* [PATCH v3 0/5] Add support for the SquashFS filesystem
@ 2020-07-21 9:22 Joao Marcos Costa
2020-07-21 9:22 ` [PATCH v3 1/5] fs/squashfs: new filesystem Joao Marcos Costa
` (5 more replies)
0 siblings, 6 replies; 15+ messages in thread
From: Joao Marcos Costa @ 2020-07-21 9:22 UTC (permalink / raw)
To: u-boot
Hello!
This series adds support for the SquashFS filesystem. For now, zlib is the
only supported compression type. This is my first contribution to U-Boot as
well as to a major Open Source project.
Changes in v3:
- Replace CONFIG_IS_ENABLED by IS_ENABLED in fs/fs.c
Changes in v2:
- Replace sqfs_ls() by U-Boot's fs_ls_generic()
- Add info. to MAINTAINERS
- Fix build failures
- Fix style problems
Best regards,
Joao Marcos Costa
Joao Marcos Costa (5):
fs/squashfs: new filesystem
fs/squashfs: add filesystem commands
include/u-boot, lib/zlib: add sources for zlib decompression
fs/squashfs: add support for zlib decompression
fs/fs.c: add symbolic link case to fs_ls_generic()
MAINTAINERS | 7 +
cmd/Kconfig | 6 +
cmd/Makefile | 1 +
cmd/sqfs.c | 42 +
common/spl/Kconfig | 9 +
fs/Kconfig | 2 +
fs/Makefile | 2 +
fs/fs.c | 18 +
fs/squashfs/Kconfig | 10 +
fs/squashfs/Makefile | 7 +
fs/squashfs/sqfs.c | 1521 +++++++++++++++++++++++++++++++
fs/squashfs/sqfs_decompressor.c | 53 ++
fs/squashfs/sqfs_decompressor.h | 58 ++
fs/squashfs/sqfs_dir.c | 107 +++
fs/squashfs/sqfs_filesystem.h | 300 ++++++
fs/squashfs/sqfs_inode.c | 142 +++
fs/squashfs/sqfs_utils.h | 49 +
include/fs.h | 1 +
include/squashfs.h | 25 +
include/u-boot/zlib.h | 32 +
lib/zlib/uncompr.c | 97 ++
lib/zlib/zlib.c | 1 +
22 files changed, 2490 insertions(+)
create mode 100644 cmd/sqfs.c
create mode 100644 fs/squashfs/Kconfig
create mode 100644 fs/squashfs/Makefile
create mode 100644 fs/squashfs/sqfs.c
create mode 100644 fs/squashfs/sqfs_decompressor.c
create mode 100644 fs/squashfs/sqfs_decompressor.h
create mode 100644 fs/squashfs/sqfs_dir.c
create mode 100644 fs/squashfs/sqfs_filesystem.h
create mode 100644 fs/squashfs/sqfs_inode.c
create mode 100644 fs/squashfs/sqfs_utils.h
create mode 100644 include/squashfs.h
create mode 100644 lib/zlib/uncompr.c
--
2.17.1
^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH v3 1/5] fs/squashfs: new filesystem
2020-07-21 9:22 [PATCH v3 0/5] Add support for the SquashFS filesystem Joao Marcos Costa
@ 2020-07-21 9:22 ` Joao Marcos Costa
2020-07-21 9:22 ` [PATCH v3 2/5] fs/squashfs: add filesystem commands Joao Marcos Costa
` (4 subsequent siblings)
5 siblings, 0 replies; 15+ messages in thread
From: Joao Marcos Costa @ 2020-07-21 9:22 UTC (permalink / raw)
To: u-boot
Add support for SquashFS filesystem. Right now, it does not support
compression but support for zlib will be added in a follow-up commit.
Signed-off-by: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
---
Changes in v3:
- Replace CONFIG_IS_ENABLED by IS_ENABLED in fs/fs.c
Changes in v2:
- Add info. to MAINTAINERS
- Replace sqfs_ls() by U-Boot's fs_ls_generic()
- Improve string parsing when resolving symlink targets
- Replace cmd_tbl_t type by the cmd_tbl struct
- Add missing <command.h> in cmd/sqfs.c
- Fix build failures
- Replace "struct disk_partition_t" by "disk_partition"
- Remove CMD_RET* from the filesystem code
- Simplify superblock handling by adding "struct squashfs_ctxt", which
encloses the complete global context (superblock, cur_dev and
cur_part_info). The struct members are set up during probe() and cleaned
up during close().
MAINTAINERS | 6 +
common/spl/Kconfig | 9 +
fs/Kconfig | 2 +
fs/Makefile | 2 +
fs/fs.c | 15 +
fs/squashfs/Kconfig | 10 +
fs/squashfs/Makefile | 7 +
fs/squashfs/sqfs.c | 1521 +++++++++++++++++++++++++++++++
fs/squashfs/sqfs_decompressor.c | 29 +
fs/squashfs/sqfs_decompressor.h | 58 ++
fs/squashfs/sqfs_dir.c | 107 +++
fs/squashfs/sqfs_filesystem.h | 300 ++++++
fs/squashfs/sqfs_inode.c | 142 +++
fs/squashfs/sqfs_utils.h | 49 +
include/fs.h | 1 +
include/squashfs.h | 25 +
16 files changed, 2283 insertions(+)
create mode 100644 fs/squashfs/Kconfig
create mode 100644 fs/squashfs/Makefile
create mode 100644 fs/squashfs/sqfs.c
create mode 100644 fs/squashfs/sqfs_decompressor.c
create mode 100644 fs/squashfs/sqfs_decompressor.h
create mode 100644 fs/squashfs/sqfs_dir.c
create mode 100644 fs/squashfs/sqfs_filesystem.h
create mode 100644 fs/squashfs/sqfs_inode.c
create mode 100644 fs/squashfs/sqfs_utils.h
create mode 100644 include/squashfs.h
diff --git a/MAINTAINERS b/MAINTAINERS
index 6316c6ca00..198fe2b89e 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -944,6 +944,12 @@ S: Maintained
F: drivers/spmi/
F: include/spmi/
+SQUASHFS
+M: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
+S: Maintained
+F: fs/squashfs/
+F: include/sqfs.h
+
TDA19988 HDMI ENCODER
M: Liviu Dudau <liviu.dudau@foss.arm.com>
S: Maintained
diff --git a/common/spl/Kconfig b/common/spl/Kconfig
index d09e52e88b..d7a32ca40f 100644
--- a/common/spl/Kconfig
+++ b/common/spl/Kconfig
@@ -577,6 +577,15 @@ config SPL_FS_EXT4
filesystem from within SPL. Support for the underlying block
device (e.g. MMC or USB) must be enabled separately.
+config SPL_FS_SQUASHFS
+ bool "Support SquashFS filesystems"
+ select FS_SQUASHFS
+ help
+ Enable support for SquashFS filesystems with SPL. This permits
+ U-Boot (or Linux in Falcon mode) to be loaded from a SquashFS
+ filesystem from within SPL. Support for the underlying block
+ device (e.g. MMC or USB) must be enabled separately.
+
config SPL_FS_FAT
bool "Support FAT filesystems"
select FS_FAT
diff --git a/fs/Kconfig b/fs/Kconfig
index 1cb9831be8..620af7f044 100644
--- a/fs/Kconfig
+++ b/fs/Kconfig
@@ -22,4 +22,6 @@ source "fs/cramfs/Kconfig"
source "fs/yaffs2/Kconfig"
+source "fs/squashfs/Kconfig"
+
endmenu
diff --git a/fs/Makefile b/fs/Makefile
index 42e669c40c..937cbcf6e8 100644
--- a/fs/Makefile
+++ b/fs/Makefile
@@ -9,6 +9,7 @@ obj-$(CONFIG_FS_LOADER) += fs.o
obj-$(CONFIG_SPL_FS_FAT) += fat/
obj-$(CONFIG_SPL_FS_EXT4) += ext4/
obj-$(CONFIG_SPL_FS_CBFS) += cbfs/
+obj-$(CONFIG_SPL_FS_SQUASHFS) += squashfs/
else
obj-y += fs.o
@@ -23,5 +24,6 @@ obj-$(CONFIG_SANDBOX) += sandbox/
obj-$(CONFIG_CMD_UBIFS) += ubifs/
obj-$(CONFIG_YAFFS2) += yaffs2/
obj-$(CONFIG_CMD_ZFS) += zfs/
+obj-$(CONFIG_FS_SQUASHFS) += squashfs/
endif
obj-y += fs_internal.o
diff --git a/fs/fs.c b/fs/fs.c
index edd8adc21b..5b31a369f7 100644
--- a/fs/fs.c
+++ b/fs/fs.c
@@ -22,6 +22,7 @@
#include <div64.h>
#include <linux/math64.h>
#include <efi_loader.h>
+#include <squashfs.h>
DECLARE_GLOBAL_DATA_PTR;
@@ -276,6 +277,20 @@ static struct fstype_info fstypes[] = {
.mkdir = fs_mkdir_unsupported,
.ln = fs_ln_unsupported,
},
+#endif
+#if IS_ENABLED(CONFIG_FS_SQUASHFS)
+ {
+ .fstype = FS_TYPE_SQUASHFS,
+ .name = "squashfs",
+ .probe = sqfs_probe,
+ .opendir = sqfs_opendir,
+ .readdir = sqfs_readdir,
+ .ls = fs_ls_generic,
+ .read = sqfs_read,
+ .size = sqfs_size,
+ .close = sqfs_close,
+ .closedir = sqfs_closedir,
+ },
#endif
{
.fstype = FS_TYPE_ANY,
diff --git a/fs/squashfs/Kconfig b/fs/squashfs/Kconfig
new file mode 100644
index 0000000000..b9772e5619
--- /dev/null
+++ b/fs/squashfs/Kconfig
@@ -0,0 +1,10 @@
+config FS_SQUASHFS
+ bool "Enable SquashFS filesystem support"
+ help
+ This provides support for reading images from SquashFS filesystem.
+ Squashfs is a compressed read-only filesystem for Linux.
+ It uses zlib, lz4, lzo, or xz compression to compress files, inodes
+ and directories. Squashfs is intended for general read-only
+ filesystem use, for archival use (i.e. in cases where a .tar.gz file
+ may be used), and in constrained block device/memory systems (e.g.
+ embedded systems) where low overhead is needed.
diff --git a/fs/squashfs/Makefile b/fs/squashfs/Makefile
new file mode 100644
index 0000000000..ba66ee821c
--- /dev/null
+++ b/fs/squashfs/Makefile
@@ -0,0 +1,7 @@
+# SPDX-License-Identifier: GPL-2.0+
+#
+
+obj-$(CONFIG_$(SPL_)FS_SQUASHFS) = sqfs.o \
+ sqfs_inode.o \
+ sqfs_dir.o \
+ sqfs_decompressor.o
diff --git a/fs/squashfs/sqfs.c b/fs/squashfs/sqfs.c
new file mode 100644
index 0000000000..76019444d8
--- /dev/null
+++ b/fs/squashfs/sqfs.c
@@ -0,0 +1,1521 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2020 Bootlin
+ *
+ * Author: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
+ *
+ * sqfs.c: SquashFS filesystem implementation
+ */
+
+#include <asm/unaligned.h>
+#include <errno.h>
+#include <fs.h>
+#include <linux/types.h>
+#include <linux/byteorder/little_endian.h>
+#include <linux/byteorder/generic.h>
+#include <memalign.h>
+#include <stdlib.h>
+#include <string.h>
+#include <squashfs.h>
+#include <part.h>
+
+#include "sqfs_decompressor.h"
+#include "sqfs_filesystem.h"
+#include "sqfs_utils.h"
+
+struct squashfs_ctxt {
+ struct disk_partition cur_part_info;
+ struct blk_desc *cur_dev;
+ struct squashfs_super_block *sblk;
+};
+
+static struct squashfs_ctxt ctxt;
+
+static int sqfs_disk_read(__u32 block, __u32 nr_blocks, void *buf)
+{
+ ulong ret;
+
+ if (!ctxt.cur_dev)
+ return -1;
+
+ ret = blk_dread(ctxt.cur_dev, ctxt.cur_part_info.start + block,
+ nr_blocks, buf);
+
+ if (ret != nr_blocks)
+ return -1;
+
+ return ret;
+}
+
+static int sqfs_read_sblk(struct squashfs_super_block **sblk)
+{
+ *sblk = malloc_cache_aligned(ctxt.cur_dev->blksz);
+ if (!*sblk)
+ return -ENOMEM;
+
+ if (sqfs_disk_read(0, 1, *sblk) != 1) {
+ free(*sblk);
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static int sqfs_count_tokens(const char *filename)
+{
+ int token_count = 1, l;
+
+ for (l = 1; l < strlen(filename); l++) {
+ if (filename[l] == '/')
+ token_count++;
+ }
+
+ /* Ignore trailing '/' in path */
+ if (filename[strlen(filename) - 1] == '/')
+ token_count--;
+
+ if (!token_count)
+ token_count = 1;
+
+ return token_count;
+}
+
+/*
+ * Calculates how many blocks are needed for the buffer used in sqfs_disk_read.
+ * The memory section (e.g. inode table) start offset and its end (i.e. the next
+ * table start) must be specified. It also calculates the offset from which to
+ * start reading the buffer.
+ */
+static int sqfs_calc_n_blks(__le64 start, __le64 end, u64 *offset)
+{
+ u64 start_, table_size;
+
+ table_size = le64_to_cpu(end) - le64_to_cpu(start);
+ start_ = le64_to_cpu(start) / ctxt.cur_dev->blksz;
+ *offset = le64_to_cpu(start) - (start_ * ctxt.cur_dev->blksz);
+
+ return DIV_ROUND_UP(table_size + *offset, ctxt.cur_dev->blksz);
+}
+
+/*
+ * Retrieves fragment block entry and returns true if the fragment block is
+ * compressed
+ */
+static int sqfs_frag_lookup(u32 inode_fragment_index,
+ struct squashfs_fragment_block_entry *e)
+{
+ u64 start, n_blks, src_len, table_offset, start_block;
+ unsigned char *metadata_buffer, *metadata, *table;
+ struct squashfs_fragment_block_entry *entries;
+ struct squashfs_super_block *sblk = ctxt.sblk;
+ unsigned long dest_len;
+ int block, offset, ret;
+ u16 header, comp_type;
+
+ comp_type = le16_to_cpu(ctxt.sblk->compression);
+
+ if (inode_fragment_index >= le32_to_cpu(sblk->fragments))
+ return -EINVAL;
+
+ start = le64_to_cpu(sblk->fragment_table_start) / ctxt.cur_dev->blksz;
+ n_blks = sqfs_calc_n_blks(sblk->fragment_table_start,
+ sblk->export_table_start, &table_offset);
+
+ /* Allocate a proper sized buffer to store the fragment index table */
+ table = malloc_cache_aligned(n_blks * ctxt.cur_dev->blksz);
+ if (!table)
+ return -ENOMEM;
+
+ if (sqfs_disk_read(start, n_blks, table) < 0) {
+ free(table);
+ return -EINVAL;
+ }
+
+ block = SQFS_FRAGMENT_INDEX(inode_fragment_index);
+ offset = SQFS_FRAGMENT_INDEX_OFFSET(inode_fragment_index);
+
+ /*
+ * Get the start offset of the metadata block that contains the right
+ * fragment block entry
+ */
+ start_block = get_unaligned((u64 *)&table[table_offset] + block);
+
+ start = start_block / ctxt.cur_dev->blksz;
+ n_blks = sqfs_calc_n_blks(cpu_to_le64(start_block),
+ sblk->fragment_table_start, &table_offset);
+
+ metadata_buffer = malloc_cache_aligned(n_blks * ctxt.cur_dev->blksz);
+ if (!metadata_buffer) {
+ ret = -ENOMEM;
+ goto free_table;
+ }
+
+ if (sqfs_disk_read(start, n_blks, metadata_buffer) < 0) {
+ ret = -EINVAL;
+ goto free_buffer;
+ }
+
+ /* Every metadata block starts with a 16-bit header */
+ header = get_unaligned((u16 *)&metadata_buffer[table_offset]);
+ metadata = &metadata_buffer[table_offset + SQFS_HEADER_SIZE];
+
+ entries = malloc(SQFS_METADATA_BLOCK_SIZE);
+ if (!entries) {
+ ret = -ENOMEM;
+ goto free_buffer;
+ }
+
+ if (SQFS_COMPRESSED_METADATA(header)) {
+ src_len = SQFS_METADATA_SIZE(header);
+ dest_len = SQFS_METADATA_BLOCK_SIZE;
+ ret = sqfs_decompress(comp_type, entries, &dest_len, metadata,
+ src_len);
+ if (ret) {
+ ret = -EINVAL;
+ goto free_entries;
+ }
+ } else {
+ memcpy(entries, metadata, dest_len);
+ }
+
+ *e = entries[offset];
+ ret = SQFS_COMPRESSED_BLOCK(e->size);
+
+free_entries:
+ free(entries);
+free_buffer:
+ free(metadata_buffer);
+free_table:
+ free(table);
+ return ret;
+}
+
+/*
+ * The entry name is a flexible array member, and we don't know its size before
+ * actually reading the entry. So we need a first copy to retrieve this size so
+ * we can finally copy the whole struct.
