unicode: Add utf8-data module

utf8data.h contains a large database table which is an auto-generated
decodification trie for the unicode normalization functions.

Allow building it into a separate module.

Based on a patch from Shreeya Patel <[e-mail]>.

Signed-off-by: Christoph Hellwig <[e-mail]>
Signed-off-by: Gabriel Krisman Bertazi <[e-mail]>

fs/unicode/Kconfig

···
 	  Say Y here to enable UTF-8 NFD normalization and NFD+CF casefolding
 	  support.
 
+config UNICODE_UTF8_DATA
+	tristate "UTF-8 normalization and casefolding tables"
+	depends on UNICODE
+	default UNICODE
+	help
+	  This contains a large table of case foldings, which can be loaded as
+	  a separate module if you say M here.  To be on the safe side stick
+	  to the default of Y.  Saying N here makes no sense, if you do not want
+	  utf8 casefolding support, disable CONFIG_UNICODE instead.
+
 config UNICODE_NORMALIZATION_SELFTEST
 	tristate "Test UTF-8 normalization support"
-	depends on UNICODE
-	default n
+	depends on UNICODE_UTF8_DATA

fs/unicode/Makefile

···
 
 obj-$(CONFIG_UNICODE) += unicode.o
 obj-$(CONFIG_UNICODE_NORMALIZATION_SELFTEST) += utf8-selftest.o
+obj-$(CONFIG_UNICODE_UTF8_DATA) += utf8data.o
 
 unicode-y := utf8-norm.o utf8-core.o
 
-$(obj)/utf8-norm.o: $(obj)/utf8data.h
+$(obj)/utf8-data.o: $(obj)/utf8data.c
 
-# In the normal build, the checked-in utf8data.h is just shipped.
+# In the normal build, the checked-in utf8data.c is just shipped.
 #
-# To generate utf8data.h from UCD, put *.txt files in this directory
+# To generate utf8data.c from UCD, put *.txt files in this directory
 # and pass REGENERATE_UTF8DATA=1 from the command line.
 ifdef REGENERATE_UTF8DATA
 
···
 		-t $(srctree)/$(src)/NormalizationTest.txt \
 		-o $@
 
-$(obj)/utf8data.h: $(obj)/mkutf8data $(filter %.txt, $(cmd_utf8data)) FORCE
+$(obj)/utf8data.c: $(obj)/mkutf8data $(filter %.txt, $(cmd_utf8data)) FORCE
 	$(call if_changed,utf8data)
 
 else
 
-$(obj)/utf8data.h: $(src)/utf8data.h_shipped FORCE
+$(obj)/utf8data.c: $(src)/utf8data.c_shipped FORCE
 	$(call if_changed,shipped)
 
 endif
 
-targets += utf8data.h
+targets += utf8data.c
 hostprogs += mkutf8data

fs/unicode/mkutf8data.c

···
 		open_fail(utf8_name, errno);
 
 	fprintf(file, "/* This file is generated code, do not edit. */\n");
-	fprintf(file, "#ifndef __INCLUDED_FROM_UTF8NORM_C__\n");
-	fprintf(file, "#error Only nls_utf8-norm.c should include this file.\n");
-	fprintf(file, "#endif\n");
 	fprintf(file, "\n");
-	fprintf(file, "static const unsigned int utf8vers = %#x;\n",
-		unicode_maxage);
+	fprintf(file, "#include <linux/module.h>\n");
+	fprintf(file, "#include <linux/kernel.h>\n");
+	fprintf(file, "#include \"utf8n.h\"\n");
 	fprintf(file, "\n");
 	fprintf(file, "static const unsigned int utf8agetab[] = {\n");
 	for (i = 0; i != ages_count; i++)
···
 		fprintf(file, "\n");
 	}
 	fprintf(file, "};\n");
+	fprintf(file, "\n");
+	fprintf(file, "struct utf8data_table utf8_data_table = {\n");
+	fprintf(file, "\t.utf8agetab = utf8agetab,\n");
+	fprintf(file, "\t.utf8agetab_size = ARRAY_SIZE(utf8agetab),\n");
+	fprintf(file, "\n");
+	fprintf(file, "\t.utf8nfdicfdata = utf8nfdicfdata,\n");
+	fprintf(file, "\t.utf8nfdicfdata_size = ARRAY_SIZE(utf8nfdicfdata),\n");
+	fprintf(file, "\n");
+	fprintf(file, "\t.utf8nfdidata = utf8nfdidata,\n");
+	fprintf(file, "\t.utf8nfdidata_size = ARRAY_SIZE(utf8nfdidata),\n");
+	fprintf(file, "\n");
+	fprintf(file, "\t.utf8data = utf8data,\n");
+	fprintf(file, "};\n");
+	fprintf(file, "EXPORT_SYMBOL_GPL(utf8_data_table);");
+	fprintf(file, "\n");
+	fprintf(file, "MODULE_LICENSE(\"GPL v2\");\n");
 	fclose(file);
 }
 

