https://github.com/torvalds/linux
Revision 4fca50d440cc5d4dc570ad5484cc0b70b381bc2a authored by Jan Kara on 08 September 2022, 09:21:24 UTC, committed by Theodore Ts'o on 22 September 2022, 02:11:34 UTC
One of the side-effects of mb_optimize_scan was that the optimized
functions to select next group to try were called even before we tried
the goal group. As a result we no longer allocate files close to
corresponding inodes as well as we don't try to expand currently
allocated extent in the same group. This results in reaim regression
with workfile.disk workload of upto 8% with many clients on my test
machine:

                     baseline               mb_optimize_scan
Hmean     disk-1       2114.16 (   0.00%)     2099.37 (  -0.70%)
Hmean     disk-41     87794.43 (   0.00%)    83787.47 *  -4.56%*
Hmean     disk-81    148170.73 (   0.00%)   135527.05 *  -8.53%*
Hmean     disk-121   177506.11 (   0.00%)   166284.93 *  -6.32%*
Hmean     disk-161   220951.51 (   0.00%)   207563.39 *  -6.06%*
Hmean     disk-201   208722.74 (   0.00%)   203235.59 (  -2.63%)
Hmean     disk-241   222051.60 (   0.00%)   217705.51 (  -1.96%)
Hmean     disk-281   252244.17 (   0.00%)   241132.72 *  -4.41%*
Hmean     disk-321   255844.84 (   0.00%)   245412.84 *  -4.08%*

Also this is causing huge regression (time increased by a factor of 5 or
so) when untarring archive with lots of small files on some eMMC storage
cards.

Fix the problem by making sure we try goal group first.

Fixes: 196e402adf2e ("ext4: improve cr 0 / cr 1 group scanning")
CC: stable@kernel.org
Reported-and-tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Tested-by: Ojaswin Mujoo <ojaswin@linux.ibm.com>
Reviewed-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
Link: https://lore.kernel.org/all/20220727105123.ckwrhbilzrxqpt24@quack3/
Link: https://lore.kernel.org/all/0d81a7c2-46b7-6010-62a4-3e6cfc1628d6@i2se.com/
Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220908092136.11770-1-jack@suse.cz
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
1 parent 7e18e42
Raw File
Tip revision: 4fca50d440cc5d4dc570ad5484cc0b70b381bc2a authored by Jan Kara on 08 September 2022, 09:21:24 UTC
ext4: make mballoc try target group first even with mb_optimize_scan
Tip revision: 4fca50d
aegis128-neon-inner.c
// SPDX-License-Identifier: GPL-2.0-or-later
/*
 * Copyright (C) 2019 Linaro, Ltd. <ard.biesheuvel@linaro.org>
 */

#ifdef CONFIG_ARM64
#include <asm/neon-intrinsics.h>

#define AES_ROUND	"aese %0.16b, %1.16b \n\t aesmc %0.16b, %0.16b"
#else
#include <arm_neon.h>

#define AES_ROUND	"aese.8 %q0, %q1 \n\t aesmc.8 %q0, %q0"
#endif

#define AEGIS_BLOCK_SIZE	16

#include <stddef.h>

extern int aegis128_have_aes_insn;

void *memcpy(void *dest, const void *src, size_t n);

struct aegis128_state {
	uint8x16_t v[5];
};

extern const uint8_t crypto_aes_sbox[];

static struct aegis128_state aegis128_load_state_neon(const void *state)
{
	return (struct aegis128_state){ {
		vld1q_u8(state),
		vld1q_u8(state + 16),
		vld1q_u8(state + 32),
		vld1q_u8(state + 48),
		vld1q_u8(state + 64)
	} };
}

static void aegis128_save_state_neon(struct aegis128_state st, void *state)
{
	vst1q_u8(state, st.v[0]);
	vst1q_u8(state + 16, st.v[1]);
	vst1q_u8(state + 32, st.v[2]);
	vst1q_u8(state + 48, st.v[3]);
	vst1q_u8(state + 64, st.v[4]);
}

static inline __attribute__((always_inline))
uint8x16_t aegis_aes_round(uint8x16_t w)
{
	uint8x16_t z = {};