+ */
+static int sqfs_read_entry(struct squashfs_directory_entry **dest, void *src)
+{
+ struct squashfs_directory_entry tmp;
+
+ memcpy(&tmp, src, sizeof(tmp));
+ /*
+ * name_size property is actually the string length - 1, so adding 2
+ * compensates this difference and adds space for the trailling null
+ * byte.
+ */
+ *dest = malloc(sizeof(tmp) + tmp.name_size + 2);
+ if (!*dest)
+ return -ENOMEM;
+
+ memcpy(*dest, src, sizeof(tmp) + tmp.name_size + 1);
+ (*dest)->name[tmp.name_size + 1] = '\0';
+
+ return 0;
+}
+
+/* Takes a token list and returns a single string with '/' as separator. */
+static char *sqfs_concat_tokens(char **token_list, int token_count)
+{
+ char *result;
+ int i, length = 0, offset = 0;
+
+ for (i = 0; i < token_count; i++)
+ length += strlen(token_list[i]) + 1;
+
+ result = malloc(length + 1);
+ result[length] = '\0';
+
+ for (i = 0; i < token_count; i++) {
+ strcpy(result + offset, token_list[i]);
+ offset += strlen(token_list[i]);
+ result[offset++] = '/';
+ }
+
+ return result;
+}
+
+/*
+ * Differently from sqfs_concat_tokens, sqfs_join writes the result into a
+ * previously allocated string, and returns the number of bytes written.
+ */
+static int sqfs_join(char **strings, char *dest, int start, int end,
+ char separator)
+{
+ int i, offset = 0;
+
+ for (i = start; i < end; i++) {
+ strcpy(dest + offset, strings[i]);
+ offset += strlen(strings[i]);
+ if (i < end - 1)
+ dest[offset++] = separator;
+ }
+
+ return offset;
+}
+
+/*
+ * Fills the given token list using its size (count) and a source string (str)
+ */
+static int sqfs_tokenize(char **tokens, int count, const char *str)
+{
+ char *aux, *strc;
+ int i, j;
+
+ strc = strdup(str);
+ if (!strc)
+ return -ENOMEM;
+
+ /* Allocate and fill base tokens list */
+ if (!strcmp(strc, "/")) {
+ tokens[0] = strdup(strc);
+ if (!tokens[0]) {
+ free(strc);
+ return -ENOMEM;
+ }
+ } else {
+ for (j = 0; j < count; j++) {
+ aux = strtok(!j ? strc : NULL, "/");
+ tokens[j] = strdup(aux);
+ if (!tokens[j]) {
+ for (i = 0; i < j; i++)
+ free(tokens[i]);
+ free(strc);
+ return -ENOMEM;
+ }
+ }
+ }
+
+ free(strc);
+
+ return 0;
+}
+
+static int sqfs_get_tokens_length(char **tokens, int count)
+{
+ int length = 0, i;
+
+ /*
+ * 1 is added to the result of strlen to consider the slash separator
+ * between the tokens
+ */
+
+ for (i = 0; i < count; i++)
+ length += strlen(tokens[i]) + 1;
+
+ return length;
+}
+
+/*
+ * Remove last 'updir + 1' tokens from the base path tokens list. This leaves us
+ * with a token list containing only the tokens needed to form the resolved
+ * path, and returns the decremented size of the token list.
+ */
+static int sqfs_clean_base_path(char **base, int count, int updir)
+{
+ int i;
+
+ for (i = count - updir - 1; i < count; i++)
+ free(base[i]);
+
+ return count - updir - 1;
+}
+
+/*
+ * Given the base ("current dir.") path and the relative one, generate the
+ * absolute path.
+ */
+static char *sqfs_get_abs_path(const char *base, const char *rel)
+{
+ char **base_tokens, **rel_tokens, *resolved = NULL, *basec, *relc;
+ int ret, bc, rc, i, updir = 0, resolved_size = 0, offset = 0;
+
+ /* create copies of parameters */
+ basec = strdup(base);
+ if (!basec)
+ return NULL;
+
+ relc = strdup(rel);
+ if (!relc)
+ goto free_basec;
+
+ /* count tokens in paths */
+ bc = sqfs_count_tokens(basec);
+ rc = sqfs_count_tokens(relc);
+ if (bc < 1 || rc < 1)
+ goto free_relc;
+
+ base_tokens = malloc(bc * sizeof(char *));
+ if (!base_tokens)
+ goto free_relc;
+
+ rel_tokens = malloc(rc * sizeof(char *));
+ if (!rel_tokens)
+ goto free_b_tokens;
+
+ /* Allocate and fill tokens list */
+ ret = sqfs_tokenize(base_tokens, bc, basec);
+ if (ret)
+ goto free_r_tokens;
+
+ sqfs_tokenize(rel_tokens, rc, relc);
+ if (ret)
+ goto free_r_tokens;
+
+ /* count '..' occurrences in target path */
+ for (i = 0; i < rc; i++) {
+ if (!strcmp(rel_tokens[i], ".."))
+ updir++;
+ }
+
+ bc = sqfs_clean_base_path(base_tokens, bc, updir);
+ if (bc < 0)
+ goto free_r_tokens;
+
+ if (!bc)
+ resolved_size++;
+
+ resolved_size += sqfs_get_tokens_length(base_tokens, bc)
+ + sqfs_get_tokens_length(rel_tokens, rc);
+
+ resolved = malloc(resolved_size + 1);
+ if (!resolved)
+ goto free_r_tokens_loop;
+
+ memset(resolved, '\0', resolved_size + 1);
+ offset += sqfs_join(base_tokens, resolved + offset, 0, bc, '/');
+ resolved[offset++] = '/';
+ offset += sqfs_join(rel_tokens, resolved + offset, updir, rc, '/');
+
+free_r_tokens_loop:
+ for (i = 0; i < rc; i++)
+ free(rel_tokens[i]);
+ for (i = 0; i < bc; i++)
+ free(base_tokens[i]);
+free_r_tokens:
+ free(rel_tokens);
+free_b_tokens:
+ free(base_tokens);
+free_relc:
+ free(relc);
+free_basec:
+ free(basec);
+
+ return resolved;
+}
+
+static char *sqfs_resolve_symlink(struct squashfs_symlink_inode *sym,
+ unsigned char *table, const char *base_path)
+{
+ char *resolved, *target;
+
+ target = malloc(sym->symlink_size + 1);
+ if (!target)
+ return NULL;
+
+ target[sym->symlink_size] = '\0';
+ /* Get target name (relative path) */
+ strncpy(target, (char *)table + sizeof(*sym), sym->symlink_size);
+
+ /* Relative -> absolute path conversion */
+ resolved = sqfs_get_abs_path(base_path, target);
+
+ free(target);
+
+ return resolved;
+}
+
+/*
+ * m_list contains each metadata block's position, and m_count is the number of
+ * elements of m_list. Those metadata blocks come from the compressed directory
+ * table.
+ */
+static int sqfs_search_dir(struct squashfs_dir_stream *dirs, char **token_list,
+ int token_count, u32 *m_list, int m_count)
+{
+ char *path, *target, **sym_tokens;
+ int j, ret, new_inode_number, offset;
+ struct squashfs_super_block *sblk = ctxt.sblk;
+ struct squashfs_symlink_inode sym;
+ struct squashfs_ldir_inode ldir;
+ struct squashfs_dir_inode dir;
+ struct fs_dir_stream *dirsp;
+ struct fs_dirent *dent;
+ unsigned char *table;
+
+ dirsp = (struct fs_dir_stream *)dirs;
+
+ /* Start by root inode */
+ table = sqfs_find_inode(dirs->inode_table, le32_to_cpu(sblk->inodes),
+ sblk->inodes, sblk->block_size);
+
+ /* root is a regular directory, not an extended one */
+ memcpy(&dir, table, sizeof(dir));
+
+ /* get directory offset in directory table */
+ offset = sqfs_dir_offset(table, m_list, m_count);
+ dirs->table = &dirs->dir_table[offset];
+
+ /* Setup directory header */
+ dirs->dir_header = malloc(SQFS_DIR_HEADER_SIZE);
+ if (!dirs->dir_header)
+ return -ENOMEM;
+
+ memcpy(dirs->dir_header, dirs->table, SQFS_DIR_HEADER_SIZE);
+
+ /* Initialize squashfs_dir_stream members */
+ dirs->table += SQFS_DIR_HEADER_SIZE;
+ dirs->size = le16_to_cpu(dir.file_size) - SQFS_DIR_HEADER_SIZE;
+ dirs->entry_count = dirs->dir_header->count + 1;
+
+ /* No path given -> root directory */
+ if (!strcmp(token_list[0], "/")) {
+ dirs->table = &dirs->dir_table[offset];
+ memcpy(&dirs->i_dir, &dir, sizeof(dir));
+ return 0;
+ }
+
+ for (j = 0; j < token_count; j++) {
+ if (!sqfs_is_dir(dir.inode_type)) {
+ printf("** Cannot find directory. **\n");
+ return -EINVAL;
+ }
+
+ while (!sqfs_readdir(dirsp, &dent)) {
+ ret = strcmp(dent->name, token_list[j]);
+ if (!ret)
+ break;
+ free(dirs->entry);
+ }
+
+ if (ret) {
+ printf("** Cannot find directory. **\n");
+ free(dirs->entry);
+ return -EINVAL;
+ }
+
+ /* Redefine inode as the found token */
+ new_inode_number = dirs->entry->inode_offset +
+ dirs->dir_header->inode_number;
+
+ /* Get reference to inode in the inode table */
+ table = sqfs_find_inode(dirs->inode_table, new_inode_number,
+ sblk->inodes, sblk->block_size);
+ memcpy(&dir, table, sizeof(dir));
+
+ /* Check for symbolic link and inode type sanity */
+ if (le16_to_cpu(dir.inode_type) == SQFS_SYMLINK_TYPE) {
+ memcpy(&sym, table, sizeof(sym));
+ path = sqfs_concat_tokens(token_list, token_count);
+ target = sqfs_resolve_symlink(&sym, table, path);
+ token_count = sqfs_count_tokens(target);
+ if (token_count < 0)
+ return -EINVAL;
+
+ sym_tokens = malloc(token_count * sizeof(char *));
+ if (!sym_tokens)
+ return -EINVAL;
+
+ /* Allocate and fill tokens list */
+ ret = sqfs_tokenize(sym_tokens, token_count, target);
+ if (ret)
+ return -EINVAL;
+ free(dirs->entry);
+
+ ret = sqfs_search_dir(dirs, sym_tokens, token_count,
+ m_list, m_count);
+ return ret;
+ } else if (!sqfs_is_dir(dir.inode_type)) {
+ printf("** Cannot find directory. **\n");
+ free(dirs->entry);
+ return -EINVAL;
+ }
+
+ /* Check if it is an extended dir. */
+ if (dir.inode_type == SQFS_LDIR_TYPE)
+ memcpy(&ldir, table, sizeof(ldir));
+
+ /* Get dir. offset into the directory table */
+ offset = sqfs_dir_offset(table, m_list, m_count);
+ dirs->table = &dirs->dir_table[offset];
+
+ /* Copy directory header */
+ memcpy(dirs->dir_header, &dirs->dir_table[offset],
+ SQFS_DIR_HEADER_SIZE);
+
+ /* Check for empty directory */
+ if (sqfs_is_empty_dir(table)) {
+ printf("Empty directory.\n");
+ free(dirs->entry);
+ return SQFS_EMPTY_DIR;
+ }
+
+ dirs->table += SQFS_DIR_HEADER_SIZE;
+ dirs->size = le16_to_cpu(dir.file_size);
+ dirs->entry_count = dirs->dir_header->count + 1;
+ dirs->size -= SQFS_DIR_HEADER_SIZE;
+ free(dirs->entry);
+ }
+
+ offset = sqfs_dir_offset(table, m_list, m_count);
+ dirs->table = &dirs->dir_table[offset];
+
+ if (dir.inode_type == SQFS_DIR_TYPE)
+ memcpy(&dirs->i_dir, &dir, sizeof(dir));
+ else
+ memcpy(&dirs->i_ldir, &ldir, sizeof(ldir));
+
+ return 0;
+}
+
+/*
+ * Inode and directory tables are stored as a series of metadata blocks, and
+ * given the compressed size of this table, we can calculate how much metadata
+ * blocks are needed to store the result of the decompression, since a
+ * decompressed metadata block should have a size of 8KiB.
+ */
+static int sqfs_count_metablks(void *table, u32 offset, int table_size)
+{
+ int count = 0, cur_size = 0, ret;
+ u32 data_size;
+ bool comp;
+
+ do {
+ ret = sqfs_read_metablock(table, offset + cur_size, &comp,
+ &data_size);
+ if (ret)
+ return -EINVAL;
+ cur_size += data_size + SQFS_HEADER_SIZE;
+ count++;
+ } while (cur_size < table_size);
+
+ return count;
+}
+
+/*
+ * Storing the metadata blocks header's positions will be useful while looking
+ * for an entry in the directory table, using the reference (index and offset)
+ * given by its inode.
+ */
+static int sqfs_get_metablk_pos(u32 *pos_list, void *table, u32 offset,
+ int metablks_count)
+{
+ u32 data_size, cur_size = 0;
+ int j, ret = 0;
+ bool comp;
+
+ if (!metablks_count)
+ return -EINVAL;
+
+ for (j = 0; j < metablks_count; j++) {
+ ret = sqfs_read_metablock(table, offset + cur_size, &comp,
+ &data_size);
+ if (ret)
+ return -EINVAL;
+
+ cur_size += data_size + SQFS_HEADER_SIZE;
+ pos_list[j] = cur_size;
+ }
+
+ return ret;
+}
+
+static int sqfs_read_inode_table(unsigned char **inode_table)
+{
+ u64 start, n_blks, table_offset, table_size;
+ int j, ret = 0, metablks_count, comp_type;
+ struct squashfs_super_block *sblk = ctxt.sblk;
+ unsigned char *src_table, *itb;
+ u32 src_len, dest_offset = 0;
+ unsigned long dest_len;
+ bool compressed;
+
+ comp_type = sblk->compression;
+ table_size = le64_to_cpu(sblk->directory_table_start -
+ sblk->inode_table_start);
+ start = sblk->inode_table_start / ctxt.cur_dev->blksz;
+ n_blks = sqfs_calc_n_blks(sblk->inode_table_start,
+ sblk->directory_table_start, &table_offset);
+
+ /* Allocate a proper sized buffer (itb) to store the inode table */
+ itb = malloc_cache_aligned(n_blks * ctxt.cur_dev->blksz);
+ if (!itb)
+ return -ENOMEM;
+
+ if (sqfs_disk_read(start, n_blks, itb) < 0) {
+ ret = -EINVAL;
+ goto free_itb;
+ }
+
+ /* Parse inode table (metadata block) header */
+ ret = sqfs_read_metablock(itb, table_offset, &compressed, &src_len);
+ if (ret) {
+ ret = -EINVAL;
+ goto free_itb;
+ }
+
+ /* Calculate size to store the whole decompressed table */
+ metablks_count = sqfs_count_metablks(itb, table_offset, table_size);
+ if (metablks_count < 1) {
+ ret = -EINVAL;
+ goto free_itb;
+ }
+
+ *inode_table = malloc(metablks_count * SQFS_METADATA_BLOCK_SIZE);
+ if (!*inode_table) {
+ ret = -ENOMEM;
+ goto free_itb;
+ }
+
+ src_table = itb + table_offset + SQFS_HEADER_SIZE;
+
+ /* Extract compressed Inode table */
+ for (j = 0; j < metablks_count; j++) {
+ sqfs_read_metablock(itb, table_offset, &compressed, &src_len);
+ if (compressed) {
+ dest_len = SQFS_METADATA_BLOCK_SIZE;
+ ret = sqfs_decompress(comp_type, *inode_table +
+ dest_offset, &dest_len,
+ src_table, src_len);
+ if (ret) {
+ free(inode_table);
+ goto free_itb;
+ }
+
+ } else {
+ memcpy(*inode_table + (j * SQFS_METADATA_BLOCK_SIZE),
+ src_table, src_len);
+ }
+
+ /*
+ * Offsets to the decompression destination, to the metadata
+ * buffer 'itb' and to the decompression source, respectively.
