Use hash func to boost file creation and lookup

Introduce a hash-based mechanism to speed up file creation and lookup
operations. The hash function enables faster access to extent and logical
block extent index, improving overall filesystem performance.

hash_code = file_hash(file_name);
extent index = hash_code / SIMPLEFS_MAX_BLOCKS_PER_EXTENT
block index = hash_code % SIMPLEFS_MAX_BLOCKS_PER_EXTENT;

Use perf to measure:
1. File Creation (random)
Legacy:
259.842753513 seconds time elapsed
23.000247000 seconds user
150.380145000 seconds sys

fnv1a:
172.464862354 seconds time elapsed
16.456851000 seconds user
117.377930000 seconds sys

full_name_hash:
222.274028604 seconds time elapsed
20.794966000 seconds user
151.941876000 seconds sys

hash64:
210.738022828 seconds time elapsed
20.578577000 seconds user
138.393533000 seconds sys

2. File Listing (random)
Legacy:
min time: 0.00171 s
max time: 0.03799 s
avg time: 0.00423332 s
tot time: 129.539510 s

fnv1a:
min time: 0.00163 s
max time: 0.07574 s
avg time: 0.00247343 s
tot time: 75.686920 s

full_name_hash:
min time: 0.00171 s
max time: 0.03588 s
avg time: 0.00305601 s
tot time: 93.514040 s

hash64:
min time: 0.00167 s
max time: 0.07705 s
avg time: 0.00261552 s
tot time: 80.034760 s

3. files Removal (Random)
Legacy :
106.921706288 seconds time elapsed
16.987883000 seconds user
91.268661000 seconds sys

fnv1a:
94.012202913 seconds time elapsed
20.617263000 seconds user
73.219170000 seconds sys

full_name_hash:
86.132655220 seconds time elapsed
19.180209000 seconds user
68.476075000 seconds sys

hash64:
85.684199320 seconds time elapsed
16.643633000 seconds user
70.499664000 seconds sys

Author: RoyWFHuang

Makefile

@@ -1,10 +1,14 @@
 obj-m += simplefs.o
-simplefs-objs := fs.o super.o inode.o file.o dir.o extent.o symlink.o
+simplefs-objs := fs.o super.o inode.o file.o dir.o extent.o symlink.o hash.o
 KDIR ?= /lib/modules/$(shell uname -r)/build
 
+CONFIG := .config
+CONFIG_H := config.h
+
 MKFS = mkfs.simplefs
+MENUCONFIG := kconfig-mconf
 
-all: $(MKFS)
+all: $(MKFS) $(CONFIG_H)
 	make -C $(KDIR) M=$(PWD) modules
 
 IMAGE ?= test.img
@@ -22,6 +26,30 @@ $(IMAGE): $(MKFS)
 	dd if=/dev/zero of=${IMAGE} bs=1M count=${IMAGESIZE}
 	./$< $(IMAGE)
 
+$(CONFIG_H): $(CONFIG)
+	@echo "/*" > $(CONFIG_H)
+	@echo " * Automatically generated file; DO NOT EDIT." >> $(CONFIG_H)
+	@echo " * Configuration" >> $(CONFIG_H)
+	@echo " */" >> $(CONFIG_H)
+	@echo "" >> $(CONFIG_H)
+	@sed \
+	    -e '/^$$/d' \
+	    -e '/^[[:space:]]*# *$$/d' \
+	    -e '/^[[:space:]]*# Automatically/d' \
+	    -e '/^[[:space:]]*# Configuration/d' \
+	    -e '/^[[:space:]]*# \(CONFIG_[A-Z0-9_]*\) is not set/ s//#undef \1/' \
+	    -e '/^[[:space:]]*CONFIG_[A-Z0-9_]*=y/ s/^[[:space:]]*\(CONFIG_[A-Z0-9_]*\)=y/#define \1 1/' \
+	    -e '/^[[:space:]]*CONFIG_[A-Z0-9_]*=[0-9]\+/ s/^[[:space:]]*\(CONFIG_[A-Z0-9_]*\)=\([0-9]*\)/#define \1 \2/' \
+	    -e '/^[[:space:]]*# / s/^[[:space:]]*# \(.*\)/\/\* \1 \*\//' \
+	    $(CONFIG) >> $(CONFIG_H)
+
+menuconfig:
+	$(MENUCONFIG) Kconfig
+
+.config:
+	@echo "Creating default .config"
+	@echo "CONFIG_SIMPLEFS_HASH_FNV1A=y" > .config
+
 journal: $(JOURNAL)
 
 $(JOURNAL):
@@ -34,6 +62,9 @@ check: all
 clean:
 	make -C $(KDIR) M=$(PWD) clean
 	rm -f *~ $(PWD)/*.ur-safe
-	rm -f $(MKFS) $(IMAGE) $(JOURNAL)
+	rm -f $(MKFS) $(IMAGE) $(JOURNAL) config.h
+
+distclean: clean
+	rm -f $(CONFIG)
 
-.PHONY: all clean journal
+.PHONY: all clean journal menuconfig distclean

dir.c

@@ -81,7 +81,7 @@ static int simplefs_iterate(struct file *dir, struct dir_context *ctx)
                 continue;
             }
 
-            for (fi = 0; fi < SIMPLEFS_FILES_PER_BLOCK;) {
+            for (fi = 0; fi < SIMPLEFS_FILES_PER_BLOCK && dblock->nr_files;) {
                 if (dblock->files[fi].inode != 0) {
                     if (offset) {
                         offset--;

hash.c

@@ -0,0 +1,40 @@
+#include <linux/crc32.h>
+#include <linux/hash.h>
+#include <linux/stringhash.h>
+#include <linux/types.h>
+#include "config.h"
+#include "simplefs.h"
+
+
+#if defined(CONFIG_SIMPLEFS_HASH_HASH64)
+static uint64_t _hash64(const char *name, size_t len)
+{
+    uint32_t val = crc32(0, name, len);
+    return hash_64(
+        val, ilog2(SIMPLEFS_MAX_EXTENTS * SIMPLEFS_MAX_BLOCKS_PER_EXTENT));
+}
+#elif defined(CONFIG_SIMPLEFS_HASH_FNV1A)
+static uint64_t _fnv1a_64(const char *str)
+{
+    uint64_t h = 0xcbf29ce484222325ULL;
+    while (*str) {
+        h ^= (unsigned char) (*str++);
+        h *= 0x100000001b3ULL;
+    }
+    return h;
+}
+#endif
+
+uint32_t simplefs_hash(struct dentry *dentry)
+{
+#if defined(CONFIG_SIMPLEFS_HASH_HASH64)
+    return _hash64(dentry->d_name.name, strlen(dentry->d_name.name));
+#elif defined(CONFIG_SIMPLEFS_HASH_FNV1A)
+    return _fnv1a_64(dentry->d_name.name);
+#elif defined(CONFIG_SIMPLEFS_HASH_FULLNAME)
+    return full_name_hash(dentry, dentry->d_name.name,
+                          strlen(dentry->d_name.name));
+#else
+#error "No hash method selected!"
+#endif
+}