| Age | Commit message (Collapse) | Author |
|---|
|
Adapt budget.c in libubifs, compared with linux kernel implementations:
1. Remove writeback related functions, there are no dirty pages/inodes
for UBIFS in userspace process.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Adapt journal.c in libubifs, compared with linux kernel implementations:
1. Remove all ubifs_jnl_XXX functions. Only keep the basic space
reservation code, fsck will add new functions for journaling
operations without using linux in-memory inode/dentry.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Adapt dir.c in libubifs, compared with linux kernel implementations:
1. Remove all functions. Only keep an empty source file, fsck will
add new functions for mkdir/link operations without using linux
in-memory inode/dentry.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Adapt auth.c in libubifs, compared with linux kernel implementations:
1. Only keep implementations used by mkfs, other implementations
are removed.
2. Adapt functions based on implementations in common/sign.c.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Adapt sb.c in libubifs, compared with linux kernel implementations:
1. Remove authentication related implementations
(eg. authenticate_sb_node), authentication is not supported in fsck
for now.
2. Remove some functions(eg. create_default_filesystem) which won't be
used in fsck/mkfs.
3. Modify ubifs_read_superblock(), remove some assignments which won't
be used in mkfs/fsck.
4. Apapt fixup_leb to ignore %-EBADMSG, subsequent steps will check
all areas carefully.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Adapt recovery subsystem(replay.c, recovery.c) in libubifs, compared with
linux kernel implementations:
1. Remove authentication related implementations
(eg. authenticate_sleb_hash), authentication is not supported in fsck
for now.
2. Add explicit type conversions(const char *) to avoid compiling
warnings.
3. Replace implementations of inode_fix_size() with ubifs_assert(0),
authentication is not supported in fsck, so this function won't
be invoked.
4. Remove unused ubifs_clean_lebs() and ubifs_write_rcvrd_mst_node().
5. Adapt fix_unclean_leb/recover_head/fix_size_in_place to ignore
%-EBADMSG, subsequent steps will check nodes in lpt/main area
carefully.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Adapt log.c in libubifs, compared with linux kernel implementations:
1. Remove debug related implementations(eg. dbg_check_bud_bytes), debug
functions are not needed by fsck, because fsck will check fs in
another way.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Adapt tnc subsystem(tnc.c,tnc_misc.c,tnc_commit.c) in libubifs, compared
with linux kernel implementations:
1. Remove debug related functions(eg. dbg_check_inode_size), debug
functions are not needed by fsck, because fsck will check fs in
another way.
2. Remove some functions(eg. ubifs_tnc_bulk_read) which won't be used
in fsck/mkfs.
3. Adapt tnc_delete and ubifs_search_zbranch to handle empty TNC case,
which could happen in fsck.
4. Don't skip checking the length of non-leaf index node's branch in
read_znode.
5. Adapt try_read_node to ignore %-EBADMSG, subsequent steps will check
nodes carefully.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Adapt lpt subsystem(lpt.c,lprops.c,lpt_commit.c) in libubifs, compared
with linux kernel implementations:
1. Remove debug related functions(eg. dbg_chk_lpt_sz, dbg_chk_pnode),
some of debug functions are not needed by fsck, because fsck will
check fs in another way.
2. Remove some functions(eg. ubifs_create_dflt_lpt) which won't be used
in fsck/mkfs.
3. Adapt do_calc_lpt_geom() to mkfs/fsck situations.
4. Adapt calc_dflt_lpt_geom to support the mkfs tool, and export it.
5. Adapt ubifs_create_lpt() according to create_lpt(mkfs), make sure that
the height of created lpt is right.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Adapt io.c in libubifs, compared with linux kernel implementations:
1. Modify io related functions(eg. ubifs_leb_read/ubifs_leb_write,
etc.), adapt them with userspace io functions lseek/read/write.
2. Remove some functions(eg. record_magic_error, ubifs_bg_wbufs_sync)
which won't be used in fsck/mkfs.
3. Replce ubifs_errc with ubifs_err, because there will be no SB_SILENT
options in mkfs/fsck.
4. Initiate wbuf->size as c->max_write_size.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Adapt super.c in libubifs, compared with linux kernel implementations:
1. It contains all definitions in common/super.c(Message printing
functions are replaced with linux kernel styles).
2. Remove some functions(eg. ubifs_iget, ubifs_dirty_inode) which won't
be used in fsck/mkfs.
3. Remove unused variables initialization in some functions(eg.
init_constants_early, init_constants_sb).
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Adapt ubifs header file in libubifs, compared with linux kernel
implementations:
1. New header file contains all definitions in common/ubifs.h
2. Remove some structures(eg. ubifs_mount_opts, bu_info) and
functions(eg. ubifs_sync_wbufs_by_inode, ubifs_jnl_XXX) which
won't be used in fsck/mkfs.