+ */
+ dest_offset += dest_len;
+ table_offset += src_len + SQFS_HEADER_SIZE;
+ src_table += src_len + SQFS_HEADER_SIZE;
+ }
+
+free_itb:
+ free(itb);
+
+ return ret;
+}
+
+static int sqfs_read_directory_table(unsigned char **dir_table, u32 **pos_list)
+{
+ u64 start, n_blks, table_offset, table_size;
+ int j, ret = 0, metablks_count = -1, comp_type;
+ struct squashfs_super_block *sblk = ctxt.sblk;
+ unsigned char *src_table, *dtb;
+ u32 src_len, dest_offset = 0;
+ unsigned long dest_len;
+ bool compressed;
+
+ comp_type = sblk->compression;
+
+ /* DIRECTORY TABLE */
+ table_size = le64_to_cpu(sblk->fragment_table_start -
+ sblk->directory_table_start);
+ start = sblk->directory_table_start / ctxt.cur_dev->blksz;
+ n_blks = sqfs_calc_n_blks(sblk->directory_table_start,
+ sblk->fragment_table_start, &table_offset);
+
+ /* Allocate a proper sized buffer (dtb) to store the directory table */
+ dtb = malloc_cache_aligned(n_blks * ctxt.cur_dev->blksz);
+ if (!dtb)
+ return -ENOMEM;
+
+ if (sqfs_disk_read(start, n_blks, dtb) < 0)
+ goto free_dtb;
+
+ /* Parse directory table (metadata block) header */
+ ret = sqfs_read_metablock(dtb, table_offset, &compressed, &src_len);
+ if (ret)
+ goto free_dtb;
+
+ /* Calculate total size to store the whole decompressed table */
+ metablks_count = sqfs_count_metablks(dtb, table_offset, table_size);
+ if (metablks_count < 1)
+ goto free_dtb;
+
+ *dir_table = malloc(metablks_count * SQFS_METADATA_BLOCK_SIZE);
+ if (!*dir_table)
+ goto free_dtb;
+
+ *pos_list = malloc(metablks_count * sizeof(u32));
+ if (!*pos_list) {
+ free(*dir_table);
+ goto free_dtb;
+ }
+
+ ret = sqfs_get_metablk_pos(*pos_list, dtb, table_offset,
+ metablks_count);
+ if (ret) {
+ metablks_count = -1;
+ free(*dir_table);
+ free(*pos_list);
+ goto free_dtb;
+ }
+
+ src_table = dtb + table_offset + SQFS_HEADER_SIZE;
+
+ /* Extract compressed Directory table */
+ dest_offset = 0;
+ for (j = 0; j < metablks_count; j++) {
+ sqfs_read_metablock(dtb, table_offset, &compressed, &src_len);
+ if (compressed) {
+ dest_len = SQFS_METADATA_BLOCK_SIZE;
+ ret = sqfs_decompress(comp_type, *dir_table +
+ (j * SQFS_METADATA_BLOCK_SIZE),
+ &dest_len, src_table, src_len);
+ if (ret) {
+ metablks_count = -1;
+ free(*dir_table);
+ goto free_dtb;
+ }
+
+ if (dest_len < SQFS_METADATA_BLOCK_SIZE) {
+ dest_offset += dest_len;
+ break;
+ }
+ } else {
+ memcpy(*dir_table + (j * SQFS_METADATA_BLOCK_SIZE),
+ src_table, src_len);
+ }
+
+ /*
+ * Offsets to the decompression destination, to the metadata
+ * buffer 'dtb' and to the decompression source, respectively.
+ */
+ dest_offset += dest_len;
+ table_offset += src_len + SQFS_HEADER_SIZE;
+ src_table += src_len + SQFS_HEADER_SIZE;
+ }
+
+free_dtb:
+ free(dtb);
+
+ return metablks_count;
+}
+
+int sqfs_opendir(const char *filename, struct fs_dir_stream **dirsp)
+{
+ unsigned char *inode_table = NULL, *dir_table = NULL;
+ int j, token_count, ret = 0, metablks_count;
+ struct squashfs_dir_stream *dirs;
+ char **token_list, *path;
+ u32 *pos_list = NULL;
+
+ dirs = malloc(sizeof(*dirs));
+ if (!dirs)
+ return -EINVAL;
+
+ ret = sqfs_read_inode_table(&inode_table);
+ if (ret)
+ return -EINVAL;
+
+ metablks_count = sqfs_read_directory_table(&dir_table, &pos_list);
+ if (metablks_count < 1)
+ return -EINVAL;
+
+ /* Tokenize filename */
+ token_count = sqfs_count_tokens(filename);
+ if (token_count < 0)
+ return -EINVAL;
+
+ path = strdup(filename);
+ if (!path)
+ return -ENOMEM;
+
+ token_list = malloc(token_count * sizeof(char *));
+ if (!token_list) {
+ ret = -EINVAL;
+ goto free_path;
+ }
+
+ /* Allocate and fill tokens list */
+ ret = sqfs_tokenize(token_list, token_count, path);
+ if (ret)
+ goto free_tokens;
+ /*
+ * ldir's (extended directory) size is greater than dir, so it works as
+ * a general solution for the malloc size, since 'i' is a union.
+ */
+ dirs->inode_table = inode_table;
+ dirs->dir_table = dir_table;
+ ret = sqfs_search_dir(dirs, token_list, token_count, pos_list,
+ metablks_count);
+ if (ret)
+ goto free_tokens;
+
+ if (le16_to_cpu(dirs->i_dir.inode_type) == SQFS_DIR_TYPE)
+ dirs->size = le16_to_cpu(dirs->i_dir.file_size);
+ else
+ dirs->size = le32_to_cpu(dirs->i_ldir.file_size);
+
+ /* Setup directory header */
+ memcpy(dirs->dir_header, dirs->table, SQFS_DIR_HEADER_SIZE);
+ dirs->entry_count = dirs->dir_header->count + 1;
+ dirs->size -= SQFS_DIR_HEADER_SIZE;
+
+ /* Setup entry */
+ dirs->entry = NULL;
+ dirs->table += SQFS_DIR_HEADER_SIZE;
+
+ *dirsp = (struct fs_dir_stream *)dirs;
+
+free_tokens:
+ for (j = 0; j < token_count; j++)
+ free(token_list[j]);
+ free(token_list);
+ free(pos_list);
+free_path:
+ free(path);
+
+ return ret;
+}
+
+int sqfs_readdir(struct fs_dir_stream *fs_dirs, struct fs_dirent **dentp)
+{
+ struct squashfs_dir_stream *dirs;
+ struct squashfs_lreg_inode lreg;
+ struct squashfs_base_inode base;
+ struct squashfs_reg_inode reg;
+ struct fs_dirent *dent;
+ int offset = 0, ret;
+
+ dirs = (struct squashfs_dir_stream *)fs_dirs;
+ if (!dirs->size) {
+ *dentp = NULL;
+ return -SQFS_STOP_READDIR;
+ }
+
+ dent = &dirs->dentp;
+
+ if (!dirs->entry_count) {
+ if (dirs->size > SQFS_DIR_HEADER_SIZE) {
+ dirs->size -= SQFS_DIR_HEADER_SIZE;
+ } else {
+ *dentp = NULL;
+ dirs->size = 0;
+ return -SQFS_STOP_READDIR;
+ }
+
+ if (dirs->size > SQFS_EMPTY_FILE_SIZE) {
+ /* Read follow-up (emitted) dir. header */
+ memcpy(dirs->dir_header, dirs->table,
+ SQFS_DIR_HEADER_SIZE);
+ dirs->entry_count = dirs->dir_header->count + 1;
+ ret = sqfs_read_entry(&dirs->entry, dirs->table +
+ SQFS_DIR_HEADER_SIZE);
+ if (ret)
+ return -SQFS_STOP_READDIR;
+
+ dirs->table += SQFS_DIR_HEADER_SIZE;
+ }
+ } else {
+ ret = sqfs_read_entry(&dirs->entry, dirs->table);
+ if (ret)
+ return -SQFS_STOP_READDIR;
+ }
+
+ memcpy(&base, &dirs->inode_table[dirs->entry->offset], sizeof(base));
+
+ /* Set entry type and size */
+ switch (dirs->entry->type) {
+ case SQFS_DIR_TYPE:
+ case SQFS_LDIR_TYPE:
+ dent->type = FS_DT_DIR;
+ break;
+ case SQFS_REG_TYPE:
+ case SQFS_LREG_TYPE:
+ /*
+ * Entries do not differentiate extended from regular types, so
+ * it needs to be verified manually.
+ */
+ if (base.inode_type == SQFS_LREG_TYPE) {
+ memcpy(&lreg, &dirs->inode_table[dirs->entry->offset],
+ sizeof(lreg));
+ dent->size = (loff_t)le64_to_cpu(lreg.file_size);
+ } else {
+ memcpy(®, &dirs->inode_table[dirs->entry->offset],
+ sizeof(reg));
+ dent->size = (loff_t)le32_to_cpu(reg.file_size);
+ }
+
+ dent->type = FS_DT_REG;
+ break;
+ case SQFS_BLKDEV_TYPE:
+ case SQFS_CHRDEV_TYPE:
+ case SQFS_LBLKDEV_TYPE:
+ case SQFS_LCHRDEV_TYPE:
+ case SQFS_FIFO_TYPE:
+ case SQFS_SOCKET_TYPE:
+ case SQFS_LFIFO_TYPE:
+ case SQFS_LSOCKET_TYPE:
+ dent->type = SQFS_MISC_ENTRY_TYPE;
+ break;
+ case SQFS_SYMLINK_TYPE:
+ case SQFS_LSYMLINK_TYPE:
+ dent->type = FS_DT_LNK;
+ break;
+ default:
+ /*
+ * This macro evaluates to 0, which usually means a successful
+ * execution, but in this case it returns 0 to stop the while
+ * loop.
+ */
+ return -SQFS_STOP_READDIR;
+ }
+
+ /* Set entry name */
+ strncpy(dent->name, dirs->entry->name, dirs->entry->name_size + 1);
+ dent->name[dirs->entry->name_size + 1] = '\0';
+
+ offset = dirs->entry->name_size + 1 + SQFS_ENTRY_BASE_LENGTH;
+ dirs->entry_count--;
+
+ /* Decrement size to be read */
+ if (dirs->size > offset)
+ dirs->size -= offset;
+ else
+ dirs->size = 0;
+
+ /* Keep a reference to the current entry before incrementing it */
+ dirs->table += offset;
+
+ *dentp = dent;
+
+ return 0;
+}
+
+int sqfs_probe(struct blk_desc *fs_dev_desc, struct disk_partition *fs_partition)
+{
+ struct squashfs_super_block *sblk;
+ int ret;
+
+ ctxt.cur_dev = fs_dev_desc;
+ ctxt.cur_part_info = *fs_partition;
+
+ ret = sqfs_read_sblk(&sblk);
+ if (ret)
+ return ret;
+
+ /* Make sure it has a valid SquashFS magic number*/
+ if (sblk->s_magic != SQFS_MAGIC_NUMBER) {
+ printf("Bad magic number for SquashFS image.\n");
+ ctxt.cur_dev = NULL;
+ return -EINVAL;
+ }
+
+ ctxt.sblk = sblk;
+
+ return 0;
+}
+
+static char *sqfs_basename(char *path)
+{
+ char *fname;
+
+ fname = path + strlen(path) - 1;
+ while (fname >= path) {
+ if (*fname == '/') {
+ fname++;
+ break;
+ }
+
+ fname--;
+ }
+
+ return fname;
+}
+
+static char *sqfs_dirname(char *path)
+{
+ char *fname;
+
+ fname = sqfs_basename(path);
+ --fname;
+ *fname = '\0';
+
+ return path;
+}
+
+/*
+ * Takes a path to file and splits it in two parts: the filename itself and the
+ * directory's path, e.g.:
+ * path: /path/to/file.txt
+ * file: file.txt
+ * dir: /path/to
+ */
+static int sqfs_split_path(char **file, char **dir, const char *path)
+{
+ char *dirc, *basec, *bname, *dname, *tmp_path;
+ int ret = 0;
+
+ /* check for first slash in path*/
+ if (path[0] == '/') {
+ tmp_path = strdup(path);
+ if (!tmp_path)
+ return -ENOMEM;
+ } else {
+ tmp_path = malloc(strlen(path) + 2);
+ if (!tmp_path)
+ return -ENOMEM;
+ tmp_path[0] = '/';
+ strcpy(tmp_path + 1, path);
+ }
+
+ /* String duplicates */
+ dirc = strdup(tmp_path);
+ if (!dirc) {
+ ret = -ENOMEM;
+ goto free_tmp;
+ }
+
+ basec = strdup(tmp_path);
+ if (!basec) {
+ ret = -ENOMEM;
+ goto free_dirc;
+ }
+
+ dname = sqfs_dirname(dirc);
+ bname = sqfs_basename(basec);
+
+ *file = strdup(bname);
+
+ if (!*file) {
+ ret = -ENOMEM;
+ goto free_basec;
+ }
+
+ if (*dname == '\0') {
+ *dir = malloc(2);
+ if (!*dir) {
+ ret = -ENOMEM;
+ goto free_basec;
+ }
+
+ (*dir)[0] = '/';
+ (*dir)[1] = '\0';
+ } else {
+ *dir = strdup(dname);
+ if (!*dir) {
+ ret = -ENOMEM;
+ goto free_basec;
+ }
+ }
+
+free_basec:
+ free(basec);
+free_dirc:
+ free(dirc);
+free_tmp:
+ free(tmp_path);
+
+ return ret;
+}
+
+static int sqfs_get_regfile_info(struct squashfs_reg_inode *reg,
+ struct squashfs_file_info *finfo,
+ struct squashfs_fragment_block_entry *fentry,
+ __le32 blksz)
+{
+ int datablk_count = 0, ret;
+
+ finfo->size = le32_to_cpu(reg->file_size);
+ finfo->offset = le32_to_cpu(reg->offset);
+ finfo->start = le32_to_cpu(reg->start_block);
+ finfo->frag = SQFS_IS_FRAGMENTED(le32_to_cpu(reg->fragment));
+
+ if (finfo->frag) {
+ datablk_count = finfo->size / le32_to_cpu(blksz);
+ ret = sqfs_frag_lookup(reg->fragment, fentry);
+ if (ret < 0)
+ return -EINVAL;
+ finfo->comp = true;
+ } else {
+ datablk_count = DIV_ROUND_UP(finfo->size, le32_to_cpu(blksz));
+ }
+
+ finfo->blk_sizes = malloc(datablk_count * sizeof(u32));
+ if (!finfo->blk_sizes)
+ return -ENOMEM;
+
+ return datablk_count;
+}
+
+static int sqfs_get_lregfile_info(struct squashfs_lreg_inode *lreg,
+ struct squashfs_file_info *finfo,
+ struct squashfs_fragment_block_entry *fentry,
+ __le32 blksz)
+{
+ int datablk_count = 0, ret;
+
+ finfo->size = le64_to_cpu(lreg->file_size);
+ finfo->offset = le32_to_cpu(lreg->offset);
+ finfo->start = le64_to_cpu(lreg->start_block);
+ finfo->frag = SQFS_IS_FRAGMENTED(le32_to_cpu(lreg->fragment));
+
+ if (finfo->frag) {
+ datablk_count = finfo->size / le32_to_cpu(blksz);
+ ret = sqfs_frag_lookup(lreg->fragment, fentry);
+ if (ret < 0)
+ return -EINVAL;
+ finfo->comp = true;
+ } else {
+ datablk_count = DIV_ROUND_UP(finfo->size, le32_to_cpu(blksz));
+ }
+
+ finfo->blk_sizes = malloc(datablk_count * sizeof(u32));
+ if (!finfo->blk_sizes)
+ return -ENOMEM;
+
+ return datablk_count;
+}
+
+int sqfs_read(const char *filename, void *buf, loff_t offset, loff_t len,
+ loff_t *actread)
+{
+ char *dir, *fragment_block, *datablock = NULL, *data_buffer = NULL;
+ char *fragment, *file, *resolved, *data;
+ u64 start, n_blks, table_size, data_offset, table_offset;
+ int ret, j, i_number, comp_type, datablk_count = 0;
+ struct squashfs_super_block *sblk = ctxt.sblk;
+ struct squashfs_fragment_block_entry frag_entry;
+ struct squashfs_file_info finfo = {0};
+ struct squashfs_symlink_inode symlink;
+ struct fs_dir_stream *dirsp = NULL;
+ struct squashfs_dir_stream *dirs;
+ struct squashfs_lreg_inode lreg;
+ struct squashfs_base_inode base;
+ struct squashfs_reg_inode reg;
+ unsigned long dest_len;
+ struct fs_dirent *dent;
+ unsigned char *ipos;
+
+ *actread = 0;
+
+ comp_type = le16_to_cpu(sblk->compression);
+
+ /*
+ * sqfs_opendir will uncompress inode and directory tables, and will
+ * return a pointer to the directory that contains the requested file.
