From f4e3f15b44f0d0e117781d194f700cc19b3a88c7 Mon Sep 17 00:00:00 2001 From: Hsuan Ting Chen Date: Thu, 6 Jul 2023 15:58:42 +0800 Subject: lib: Introduce new parsing rules for ux_locales.c Introduce new parsing rules for ux_locales.c:ux_locales_get_text(): * Add a version byte: PRERAM_LOCALES_VERSION_BYTE in the beginning. This provides more flexibility if we want to change the format of preram_locales region. * Add a new delimiter 0x01 between two string_names. This could fix the issue that 'string_name' and 'localized_string' might be the same. Also fix two bugs: 1. We would search for the language ID exceeding the range of current string_name. 2. In 'move_next()', we would exceed the 'size' due to the unconditional increase of offset. Finally, make some minor improvements to some existing comments. BUG=b:264666392, b:289995591 BRANCH=brya TEST=emerge-brya coreboot chromeos-bootimage Signed-off-by: Hsuan Ting Chen Change-Id: Ic0916a0badd7071fa2c43ee9cfc76ca5e79dbf8f Reviewed-on: https://review.coreboot.org/c/coreboot/+/76320 Tested-by: build bot (Jenkins) Reviewed-by: Yu-Ping Wu --- src/lib/ux_locales.c | 91 ++++++++++++++++++++++++++++++++-------------------- 1 file changed, 57 insertions(+), 34 deletions(-) (limited to 'src') diff --git a/src/lib/ux_locales.c b/src/lib/ux_locales.c index 86126195b4..a2e38fc1dc 100644 --- a/src/lib/ux_locales.c +++ b/src/lib/ux_locales.c @@ -10,8 +10,15 @@ #define LANG_ID_MAX 100 #define LANG_ID_LEN 3 + +#define PRERAM_LOCALES_VERSION_BYTE 0x01 #define PRERAM_LOCALES_NAME "preram_locales" +/* We need different delimiters to deal with the case where 'string_name' is the same as + 'localized_string'. */ +#define DELIM_STR 0x00 +#define DELIM_NAME 0x01 + /* * Devices which support early vga have the capability to show localized text in * Code Page 437 encoding. (see src/drivers/pc80/vga/vga_font_8x16.c) @@ -19,9 +26,11 @@ * preram_locales located in CBFS is an uncompressed file located in either RO * or RW CBFS. It contains the localization information in the following format: * + * [PRERAM_LOCALES_VERSION_BYTE] * [string_name_1] [\x00] * [language_id_1] [\x00] [localized_string_1] [\x00] * [language_id_2] [\x00] [localized_string_2] [\x00] ... + * [\x01] * [string_name_2] [\x00] ... * * This file contains tools to locate the file and search for localized strings @@ -62,31 +71,36 @@ static void *locales_get_map(size_t *size_out, bool unmap) return cached_state.data; } -/* Move to the next string in the data. Strings are separated by 0x00. */ -static size_t move_next(const char *data, size_t offset, size_t size) +/* Move to the next string in the data. Strings are separated by delim. */ +static size_t move_next(const char *data, size_t offset, size_t size, char delim) { - if (offset >= size) - return size; - while (offset < size && data[offset] != '\0') + while (offset < size && data[offset] != delim) + offset++; + /* If we found delim, move to the start of the next string. */ + if (offset < size) offset++; - offset++; return offset; } -/* Find the next occurrence of the specific string. */ +/* Find the next occurrence of the specific string. Strings are separated by delim. */ static size_t search_for(const char *data, size_t offset, size_t size, - const char *str) + const char *str, char delim) { - if (offset >= size) - return size; while (offset < size) { if (!strncmp(data + offset, str, size - offset)) return offset; - offset = move_next(data, offset, size); + offset = move_next(data, offset, size, delim); } return size; } +/* Find the next occurrence of the string_name, which should always follow a DELIM_NAME. */ +static inline size_t search_for_name(const char *data, size_t offset, size_t size, + const char *name) +{ + return search_for(data, offset, size, name, DELIM_NAME); +} + /* Find the next occurrence of the integer ID, where ID is less than 100. */ static size_t search_for_id(const char *data, size_t offset, size_t size, int id) @@ -95,17 +109,18 @@ static size_t search_for_id(const char *data, size_t offset, size_t size, return offset; char int_to_str[LANG_ID_LEN] = {}; snprintf(int_to_str, LANG_ID_LEN, "%d", id); - return search_for(data, offset, size, int_to_str); + return search_for(data, offset, size, int_to_str, DELIM_STR); } const char *ux_locales_get_text(const char *name) { const char *data; - size_t size, offset, name_offset, next; + size_t size, offset, name_offset, next_name_offset, next; uint32_t lang_id; + unsigned char version; data = locales_get_map(&size, false); - if (!data) { + if (!data || size == 0) { printk(BIOS_ERR, "%s: %s not found.\n", __func__, PRERAM_LOCALES_NAME); return NULL; @@ -123,40 +138,48 @@ const char *ux_locales_get_text(const char *name) printk(BIOS_INFO, "%s: Search for %s with language ID: %u\n", __func__, name, lang_id); - /* Search for name. */ - offset = search_for(data, 0, size, name); + /* Check if the version byte is the expected version. */ + version = (unsigned char)data[0]; + if (version != PRERAM_LOCALES_VERSION_BYTE) { + printk(BIOS_ERR, "%s: The version %u is not the expected one %u\n", + __func__, version, PRERAM_LOCALES_VERSION_BYTE); + return NULL; + } + + /* Search for name. Skip the version byte. */ + offset = search_for_name(data, 1, size, name); if (offset >= size) { printk(BIOS_ERR, "%s: Name %s not found.\n", __func__, name); return NULL; } name_offset = offset; - /* Search for language ID. */ - offset = search_for_id(data, name_offset, size, lang_id); - /* Language ID not supported; fallback to English if the current - * language is not English (0). */ - if (offset >= size) { - /* - * Since we only support a limited charset, it is very normal - * that a language is not supported and we fallback here - * silently. - */ + /* Search for language ID. We should not search beyond the range of the current + string_name. */ + next_name_offset = move_next(data, offset, size, DELIM_NAME); + assert(next_name_offset <= size); + offset = search_for_id(data, name_offset, next_name_offset, lang_id); + /* Language ID not supported; fallback to English if the current language is not + English (0). */ + if (offset >= next_name_offset) { + /* Since we only support a limited charset, it is very normal that a language + is not supported and we fallback here silently. */ if (lang_id != 0) - offset = search_for_id(data, name_offset, size, 0); - if (offset >= size) { - printk(BIOS_ERR, "%s: Neither %d nor 0 found.\n", - __func__, lang_id); + offset = search_for_id(data, name_offset, next_name_offset, 0); + if (offset >= next_name_offset) { + printk(BIOS_ERR, "%s: Neither %d nor 0 found.\n", __func__, lang_id); return NULL; } } - offset = move_next(data, offset, size); - if (offset >= size) + /* Move to the corresponding localized_string. */ + offset = move_next(data, offset, next_name_offset, DELIM_STR); + if (offset >= next_name_offset) return NULL; /* Validity check that the returned string must be NULL terminated. */ - next = move_next(data, offset, size) - 1; - if (next >= size || data[next] != '\0') { + next = move_next(data, offset, next_name_offset, DELIM_STR) - 1; + if (next >= next_name_offset || data[next] != '\0') { printk(BIOS_ERR, "%s: %s is not NULL terminated.\n", __func__, PRERAM_LOCALES_NAME); return NULL; -- cgit v1.2.3