#ifdef CONFIG_ARM64
	if (!__builtin_expect(aegis128_have_aes_insn, 1)) {
		static const uint8_t shift_rows[] = {
			0x0, 0x5, 0xa, 0xf, 0x4, 0x9, 0xe, 0x3,
			0x8, 0xd, 0x2, 0x7, 0xc, 0x1, 0x6, 0xb,
		};
		static const uint8_t ror32by8[] = {
			0x1, 0x2, 0x3, 0x0, 0x5, 0x6, 0x7, 0x4,
			0x9, 0xa, 0xb, 0x8, 0xd, 0xe, 0xf, 0xc,
		};
		uint8x16_t v;

		// shift rows
		w = vqtbl1q_u8(w, vld1q_u8(shift_rows));

		// sub bytes
#ifndef CONFIG_CC_IS_GCC
		v = vqtbl4q_u8(vld1q_u8_x4(crypto_aes_sbox), w);
		v = vqtbx4q_u8(v, vld1q_u8_x4(crypto_aes_sbox + 0x40), w - 0x40);
		v = vqtbx4q_u8(v, vld1q_u8_x4(crypto_aes_sbox + 0x80), w - 0x80);
		v = vqtbx4q_u8(v, vld1q_u8_x4(crypto_aes_sbox + 0xc0), w - 0xc0);
#else
		asm("tbl %0.16b, {v16.16b-v19.16b}, %1.16b" : "=w"(v) : "w"(w));
		w -= 0x40;
		asm("tbx %0.16b, {v20.16b-v23.16b}, %1.16b" : "+w"(v) : "w"(w));
		w -= 0x40;
		asm("tbx %0.16b, {v24.16b-v27.16b}, %1.16b" : "+w"(v) : "w"(w));
		w -= 0x40;
		asm("tbx %0.16b, {v28.16b-v31.16b}, %1.16b" : "+w"(v) : "w"(w));
#endif

		// mix columns
		w = (v << 1) ^ (uint8x16_t)(((int8x16_t)v >> 7) & 0x1b);
		w ^= (uint8x16_t)vrev32q_u16((uint16x8_t)v);
		w ^= vqtbl1q_u8(v ^ w, vld1q_u8(ror32by8));

		return w;
	}
#endif

	/*
	 * We use inline asm here instead of the vaeseq_u8/vaesmcq_u8 intrinsics
	 * to force the compiler to issue the aese/aesmc instructions in pairs.
	 * This is much faster on many cores, where the instruction pair can
	 * execute in a single cycle.
	 */
	asm(AES_ROUND : "+w"(w) : "w"(z));
	return w;
}

static inline __attribute__((always_inline))
struct aegis128_state aegis128_update_neon(struct aegis128_state st,
					   uint8x16_t m)
{
	m       ^= aegis_aes_round(st.v[4]);
	st.v[4] ^= aegis_aes_round(st.v[3]);
	st.v[3] ^= aegis_aes_round(st.v[2]);
	st.v[2] ^= aegis_aes_round(st.v[1]);
	st.v[1] ^= aegis_aes_round(st.v[0]);
	st.v[0] ^= m;

	return st;
}

static inline __attribute__((always_inline))
void preload_sbox(void)
{
	if (!IS_ENABLED(CONFIG_ARM64) ||
	    !IS_ENABLED(CONFIG_CC_IS_GCC) ||
	    __builtin_expect(aegis128_have_aes_insn, 1))
		return;