+ */
+ sqfs_split_path(&file, &dir, filename);
+ ret = sqfs_opendir(dir, &dirsp);
+ if (ret) {
+ sqfs_closedir(dirsp);
+ goto free_paths;
+ }
+
+ dirs = (struct squashfs_dir_stream *)dirsp;
+
+ /* For now, only regular files are able to be loaded */
+ while (!sqfs_readdir(dirsp, &dent)) {
+ ret = strcmp(dent->name, file);
+ if (!ret)
+ break;
+
+ free(dirs->entry);
+ }
+
+ if (ret) {
+ printf("File not found.\n");
+ *actread = 0;
+ sqfs_closedir(dirsp);
+ ret = -ENOENT;
+ goto free_paths;
+ }
+
+ i_number = dirs->dir_header->inode_number + dirs->entry->inode_offset;
+ ipos = sqfs_find_inode(dirs->inode_table, i_number, sblk->inodes,
+ sblk->block_size);
+
+ memcpy(&base, ipos, sizeof(base));
+ switch (base.inode_type) {
+ case SQFS_REG_TYPE:
+ memcpy(®, ipos, sizeof(reg));
+ datablk_count = sqfs_get_regfile_info(®, &finfo, &frag_entry,
+ sblk->block_size);
+ if (datablk_count < 0) {
+ ret = -EINVAL;
+ goto free_paths;
+ }
+ memcpy(finfo.blk_sizes, ipos + sizeof(reg),
+ datablk_count * sizeof(u32));
+ break;
+ case SQFS_LREG_TYPE:
+ memcpy(&lreg, ipos, sizeof(lreg));
+ datablk_count = sqfs_get_lregfile_info(&lreg, &finfo,
+ &frag_entry,
+ sblk->block_size);
+ if (datablk_count < 0) {
+ ret = -EINVAL;
+ goto free_paths;
+ }
+ memcpy(finfo.blk_sizes, ipos + sizeof(lreg),
+ datablk_count * sizeof(u32));
+ break;
+ case SQFS_SYMLINK_TYPE:
+ case SQFS_LSYMLINK_TYPE:
+ memcpy(&symlink, ipos, sizeof(symlink));
+ resolved = sqfs_resolve_symlink(&symlink, ipos, filename);
+ ret = sqfs_read(resolved, buf, offset, len, actread);
+ free(resolved);
+ goto free_paths;
+ case SQFS_BLKDEV_TYPE:
+ case SQFS_CHRDEV_TYPE:
+ case SQFS_LBLKDEV_TYPE:
+ case SQFS_LCHRDEV_TYPE:
+ case SQFS_FIFO_TYPE:
+ case SQFS_SOCKET_TYPE:
+ case SQFS_LFIFO_TYPE:
+ case SQFS_LSOCKET_TYPE:
+ default:
+ printf("Unsupported entry type\n");
+ ret = -EINVAL;
+ goto free_paths;
+ }
+
+ /* If the user specifies a length, check its sanity */
+ if (len) {
+ if (len > finfo.size) {
+ ret = -EINVAL;
+ goto free_paths;
+ }
+
+ finfo.size = len;
+ }
+
+ if (datablk_count) {
+ data_offset = finfo.start;
+ datablock = malloc(sblk->block_size);
+ if (!datablock) {
+ ret = -ENOMEM;
+ goto free_paths;
+ }
+ }
+
+ for (j = 0; j < datablk_count; j++) {
+ start = data_offset / ctxt.cur_dev->blksz;
+ table_size = SQFS_BLOCK_SIZE(finfo.blk_sizes[j]);
+ table_offset = data_offset - (start * ctxt.cur_dev->blksz);
+ n_blks = DIV_ROUND_UP(table_size + table_offset,
+ ctxt.cur_dev->blksz);
+
+ data_buffer = malloc_cache_aligned(n_blks * ctxt.cur_dev->blksz);
+
+ if (!data_buffer) {
+ ret = -ENOMEM;
+ goto free_datablk;
+ }
+
+ ret = sqfs_disk_read(start, n_blks, data_buffer);
+ if (ret < 0) {
+ /*
+ * Possible causes: too many data blocks or too large
+ * SquashFS block size. Tip: re-compile the SquashFS
+ * image with mksquashfs's -b <block_size> option.
+ */
+ printf("Error: too many data blocks to be read.\n");
+ goto free_buffer;
+ }
+
+ data = data_buffer + table_offset;
+
+ /* Load the data */
+ if (SQFS_COMPRESSED_BLOCK(finfo.blk_sizes[j])) {
+ dest_len = sblk->block_size;
+ ret = sqfs_decompress(comp_type, datablock, &dest_len,
+ data, table_size);
+ if (ret)
+ goto free_buffer;
+
+ memcpy(buf + offset + *actread, datablock, dest_len);
+ *actread += dest_len;
+ } else {
+ memcpy(buf + offset + *actread, data, table_size);
+ *actread += table_size;
+ }
+
+ data_offset += table_size;
+ }
+
+ free(finfo.blk_sizes);
+
+ /*
+ * There is no need to continue if the file is not fragmented.
+ */
+ if (!finfo.frag) {
+ ret = 0;
+ goto free_buffer;
+ }
+
+ start = frag_entry.start / ctxt.cur_dev->blksz;
+ table_size = SQFS_BLOCK_SIZE(frag_entry.size);
+ table_offset = frag_entry.start - (start * ctxt.cur_dev->blksz);
+ n_blks = DIV_ROUND_UP(table_size + table_offset, ctxt.cur_dev->blksz);
+
+ fragment = malloc_cache_aligned(n_blks * ctxt.cur_dev->blksz);
+
+ if (!fragment) {
+ ret = -ENOMEM;
+ goto free_buffer;
+ }
+
+ ret = sqfs_disk_read(start, n_blks, fragment);
+ if (ret < 0)
+ goto free_fragment;
+
+ /* File compressed and fragmented */
+ if (finfo.frag && finfo.comp) {
+ dest_len = sblk->block_size;
+ fragment_block = malloc(sblk->block_size);
+ if (!fragment_block) {
+ ret = -ENOMEM;
+ goto free_fragment;
+ }
+
+ ret = sqfs_decompress(comp_type, fragment_block, &dest_len,
+ (void *)fragment + table_offset,
+ frag_entry.size);
+ if (ret) {
+ free(fragment_block);
+ goto free_fragment;
+ }
+
+ for (j = offset + *actread; j < finfo.size; j++) {
+ memcpy(buf + j, &fragment_block[finfo.offset + j], 1);
+ (*actread)++;
+ }
+
+ free(fragment_block);
+
+ } else if (finfo.frag && !finfo.comp) {
+ fragment_block = (void *)fragment + table_offset;
+
+ for (j = offset + *actread; j < finfo.size; j++) {
+ memcpy(buf + j, &fragment_block[finfo.offset + j], 1);
+ (*actread)++;
+ }
+ }
+
+free_fragment:
+ free(fragment);
+free_buffer:
+ if (datablk_count)
+ free(data_buffer);
+free_datablk:
+ if (datablk_count)
+ free(datablock);
+free_paths:
+ free(file);
+ free(dir);
+
+ return ret;
+}
+
+int sqfs_size(const char *filename, loff_t *size)
+{
+ struct squashfs_super_block *sblk = ctxt.sblk;
+ struct squashfs_symlink_inode symlink;
+ struct fs_dir_stream *dirsp = NULL;
+ struct squashfs_base_inode base;
+ struct squashfs_dir_stream *dirs;
+ struct squashfs_lreg_inode lreg;
+ struct squashfs_reg_inode reg;
+ char *dir, *file, *resolved;
+ struct fs_dirent *dent;
+ unsigned char *ipos;
+ int ret, i_number;
+
+ sqfs_split_path(&file, &dir, filename);
+ /*
+ * sqfs_opendir will uncompress inode and directory tables, and will
+ * return a pointer to the directory that contains the requested file.
+ */
+ ret = sqfs_opendir(dir, &dirsp);
+ if (ret) {
+ sqfs_closedir(dirsp);
+ goto free_strings;
+ }
+
+ dirs = (struct squashfs_dir_stream *)dirsp;
+
+ while (!sqfs_readdir(dirsp, &dent)) {
+ ret = strcmp(dent->name, file);
+ if (!ret)
+ break;
+ free(dirs->entry);
+ }
+
+ if (ret) {
+ printf("File not found.\n");
+ sqfs_closedir(dirsp);
+ goto free_strings;
+ }
+
+ i_number = dirs->dir_header->inode_number + dirs->entry->inode_offset;
+ ipos = sqfs_find_inode(dirs->inode_table, i_number, sblk->inodes,
+ sblk->block_size);
+ free(dirs->entry);
+
+ memcpy(&base, ipos, sizeof(base));
+ switch (base.inode_type) {
+ case SQFS_REG_TYPE:
+ memcpy(®, ipos, sizeof(reg));
+ *size = reg.file_size;
+ break;
+ case SQFS_LREG_TYPE:
+ memcpy(&lreg, ipos, sizeof(lreg));
+ *size = lreg.file_size;
+ break;
+ case SQFS_SYMLINK_TYPE:
+ case SQFS_LSYMLINK_TYPE:
+ memcpy(&symlink, ipos, sizeof(symlink));
+ resolved = sqfs_resolve_symlink(&symlink, ipos, filename);
+ ret = sqfs_size(resolved, size);
+ free(resolved);
+ break;
+ case SQFS_BLKDEV_TYPE:
+ case SQFS_CHRDEV_TYPE:
+ case SQFS_LBLKDEV_TYPE:
+ case SQFS_LCHRDEV_TYPE:
+ case SQFS_FIFO_TYPE:
+ case SQFS_SOCKET_TYPE:
+ case SQFS_LFIFO_TYPE:
+ case SQFS_LSOCKET_TYPE:
+ default:
+ printf("Unable to recover entry's size.\n");
+ *size = 0;
+ ret = -EINVAL;
+ break;
+ }
+
+free_strings:
+ free(dir);
+ free(file);
+
+ sqfs_closedir(dirsp);
+
+ return ret;
+}
+
+void sqfs_close(void)
+{
+ free(ctxt.sblk);
+ ctxt.cur_dev = NULL;
+}
+
+void sqfs_closedir(struct fs_dir_stream *dirs)
+{
+ struct squashfs_dir_stream *sqfs_dirs;
+
+ sqfs_dirs = (struct squashfs_dir_stream *)dirs;
+ free(sqfs_dirs->inode_table);
+ free(sqfs_dirs->dir_table);
+ free(sqfs_dirs->dir_header);
+}
diff --git a/fs/squashfs/sqfs_decompressor.c b/fs/squashfs/sqfs_decompressor.c
new file mode 100644
index 0000000000..a899a5704b
--- /dev/null
+++ b/fs/squashfs/sqfs_decompressor.c
@@ -0,0 +1,29 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2020 Bootlin
+ *
+ * Author: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
+ */
+
+#include <errno.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "sqfs_decompressor.h"
+#include "sqfs_filesystem.h"
+#include "sqfs_utils.h"
+
+int sqfs_decompress(u16 comp_type, void *dest, unsigned long *dest_len,
+ void *source, u32 lenp)
+{
+ int ret = 0;
+
+ switch (comp_type) {
+ default:
+ printf("Error: unknown compression type.\n");
+ return -EINVAL;
+ }
+
+ return ret;
+}
diff --git a/fs/squashfs/sqfs_decompressor.h b/fs/squashfs/sqfs_decompressor.h
new file mode 100644
index 0000000000..378965dda8
--- /dev/null
+++ b/fs/squashfs/sqfs_decompressor.h
@@ -0,0 +1,58 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2020 Bootlin
+ *
+ * Author: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
+ */
+
+#ifndef SQFS_DECOMPRESSOR_H
+#define SQFS_DECOMPRESSOR_H
+
+#include <stdint.h>
+
+#define SQFS_COMP_ZLIB 1
+#define SQFS_COMP_LZMA 2
+#define SQFS_COMP_LZO 3
+#define SQFS_COMP_XZ 4
+#define SQFS_COMP_LZ4 5
+#define SQFS_COMP_ZSTD 6
+
+/* LZMA does not support any compression options */
+
+struct squashfs_gzip_opts {
+ u32 compression_level;
+ u16 window_size;
+ u16 strategies;
+};
+
+struct squashfs_xz_opts {
+ u32 dictionary_size;
+ u32 executable_filters;
+};
+
+struct squashfs_lz4_opts {
+ u32 version;
+ u32 flags;
+};
+
+struct squashfs_zstd_opts {
+ u32 compression_level;
+};
+
+struct squashfs_lzo_opts {
+ u32 algorithm;
+ u32 level;
+};
+
+union squashfs_compression_opts {
+ struct squashfs_gzip_opts *gzip;
+ struct squashfs_xz_opts *xz;
+ struct squashfs_lz4_opts *lz4;
+ struct squashfs_zstd_opts *zstd;
+ struct squashfs_lzo_opts *lzo;
+};
+
+int sqfs_decompress(u16 comp_type, void *dest, unsigned long *dest_len,
+ void *source, u32 lenp);
+
+#endif /* SQFS_DECOMPRESSOR_H */
diff --git a/fs/squashfs/sqfs_dir.c b/fs/squashfs/sqfs_dir.c
new file mode 100644
index 0000000000..4a72191f88
--- /dev/null
+++ b/fs/squashfs/sqfs_dir.c
@@ -0,0 +1,107 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2020 Bootlin
+ *
+ * Author: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
+ */
+
+#include <errno.h>
+#include <linux/types.h>
+#include <linux/byteorder/little_endian.h>
+#include <linux/byteorder/generic.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "sqfs_filesystem.h"
+#include "sqfs_utils.h"
+
+bool sqfs_is_dir(__le16 type)
+{
+ return (le16_to_cpu(type) == SQFS_DIR_TYPE) ||
+ (le16_to_cpu(type) == SQFS_LDIR_TYPE);
+}
+
+/*
+ * Receives a pointer (void *) to a position in the inode table containing the
+ * directory's inode. Returns directory inode offset into the directory table.
+ * m_list contains each metadata block's position, and m_count is the number of
+ * elements of m_list. Those metadata blocks come from the compressed directory
+ * table.
