Add current time to the PID to improve the pseudo-random number
generator seeding. Also, use 'rand()' instead of 'random()', because
'srand()' is for 'rand()'.
Michael Roth [Wed, 23 Sep 2009 13:01:27 +0000 (15:01 +0200)]
ubiformat/ubinize: Don't randomize 0xFFFFFFFF sequence number
args.image_seq is of type uint32_t and was initialized to -1 which
becomes 0xFFFFFFFF in this case. Later the value -1 was used as a flag
that args.image_seq should be replaced with a randomized value.
With the option --image-seq (-Q) a user could provide any sequence
number at will.
But when the user provided sequence number was 0xFFFFFFFF this was
understood effectivly as -1 and got overridden by a randomized
sequence number.
So this patch change the programm flow to respect the principle of
least surprise and never use a randomized sequence number when the
user provide one at own will.
Signed-off-by: Michael Roth <mroth@nessie.de> Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Artem Bityutskiy [Mon, 24 Aug 2009 12:41:21 +0000 (15:41 +0300)]
ubinize: fix 64-bit image sequence number confusion
UBI image sequence number which we store in EC headers is 32 bits,
not 64-bits. I was confused when noticed that the 'image_seq'
variable had type 'unsigned long long'. Turn it into a 'uint32_t'
type.
Artem Bityutskiy [Mon, 24 Aug 2009 12:24:41 +0000 (15:24 +0300)]
mkfs.ubifs: do not forget to torture PEB
We added PEB torturing for the case when we fail during flashing
an image, but forgot to add it for the case when we are formating
the flash. Add the forgottent torturing call.
Jehan Bing [Thu, 6 Aug 2009 00:40:55 +0000 (17:40 -0700)]
nandwrite: unified reading from standard input and from file - part 3
Nandwrite tries to use lseek() when failing to write on a page. lseek()
will fail when used on the standard input so nandwrite fails. This code
replaces lseek with a buffer.
When the data is read, it is put in a buffer (filebuf). This buffer is
reset at each block boundary. So a "seek" just means reading from the
beginning of the buffer. writebuf and oobreadbuf are now just pointers
to locations in filebuf.
Daniel Neukomm [Tue, 11 Aug 2009 08:33:23 +0000 (11:33 +0300)]
mkfs.ubifs: correct devtable increment handling
With the device table one can add /dev entries to the root file system
image. The device table file contains among others the fields minor,
start, increment and count. If there is an entry with minor=0 start=0
increment=32 and count=4 the mkfs.ubifs makes 128 device entries, with
minor numbers from 0 to 127. The correct version makes 4 entries with
minor number 0,32,64,96.
/dev/mtd c 640 0 0 90 0 0 2 7
This gives 14 devices /dev/mtdXX instead of 7 devices.
Due to this error mtd_debug info /dev/mtd3 delivers the information of
/dev/mtd1 instead of.
Signed-off-by: Daniel Neukomm <Neukomm@vtxmail.ch> Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Timo Juhani Lindfors [Mon, 27 Jul 2009 21:05:30 +0000 (00:05 +0300)]
jffs2dump: rewrite --help output to be more suitable for help2man
$ help2man /usr/sbin/jffs2dump | man /dev/stdin
currently creates badly formatted output like
-b --bigendian image is big endian
-l --littleendian image is little
endian -c --content dump image
contents -e fname --endianconvert=fname convert
image endianness, output to file fname -r --recalc-
This patch modifies the --help output for jffs2dump to be more
suitable for help2man and thus makes it possible to easily create
manual pages that stay up-to-date.
Signed-off-by: Timo Juhani Lindfors <timo.lindfors@iki.fi> Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
UBI now supports the image sequence number feature which
prevents UBI from mistakingly accepting half-written images,
if the image was written on top of an older image. Support
this feature in ubinize and make it pick a random
number for the UBI image sequence number.
It's already about 2 years since the "-d" option was declared
deprecated. UBI utils have printed a warning about this. Now
it is time to kill this option completely.
The affected utilities are ubimkvol, ubirmvol and ubiupdatevol.
Their versions are also increased.
Mike Frysinger [Thu, 2 Jul 2009 01:15:03 +0000 (21:15 -0400)]
ubi-utils: fix compilation errors when using CPPFLAGS
The CPPFLAGS in the env changes how make initializes/appends CPPFLAGS
in sub-makes. This causes the old ubi utils to find the newer headers
at ubi-utils/include/ before the ubi-utils/old-utils/inc/.
