[PATCH bpf-next] bpf: compute hashes in bloom filter similar to hashmap

[PATCH bpf-next] bpf: compute hashes in bloom filter similar to hashmap

[Anton Protopopov]

If the value size in a bloom filter is a multiple of 4, then the jhash2()
function is used to compute hashes. The length parameter of this function
equals to the number of 32-bit words in input. Compute it in the hot path
instead of pre-computing it, as this is translated to one extra shift to
divide the length by four vs. one extra memory load of a pre-computed length.

Signed-off-by: Anton Protopopov <aspsk@isovalent.com>
---
 kernel/bpf/bloom_filter.c | 17 ++---------------
 1 file changed, 2 insertions(+), 15 deletions(-)

diff --git a/kernel/bpf/bloom_filter.c b/kernel/bpf/bloom_filter.c
index db19784601a7..540331b610a9 100644
--- a/kernel/bpf/bloom_filter.c
+++ b/kernel/bpf/bloom_filter.c
@@ -16,13 +16,6 @@ struct bpf_bloom_filter {
 	struct bpf_map map;
 	u32 bitset_mask;
 	u32 hash_seed;
-	/* If the size of the values in the bloom filter is u32 aligned,
-	 * then it is more performant to use jhash2 as the underlying hash
-	 * function, else we use jhash. This tracks the number of u32s
-	 * in an u32-aligned value size. If the value size is not u32 aligned,
-	 * this will be 0.
-	 */
-	u32 aligned_u32_count;
 	u32 nr_hash_funcs;
 	unsigned long bitset[];
 };
@@ -32,9 +25,8 @@ static u32 hash(struct bpf_bloom_filter *bloom, void *value,
 {
 	u32 h;
 
-	if (bloom->aligned_u32_count)
-		h = jhash2(value, bloom->aligned_u32_count,
-			   bloom->hash_seed + index);
+	if (likely(value_size % 4 == 0))
+		h = jhash2(value, value_size / 4, bloom->hash_seed + index);
 	else
 		h = jhash(value, value_size, bloom->hash_seed + index);
 
@@ -152,11 +144,6 @@ static struct bpf_map *bloom_map_alloc(union bpf_attr *attr)
 	bloom->nr_hash_funcs = nr_hash_funcs;
 	bloom->bitset_mask = bitset_mask;
 
-	/* Check whether the value size is u32-aligned */
-	if ((attr->value_size & (sizeof(u32) - 1)) == 0)
-		bloom->aligned_u32_count =
-			attr->value_size / sizeof(u32);
-
 	if (!(attr->map_flags & BPF_F_ZERO_SEED))
 		bloom->hash_seed = get_random_u32();
 
-- 
2.34.1

Re: [PATCH bpf-next] bpf: compute hashes in bloom filter similar to hashmap

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to bpf/bpf-next.git (master)
by Alexei Starovoitov <ast@kernel.org>:

On Sun,  2 Apr 2023 11:43:40 +0000 you wrote:
> If the value size in a bloom filter is a multiple of 4, then the jhash2()
> function is used to compute hashes. The length parameter of this function
> equals to the number of 32-bit words in input. Compute it in the hot path
> instead of pre-computing it, as this is translated to one extra shift to
> divide the length by four vs. one extra memory load of a pre-computed length.
> 
> Signed-off-by: Anton Protopopov <aspsk@isovalent.com>
> 
> [...]

Here is the summary with links:
  - [bpf-next] bpf: compute hashes in bloom filter similar to hashmap
    https://git.kernel.org/bpf/bpf-next/c/92b2e810f0d3

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html