+ */
+int sqfs_dir_offset(void *dir_i, u32 *m_list, int m_count)
+{
+ struct squashfs_base_inode base;
+ struct squashfs_ldir_inode ldir;
+ struct squashfs_dir_inode dir;
+ u32 start_block;
+ u16 offset;
+ int j;
+
+ memcpy(&base, dir_i, sizeof(base));
+
+ switch (base.inode_type) {
+ case SQFS_DIR_TYPE:
+ memcpy(&dir, dir_i, sizeof(dir));
+ start_block = le32_to_cpu(dir.start_block);
+ offset = le16_to_cpu(dir.offset);
+ break;
+ case SQFS_LDIR_TYPE:
+ memcpy(&ldir, dir_i, sizeof(ldir));
+ start_block = le32_to_cpu(ldir.start_block);
+ offset = le16_to_cpu(ldir.offset);
+ break;
+ default:
+ printf("Error: this is not a directory.\n");
+ return -EINVAL;
+ }
+
+ for (j = 0; j < m_count; j++) {
+ if (m_list[j] == start_block)
+ return (++j * SQFS_METADATA_BLOCK_SIZE) + offset;
+ }
+
+ if (start_block == 0)
+ return offset;
+
+ printf("Error: invalid inode reference to directory table.\n");
+
+ return -EINVAL;
+}
+
+bool sqfs_is_empty_dir(void *dir_i)
+{
+ struct squashfs_base_inode *base;
+ struct squashfs_ldir_inode *ldir;
+ struct squashfs_dir_inode *dir;
+ u32 file_size;
+
+ base = malloc(sizeof(*base));
+ if (!base)
+ return errno;
+
+ memcpy(base, dir_i, sizeof(*base));
+
+ switch (le16_to_cpu(base->inode_type)) {
+ case SQFS_DIR_TYPE:
+ dir = malloc(sizeof(*dir));
+ memcpy(dir, dir_i, sizeof(*dir));
+ file_size = le16_to_cpu(dir->file_size);
+ free(dir);
+ break;
+ case SQFS_LDIR_TYPE:
+ ldir = malloc(sizeof(*ldir));
+ memcpy(ldir, dir_i, sizeof(*ldir));
+ file_size = le32_to_cpu(ldir->file_size);
+ free(ldir);
+ break;
+ default:
+ printf("Error: this is not a directory.\n");
+ free(base);
+ return false;
+ }
+
+ free(base);
+
+ return file_size == SQFS_EMPTY_FILE_SIZE;
+}
diff --git a/fs/squashfs/sqfs_filesystem.h b/fs/squashfs/sqfs_filesystem.h
new file mode 100644
index 0000000000..cbdaa9f1e4
--- /dev/null
+++ b/fs/squashfs/sqfs_filesystem.h
@@ -0,0 +1,300 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2020 Bootlin
+ *
+ * Author: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
+ */
+
+#ifndef SQFS_FILESYSTEM_H
+#define SQFS_FILESYSTEM_H
+
+#include <asm/unaligned.h>
+#include <stdint.h>
+#include <fs.h>
+
+#define SQFS_UNCOMPRESSED_DATA 0x0002
+#define SQFS_MAGIC_NUMBER 0x73717368
+/* The three first members of squashfs_dir_index make a total of 12 bytes */
+#define SQFS_DIR_INDEX_BASE_LENGTH 12
+/* size of metadata (inode and directory) blocks */
+#define SQFS_METADATA_BLOCK_SIZE 8192
+/* Max. number of fragment entries in a metadata block is 512 */
+#define SQFS_MAX_ENTRIES 512
+/* Metadata blocks start by a 2-byte length header */
+#define SQFS_HEADER_SIZE 2
+#define SQFS_LREG_INODE_MIN_SIZE 56
+#define SQFS_DIR_HEADER_SIZE 12
+#define SQFS_MISC_ENTRY_TYPE -1
+#define SQFS_EMPTY_FILE_SIZE 3
+#define SQFS_STOP_READDIR 1
+#define SQFS_EMPTY_DIR -1
+/*
+ * A directory entry object has a fixed length of 8 bytes, corresponding to its
+ * first four members, plus the size of the entry name, which is equal to
+ * 'entry_name' + 1 bytes.
+ */
+#define SQFS_ENTRY_BASE_LENGTH 8
+/* Inode types */
+#define SQFS_DIR_TYPE 1
+#define SQFS_REG_TYPE 2
+#define SQFS_SYMLINK_TYPE 3
+#define SQFS_BLKDEV_TYPE 4
+#define SQFS_CHRDEV_TYPE 5
+#define SQFS_FIFO_TYPE 6
+#define SQFS_SOCKET_TYPE 7
+#define SQFS_LDIR_TYPE 8
+#define SQFS_LREG_TYPE 9
+#define SQFS_LSYMLINK_TYPE 10
+#define SQFS_LBLKDEV_TYPE 11
+#define SQFS_LCHRDEV_TYPE 12
+#define SQFS_LFIFO_TYPE 13
+#define SQFS_LSOCKET_TYPE 14
+
+struct squashfs_super_block {
+ __le32 s_magic;
+ __le32 inodes;
+ __le32 mkfs_time;
+ __le32 block_size;
+ __le32 fragments;
+ __le16 compression;
+ __le16 block_log;
+ __le16 flags;
+ __le16 no_ids;
+ __le16 s_major;
+ __le16 s_minor;
+ __le64 root_inode;
+ __le64 bytes_used;
+ __le64 id_table_start;
+ __le64 xattr_id_table_start;
+ __le64 inode_table_start;
+ __le64 directory_table_start;
+ __le64 fragment_table_start;
+ __le64 export_table_start;
+};
+
+struct squashfs_directory_index {
+ u32 index;
+ u32 start;
+ u32 size;
+ char name[0];
+};
+
+struct squashfs_base_inode {
+ __le16 inode_type;
+ __le16 mode;
+ __le16 uid;
+ __le16 guid;
+ __le32 mtime;
+ __le32 inode_number;
+};
+
+struct squashfs_ipc_inode {
+ __le16 inode_type;
+ __le16 mode;
+ __le16 uid;
+ __le16 guid;
+ __le32 mtime;
+ __le32 inode_number;
+ __le32 nlink;
+};
+
+struct squashfs_lipc_inode {
+ __le16 inode_type;
+ __le16 mode;
+ __le16 uid;
+ __le16 guid;
+ __le32 mtime;
+ __le32 inode_number;
+ __le32 nlink;
+ __le32 xattr;
+};
+
+struct squashfs_dev_inode {
+ __le16 inode_type;
+ __le16 mode;
+ __le16 uid;
+ __le16 guid;
+ __le32 mtime;
+ __le32 inode_number;
+ __le32 nlink;
+ __le32 rdev;
+};
+
+struct squashfs_ldev_inode {
+ __le16 inode_type;
+ __le16 mode;
+ __le16 uid;
+ __le16 guid;
+ __le32 mtime;
+ __le32 inode_number;
+ __le32 nlink;
+ __le32 rdev;
+ __le32 xattr;
+};
+
+struct squashfs_symlink_inode {
+ __le16 inode_type;
+ __le16 mode;
+ __le16 uid;
+ __le16 guid;
+ __le32 mtime;
+ __le32 inode_number;
+ __le32 nlink;
+ __le32 symlink_size;
+ char symlink[0];
+};
+
+struct squashfs_reg_inode {
+ __le16 inode_type;
+ __le16 mode;
+ __le16 uid;
+ __le16 guid;
+ __le32 mtime;
+ __le32 inode_number;
+ __le32 start_block;
+ __le32 fragment;
+ __le32 offset;
+ __le32 file_size;
+ __le32 block_list[0];
+};
+
+struct squashfs_lreg_inode {
+ __le16 inode_type;
+ __le16 mode;
+ __le16 uid;
+ __le16 guid;
+ __le32 mtime;
+ __le32 inode_number;
+ __le64 start_block;
+ __le64 file_size;
+ __le64 sparse;
+ __le32 nlink;
+ __le32 fragment;
+ __le32 offset;
+ __le32 xattr;
+ __le32 block_list[0];
+};
+
+struct squashfs_dir_inode {
+ __le16 inode_type;
+ __le16 mode;
+ __le16 uid;
+ __le16 guid;
+ __le32 mtime;
+ __le32 inode_number;
+ __le32 start_block;
+ __le32 nlink;
+ __le16 file_size;
+ __le16 offset;
+ __le32 parent_inode;
+};
+
+struct squashfs_ldir_inode {
+ __le16 inode_type;
+ __le16 mode;
+ __le16 uid;
+ __le16 guid;
+ __le32 mtime;
+ __le32 inode_number;
+ __le32 nlink;
+ __le32 file_size;
+ __le32 start_block;
+ __le32 parent_inode;
+ __le16 i_count;
+ __le16 offset;
+ __le32 xattr;
+ struct squashfs_directory_index index[0];
+};
+
+union squashfs_inode {
+ struct squashfs_base_inode *base;
+ struct squashfs_dev_inode *dev;
+ struct squashfs_ldev_inode *ldev;
+ struct squashfs_symlink_inode *symlink;
+ struct squashfs_reg_inode *reg;
+ struct squashfs_lreg_inode *lreg;
+ struct squashfs_dir_inode *dir;
+ struct squashfs_ldir_inode *ldir;
+ struct squashfs_ipc_inode *ipc;
+ struct squashfs_lipc_inode *lipc;
+};
+
+struct squashfs_directory_entry {
+ u16 offset;
+ u16 inode_offset;
+ u16 type;
+ u16 name_size;
+ char name[0];
+};
+
+struct squashfs_directory_header {
+ u32 count;
+ u32 start;
+ u32 inode_number;
+};
+
+struct squashfs_fragment_block_entry {
+ u64 start;
+ u32 size;
+ u32 _unused;
+};
+
+struct squashfs_dir_stream {
+ struct fs_dir_stream fs_dirs;
+ struct fs_dirent dentp;
+ /*
+ * 'size' is the uncompressed size of the entire listing, including
+ * headers. 'entry_count' is the number of entries following a
+ * specific header. Both variables are decremented in sqfs_readdir() so
+ * the function knows when the end of the directory is reached.
+ */
+ size_t size;
+ int entry_count;
+ /* SquashFS structures */
+ struct squashfs_directory_header *dir_header;
+ struct squashfs_directory_entry *entry;
+ /*
+ * 'table' points to a position into the directory table. Both 'table'
+ * and 'inode' are defined for the first time in sqfs_opendir().
+ * 'table's value changes in sqfs_readdir().
+ */
+ unsigned char *table;
+ union squashfs_inode i;
+ struct squashfs_dir_inode i_dir;
+ struct squashfs_ldir_inode i_ldir;
+ /*
+ * References to the tables' beginnings. They are assigned in
+ * sqfs_opendir() and freed in sqfs_closedir().
+ */
+ unsigned char *inode_table;
+ unsigned char *dir_table;
+};
+
+struct squashfs_file_info {
+ /* File size in bytes (uncompressed) */
+ size_t size;
+ /* Reference to list of data blocks's sizes */
+ u32 *blk_sizes;
+ /* Offset into the fragment block */
+ u32 offset;
+ /* Offset in which the data blocks begin */
+ u64 start;
+ /* Is file fragmented? */
+ bool frag;
+ /* Compressed fragment */
+ bool comp;
+};
+
+void *sqfs_find_inode(void *inode_table, int inode_number, __le32 inode_count,
+ __le32 block_size);
+
+int sqfs_dir_offset(void *dir_i, u32 *m_list, int m_count);
+
+int sqfs_read_metablock(unsigned char *file_mapping, int offset,
+ bool *compressed, u32 *data_size);
+
+bool sqfs_is_empty_dir(void *dir_i);
+
+bool sqfs_is_dir(__le16 type);
+
+#endif /* SQFS_FILESYSTEM_H */
diff --git a/fs/squashfs/sqfs_inode.c b/fs/squashfs/sqfs_inode.c
new file mode 100644
index 0000000000..dbde74a750
--- /dev/null
+++ b/fs/squashfs/sqfs_inode.c
@@ -0,0 +1,142 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2020 Bootlin
+ *
+ * Author: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
+ */
+
+#include <asm/unaligned.h>
+#include <errno.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "sqfs_decompressor.h"
+#include "sqfs_filesystem.h"
+#include "sqfs_utils.h"
+
+/*
+ * Given the uncompressed inode table, the inode to be found and the number of
+ * inodes in the table, return inode position in case of success.
+ */
+void *sqfs_find_inode(void *inode_table, int inode_number, __le32 inode_count,
+ __le32 block_size)
+{
+ int k, l, blk_list_size = 0, offset = 0, index_list_size = 0;
+ struct squashfs_symlink_inode symlink;
+ struct squashfs_base_inode base;
+ struct squashfs_ldev_inode ldev;
+ struct squashfs_lreg_inode lreg;
+ struct squashfs_ldir_inode ldir;
+ struct squashfs_lipc_inode lipc;
+ struct squashfs_dev_inode dev;
+ struct squashfs_reg_inode reg;
+ struct squashfs_dir_inode dir;
+ struct squashfs_ipc_inode ipc;
+
+ if (!inode_table) {
+ printf("%s: Invalid pointer to inode table.\n", __func__);
+ return NULL;
+ }
+
+ for (k = 0; k < le32_to_cpu(inode_count); k++) {
+ memcpy(&base, inode_table + offset, sizeof(base));
+ if (le32_to_cpu(base.inode_number) == inode_number)
+ return (void *)(inode_table + offset);
+
+ switch (base.inode_type) {
+ case SQFS_DIR_TYPE:
+ memcpy(&dir, inode_table + offset, sizeof(dir));
+ offset += sizeof(dir);
+ break;
+ case SQFS_REG_TYPE:
+ memcpy(®, inode_table + offset, sizeof(reg));
+ if (SQFS_IS_FRAGMENTED(reg.fragment)) {
+ blk_list_size = (reg.file_size / block_size);
+ } else {
+ blk_list_size = DIV_ROUND_UP(reg.file_size,
+ block_size);
+ }
+
+ offset += sizeof(reg) + blk_list_size * sizeof(u32);
+ break;
+ case SQFS_LDIR_TYPE:
+ memcpy(&ldir, inode_table + offset, sizeof(ldir));
+ if (ldir.i_count == 0) {
+ offset += sizeof(ldir);
+ break;
+ }
+
+ for (l = 0; l < ldir.i_count + 1; l++)
+ index_list_size += ldir.index[l].size + 1;
+
+ offset += sizeof(ldir) + index_list_size +
+ (ldir.i_count + 1) * SQFS_DIR_INDEX_BASE_LENGTH;
+ index_list_size = 0;
+ break;
+ case SQFS_LREG_TYPE:
+ memcpy(&lreg, inode_table + offset, sizeof(lreg));
+ if (lreg.fragment == 0xFFFFFFFF) {
+ blk_list_size = DIV_ROUND_UP(lreg.file_size,
+ block_size);
+ } else {
+ blk_list_size = (lreg.file_size / block_size);
+ }
+
+ offset += sizeof(lreg) + blk_list_size * sizeof(u32);
+ break;
+ case SQFS_SYMLINK_TYPE:
+ case SQFS_LSYMLINK_TYPE:
+ memcpy(&symlink, inode_table + offset, sizeof(symlink));
+ offset += sizeof(symlink) + symlink.symlink_size;
+ break;
+ case SQFS_BLKDEV_TYPE:
+ case SQFS_CHRDEV_TYPE:
+ memcpy(&dev, inode_table + offset, sizeof(dev));
+ offset += sizeof(dev);
+ break;
+ case SQFS_LBLKDEV_TYPE:
+ case SQFS_LCHRDEV_TYPE:
+ memcpy(&ldev, inode_table + offset, sizeof(ldev));
+ offset += sizeof(ldev);
+ break;
+ case SQFS_FIFO_TYPE:
+ case SQFS_SOCKET_TYPE:
+ memcpy(&ipc, inode_table + offset, sizeof(ipc));
+ offset += sizeof(ipc);
+ break;
+ case SQFS_LFIFO_TYPE:
+ case SQFS_LSOCKET_TYPE:
+ memcpy(&lipc, inode_table + offset, sizeof(lipc));
+ offset += sizeof(lipc);
+ break;
+ default:
+ printf("Error while searching inode: unknown type.\n");
+ return NULL;
+ }
+ }
+
+ printf("Inode not found.\n");
+
+ return NULL;
+}
+
+int sqfs_read_metablock(unsigned char *file_mapping, int offset,
+ bool *compressed, u32 *data_size)
+{
+ unsigned char *data;
+ u16 header;
+
+ data = file_mapping + offset;
+ header = get_unaligned((u16 *)data);
+ *compressed = SQFS_COMPRESSED_METADATA(header);
+ *data_size = SQFS_METADATA_SIZE(header);
+
+ if (*data_size > SQFS_METADATA_BLOCK_SIZE) {
+ printf("Invalid metatada block size: %d bytes.\n", *data_size);
+ return -EINVAL;
+ }
+
+ return 0;
+}
diff --git a/fs/squashfs/sqfs_utils.h b/fs/squashfs/sqfs_utils.h
new file mode 100644
index 0000000000..1260abe22b
--- /dev/null
+++ b/fs/squashfs/sqfs_utils.h
@@ -0,0 +1,49 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2020 Bootlin
+ *
+ * Author: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
+ */
+
+#ifndef SQFS_UTILS_H
+#define SQFS_UTILS_H
+
+#include <linux/bitops.h>
+#include <linux/kernel.h>
+#include <stdbool.h>
+
+#define SQFS_FRAGMENT_INDEX_OFFSET(A) ((A) % SQFS_MAX_ENTRIES)
+#define SQFS_FRAGMENT_INDEX(A) ((A) / SQFS_MAX_ENTRIES)
+#define SQFS_BLOCK_SIZE(A) ((A) & GENMASK(23, 0))
+#define SQFS_CHECK_FLAG(flag, bit) (((flag) >> (bit)) & 1)
+/* Useful for both fragment and data blocks */
+#define SQFS_COMPRESSED_BLOCK(A) (!((A) & BIT(24)))
+/* SQFS_COMPRESSED_DATA strictly used with super block's 'flags' member */
+#define SQFS_COMPRESSED_DATA(A) (!((A) & 0x0002))
+#define SQFS_IS_FRAGMENTED(A) ((A) != 0xFFFFFFFF)
+/*
+ * These two macros work as getters for a metada block header, retrieving the
+ * data size and if it is compressed/uncompressed
+ */
+#define SQFS_COMPRESSED_METADATA(A) (!((A) & BIT(15)))
+#define SQFS_METADATA_SIZE(A) ((A) & GENMASK(14, 0))
+
+struct squashfs_super_block_flags {
+ /* check: unused
+ * uncompressed_ids: not supported
+ */
+ bool uncompressed_inodes;
+ bool uncompressed_data;
+ bool check;
+ bool uncompressed_frags;
+ bool no_frags;
+ bool always_frags;
+ bool duplicates;
+ bool exportable;
+ bool uncompressed_xattrs;
+ bool no_xattrs;
+ bool compressor_options;
+ bool uncompressed_ids;
+};
+
+#endif /* SQFS_UTILS_H */
diff --git a/include/fs.h b/include/fs.h
index b08b1f40c5..0794b50d10 100644
--- a/include/fs.h
+++ b/include/fs.h
@@ -15,6 +15,7 @@ struct cmd_tbl;
#define FS_TYPE_SANDBOX 3
#define FS_TYPE_UBIFS 4
#define FS_TYPE_BTRFS 5
+#define FS_TYPE_SQUASHFS 6
struct blk_desc;
diff --git a/include/squashfs.h b/include/squashfs.h
new file mode 100644
index 0000000000..819cf8c2da
--- /dev/null
+++ b/include/squashfs.h
@@ -0,0 +1,25 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2020 Bootlin
+ *
+ * Author: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
+ *
+ * squashfs.h: SquashFS filesystem implementation.