fs/unicode/utf8-core.c

···
 }
 EXPORT_SYMBOL(utf8_normalize);
 
+static const struct utf8data *find_table_version(const struct utf8data *table,
+		size_t nr_entries, unsigned int version)
+{
+	size_t i = nr_entries - 1;
+
+	while (version < table[i].maxage)
+		i--;
+	if (version > table[i].maxage)
+		return NULL;
+	return &table[i];
+}
+
 struct unicode_map *utf8_load(unsigned int version)
 {
 	struct unicode_map *um;
-
-	if (!utf8version_is_supported(version))
-		return ERR_PTR(-EINVAL);
 
 	um = kzalloc(sizeof(struct unicode_map), GFP_KERNEL);
 	if (!um)
 		return ERR_PTR(-ENOMEM);
 	um->version = version;
-	um->ntab[UTF8_NFDI] = utf8nfdi(version);
+
+	um->tables = symbol_request(utf8_data_table);
+	if (!um->tables)
+		goto out_free_um;
+
+	if (!utf8version_is_supported(um, version))
+		goto out_symbol_put;
+	um->ntab[UTF8_NFDI] = find_table_version(um->tables->utf8nfdidata,
+			um->tables->utf8nfdidata_size, um->version);
 	if (!um->ntab[UTF8_NFDI])
-		goto out_free_um;
-	um->ntab[UTF8_NFDICF] = utf8nfdicf(version);
+		goto out_symbol_put;
+	um->ntab[UTF8_NFDICF] = find_table_version(um->tables->utf8nfdicfdata,
+			um->tables->utf8nfdicfdata_size, um->version);
 	if (!um->ntab[UTF8_NFDICF])
-		goto out_free_um;
+		goto out_symbol_put;
 	return um;
 
+out_symbol_put:
+	symbol_put(um->tables);
 out_free_um:
 	kfree(um);
 	return ERR_PTR(-EINVAL);
···
 
 void utf8_unload(struct unicode_map *um)
 {
-	kfree(um);
+	if (um) {
+		symbol_put(utf8_data_table);
+		kfree(um);
+	}
 }
 EXPORT_SYMBOL(utf8_unload);
 

fs/unicode/utf8-norm.c

···
 
 #include "utf8n.h"
 
-struct utf8data {
-	unsigned int maxage;
-	unsigned int offset;
-};
-
-#define __INCLUDED_FROM_UTF8NORM_C__
-#include "utf8data.h"
-#undef __INCLUDED_FROM_UTF8NORM_C__
-
-int utf8version_is_supported(unsigned int version)
+int utf8version_is_supported(const struct unicode_map *um, unsigned int version)
 {
-	int i = ARRAY_SIZE(utf8agetab) - 1;
+	int i = um->tables->utf8agetab_size - 1;
 
