erofs: Support the new bloom filter erofs compat feature

This matches: https://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs.git/commit/?h=dev&id=fd73a4395d477ae134f319f7368a9f8a6264fd8b We set the COMPAT_XATTR_FILTER flag on the superblock, and we add to each xattr header a 32bit bloom filter which defaults to all bits set, and then we compute a hash for each key modulo 32, and unset that bit in the bloom filter. Older kernels will ignore this, but newer kernels will use it to make xattr lookup more efficient. NOTE: This is a format-breaking change, so we also update the test checksums. We should merge this before freezing the format. Signed-off-by: Alexander Larsson <[email protected]>
containers · Aug 25, 2023 · 2993e34 · 2993e34
1 parent 7a94084
commit 2993e34
Show file tree

Hide file tree

Showing 6 changed files with 33 additions and 7 deletions.
diff --git a/libcomposefs/lcfs-writer-erofs.c b/libcomposefs/lcfs-writer-erofs.c
@@ -36,7 +36,10 @@
 #include <assert.h>
 #include <linux/fsverity.h>
 
-/* The xxh32 hash function is copied from xxhash with this copyright:
+/* The xxh32 hash function is copied from the linux kernel at:
+ *  https://github.com/torvalds/linux/blob/d89775fc929c5a1d91ed518a71b456da0865e5ff/lib/xxhash.c
+ *
+ * The original copyright is:
  *
  * xxHash - Extremely Fast Hash algorithm
  * Copyright (C) 2012-2016, Yann Collet.
@@ -544,6 +547,26 @@ static void compute_erofs_xattr_counts(struct lcfs_node_s *node,
 	*unshared_xattrs_size_out = unshared_xattrs_size;
 }
 
+static uint32_t compute_erofs_xattr_filter(struct lcfs_node_s *node)
+{
+	uint32_t name_filter = 0;
+
+	for (size_t i = 0; i < node->n_xattrs; i++) {
+		struct lcfs_xattr_s *xattr = &node->xattrs[i];
+		uint32_t name_filter_bit;
+		uint8_t index;
+		char *key;
+
+		index = xattr_erofs_entry_index(xattr, &key);
+		name_filter_bit =
+			xxh32(key, strlen(key), EROFS_XATTR_FILTER_SEED + index) &
+			(EROFS_XATTR_FILTER_BITS - 1);
+		name_filter |= 1UL << name_filter_bit;
+	}
+
+	return EROFS_XATTR_FILTER_DEFAULT & ~name_filter;
+}
+
 static uint64_t compute_erofs_inode_padding_for_tail(struct lcfs_node_s *node,
 						     uint64_t pos, size_t inode_size,
 						     size_t xattr_size)
@@ -878,6 +901,8 @@ static int write_erofs_inode_data(struct lcfs_ctx_s *ctx, struct lcfs_node_s *no
 	if (xattr_size) {
 		struct erofs_xattr_ibody_header xattr_header = { 0 };
 		xattr_header.h_shared_count = n_shared_xattrs;
+		xattr_header.h_name_filter =
+			lcfs_u32_to_file(compute_erofs_xattr_filter(node));
 
 		ret = lcfs_write(ctx, &xattr_header, sizeof(xattr_header));
 		if (ret < 0)
@@ -1218,7 +1243,8 @@ int lcfs_write_erofs_to(struct lcfs_ctx_s *ctx)
 	if (ret < 0)
 		return ret;
 
-	superblock.feature_compat = lcfs_u32_to_file(EROFS_FEATURE_COMPAT_MTIME);
+	superblock.feature_compat = lcfs_u32_to_file(
+		EROFS_FEATURE_COMPAT_MTIME | EROFS_FEATURE_COMPAT_XATTR_FILTER);
 	superblock.inos = lcfs_u64_to_file(ctx->num_inodes);
 
 	superblock.build_time = lcfs_u64_to_file(ctx->min_mtim_sec);

diff --git a/tools/test-assets/config-with-hard-link.json.gz.sha256_erofs b/tools/test-assets/config-with-hard-link.json.gz.sha256_erofs
@@ -1 +1 @@
-ba8b921eb4600eca90457c7b1f6139b59a3a4ca110e0fcd76c3cb8d681dfb60c
+d67e775ced030711b36d70d315ddf7150214312fd18b9345f98a5572035a2076
diff --git a/tools/test-assets/config.json.gz.sha256_erofs b/tools/test-assets/config.json.gz.sha256_erofs
@@ -1 +1 @@
-e48676802b93e656265343e71c7d63db472994d8253e591ea3757b0e5b37c61f
+f93737bda1b19d79c8fe2828615b9dd24e4ec5d97410a1ce6e3716b1a778f00f
diff --git a/tools/test-assets/cs9-x86_64-developer.json.gz.sha256_erofs b/tools/test-assets/cs9-x86_64-developer.json.gz.sha256_erofs
@@ -1 +1 @@
-3fe5370d0fed684ed3852870422c3a93273c612ef9628099ea225a5b17741ae0
+8c61b21852d6abfdd3a7a3da71a5ccb56fae2ad099bebe7d564fcaa842cb9532
diff --git a/tools/test-assets/cs9-x86_64-minimal.json.gz.sha256_erofs b/tools/test-assets/cs9-x86_64-minimal.json.gz.sha256_erofs
@@ -1 +1 @@
-e50e762f77d6656f4d48354d3ea127ef091521a6ad00d63fc727c2ad0b2e8174
+a168966c314b0b27c8e25270e6f20838a4736a888e5de97993d15d046c797ba3
diff --git a/tools/test-assets/f36-x86_64-silverblue.json.gz.sha256_erofs b/tools/test-assets/f36-x86_64-silverblue.json.gz.sha256_erofs
@@ -1 +1 @@
-40afd104079a9886ab53c6a9fea2b2e76e24f36add35dd4ccb1b18b85151ae52
+34bade1be8cef44439944f94a0a7d1421e0661660a506ef0c744135c79cc8ef0