+ */
+
+#ifndef _SQFS_H_
+#define _SQFS_H_
+
+struct disk_partition;
+
+int sqfs_opendir(const char *filename, struct fs_dir_stream **dirsp);
+int sqfs_readdir(struct fs_dir_stream *dirs, struct fs_dirent **dentp);
+int sqfs_probe(struct blk_desc *fs_dev_desc,
+ struct disk_partition *fs_partition);
+int sqfs_read(const char *filename, void *buf, loff_t offset,
+ loff_t len, loff_t *actread);
+int sqfs_size(const char *filename, loff_t *size);
+void sqfs_close(void);
+void sqfs_closedir(struct fs_dir_stream *dirs);
+
+#endif /* SQFS_H */
--
2.17.1
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [PATCH v3 2/5] fs/squashfs: add filesystem commands
2020-07-21 9:22 [PATCH v3 0/5] Add support for the SquashFS filesystem Joao Marcos Costa
2020-07-21 9:22 ` [PATCH v3 1/5] fs/squashfs: new filesystem Joao Marcos Costa
@ 2020-07-21 9:22 ` Joao Marcos Costa
2020-07-21 9:22 ` [PATCH v3 3/5] include/u-boot, lib/zlib: add sources for zlib decompression Joao Marcos Costa
` (3 subsequent siblings)
5 siblings, 0 replies; 15+ messages in thread
From: Joao Marcos Costa @ 2020-07-21 9:22 UTC (permalink / raw)
To: u-boot
Add 'ls' and 'load' commands.
Signed-off-by: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
---
Changes in v3:
- No changes since v2.
Changes in v2:
- Add info. to MAINTAINERS
- Add 'static' keyword to do_sqfs_*(). These functions' prototypes
are no longer needed in include/fs.h.
- Change commit message.
MAINTAINERS | 1 +
cmd/Kconfig | 6 ++++++
cmd/Makefile | 1 +
cmd/sqfs.c | 42 ++++++++++++++++++++++++++++++++++++++++++
4 files changed, 50 insertions(+)
create mode 100644 cmd/sqfs.c
diff --git a/MAINTAINERS b/MAINTAINERS
index 198fe2b89e..9a5f330148 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -949,6 +949,7 @@ M: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
S: Maintained
F: fs/squashfs/
F: include/sqfs.h
+F: cmd/sqfs.c
TDA19988 HDMI ENCODER
M: Liviu Dudau <liviu.dudau@foss.arm.com>
diff --git a/cmd/Kconfig b/cmd/Kconfig
index bfe6c163dc..c8f3d34b88 100644
--- a/cmd/Kconfig
+++ b/cmd/Kconfig
@@ -2059,6 +2059,12 @@ config CMD_FAT
help
Support for the FAT fs
+config CMD_SQUASHFS
+ bool "SquashFS command support"
+ select FS_SQUASHFS
+ help
+ Enables SquashFS filesystem commands (e.g. load, ls).
+
config CMD_FS_GENERIC
bool "filesystem commands"
help
diff --git a/cmd/Makefile b/cmd/Makefile
index 7952138dc2..ef2da1c928 100644
--- a/cmd/Makefile
+++ b/cmd/Makefile
@@ -62,6 +62,7 @@ obj-$(CONFIG_CMD_EXT4) += ext4.o
obj-$(CONFIG_CMD_EXT2) += ext2.o
obj-$(CONFIG_CMD_FAT) += fat.o
obj-$(CONFIG_CMD_FDT) += fdt.o
+obj-$(CONFIG_CMD_SQUASHFS) += sqfs.o
obj-$(CONFIG_CMD_FLASH) += flash.o
obj-$(CONFIG_CMD_FPGA) += fpga.o
obj-$(CONFIG_CMD_FPGAD) += fpgad.o
diff --git a/cmd/sqfs.c b/cmd/sqfs.c
new file mode 100644
index 0000000000..107038c4cf
--- /dev/null
+++ b/cmd/sqfs.c
@@ -0,0 +1,42 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2020 Bootlin
+ *
+ * Author: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
+ *
+ * squashfs.c: implements SquashFS related commands
+ */
+
+#include <command.h>
+#include <fs.h>
+#include <squashfs.h>
+
+static int do_sqfs_ls(struct cmd_tbl *cmdtp, int flag, int argc, char * const argv[])
+{
+ return do_ls(cmdtp, flag, argc, argv, FS_TYPE_SQUASHFS);
+}
+
+U_BOOT_CMD(sqfsls, 4, 1, do_sqfs_ls,
+ "List files in directory. Default: root (/).",
+ "<interface> [<dev[:part]>] [directory]\n"
+ " - list files from 'dev' on 'interface' in 'directory'\n"
+);
+
+static int do_sqfs_load(struct cmd_tbl *cmdtp, int flag, int argc, char * const argv[])
+{
+ return do_load(cmdtp, flag, argc, argv, FS_TYPE_SQUASHFS);
+}
+
+U_BOOT_CMD(sqfsload, 7, 0, do_sqfs_load,
+ "load binary file from a SquashFS filesystem",
+ "<interface> [<dev[:part]> [<addr> [<filename> [bytes [pos]]]]]\n"
+ " - Load binary file 'filename' from 'dev' on 'interface'\n"
+ " to address 'addr' from SquashFS filesystem.\n"
+ " 'pos' gives the file position to start loading from.\n"
+ " If 'pos' is omitted, 0 is used. 'pos' requires 'bytes'.\n"
+ " 'bytes' gives the size to load. If 'bytes' is 0 or omitted,\n"
+ " the load stops on end of file.\n"
+ " If either 'pos' or 'bytes' are not aligned to\n"
+ " ARCH_DMA_MINALIGN then a misaligned buffer warning will\n"
+ " be printed and performance will suffer for the load."
+);
--
2.17.1
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [PATCH v3 3/5] include/u-boot, lib/zlib: add sources for zlib decompression
2020-07-21 9:22 [PATCH v3 0/5] Add support for the SquashFS filesystem Joao Marcos Costa
2020-07-21 9:22 ` [PATCH v3 1/5] fs/squashfs: new filesystem Joao Marcos Costa
2020-07-21 9:22 ` [PATCH v3 2/5] fs/squashfs: add filesystem commands Joao Marcos Costa
@ 2020-07-21 9:22 ` Joao Marcos Costa
2020-07-29 15:27 ` Tom Rini
2020-07-21 9:22 ` [PATCH v3 4/5] fs/squashfs: add support " Joao Marcos Costa
` (2 subsequent siblings)
5 siblings, 1 reply; 15+ messages in thread
From: Joao Marcos Costa @ 2020-07-21 9:22 UTC (permalink / raw)
To: u-boot
Add zlib (v1.2.11) uncompr() function to U-Boot. SquashFS depends on
this function to decompress data from a raw disk image. The actual
support for zlib into SquashFS sources will be added in a follow-up
commit.
Signed-off-by: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
---
Changes in v3:
- No changes since v2.
Changes in v2:
- Changed commit title and message, which were wrong and/or misleading
in v1.
include/u-boot/zlib.h | 32 ++++++++++++++
lib/zlib/uncompr.c | 97 +++++++++++++++++++++++++++++++++++++++++++
lib/zlib/zlib.c | 1 +
3 files changed, 130 insertions(+)
create mode 100644 lib/zlib/uncompr.c
diff --git a/include/u-boot/zlib.h b/include/u-boot/zlib.h
index e23ceb50ca..a33cc8780d 100644
--- a/include/u-boot/zlib.h
+++ b/include/u-boot/zlib.h
@@ -110,6 +110,12 @@ extern "C" {
# define voidp z_voidp
#endif
+#if defined(ZLIB_CONST) && !defined(z_const)
+# define z_const const
+#else
+# define z_const
+#endif
+
#if defined(__MSDOS__) && !defined(MSDOS)
# define MSDOS
#endif
@@ -710,6 +716,32 @@ ZEXTERN uInt ZEXPORT crc32 OF((uInt crc, const Bytef *buf, uInt len));
if (crc != original_crc) error();
*/
+ZEXTERN int ZEXPORT uncompress OF((Bytef *dest, uLongf *destLen,
+ const Bytef *source, uLong sourceLen));
+/*
+ Decompresses the source buffer into the destination buffer. sourceLen is
+ the byte length of the source buffer. Upon entry, destLen is the total size
+ of the destination buffer, which must be large enough to hold the entire
+ uncompressed data. (The size of the uncompressed data must have been saved
+ previously by the compressor and transmitted to the decompressor by some
+ mechanism outside the scope of this compression library.) Upon exit, destLen
+ is the actual size of the uncompressed data.
+
+ uncompress returns Z_OK if success, Z_MEM_ERROR if there was not
+ enough memory, Z_BUF_ERROR if there was not enough room in the output
+ buffer, or Z_DATA_ERROR if the input data was corrupted or incomplete. In
+ the case where there is not enough room, uncompress() will fill the output
+ buffer with the uncompressed data up to that point.
+*/
+
+ZEXTERN int ZEXPORT uncompress2 OF((Bytef *dest, uLongf *destLen,
+ const Bytef *source, uLong *sourceLen));
+/*
+ Same as uncompress, except that sourceLen is a pointer, where the
+ length of the source is *sourceLen. On return, *sourceLen is the number of
+ source bytes consumed.
+*/
+
ZEXTERN int ZEXPORT inflateInit2_ OF((z_streamp strm, int windowBits,
const char *version, int stream_size));
#define inflateInit(strm) \
diff --git a/lib/zlib/uncompr.c b/lib/zlib/uncompr.c
new file mode 100644
index 0000000000..21e93933b2
--- /dev/null
+++ b/lib/zlib/uncompr.c
@@ -0,0 +1,97 @@
+/* uncompr.c -- decompress a memory buffer
+ * Copyright (C) 1995-2003, 2010, 2014, 2016 Jean-loup Gailly, Mark Adler
+ * For conditions of distribution and use, see copyright notice in zlib.h
+ */
+
+/* @(#) $Id$ */
+
+#define ZLIB_INTERNAL
+#include "zlib.h"
+
+/* ===========================================================================
+ Decompresses the source buffer into the destination buffer. *sourceLen is
+ the byte length of the source buffer. Upon entry, *destLen is the total size
+ of the destination buffer, which must be large enough to hold the entire
+ uncompressed data. (The size of the uncompressed data must have been saved
+ previously by the compressor and transmitted to the decompressor by some
+ mechanism outside the scope of this compression library.) Upon exit,
+ *destLen is the size of the decompressed data and *sourceLen is the number
+ of source bytes consumed. Upon return, source + *sourceLen points to the
+ first unused input byte.
+
+ uncompress returns Z_OK if success, Z_MEM_ERROR if there was not enough
+ memory, Z_BUF_ERROR if there was not enough room in the output buffer, or
+ Z_DATA_ERROR if the input data was corrupted, including if the input data is
+ an incomplete zlib stream.
+*/
+int ZEXPORT uncompress2(dest, destLen, source, sourceLen)
+ Bytef *dest;
+ uLongf *destLen;
+ const Bytef *source;
+ uLong *sourceLen;
+
+{
+ z_stream stream;
+ int err;
+ const uInt max = (uInt)-1;
+ uLong len, left;
+ /* for detection of incomplete stream when *destLen == 0 */
+ Byte buf[1];
+
+ len = *sourceLen;
+ if (*destLen) {
+ left = *destLen;
+ *destLen = 0;
+ } else {
+ left = 1;
+ dest = buf;
+ }
+
+ stream.next_in = (z_const Bytef *)source;
+ stream.avail_in = 0;
+ stream.zalloc = (alloc_func)0;
+ stream.zfree = (free_func)0;
+ stream.opaque = (voidpf)0;
+
+ err = inflateInit(&stream);
+ if (err != Z_OK)
+ return err;
+
+ stream.next_out = dest;
+ stream.avail_out = 0;
+
+ do {
+ if (stream.avail_out == 0) {
+ stream.avail_out = left > (uLong)max ? max : (uInt)left;
+ left -= stream.avail_out;
+ }
+
+ if (stream.avail_in == 0) {
+ stream.avail_in = len > (uLong)max ? max : (uInt)len;
+ len -= stream.avail_in;
+ }
+
+ err = inflate(&stream, Z_NO_FLUSH);
+ } while (err == Z_OK);
+
+ *sourceLen -= len + stream.avail_in;
+ if (dest != buf)
+ *destLen = stream.total_out;
+ else if (stream.total_out && err == Z_BUF_ERROR)
+ left = 1;
+
+ inflateEnd(&stream);
+ return err == Z_STREAM_END ? Z_OK :
+ err == Z_NEED_DICT ? Z_DATA_ERROR :
+ err == Z_BUF_ERROR && left + stream.avail_out ? Z_DATA_ERROR :
+ err;
+}
+
+int ZEXPORT uncompress(dest, destLen, source, sourceLen)
+ Bytef *dest;
+ uLongf *destLen;
+ const Bytef *source;
+ uLong sourceLen;
+{
+ return uncompress2(dest, destLen, source, &sourceLen);
+}
diff --git a/lib/zlib/zlib.c b/lib/zlib/zlib.c
index 7e1570292c..05a2734086 100644
--- a/lib/zlib/zlib.c
+++ b/lib/zlib/zlib.c
@@ -30,3 +30,4 @@
#include "inflate.c"
#include "zutil.c"
#include "adler32.c"
+#include "uncompr.c"
--
2.17.1
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [PATCH v3 4/5] fs/squashfs: add support for zlib decompression
2020-07-21 9:22 [PATCH v3 0/5] Add support for the SquashFS filesystem Joao Marcos Costa
` (2 preceding siblings ...)