	asm("ld1	{v16.16b-v19.16b}, [%0], #64	\n\t"
	    "ld1	{v20.16b-v23.16b}, [%0], #64	\n\t"
	    "ld1	{v24.16b-v27.16b}, [%0], #64	\n\t"
	    "ld1	{v28.16b-v31.16b}, [%0]		\n\t"
	    :: "r"(crypto_aes_sbox));
}

void crypto_aegis128_init_neon(void *state, const void *key, const void *iv)
{
	static const uint8_t const0[] = {
		0x00, 0x01, 0x01, 0x02, 0x03, 0x05, 0x08, 0x0d,
		0x15, 0x22, 0x37, 0x59, 0x90, 0xe9, 0x79, 0x62,
	};
	static const uint8_t const1[] = {
		0xdb, 0x3d, 0x18, 0x55, 0x6d, 0xc2, 0x2f, 0xf1,
		0x20, 0x11, 0x31, 0x42, 0x73, 0xb5, 0x28, 0xdd,
	};
	uint8x16_t k = vld1q_u8(key);
	uint8x16_t kiv = k ^ vld1q_u8(iv);
	struct aegis128_state st = {{
		kiv,
		vld1q_u8(const1),
		vld1q_u8(const0),
		k ^ vld1q_u8(const0),
		k ^ vld1q_u8(const1),
	}};
	int i;

	preload_sbox();

	for (i = 0; i < 5; i++) {
		st = aegis128_update_neon(st, k);
		st = aegis128_update_neon(st, kiv);
	}
	aegis128_save_state_neon(st, state);
}

void crypto_aegis128_update_neon(void *state, const void *msg)
{
	struct aegis128_state st = aegis128_load_state_neon(state);

	preload_sbox();

	st = aegis128_update_neon(st, vld1q_u8(msg));

	aegis128_save_state_neon(st, state);
}

#ifdef CONFIG_ARM
/*
 * AArch32 does not provide these intrinsics natively because it does not
 * implement the underlying instructions. AArch32 only provides 64-bit
 * wide vtbl.8/vtbx.8 instruction, so use those instead.
 */
static uint8x16_t vqtbl1q_u8(uint8x16_t a, uint8x16_t b)
{
	union {
		uint8x16_t	val;
		uint8x8x2_t	pair;
	} __a = { a };

	return vcombine_u8(vtbl2_u8(__a.pair, vget_low_u8(b)),
			   vtbl2_u8(__a.pair, vget_high_u8(b)));
}

static uint8x16_t vqtbx1q_u8(uint8x16_t v, uint8x16_t a, uint8x16_t b)
{
	union {
		uint8x16_t	val;
		uint8x8x2_t	pair;
	} __a = { a };

	return vcombine_u8(vtbx2_u8(vget_low_u8(v), __a.pair, vget_low_u8(b)),
			   vtbx2_u8(vget_high_u8(v), __a.pair, vget_high_u8(b)));
}

static int8_t vminvq_s8(int8x16_t v)
{
	int8x8_t s = vpmin_s8(vget_low_s8(v), vget_high_s8(v));

	s = vpmin_s8(s, s);
	s = vpmin_s8(s, s);
	s = vpmin_s8(s, s);

	return vget_lane_s8(s, 0);
}
#endif

static const uint8_t permute[] __aligned(64) = {
	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
	 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12, 13, 14, 15,
	-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
};

void crypto_aegis128_encrypt_chunk_neon(void *state, void *dst, const void *src,
					unsigned int size)
{
	struct aegis128_state st = aegis128_load_state_neon(state);
	const int short_input = size < AEGIS_BLOCK_SIZE;
	uint8x16_t msg;

	preload_sbox();

	while (size >= AEGIS_BLOCK_SIZE) {
		uint8x16_t s = st.v[1] ^ (st.v[2] & st.v[3]) ^ st.v[4];

		msg = vld1q_u8(src);
		st = aegis128_update_neon(st, msg);
		msg ^= s;
		vst1q_u8(dst, msg);