Signed-off-by: Mike Frysinger <vapier.adi@gmail.com> Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Jehan Bing [Mon, 8 Jun 2009 20:43:26 +0000 (13:43 -0700)]
nandwrite: fix error handling
Artem Bityutskiy wrote:
> Yes, write and erase failure mean that the erasblock is bad. But I think
> marking a block as bad straight away is just dangerous. Who knows may be
> this is a small glitch in a bus, or a software bug, or some-one
> corrupted driver's memory, or whatever. This is why UBI is doing
> eraseblock torturing before marking it as bad. And it is very careful
> about error codes - only EIO code is considered as a reason to mark an
> eraseblock as bad.
Fixed broken behavior in case of write failure. More specifically:
- Only try to mark a block bad if the errors are EIO. Other errors
will abort the tool.
- Also abort the tool if the marking fails instead of ignoring it.
Jehan Bing [Sat, 6 Jun 2009 01:45:53 +0000 (18:45 -0700)]
nandwrite: amend loop condition
If the file contains only a few bytes in the last page and the
--oob option is selected, the loop may exit early
(readlen < meminfo.oobsize).
Most of the time it will still work though because the code
tries to read the whole OOB in one chunk.
Artem Bityutskiy [Tue, 2 Jun 2009 13:11:35 +0000 (16:11 +0300)]
libmts: recognize pre-MTD-sysfs kernels better
Not-so-old linux kernel like 2.6.29 do already have "/sys/class/mtd/mtdX"
directories, while very old kernels do not. But in 2.6.29 these
directories do not contain any information.
Anyway, the logic in libmtd which checked whether the system
supports sysfs was broken, because it assumed that old systems
do not even have "/sys/class/mtd/" directory.
Fix this problem by checking for "/sys/class/mtd/mtdX/name". If this
is present - the system really has sysfs support.
Corentin Chary [Tue, 26 May 2009 13:08:36 +0000 (15:08 +0200)]
mkfs.ubifs: use libubi to format UBI volume
libubi is now used to format directly UBI volume.
Typing mkfs.ubifs /dev/ubi0_0 is now possible.
dtypes should be ok as they are taken from UBIFS code.
Artem Bityutskiy [Fri, 8 May 2009 09:33:56 +0000 (12:33 +0300)]
libubi: do not use udevsettle
Not sure why, but udevsettle does not work when threads are
used. In the io_paral test I experience the problem when
udevsettle opens UBI volumes and prevents tests to open
them (EBUSY is returned).
Rename 'struct mtd_info' to 'struct mtd_dev_info' to reflect the
fact that it provides information about a specific MTD device.
Also, this is more consistent with libubi.
ubiformat: mtd0 (NAND), size 134217728 bytes (128.0 MiB), 16384 eraseblocks of 16384 bytes (16.0 KiB), min. I/O size 512 bytes
libscan: scanning eraseblock 8191 -- 100 % complete
ubiformat: 8190 eraseblocks have valid erase counter, mean value is 9
ubiformat: 2 eraseblocks are supposedly empty
ubiformat: warning!: VID header and data offsets on flash are 512 and 1024, which is different to calculated offsets 256 and 512
ubiformat: use new offsets 512 and 1024? (yes/no) no
Segmentation fault
The reason is that volume table size is calculated for 256/512
layout, and when user chooses 512/1024 - it is not re-calculated
which leads to segfault in 'ubigen_write_layout_vol()'.
This patch also fixes the message - in the above output new
offsets are 256/512, not 512/1024.
Also, this patch adds explicit printing of the selected
positions.
Remove the fd field from the mtd information data structure,
because libmtd does not really know the device node file name,
and serves only as a place to save the descriptor. The callers
should find a better place.
This patch improves code readability and prepares for further
changes.
Mike Frysinger [Tue, 31 Mar 2009 04:09:40 +0000 (00:09 -0400)]
ubi-utils: tweak vpath handling
Building out of tree currently fails with ubi-utils because make gets
confused and can't figure out the vpathed .c files. Using VPATH rather
than vpath seems to work around it.