2020-07-21 9:22 ` [PATCH v3 3/5] include/u-boot, lib/zlib: add sources for zlib decompression Joao Marcos Costa
@ 2020-07-21 9:22 ` Joao Marcos Costa
2020-07-21 9:22 ` [PATCH v3 5/5] fs/fs.c: add symbolic link case to fs_ls_generic() Joao Marcos Costa
2020-07-29 12:54 ` [PATCH v3 0/5] Add support for the SquashFS filesystem Joao Marcos Costa
5 siblings, 0 replies; 15+ messages in thread
From: Joao Marcos Costa @ 2020-07-21 9:22 UTC (permalink / raw)
To: u-boot
Add call to zlib's 'uncompress' function. Add function to display the
right error message depending on the decompression's return value.
Signed-off-by: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
---
Changes in v3:
- No changes since v2.
Changes in v2:
- No changes since v1.
fs/squashfs/sqfs_decompressor.c | 24 ++++++++++++++++++++++++
1 file changed, 24 insertions(+)
diff --git a/fs/squashfs/sqfs_decompressor.c b/fs/squashfs/sqfs_decompressor.c
index a899a5704b..a8c37f73b7 100644
--- a/fs/squashfs/sqfs_decompressor.c
+++ b/fs/squashfs/sqfs_decompressor.c
@@ -9,17 +9,41 @@
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
+#include <u-boot/zlib.h>
#include "sqfs_decompressor.h"
#include "sqfs_filesystem.h"
#include "sqfs_utils.h"
+static void zlib_decompression_status(int ret)
+{
+ switch (ret) {
+ case Z_BUF_ERROR:
+ printf("Error: 'dest' buffer is not large enough.\n");
+ break;
+ case Z_DATA_ERROR:
+ printf("Error: corrupted compressed data.\n");
+ break;
+ case Z_MEM_ERROR:
+ printf("Error: insufficient memory.\n");
+ break;
+ }
+}
+
int sqfs_decompress(u16 comp_type, void *dest, unsigned long *dest_len,
void *source, u32 lenp)
{
int ret = 0;
switch (comp_type) {
+ case SQFS_COMP_ZLIB:
+ ret = uncompress(dest, dest_len, source, lenp);
+ if (ret) {
+ zlib_decompression_status(ret);
+ return -EINVAL;
+ }
+
+ break;
default:
printf("Error: unknown compression type.\n");
return -EINVAL;
--
2.17.1
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [PATCH v3 5/5] fs/fs.c: add symbolic link case to fs_ls_generic()
2020-07-21 9:22 [PATCH v3 0/5] Add support for the SquashFS filesystem Joao Marcos Costa
` (3 preceding siblings ...)
2020-07-21 9:22 ` [PATCH v3 4/5] fs/squashfs: add support " Joao Marcos Costa
@ 2020-07-21 9:22 ` Joao Marcos Costa
2020-07-29 12:54 ` [PATCH v3 0/5] Add support for the SquashFS filesystem Joao Marcos Costa
5 siblings, 0 replies; 15+ messages in thread
From: Joao Marcos Costa @ 2020-07-21 9:22 UTC (permalink / raw)
To: u-boot
Adds an 'else if' statement inside the loop to check for symbolic links.
Signed-off-by: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
---
Changes in v3:
- No changes since v2.
Changes in v2:
- This patch was added in v2 and did not exist in the last version.
fs/fs.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/fs/fs.c b/fs/fs.c
index 5b31a369f7..17e4bc33f7 100644
--- a/fs/fs.c
+++ b/fs/fs.c
@@ -59,6 +59,9 @@ static int fs_ls_generic(const char *dirname)
if (dent->type == FS_DT_DIR) {
printf(" %s/\n", dent->name);
ndirs++;
+ } else if (dent->type == FS_DT_LNK) {
+ printf(" <SYM> %s\n", dent->name);
+ nfiles++;
} else {
printf(" %8lld %s\n", dent->size, dent->name);
nfiles++;
--
2.17.1
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [PATCH v3 0/5] Add support for the SquashFS filesystem
2020-07-21 9:22 [PATCH v3 0/5] Add support for the SquashFS filesystem Joao Marcos Costa
` (4 preceding siblings ...)
2020-07-21 9:22 ` [PATCH v3 5/5] fs/fs.c: add symbolic link case to fs_ls_generic() Joao Marcos Costa
@ 2020-07-29 12:54 ` Joao Marcos Costa
2020-07-29 15:21 ` Tom Rini
5 siblings, 1 reply; 15+ messages in thread
From: Joao Marcos Costa @ 2020-07-29 12:54 UTC (permalink / raw)
To: u-boot
On Tue, 21 Jul 2020 11:22:54 +0200
Joao Marcos Costa <joaomarcos.costa@bootlin.com> wrote:
> Hello!
>
> This series adds support for the SquashFS filesystem. For now, zlib
> is the only supported compression type. This is my first contribution
> to U-Boot as well as to a major Open Source project.
>
> Changes in v3:
> - Replace CONFIG_IS_ENABLED by IS_ENABLED in fs/fs.c
> Changes in v2:
> - Replace sqfs_ls() by U-Boot's fs_ls_generic()
> - Add info. to MAINTAINERS
> - Fix build failures
> - Fix style problems
>
> Best regards,
> Joao Marcos Costa
>
> Joao Marcos Costa (5):
> fs/squashfs: new filesystem
> fs/squashfs: add filesystem commands
> include/u-boot, lib/zlib: add sources for zlib decompression
> fs/squashfs: add support for zlib decompression
> fs/fs.c: add symbolic link case to fs_ls_generic()
>
> MAINTAINERS | 7 +
> cmd/Kconfig | 6 +
> cmd/Makefile | 1 +
> cmd/sqfs.c | 42 +
> common/spl/Kconfig | 9 +
> fs/Kconfig | 2 +
> fs/Makefile | 2 +
> fs/fs.c | 18 +
> fs/squashfs/Kconfig | 10 +
> fs/squashfs/Makefile | 7 +
> fs/squashfs/sqfs.c | 1521
> +++++++++++++++++++++++++++++++ fs/squashfs/sqfs_decompressor.c |
> 53 ++ fs/squashfs/sqfs_decompressor.h | 58 ++
> fs/squashfs/sqfs_dir.c | 107 +++
> fs/squashfs/sqfs_filesystem.h | 300 ++++++
> fs/squashfs/sqfs_inode.c | 142 +++
> fs/squashfs/sqfs_utils.h | 49 +
> include/fs.h | 1 +
> include/squashfs.h | 25 +
> include/u-boot/zlib.h | 32 +
> lib/zlib/uncompr.c | 97 ++
> lib/zlib/zlib.c | 1 +
> 22 files changed, 2490 insertions(+)
> create mode 100644 cmd/sqfs.c
> create mode 100644 fs/squashfs/Kconfig
> create mode 100644 fs/squashfs/Makefile
> create mode 100644 fs/squashfs/sqfs.c
> create mode 100644 fs/squashfs/sqfs_decompressor.c
> create mode 100644 fs/squashfs/sqfs_decompressor.h
> create mode 100644 fs/squashfs/sqfs_dir.c
> create mode 100644 fs/squashfs/sqfs_filesystem.h
> create mode 100644 fs/squashfs/sqfs_inode.c
> create mode 100644 fs/squashfs/sqfs_utils.h
> create mode 100644 include/squashfs.h
> create mode 100644 lib/zlib/uncompr.c
>
Hello again,
I am currently writing tests for the SquashFS support. I wrote two
Python scripts to test 'ls' and 'load' commands, but I need a SquashFS
image to do so, and I would like to know what is the better approach
for this:
- Add a compiled SquashFS image into the same directory as the tests,
i.e.: test/py/tests/test_fs/test_squashfs. This is a +-7MB sized binary
file.
- Use the scripts to generate an image, with mksquashfs. However,
mksquashfs becomes a dependency.
Best regards,
Joao Marcos
^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH v3 0/5] Add support for the SquashFS filesystem
2020-07-29 12:54 ` [PATCH v3 0/5] Add support for the SquashFS filesystem Joao Marcos Costa
@ 2020-07-29 15:21 ` Tom Rini
2020-07-29 16:02 ` Joao Marcos Costa
0 siblings, 1 reply; 15+ messages in thread
From: Tom Rini @ 2020-07-29 15:21 UTC (permalink / raw)
To: u-boot
On Wed, Jul 29, 2020 at 02:54:56PM +0200, Joao Marcos Costa wrote:
> On Tue, 21 Jul 2020 11:22:54 +0200
> Joao Marcos Costa <joaomarcos.costa@bootlin.com> wrote:
>
> > Hello!
> >
> > This series adds support for the SquashFS filesystem. For now, zlib
> > is the only supported compression type. This is my first contribution
> > to U-Boot as well as to a major Open Source project.
> >
> > Changes in v3:
> > - Replace CONFIG_IS_ENABLED by IS_ENABLED in fs/fs.c
> > Changes in v2:
> > - Replace sqfs_ls() by U-Boot's fs_ls_generic()
> > - Add info. to MAINTAINERS
> > - Fix build failures
> > - Fix style problems
> >
> > Best regards,
> > Joao Marcos Costa
> >
> > Joao Marcos Costa (5):
> > fs/squashfs: new filesystem
> > fs/squashfs: add filesystem commands
> > include/u-boot, lib/zlib: add sources for zlib decompression
> > fs/squashfs: add support for zlib decompression
> > fs/fs.c: add symbolic link case to fs_ls_generic()
> >
> > MAINTAINERS | 7 +
> > cmd/Kconfig | 6 +
> > cmd/Makefile | 1 +
> > cmd/sqfs.c | 42 +
> > common/spl/Kconfig | 9 +
> > fs/Kconfig | 2 +
> > fs/Makefile | 2 +
> > fs/fs.c | 18 +
> > fs/squashfs/Kconfig | 10 +
> > fs/squashfs/Makefile | 7 +
> > fs/squashfs/sqfs.c | 1521
> > +++++++++++++++++++++++++++++++ fs/squashfs/sqfs_decompressor.c |
> > 53 ++ fs/squashfs/sqfs_decompressor.h | 58 ++
> > fs/squashfs/sqfs_dir.c | 107 +++
> > fs/squashfs/sqfs_filesystem.h | 300 ++++++
> > fs/squashfs/sqfs_inode.c | 142 +++
> > fs/squashfs/sqfs_utils.h | 49 +
> > include/fs.h | 1 +
> > include/squashfs.h | 25 +
> > include/u-boot/zlib.h | 32 +
> > lib/zlib/uncompr.c | 97 ++
> > lib/zlib/zlib.c | 1 +
> > 22 files changed, 2490 insertions(+)
> > create mode 100644 cmd/sqfs.c
> > create mode 100644 fs/squashfs/Kconfig
> > create mode 100644 fs/squashfs/Makefile
> > create mode 100644 fs/squashfs/sqfs.c
> > create mode 100644 fs/squashfs/sqfs_decompressor.c
> > create mode 100644 fs/squashfs/sqfs_decompressor.h
> > create mode 100644 fs/squashfs/sqfs_dir.c
> > create mode 100644 fs/squashfs/sqfs_filesystem.h
> > create mode 100644 fs/squashfs/sqfs_inode.c
> > create mode 100644 fs/squashfs/sqfs_utils.h
> > create mode 100644 include/squashfs.h
> > create mode 100644 lib/zlib/uncompr.c
> >
>
> Hello again,
>
> I am currently writing tests for the SquashFS support. I wrote two
> Python scripts to test 'ls' and 'load' commands, but I need a SquashFS
> image to do so, and I would like to know what is the better approach
> for this:
> - Add a compiled SquashFS image into the same directory as the tests,
> i.e.: test/py/tests/test_fs/test_squashfs. This is a +-7MB sized binary
> file.
> - Use the scripts to generate an image, with mksquashfs. However,
> mksquashfs becomes a dependency.
We should dynamically make the images, just like ext* and fat. The
tests should skip when the required tools aren't found, in the same
manner we do for other fs tests. Thanks!
--
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 659 bytes
Desc: not available
URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200729/ed3c01b8/attachment.sig>
^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH v3 3/5] include/u-boot, lib/zlib: add sources for zlib decompression
2020-07-21 9:22 ` [PATCH v3 3/5] include/u-boot, lib/zlib: add sources for zlib decompression Joao Marcos Costa
@ 2020-07-29 15:27 ` Tom Rini
2020-07-29 16:00 ` Joao Marcos Costa
0 siblings, 1 reply; 15+ messages in thread
From: Tom Rini @ 2020-07-29 15:27 UTC (permalink / raw)
To: u-boot
On Tue, Jul 21, 2020 at 11:22:57AM +0200, Joao Marcos Costa wrote:
> Add zlib (v1.2.11) uncompr() function to U-Boot. SquashFS depends on
> this function to decompress data from a raw disk image. The actual
> support for zlib into SquashFS sources will be added in a follow-up
> commit.
>
> Signed-off-by: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
> ---
> Changes in v3:
> - No changes since v2.
> Changes in v2:
> - Changed commit title and message, which were wrong and/or misleading
> in v1.
> include/u-boot/zlib.h | 32 ++++++++++++++
> lib/zlib/uncompr.c | 97 +++++++++++++++++++++++++++++++++++++++++++
> lib/zlib/zlib.c | 1 +
> 3 files changed, 130 insertions(+)
> create mode 100644 lib/zlib/uncompr.c
Can you confirm that on a platform where we aren't enabling squashfs the
new zlib functionality ends up being discarded at link time? Thanks!
--
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 659 bytes
Desc: not available
URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200729/28864363/attachment.sig>
^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH v3 3/5] include/u-boot, lib/zlib: add sources for zlib decompression
2020-07-29 15:27 ` Tom Rini
@ 2020-07-29 16:00 ` Joao Marcos Costa
2020-07-29 16:54 ` Tom Rini
0 siblings, 1 reply; 15+ messages in thread
From: Joao Marcos Costa @ 2020-07-29 16:00 UTC (permalink / raw)
To: u-boot
On Wed, 29 Jul 2020 11:27:45 -0400
Tom Rini <trini@konsulko.com> wrote:
> On Tue, Jul 21, 2020 at 11:22:57AM +0200, Joao Marcos Costa wrote:
>
> > Add zlib (v1.2.11) uncompr() function to U-Boot. SquashFS depends on
> > this function to decompress data from a raw disk image. The actual
> > support for zlib into SquashFS sources will be added in a follow-up
> > commit.
> >
> > Signed-off-by: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
> > ---
> > Changes in v3:
> > - No changes since v2.
> > Changes in v2:
> > - Changed commit title and message, which were wrong and/or
> > misleading in v1.
> > include/u-boot/zlib.h | 32 ++++++++++++++
> > lib/zlib/uncompr.c | 97
> > +++++++++++++++++++++++++++++++++++++++++++ lib/zlib/zlib.c
> > | 1 + 3 files changed, 130 insertions(+)
> > create mode 100644 lib/zlib/uncompr.c
>
> Can you confirm that on a platform where we aren't enabling squashfs
> the new zlib functionality ends up being discarded at link time?
> Thanks!
>
I dumped lib/zlib/zlib.o with readelf and the new functionality is not
discarded, even though the SquashFS support is not enabled, but this
issue will be fixed in the next version (v4). Thank you!
