page), tags);
- set_page_mte_tagged(page);
- }
+ _mte_restore_tags(tags, page);
}
void mte_invalidate_tags(int type, pgoff_t offset)
@@ -64,7 +61,7 @@ void mte_invalidate_tags(int type, pgoff_t offset)
swp_entry_t entry = swp_entry(type, offset);
void *tags = xa_erase(&mte_pages, entry.val);
- mte_free_tag_storage(tags);
+ _mte_free_saved_tags(tags);
}
void mte_invalidate_tags_area(int type)
@@ -78,7 +75,7 @@ void mte_invalidate_tags_area(int type)
xa_lock(&mte_pages);
xas_for_each(&xa_state, tags, last_entry.val - 1) {
__xa_erase(&mte_pages, xa_state.xa_index);
- mte_free_tag_storage(tags);
+ _mte_free_saved_tags(tags);
}
xa_unlock(&mte_pages);
}
diff --git a/arch/arm64/mm/mteswap.h b/arch/arm64/mm/mteswap.h
new file mode 100644
index 0000000000000..bf25f2b3e75a4
--- /dev/null
+++ b/arch/arm64/mm/mteswap.h
@@ -0,0 +1,12 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifndef ARCH_ARM64_MM_MTESWAP_H_
+#define ARCH_ARM64_MM_MTESWAP_H_
+
+#include <linux/mm_types.h>
+
+void *_mte_alloc_and_save_tags(struct page *page);
+void _mte_free_saved_tags(void *tags);
+void _mte_restore_tags(void *tags, struct page *page);
+
+#endif // ARCH_ARM64_MM_MTESWAP_H_
diff --git a/arch/arm64/mm/mteswap_comp.c b/arch/arm64/mm/mteswap_comp.c
new file mode 100644
index 0000000000000..bc2591f8d4f35
--- /dev/null
+++ b/arch/arm64/mm/mteswap_comp.c
@@ -0,0 +1,50 @@
+// SPDX-License-Identifier: GPL-2.0
+
+/* MTE tag storage management with EA0 compression. */
+
+#include <asm/mte.h>
+#include <asm/mtecomp.h>
+#include <linux/pagemap.h>
+#include <linux/slab.h>
+#include <linux/swap.h>
+#include <linux/swapops.h>
+#include <linux/xarray.h>
+
+#include "mteswap.h"
+
+void *_mte_alloc_and_save_tags(struct page *page)
+{
+ u8 tags[128];
+ u64 handle;
+
+ mte_save_page_tags(page_address(page), tags);
+ handle = ea0_compress(tags);
+ return xa_mk_value(handle);
+}
+
+void _mte_free_saved_tags(void *storage)
+{
+ unsigned long handle = xa_to_value(storage);
+ int size;
+
+ if (!handle)
+ return;
+ size = ea0_storage_size(handle);
+ ea0_release_handle(handle);
+}
+
+void _mte_restore_tags(void *tags, struct page *page)
+{
+ u64 handle = xa_to_value(tags);
+ u8 tags_decomp[128];
+
+ if (!handle)
+ return;
+
+ if (try_page_mte_tagging(page)) {
+ if (!ea0_decompress(handle, tags_decomp))
+ return;
+ mte_restore_page_tags(page_address(page), tags_decomp);
+ set_page_mte_tagged(page);
+ }
+}
diff --git a/arch/arm64/mm/mteswap_nocomp.c b/arch/arm64/mm/mteswap_nocomp.c
new file mode 100644
index 0000000000000..efcdac88b342d
--- /dev/null
+++ b/arch/arm64/mm/mteswap_nocomp.c
@@ -0,0 +1,37 @@
+// SPDX-License-Identifier: GPL-2.0
+
+/* MTE tag storage management without compression support. */
+
+#include <asm/mte.h>
+#include <linux/pagemap.h>
+#include <linux/slab.h>
+#include <linux/swap.h>
+#include <linux/swapops.h>
+#include <linux/xarray.h>
+
+#include "mteswap.h"
+
+void *_mte_alloc_and_save_tags(struct page *page)
+{
+ void *storage;
+
+ storage = mte_allocate_tag_storage();
+ if (!storage)
+ return NULL;
+
+ mte_save_page_tags(page_address(page), storage);
+ return storage;
+}
+
+void _mte_free_saved_tags(void *storage)
+{
+ mte_free_tag_storage(storage);
+}
+
+void _mte_restore_tags(void *tags, struct page *page)
+{
+ if (try_page_mte_tagging(page)) {
+ mte_restore_page_tags(page_address(page), tags);
+ set_page_mte_tagged(page);
+ }
+}
--
2.41.0.255.g8b1d071c50-goog
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 8+ messages in thread