3. Modify some authentication related functions, keep functions
that mkfs may need, other functions are defined as empty.
4. Move ubifs message printing functions(ubifs_err/warn/msg) which
are originally defined in misc.c into ubifs.h.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
There are several new libs(eg. atomic, kmem, bitops. etc.) have
been imported into ubifs-utils, add descriptions for these libs
in README.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Add common definitions in linux kernel, which are used in UBIFS linux
kernel libs.
This is a preparation for replacing implementation of UBIFS utils with
linux kernel libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Add ubi_leb_map() implementation, it is used in UBIFS linux kernel libs.
This is a preparation for replacing implementation of UBIFS utils with
linux kernel libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Add linux hexdump implementations lib, because function print_hex_dump()
is used in UBIFS linux kernel libs.
This is a preparation for replacing implementation of UBIFS utils with
linux kernel libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Add implementations for linux kernel printing functions, because these
functions(eg. pr_debug, pr_err, etc.) are widely used in UBIFS linux
kernel libs. No need to define multiple levels in dbg_msg for debuging,
just replace dbg_msg with pr_debug. Now, there are five levels of
printing messages:
0 - none
1 - error message
2 - warning message [default]
3 - notice message
4 - debug message
This is a preparation for replacing implementation of UBIFS utils with
linux kernel libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Add sorting implementations, because the sorting function
is used in UBIFS linux kernel libs.
This is a preparation for replacing implementation of UBIFS utils with
linux kernel libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Add rwsem implementations, because there are some rwsems
(eg. c->commit_sem) used in UBIFS linux kernel libs.
The rwsem is implemented based on pthread mutex.
This is a preparation for replacing implementation of UBIFS utils with
linux kernel libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Add mutexlock implementations, because there are some mutexlocks
(eg. c->tnc_mutex, c->log_mutex) used in UBIFS linux kernel libs.
The mutexlock is implemented based on pthread mutex.
This is a preparation for replacing implementation of UBIFS utils with
linux kernel libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Add spinlock implementations, because there are some spinlocks
(eg. c->cnt_lock, c->buds_lock) used in UBIFS linux kernel libs.
The spinlock is implemented based on pthread mutex.
This is a preparation for replacing implementation of UBIFS utils with
linux kernel libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Add bit operations implementations, because there are some bit operations
(eg. __set_bit, test_bit) used in UBIFS linux kernel libs.
This is a preparation for replacing implementation of UBIFS utils with
linux kernel libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Add atomic implementations, because there are some atomic operations
(eg. atomic_long_xxx) used in UBIFS linux kernel libs.
This is a preparation for replacing implementation of UBIFS utils with
linux kernel libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Add linux kernel memory allocation implementations, because there are many
memory allocations(eg. kmalloc, kzalloc) used in UBIFS linux kernel libs.
This is a preparation for replacing implementation of UBIFS utils with
linux kernel libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Add linux kernel error conversion definitions, because there are
many error type conversions(eg. PTR_ERR, ERR_PTR) used in UBIFS
linux kernel libs.
This is a preparation for replacing implementation of UBIFS utils with
linux kernel libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Add linux type definitions, because there are many types
(eg. u8/u16/u64) used in UBIFS linux kernel libs. Besides
move type conversions (eg. cpu_to_le16, cpu_to_le32, etc.)
into type definitions header file.
This is a preparation for replacing implementation of UBIFS utils with
linux kernel libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Add compiler attributes implementations, such as __packed, __unused,
__const, etc., which could be used in linux kernel libs. Besides, change
some existing attributes into linux kernel style.
This is a preparation for replacing implementation of UBIFS utils with
linux kernel libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Import UBIFS libs from linux kernel. Next patches will replace ubifs
related source code with implementation of linux kernel, which makes
userspace implementation be same with linux kernel, then fsck.ubifs
can resuse the code.
Notice: lpt.c is modified with [1] applied. ubifs.h and orphan.c are
modified with [2] applied, journal.c is modified with [3] reverted(
because fsck runs in a single thread, so waitqueue is not needed to
be implemented in userspace.).
[1] https://lore.kernel.org/linux-mtd/20231228014112.2836317-13-chengzhihao1@huawei.com/
[2] https://lore.kernel.org/linux-mtd/20240410073751.2522830-1-chengzhihao1@huawei.com/
[3] https://lore.kernel.org/linux-mtd/20240122063103.359501-1-chengzhihao1@huawei.com/
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
new source file
Move UBI opening/closing/volume_check_empty functions into common/super.c.
These functions will be used in fsck.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Embed new members 'dev_fd' and 'libubi' into ubifs_info structure, so
that global variable 'ubi', 'out_fd' and 'out_ubi' could be removed
from mkfs.ubifs.c. Besides, add parameter in check_volume_empty().