		size -= AEGIS_BLOCK_SIZE;
		src += AEGIS_BLOCK_SIZE;
		dst += AEGIS_BLOCK_SIZE;
	}

	if (size > 0) {
		uint8x16_t s = st.v[1] ^ (st.v[2] & st.v[3]) ^ st.v[4];
		uint8_t buf[AEGIS_BLOCK_SIZE];
		const void *in = src;
		void *out = dst;
		uint8x16_t m;

		if (__builtin_expect(short_input, 0))
			in = out = memcpy(buf + AEGIS_BLOCK_SIZE - size, src, size);

		m = vqtbl1q_u8(vld1q_u8(in + size - AEGIS_BLOCK_SIZE),
			       vld1q_u8(permute + 32 - size));

		st = aegis128_update_neon(st, m);

		vst1q_u8(out + size - AEGIS_BLOCK_SIZE,
			 vqtbl1q_u8(m ^ s, vld1q_u8(permute + size)));

		if (__builtin_expect(short_input, 0))
			memcpy(dst, out, size);
		else
			vst1q_u8(out - AEGIS_BLOCK_SIZE, msg);
	}

	aegis128_save_state_neon(st, state);
}

void crypto_aegis128_decrypt_chunk_neon(void *state, void *dst, const void *src,
					unsigned int size)
{
	struct aegis128_state st = aegis128_load_state_neon(state);
	const int short_input = size < AEGIS_BLOCK_SIZE;
	uint8x16_t msg;

	preload_sbox();

	while (size >= AEGIS_BLOCK_SIZE) {
		msg = vld1q_u8(src) ^ st.v[1] ^ (st.v[2] & st.v[3]) ^ st.v[4];
		st = aegis128_update_neon(st, msg);
		vst1q_u8(dst, msg);

		size -= AEGIS_BLOCK_SIZE;
		src += AEGIS_BLOCK_SIZE;
		dst += AEGIS_BLOCK_SIZE;
	}

	if (size > 0) {
		uint8x16_t s = st.v[1] ^ (st.v[2] & st.v[3]) ^ st.v[4];
		uint8_t buf[AEGIS_BLOCK_SIZE];
		const void *in = src;
		void *out = dst;
		uint8x16_t m;

		if (__builtin_expect(short_input, 0))
			in = out = memcpy(buf + AEGIS_BLOCK_SIZE - size, src, size);

		m = s ^ vqtbx1q_u8(s, vld1q_u8(in + size - AEGIS_BLOCK_SIZE),
				   vld1q_u8(permute + 32 - size));

		st = aegis128_update_neon(st, m);

		vst1q_u8(out + size - AEGIS_BLOCK_SIZE,
			 vqtbl1q_u8(m, vld1q_u8(permute + size)));

		if (__builtin_expect(short_input, 0))
			memcpy(dst, out, size);
		else
			vst1q_u8(out - AEGIS_BLOCK_SIZE, msg);
	}

	aegis128_save_state_neon(st, state);
}

int crypto_aegis128_final_neon(void *state, void *tag_xor,
			       unsigned int assoclen,
			       unsigned int cryptlen,
			       unsigned int authsize)
{
	struct aegis128_state st = aegis128_load_state_neon(state);
	uint8x16_t v;
	int i;

	preload_sbox();

	v = st.v[3] ^ (uint8x16_t)vcombine_u64(vmov_n_u64(8ULL * assoclen),
					       vmov_n_u64(8ULL * cryptlen));

	for (i = 0; i < 7; i++)
		st = aegis128_update_neon(st, v);

	v = st.v[0] ^ st.v[1] ^ st.v[2] ^ st.v[3] ^ st.v[4];

	if (authsize > 0) {
		v = vqtbl1q_u8(~vceqq_u8(v, vld1q_u8(tag_xor)),
			       vld1q_u8(permute + authsize));

		return vminvq_s8((int8x16_t)v);
	}

	vst1q_u8(tag_xor, v);
	return 0;
}
back to top