Skip to main content

Browse the archive

Revision e03a5125ec7bd1d4e8b062816a8813436876dc7c authored by Nicholas Piggin on 12 July 2019, 03:59:12 UTC, committed by Linus Torvalds on 12 July 2019, 18:05:46 UTC 
mm/large system hash: clear hashdist when only one node with memory is booted
 
CONFIG_NUMA on 64-bit CPUs currently enables hashdist unconditionally even
when booting on single node machines.  This causes the large system hashes
to be allocated with vmalloc, and mapped with small pages.

This change clears hashdist if only one node has come up with memory.

This results in the important large inode and dentry hashes using memblock
allocations.  All others are within 4MB size up to about 128GB of RAM,
which allows them to be allocated from the linear map on most non-NUMA
images.

Other big hashes like futex and TCP should eventually be moved over to the
same style of allocation as those vfs caches that use HASH_EARLY if
!hashdist, so they don't exceed MAX_ORDER on very large non-NUMA images.

This brings dTLB misses for linux kernel tree `git diff` from ~45,000 to
~8,000 on a Kaby Lake KVM guest with 8MB dentry hash and mitigations=off
(performance is in the noise, under 1% difference, page tables are likely
to be well cached for this workload).

Link: http://lkml.kernel.org/r/20190605144814.29319-2-npiggin@gmail.com
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1 parent ec11408
Raw File
crc4.c 
// SPDX-License-Identifier: GPL-2.0-only
/*
 * crc4.c - simple crc-4 calculations.
 */

#include <linux/crc4.h>
#include <linux/module.h>

static const uint8_t crc4_tab[] = {
	0x0, 0x7, 0xe, 0x9, 0xb, 0xc, 0x5, 0x2,
	0x1, 0x6, 0xf, 0x8, 0xa, 0xd, 0x4, 0x3,
};

/**
 * crc4 - calculate the 4-bit crc of a value.
 * @c:    starting crc4
 * @x:    value to checksum
 * @bits: number of bits in @x to checksum
 *
 * Returns the crc4 value of @x, using polynomial 0b10111.
 *
 * The @x value is treated as left-aligned, and bits above @bits are ignored
 * in the crc calculations.
 */
uint8_t crc4(uint8_t c, uint64_t x, int bits)
{
	int i;

	/* mask off anything above the top bit */
	x &= (1ull << bits) - 1;

	/* Align to 4-bits */
	bits = (bits + 3) & ~0x3;

	/* Calculate crc4 over four-bit nibbles, starting at the MSbit */
	for (i = bits - 4; i >= 0; i -= 4)
		c = crc4_tab[c ^ ((x >> i) & 0xf)];

	return c;
}
EXPORT_SYMBOL_GPL(crc4);

MODULE_DESCRIPTION("CRC4 calculations");
MODULE_LICENSE("GPL");
back to top