Next patch will extract UBI opening/closing/check_volume_empty functions
into a new source file, these functions will be used in fsck.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Embed new member 'dev_name' into 'ubifs_info' structure, then global
variable 'output' can be removed from mkfs.ubifs.c. Next patches will
import UBIFS libs from linux kernel, which could print messages that
contain device information, so this patch can distinguish different
devices according to messages.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Embed new member 'debug_level' into 'ubifs_info' structure, then global
variable 'debug_level' can be removed from mkfs.ubifs.c. Next patches
will import UBIFS libs from linux kernel, which could print messages
with different levels, then 'debug_level' can be used to control which
level messages should be printed.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Functions 'err_msg' and 'sys_err_msg' are almost same with 'errmsg' and
'sys_errmsg', since 'errmsg' and 'sys_errmsg' can print programe name,
so replace error message printing functions (ubifs-utils) with common
lib functions(include/common.h).
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
PROGRAM_NAME is defined as a const string "mkfs.ubifs", which won't be
suitable for fsck.ubifs. Add 'program_name' member in ubifs_info
structure, define PROGRAM_NAME as ubifs_info.program_name. Then, error
messages will display right program name if fsck.ubifs is supported.
Besides, add 'program_type' member in ubifs_info structure to identify
which current program type is, without comparing program name.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Header file mkfs.ubifs.h is included in other modules(eg. compr.c, lpt.c,
fscrypt.h, sign.c), decouple it out of other modules.
There are two parts in mkfs.ubifs.h:
1. common functions, for example dbg_msg, err_msg and write_leb, move
these functions into common/defs.h and common/ubifs.h.
2. devtable related definations, move them into a new header file
common/devtable.h.
Splitting common functions from mkfs.ubifs.h is also a preparation for
importing libubifs(from linux kernel) to replace current UBIFS libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Split common source files into common dir from mkfs.ubifs, this is a
preparation for importing libubifs(from linux kernel) to replace
current UBIFS libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Make sure that the first item is parsed correctly. One issue otherwise is
a null pointer dereference when the first item is not found.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Reviewed-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Currently the read throuput test tries:
- 1 page
- 2 pages
- 1 block (64 or more pages, usually)
But it might be interesting to see how the speed gradually increases,
eg. testing all number of pages from 1 to maybe 16, and then
arbitrarilly 32 and 64.
Let's add a -C parameter to enable this additional test.
The 2-page read/write tests are also moved under this new option.
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
The read_eraseblock_by_2pages() has been generalized so it became
read_eraseblock_by_npages(), but there is no limitation (besides 0)
regarding the number of pages. Hence, drop the _by_page() helper and
replace it with the _by_npages(), using 'npages = 1'.
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Right now there are only 2 pages that may be read continuously, but why
not trying more? At least when the continuous feature is out, this type
of benchmarking will be interesting. In order to facilitate later
additions, lets make this helper more generic and accept a global
'npages' variable as parameter (because this function is called in a
macro, it is simpler like that).
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
So far speed calculations have only be done 1 page at a time or 2 pages
at a time in a block; so basically all the block was always read because
all blocks are multiple of 2. But in the future, if we want to extend
the number of pages in a single read, the final number of pages actually
read might be less than an erase block size, hence failing the throuput
calculations.
Make the number of pages in a set explicit.
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
I know no device without a multiple of 2 number of pages in each
block. Even though it might be the case, it is clearly not a big deal
and we don't really care about reading the last page, we are doing a
speed benchmark; so as long as the throughput calculation knows how much
data has been read it's fine. Eitherway, I don't think we ever have
fallen in this block because we would read the content of two pages (so
one past the block) and put it in a page-wide buffer, which would
probably lead to an out-of-bound abort.
Just drop the block.
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
In order to trigger a continuous read, the user needs to request at
least two pages at the same time. So far the tool would only read single
pages, so let's extend its capabilities to test continuous read output
when the -c option is passed.
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
'pagesize' is used for two purposes:
- accessing the size of a page
- getting the size of the test buffer, which happen to be the size of a
page for now.
Use an intermediate variable when getting the size of the test buffer,
as we will later increase its size.
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
In order to trigger a continuous read, the user needs to request more
than one data page. So far the tool would split the length into page
chunks. This is no longer the case when the -C option is passed (-c is
already used for the canonical output style).
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Using bs when skipping the bad sector is abusive as what we want is
using the size of a block and the size of a page. The fact that bs
currently is the size of a page is misleading here, has I intend to make
this amount grow.
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
The read buffer size happen to be as big as the bs variable, but this is
going to change. When accessing the buffer size, use a specific variable
instead.
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
|
Current rbtree implementation code is put under jffs utils, extract it
into common lib, and add more rbtree operations(eg. rb_first_postorder,
rb_next_postorder, etc.).
This is a preparation for replacing implementation of UBIFS utils with
linux kernel libs.
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>
|