-	while (i >= 0 && utf8agetab[i] != 0) {
-		if (version == utf8agetab[i])
+	while (i >= 0 && um->tables->utf8agetab[i] != 0) {
+		if (version == um->tables->utf8agetab[i])
 			return 1;
 		i--;
 	}
···
  * underlying datatype: unsigned char.
  *
  * leaf[0]: The unicode version, stored as a generation number that is
- *          an index into utf8agetab[].  With this we can filter code
+ *          an index into ->utf8agetab[].  With this we can filter code
  *          points based on the unicode version in which they were
  *          defined.  The CCC of a non-defined code point is 0.
  * leaf[1]: Canonical Combining Class. During normalization, we need
···
 		enum utf8_normalization n, unsigned char *hangul, const char *s,
 		size_t len)
 {
-	utf8trie_t	*trie = utf8data + um->ntab[n]->offset;
+	utf8trie_t	*trie = um->tables->utf8data + um->ntab[n]->offset;
 	int		offlen;
 	int		offset;
 	int		mask;
···
 		leaf = utf8nlookup(um, n, hangul, s, len);
 		if (!leaf)
 			return -1;
-		if (utf8agetab[LEAF_GEN(leaf)] > um->ntab[n]->maxage)
+		if (um->tables->utf8agetab[LEAF_GEN(leaf)] >
+		    um->ntab[n]->maxage)
 			ret += utf8clen(s);
 		else if (LEAF_CCC(leaf) == DECOMPOSE)
 			ret += strlen(LEAF_STR(leaf));
···
 
 		ccc = LEAF_CCC(leaf);
 		/* Characters that are too new have CCC 0. */
-		if (utf8agetab[LEAF_GEN(leaf)] >
+		if (u8c->um->tables->utf8agetab[LEAF_GEN(leaf)] >
 		    u8c->um->ntab[u8c->n]->maxage) {
 			ccc = STOPPER;
 		} else if (ccc == DECOMPOSE) {
···
 	}
 }
 EXPORT_SYMBOL(utf8byte);
-
-const struct utf8data *utf8nfdi(unsigned int maxage)
-{
-	int i = ARRAY_SIZE(utf8nfdidata) - 1;
-
-	while (maxage < utf8nfdidata[i].maxage)
-		i--;
-	if (maxage > utf8nfdidata[i].maxage)
-		return NULL;
-	return &utf8nfdidata[i];
-}
-
-const struct utf8data *utf8nfdicf(unsigned int maxage)
-{
-	int i = ARRAY_SIZE(utf8nfdicfdata) - 1;
-
-	while (maxage < utf8nfdicfdata[i].maxage)
-		i--;
-	if (maxage > utf8nfdicfdata[i].maxage)
-		return NULL;
-	return &utf8nfdicfdata[i];
-}

fs/unicode/utf8-selftest.c

···
 	}
 }
 
-static void check_supported_versions(void)
+static void check_supported_versions(struct unicode_map *um)
 {
 	/* Unicode 7.0.0 should be supported. */
-	test(utf8version_is_supported(UNICODE_AGE(7, 0, 0)));
+	test(utf8version_is_supported(um, UNICODE_AGE(7, 0, 0)));
 
 	/* Unicode 9.0.0 should be supported. */
-	test(utf8version_is_supported(UNICODE_AGE(9, 0, 0)));
+	test(utf8version_is_supported(um, UNICODE_AGE(9, 0, 0)));
 
 	/* Unicode 1x.0.0 (the latest version) should be supported. */
-	test(utf8version_is_supported(UTF8_LATEST));
+	test(utf8version_is_supported(um, UTF8_LATEST));
 