^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH v3 0/5] Add support for the SquashFS filesystem
2020-07-29 15:21 ` Tom Rini
@ 2020-07-29 16:02 ` Joao Marcos Costa
0 siblings, 0 replies; 15+ messages in thread
From: Joao Marcos Costa @ 2020-07-29 16:02 UTC (permalink / raw)
To: u-boot
On Wed, 29 Jul 2020 11:21:26 -0400
Tom Rini <trini@konsulko.com> wrote:
> On Wed, Jul 29, 2020 at 02:54:56PM +0200, Joao Marcos Costa wrote:
> > On Tue, 21 Jul 2020 11:22:54 +0200
> > Joao Marcos Costa <joaomarcos.costa@bootlin.com> wrote:
> >
> > > Hello!
> > >
> > > This series adds support for the SquashFS filesystem. For now,
> > > zlib is the only supported compression type. This is my first
> > > contribution to U-Boot as well as to a major Open Source project.
> > >
> > > Changes in v3:
> > > - Replace CONFIG_IS_ENABLED by IS_ENABLED in fs/fs.c
> > > Changes in v2:
> > > - Replace sqfs_ls() by U-Boot's fs_ls_generic()
> > > - Add info. to MAINTAINERS
> > > - Fix build failures
> > > - Fix style problems
> > >
> > > Best regards,
> > > Joao Marcos Costa
> > >
> > > Joao Marcos Costa (5):
> > > fs/squashfs: new filesystem
> > > fs/squashfs: add filesystem commands
> > > include/u-boot, lib/zlib: add sources for zlib decompression
> > > fs/squashfs: add support for zlib decompression
> > > fs/fs.c: add symbolic link case to fs_ls_generic()
> > >
> > > MAINTAINERS | 7 +
> > > cmd/Kconfig | 6 +
> > > cmd/Makefile | 1 +
> > > cmd/sqfs.c | 42 +
> > > common/spl/Kconfig | 9 +
> > > fs/Kconfig | 2 +
> > > fs/Makefile | 2 +
> > > fs/fs.c | 18 +
> > > fs/squashfs/Kconfig | 10 +
> > > fs/squashfs/Makefile | 7 +
> > > fs/squashfs/sqfs.c | 1521
> > > +++++++++++++++++++++++++++++++ fs/squashfs/sqfs_decompressor.c |
> > > 53 ++ fs/squashfs/sqfs_decompressor.h | 58 ++
> > > fs/squashfs/sqfs_dir.c | 107 +++
> > > fs/squashfs/sqfs_filesystem.h | 300 ++++++
> > > fs/squashfs/sqfs_inode.c | 142 +++
> > > fs/squashfs/sqfs_utils.h | 49 +
> > > include/fs.h | 1 +
> > > include/squashfs.h | 25 +
> > > include/u-boot/zlib.h | 32 +
> > > lib/zlib/uncompr.c | 97 ++
> > > lib/zlib/zlib.c | 1 +
> > > 22 files changed, 2490 insertions(+)
> > > create mode 100644 cmd/sqfs.c
> > > create mode 100644 fs/squashfs/Kconfig
> > > create mode 100644 fs/squashfs/Makefile
> > > create mode 100644 fs/squashfs/sqfs.c
> > > create mode 100644 fs/squashfs/sqfs_decompressor.c
> > > create mode 100644 fs/squashfs/sqfs_decompressor.h
> > > create mode 100644 fs/squashfs/sqfs_dir.c
> > > create mode 100644 fs/squashfs/sqfs_filesystem.h
> > > create mode 100644 fs/squashfs/sqfs_inode.c
> > > create mode 100644 fs/squashfs/sqfs_utils.h
> > > create mode 100644 include/squashfs.h
> > > create mode 100644 lib/zlib/uncompr.c
> > >
> >
> > Hello again,
> >
> > I am currently writing tests for the SquashFS support. I wrote two
> > Python scripts to test 'ls' and 'load' commands, but I need a
> > SquashFS image to do so, and I would like to know what is the
> > better approach for this:
> > - Add a compiled SquashFS image into the same directory as the
> > tests, i.e.: test/py/tests/test_fs/test_squashfs. This is a +-7MB
> > sized binary file.
> > - Use the scripts to generate an image, with mksquashfs. However,
> > mksquashfs becomes a dependency.
>
> We should dynamically make the images, just like ext* and fat. The
> tests should skip when the required tools aren't found, in the same
> manner we do for other fs tests. Thanks!
>
Ok, thank you!
Best regards,
Joao Marcos
^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH v3 3/5] include/u-boot, lib/zlib: add sources for zlib decompression
2020-07-29 16:00 ` Joao Marcos Costa
@ 2020-07-29 16:54 ` Tom Rini
2020-07-30 12:10 ` Joao Marcos Costa
0 siblings, 1 reply; 15+ messages in thread
From: Tom Rini @ 2020-07-29 16:54 UTC (permalink / raw)
To: u-boot
On Wed, Jul 29, 2020 at 06:00:57PM +0200, Joao Marcos Costa wrote:
> On Wed, 29 Jul 2020 11:27:45 -0400
> Tom Rini <trini@konsulko.com> wrote:
>
> > On Tue, Jul 21, 2020 at 11:22:57AM +0200, Joao Marcos Costa wrote:
> >
> > > Add zlib (v1.2.11) uncompr() function to U-Boot. SquashFS depends on
> > > this function to decompress data from a raw disk image. The actual
> > > support for zlib into SquashFS sources will be added in a follow-up
> > > commit.
> > >
> > > Signed-off-by: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
> > > ---
> > > Changes in v3:
> > > - No changes since v2.
> > > Changes in v2:
> > > - Changed commit title and message, which were wrong and/or
> > > misleading in v1.
> > > include/u-boot/zlib.h | 32 ++++++++++++++
> > > lib/zlib/uncompr.c | 97
> > > +++++++++++++++++++++++++++++++++++++++++++ lib/zlib/zlib.c
> > > | 1 + 3 files changed, 130 insertions(+)
> > > create mode 100644 lib/zlib/uncompr.c
> >
> > Can you confirm that on a platform where we aren't enabling squashfs
> > the new zlib functionality ends up being discarded at link time?
> > Thanks!
> >
>
> I dumped lib/zlib/zlib.o with readelf and the new functionality is not
> discarded, even though the SquashFS support is not enabled, but this
> issue will be fixed in the next version (v4). Thank you!
Well, don't dump lib/zlib/zlib.o, that won't say. Check u-boot.map :)
--
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 659 bytes
Desc: not available
URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200729/b7e9d8a4/attachment.sig>
^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH v3 3/5] include/u-boot, lib/zlib: add sources for zlib decompression
2020-07-29 16:54 ` Tom Rini
@ 2020-07-30 12:10 ` Joao Marcos Costa
2020-07-30 12:17 ` Tom Rini
0 siblings, 1 reply; 15+ messages in thread
From: Joao Marcos Costa @ 2020-07-30 12:10 UTC (permalink / raw)
To: u-boot
On Wed, 29 Jul 2020 12:54:23 -0400
Tom Rini <trini@konsulko.com> wrote:
> On Wed, Jul 29, 2020 at 06:00:57PM +0200, Joao Marcos Costa wrote:
> > On Wed, 29 Jul 2020 11:27:45 -0400
> > Tom Rini <trini@konsulko.com> wrote:
> >
> > > On Tue, Jul 21, 2020 at 11:22:57AM +0200, Joao Marcos Costa wrote:
> > >
> > > > Add zlib (v1.2.11) uncompr() function to U-Boot. SquashFS
> > > > depends on this function to decompress data from a raw disk
> > > > image. The actual support for zlib into SquashFS sources will
> > > > be added in a follow-up commit.
> > > >
> > > > Signed-off-by: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
> > > > ---
> > > > Changes in v3:
> > > > - No changes since v2.
> > > > Changes in v2:
> > > > - Changed commit title and message, which were wrong
> > > > and/or misleading in v1.
> > > > include/u-boot/zlib.h | 32 ++++++++++++++
> > > > lib/zlib/uncompr.c | 97
> > > > +++++++++++++++++++++++++++++++++++++++++++ lib/zlib/zlib.c
> > > > | 1 + 3 files changed, 130 insertions(+)
> > > > create mode 100644 lib/zlib/uncompr.c
> > >
> > > Can you confirm that on a platform where we aren't enabling
> > > squashfs the new zlib functionality ends up being discarded at
> > > link time? Thanks!
> > >
> >
> > I dumped lib/zlib/zlib.o with readelf and the new functionality is
> > not discarded, even though the SquashFS support is not enabled, but
> > this issue will be fixed in the next version (v4). Thank you!
>
> Well, don't dump lib/zlib/zlib.o, that won't say. Check u-boot.map :)
>
Thanks for the tip! Now I can confirm that the functionality is
discarded when squashfs is not enabled. I just would like to know if I
should send the tests in the same patch as the SquashFS commands, or
should it be into a separate patch?
^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH v3 3/5] include/u-boot, lib/zlib: add sources for zlib decompression
2020-07-30 12:10 ` Joao Marcos Costa
@ 2020-07-30 12:17 ` Tom Rini
2020-07-30 13:27 ` Joao Marcos Costa
0 siblings, 1 reply; 15+ messages in thread
From: Tom Rini @ 2020-07-30 12:17 UTC (permalink / raw)
To: u-boot
On Thu, Jul 30, 2020 at 02:10:03PM +0200, Joao Marcos Costa wrote:
> On Wed, 29 Jul 2020 12:54:23 -0400
> Tom Rini <trini@konsulko.com> wrote:
>
> > On Wed, Jul 29, 2020 at 06:00:57PM +0200, Joao Marcos Costa wrote:
> > > On Wed, 29 Jul 2020 11:27:45 -0400
> > > Tom Rini <trini@konsulko.com> wrote:
> > >
> > > > On Tue, Jul 21, 2020 at 11:22:57AM +0200, Joao Marcos Costa wrote:
> > > >
> > > > > Add zlib (v1.2.11) uncompr() function to U-Boot. SquashFS
> > > > > depends on this function to decompress data from a raw disk
> > > > > image. The actual support for zlib into SquashFS sources will
> > > > > be added in a follow-up commit.
> > > > >
> > > > > Signed-off-by: Joao Marcos Costa <joaomarcos.costa@bootlin.com>
> > > > > ---
> > > > > Changes in v3:
> > > > > - No changes since v2.
> > > > > Changes in v2:
> > > > > - Changed commit title and message, which were wrong
> > > > > and/or misleading in v1.
> > > > > include/u-boot/zlib.h | 32 ++++++++++++++
> > > > > lib/zlib/uncompr.c | 97
> > > > > +++++++++++++++++++++++++++++++++++++++++++ lib/zlib/zlib.c
> > > > > | 1 + 3 files changed, 130 insertions(+)
> > > > > create mode 100644 lib/zlib/uncompr.c
> > > >
> > > > Can you confirm that on a platform where we aren't enabling
> > > > squashfs the new zlib functionality ends up being discarded at
> > > > link time? Thanks!
> > > >
> > >
> > > I dumped lib/zlib/zlib.o with readelf and the new functionality is
> > > not discarded, even though the SquashFS support is not enabled, but
> > > this issue will be fixed in the next version (v4). Thank you!
> >
> > Well, don't dump lib/zlib/zlib.o, that won't say. Check u-boot.map :)
> >
>
> Thanks for the tip! Now I can confirm that the functionality is
> discarded when squashfs is not enabled. I just would like to know if I
> should send the tests in the same patch as the SquashFS commands, or
> should it be into a separate patch?
A separate patch, and are you able to run them on sandbox then? So it
should also enable squashfs on that target. Thanks!
--
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 659 bytes
Desc: not available
URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200730/3294f73f/attachment.sig>
^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH v3 3/5] include/u-boot, lib/zlib: add sources for zlib decompression
2020-07-30 12:17 ` Tom Rini
@ 2020-07-30 13:27 ` Joao Marcos Costa
0 siblings, 0 replies; 15+ messages in thread
From: Joao Marcos Costa @ 2020-07-30 13:27 UTC (permalink / raw)
To: u-boot
On Thu, 30 Jul 2020 08:17:59 -0400
Tom Rini <trini@konsulko.com> wrote:
> On Thu, Jul 30, 2020 at 02:10:03PM +0200, Joao Marcos Costa wrote:
> > On Wed, 29 Jul 2020 12:54:23 -0400
> > Tom Rini <trini@konsulko.com> wrote:
> >
> > > On Wed, Jul 29, 2020 at 06:00:57PM +0200, Joao Marcos Costa
> > > wrote:
> > > > On Wed, 29 Jul 2020 11:27:45 -0400
> > > > Tom Rini <trini@konsulko.com> wrote:
> > > >
> > > > > On Tue, Jul 21, 2020 at 11:22:57AM +0200, Joao Marcos Costa
> > > > > wrote:
> > > > > > Add zlib (v1.2.11) uncompr() function to U-Boot. SquashFS
> > > > > > depends on this function to decompress data from a raw disk
> > > > > > image. The actual support for zlib into SquashFS sources
> > > > > > will be added in a follow-up commit.
> > > > > >
> > > > > > Signed-off-by: Joao Marcos Costa
> > > > > > <joaomarcos.costa@bootlin.com> ---
> > > > > > Changes in v3:
> > > > > > - No changes since v2.
> > > > > > Changes in v2:
> > > > > > - Changed commit title and message, which were wrong
> > > > > > and/or misleading in v1.
> > > > > > include/u-boot/zlib.h | 32 ++++++++++++++
> > > > > > lib/zlib/uncompr.c | 97
> > > > > > +++++++++++++++++++++++++++++++++++++++++++ lib/zlib/zlib.c
> > > > > > | 1 + 3 files changed, 130 insertions(+)
> > > > > > create mode 100644 lib/zlib/uncompr.c
> > > > >
> > > > > Can you confirm that on a platform where we aren't enabling
> > > > > squashfs the new zlib functionality ends up being discarded at
> > > > > link time? Thanks!
> > > > >
> > > >
> > > > I dumped lib/zlib/zlib.o with readelf and the new functionality
> > > > is not discarded, even though the SquashFS support is not
> > > > enabled, but this issue will be fixed in the next version (v4).
> > > > Thank you!
> > >
> > > Well, don't dump lib/zlib/zlib.o, that won't say. Check
> > > u-boot.map :)
> >
> > Thanks for the tip! Now I can confirm that the functionality is
> > discarded when squashfs is not enabled. I just would like to know
> > if I should send the tests in the same patch as the SquashFS
> > commands, or should it be into a separate patch?
>
> A separate patch, and are you able to run them on sandbox then? So it
> should also enable squashfs on that target. Thanks!
>
This will be addressed in v4. Thanks!
^ permalink raw reply [flat|nested] 15+ messages in thread
end of thread, other threads:[~2020-07-30 13:27 UTC | newest]
Thread overview: 15+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-07-21 9:22 [PATCH v3 0/5] Add support for the SquashFS filesystem Joao Marcos Costa
2020-07-21 9:22 ` [PATCH v3 1/5] fs/squashfs: new filesystem Joao Marcos Costa
2020-07-21 9:22 ` [PATCH v3 2/5] fs/squashfs: add filesystem commands Joao Marcos Costa
2020-07-21 9:22 ` [PATCH v3 3/5] include/u-boot, lib/zlib: add sources for zlib decompression Joao Marcos Costa
2020-07-29 15:27 ` Tom Rini
2020-07-29 16:00 ` Joao Marcos Costa
2020-07-29 16:54 ` Tom Rini
2020-07-30 12:10 ` Joao Marcos Costa
2020-07-30 12:17 ` Tom Rini
2020-07-30 13:27 ` Joao Marcos Costa
2020-07-21 9:22 ` [PATCH v3 4/5] fs/squashfs: add support " Joao Marcos Costa
2020-07-21 9:22 ` [PATCH v3 5/5] fs/fs.c: add symbolic link case to fs_ls_generic() Joao Marcos Costa
2020-07-29 12:54 ` [PATCH v3 0/5] Add support for the SquashFS filesystem Joao Marcos Costa
2020-07-29 15:21 ` Tom Rini
2020-07-29 16:02 ` Joao Marcos Costa
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.