Mike Frysinger [Thu, 26 Feb 2009 08:30:06 +0000 (03:30 -0500)]
make sure compiler supports warning flags
Some compilers (like gcc-3.3) don't support all the newer -W flags that we
are using. So import the compiler check found in the kernel and test each
flag we add. The := is important so we only do the compiler tests once
per `make` rather than every time we compile a file.
Signed-off-by: Mike Frysinger <vapier@gentoo.org> Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Sebastian Andrzej Siewior [Mon, 23 Feb 2009 21:38:54 +0000 (22:38 +0100)]
mkfs.jffs2: fix lzo usage on 64bit systems
the compress size parameter in lzo is defined in the header file as lzo_uint.
This looks very much like uint32_t, I know, but is defined as unsigned long.
So on 64bit LE systems we zero some bytes near by and on BE systems we
get a size of zero.
Signed-off-by: Sebastian Andrzej Siewior <sebastian@breakpoint.cc> Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Adrian Hunter [Thu, 12 Feb 2009 08:53:18 +0000 (10:53 +0200)]
mkfs.ubifs: eliminate compiler warnings
The warnings were:
lpt.c: In function ‘create_lpt’:
lpt.c:552: warning: format ‘%d’ expects type ‘int’, but argument 3 has type ‘long long int’
mkfs.ubifs.c: In function ‘do_openat’:
mkfs.ubifs.c:251: warning: ignoring return value of ‘chdir’, declared with attribute warn_unused_result
mkfs.ubifs.c: In function ‘get_options’:
mkfs.ubifs.c:542: warning: format not a string literal and no format arguments
Signed-off-by: Adrian Hunter <ext-adrian.hunter@nokia.com> Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
linux-mtd-owner@lists.infradead.org [Wed, 11 Feb 2009 09:05:39 +0000 (04:05 -0500)]
Artem Bityutskiy said once:
|I mean, IMO most of these warnings are not really sane, and cleaning
|that up just adds mess - you stop understanding why is something signed
|or unsigned :-)
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
linux-mtd-owner@lists.infradead.org [Wed, 11 Feb 2009 09:04:39 +0000 (04:04 -0500)]
I have here a mtd part which is 3 GiB with a flash page size of 256KiB.
The 2GiB limit is at erase block 8192. In mtd_is_bad() the computation
for the MEMGETBADBLOCK ioctl() looks like the following:
| seek = eb * mtd->eb_size;
with both eb and mtd->eb_size being a signed int results in seek being a
signed result.
The _FILE_OFFSET_BITS=64 define is required to switch off_t from 32bit
to 64bit an 32bit systems. This is required in order to keep using
lseek() as lseek64 on 32bit system. Without this change lseek() in
mtd_read() is called with a 32bit value with most significat bit set and
the kernel performs a sign extension for the 64bit value which is used
in the mtd layer.
The last change also changes the size of the parameter which is passed
to the MEMGETBADBLOCK ioctl() from 32 to 64bit. The counter part in
kernel is also defined as loff_t which is of type __kernel_loff_t and
this is "long long". So this must have been broken for a while unless I
missed something.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Artem Bityutskiy [Wed, 28 Jan 2009 12:19:43 +0000 (14:19 +0200)]
mkfs.ubifs: support reserved space allocation
UBIFS may reserve some amount of flash space for the super-user,
ant this amount is stored in the superblock. Currently mkfs.ubifs
always puts 0 there. This patch adds -R option which may be used
to specify size of the reserved space.
Sebastian Andrzej Siewior [Wed, 17 Dec 2008 10:15:54 +0000 (11:15 +0100)]
ubiupdatevol: fix -t parameter
The execution of
|./ubiupdatevol /dev/ubi0 -t
will fail because 'argv[optind + 1]' is undefined and the later executed
'strcmp(args.img, "-")' will segfault.
So I can hack around and supply a dummy image or fix it that way.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Mike Frysinger [Mon, 12 Jan 2009 16:08:03 +0000 (11:08 -0500)]
common.mk: tweak rules to workaround make-3.80 bugs
I got some reports from people who use make-3.80 that mtd-utils wasn't
building correctly the first time. Turns out that older versions of make
misbehave with pattern rules and full paths. So I've tweaked the code a
little to work with make-3.80 and make-3.81 (the latest release).
Also, I added a small optimization to avoid running `mkdir` when building
in-tree.
Signed-off-by: Mike Frysinger <vapier@gentoo.org> Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
projects / mtd-utils.git / log