 	/* Next versions don't exist. */
-	test(!utf8version_is_supported(UNICODE_AGE(13, 0, 0)));
-	test(!utf8version_is_supported(UNICODE_AGE(0, 0, 0)));
-	test(!utf8version_is_supported(UNICODE_AGE(-1, -1, -1)));
+	test(!utf8version_is_supported(um, UNICODE_AGE(13, 0, 0)));
+	test(!utf8version_is_supported(um, UNICODE_AGE(0, 0, 0)));
+	test(!utf8version_is_supported(um, UNICODE_AGE(-1, -1, -1)));
 }
 
 static int __init init_test_ucd(void)
···
 		return PTR_ERR(um);
 	}
 
-	check_supported_versions();
+	check_supported_versions(um);
 	check_utf8_nfdi(um);
 	check_utf8_nfdicf(um);
 	check_utf8_comparisons(um);

fs/unicode/utf8data.h_shipped fs/unicode/utf8data.c_shipped

···
 /* This file is generated code, do not edit. */
-#ifndef __INCLUDED_FROM_UTF8NORM_C__
-#error Only nls_utf8-norm.c should include this file.
-#endif
 
-static const unsigned int utf8vers = 0xc0100;
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include "utf8n.h"
 
 static const unsigned int utf8agetab[] = {
 	0,
···
 	0x52,0x04,0x00,0x00,0x11,0x04,0x00,0x00,0x02,0x00,0xcf,0x86,0xcf,0x06,0x02,0x00,
 	0x81,0x80,0xcf,0x86,0x85,0x84,0xcf,0x86,0xcf,0x06,0x02,0x00,0x00,0x00,0x00,0x00
 };
+
+struct utf8data_table utf8_data_table = {
+	.utf8agetab = utf8agetab,
+	.utf8agetab_size = ARRAY_SIZE(utf8agetab),
+
+	.utf8nfdicfdata = utf8nfdicfdata,
+	.utf8nfdicfdata_size = ARRAY_SIZE(utf8nfdicfdata),
+
+	.utf8nfdidata = utf8nfdidata,
+	.utf8nfdidata_size = ARRAY_SIZE(utf8nfdidata),
+
+	.utf8data = utf8data,
+};
+EXPORT_SYMBOL_GPL(utf8_data_table);
+MODULE_LICENSE("GPL v2");

fs/unicode/utf8n.h

···
 #include <linux/module.h>
 #include <linux/unicode.h>
 
-int utf8version_is_supported(unsigned int version);
-
-/*
- * Look for the correct const struct utf8data for a unicode version.
- * Returns NULL if the version requested is too new.
- *
- * Two normalization forms are supported: nfdi and nfdicf.
- *
- * nfdi:
- *  - Apply unicode normalization form NFD.
- *  - Remove any Default_Ignorable_Code_Point.
- *
- * nfdicf:
- *  - Apply unicode normalization form NFD.
- *  - Remove any Default_Ignorable_Code_Point.
- *  - Apply a full casefold (C + F).
- */
-extern const struct utf8data *utf8nfdi(unsigned int maxage);
-extern const struct utf8data *utf8nfdicf(unsigned int maxage);
+int utf8version_is_supported(const struct unicode_map *um, unsigned int version);
 
 /*
  * Determine the length of the normalized from of the string,
···
  * Returns -1 if the string being normalized is not valid UTF-8.
  */
 extern int utf8byte(struct utf8cursor *u8c);
+
+struct utf8data {
+	unsigned int maxage;
+	unsigned int offset;
+};
+
+struct utf8data_table {
+	const unsigned int *utf8agetab;
+	int utf8agetab_size;
+
+	const struct utf8data *utf8nfdicfdata;
+	int utf8nfdicfdata_size;
+
+	const struct utf8data *utf8nfdidata;
+	int utf8nfdidata_size;
+
+	const unsigned char *utf8data;
+};
+
+extern struct utf8data_table utf8_data_table;
 
 #endif /* UTF8NORM_H */

include/linux/unicode.h

···
 #include <linux/dcache.h>
 
 struct utf8data;
+struct utf8data_table;
 
 #define UNICODE_MAJ_SHIFT		16
 #define UNICODE_MIN_SHIFT		8
···
 struct unicode_map {
 	unsigned int version;
 	const struct utf8data *ntab[UTF8_NMAX];
+	const struct utf8data_table *tables;
 };
 
 int utf8_validate(const struct unicode_map *um, const struct qstr *str);