On 10/01/2011 09:46 AM, Stuart Hughes wrote:
>
> Hi Jurgen,
>
> Thanks for the patch. I'll try to get that in as time permits.
>
> Please send your UBU and I'll try to get that in too.
>
In attach 2 cvs unified diffs and a new file.
This adds UBI support.
Mark: you need a recent mtd-utils on your PC (=host)! I compiled it
myself: v1.4.6.
And also the mtd-utils on the target be recent if you want to run ubi
commands in linux, e.g. ubiupdatevol!
So I first cross-compiled mtd-utils 1.4.6 (already mailed it), but now I
just use the mtd-utils tools inside busybox. It does not have ubinfo,
but the necessary tools are there and work.
Enjoy,
Jürgen
>
>
> Please post your imx27pdk, maybe someone else on the list would have
> time to clean it up and re-submit it. Even if not, it could be helpful
> to others.
>
> Regards, Stuart
>
[snip]
--
Jürgen Lambrecht
R&D Associate
Tel: +32 (0)51 303045 Fax: +32 (0)51 310670
http://www.televic-rail.com
Televic Rail NV - Leo Bekaertlaan 1 - 8870 Izegem - Belgium
Company number 0825.539.581 - RPR Kortrijk
Index: config/userspace/deployment.lkc
===================================================================
RCS file: /sources/ltib/ltib/config/userspace/deployment.lkc,v
retrieving revision 1.6
diff -U 3 -r1.6 deployment.lkc
--- config/userspace/deployment.lkc 28 Dec 2009 11:00:18 -0000 1.6
+++ config/userspace/deployment.lkc 3 Oct 2011 13:10:27 -0000
@@ -20,6 +20,28 @@
help
build a yaffs2 flash filesystem image
+ config DEPLOYMENT_UBIFS
+ bool "ubifs"
+ help
+ build a ubi flash filesystem image
+ Apart from the options below, also the /bin/ubifs_ubinize_tmpl.ini
+ file is used to create a partition with 1 rootfs volume
+ (http://www.linux-mtd.infradead.org/faq/ubi.html#L_ubi_mkimg):
+ * image: name of the image file; automatically filled in
+ * vol_size: volume size : for fixed size; or NA with autoresize
+ flag, so left away here (if absent, the volume size is assumed to
+ be equivalent to the size of the image file).
+ * vol_type: 'dynamic' for RW and 'static' for RO
+ * vol_name: name used during mounting, so must match the name given
+ by the bootloader to the kernel!
+ * vol_alignment: put to 1; this is only useful, for example, when
+ one wants to maintain a block device on top of an UBI volume. In
+ this case, it is desirable to fit an integer number of blocks in
+ logical eraseblocks of this UBI volume.
+ * vol_flags: autoresize flag: then autoresize if you want to use
+ all available LEBs (because some PEBs are bad, it is not possible
+ to know in forehand what the maximum available LEBs are)
+
config DEPLOYMENT_RAMDISK
bool "ext2.gz ramdisk"
help
@@ -56,6 +78,7 @@
if CAP_DEPLOYMENT_ELF && (DEPLOYMENT_JFFS2 \
|| DEPLOYMENT_YAFFS2 \
+ || DEPLOYMENT_UBIFS \
|| DEPLOYMENT_RAMDISK)
config DEPLOYMENT_ELF
bool "Create a combined ELF image"
@@ -75,6 +98,7 @@
string
default JFFS2 if DEPLOYMENT_JFFS2
default YAFFS2 if DEPLOYMENT_YAFFS2
+ default UBIFS if DEPLOYMENT_UBIFS
default RAMDISK if DEPLOYMENT_RAMDISK
default CRAMFS if DEPLOYMENT_CRAMFS
default ROMFS if DEPLOYMENT_ROMFS
@@ -107,6 +131,96 @@
endif
+if DEPLOYMENT_UBIFS
+
+config DEPLOYMENT_MIN_I0_UNIT_SIZE
+ string "ubifs minimum I/O unit size"
+ default 2048
+ help
+ This set the UBIFS minimum I/O unit size in Byte.
+ Is usually 1B for NOR flashes, and agrees with the page size of NAND
+ flashes (without spare).
+ See http://www.linux-mtd.infradead.org/doc/ubi.html#L_min_io_unit
+
+config DEPLOYMENT_SUB_PAGE_SIZE
+ string "ubifs sub-page size"
+ default 512
+ help
+ This set the UBIFS sub-page size in Byte - only for NAND flash.
+ In practice, this is the minimum unit size for ECC calculation. See
+ http://www.linux-mtd.infradead.org/doc/ubi.html#L_subpage.
+ Easiest way to discover this (and all) parameter is to run linux on
+ your board (from NFS or JFFS2), keep a spare partition in flash, and do
+ 'ubiformat' and 'ubiattach'; the latter will show all parameters.
+
+config DEPLOYMENT_PEB_SIZE
+ string "ubifs physical erase block (PEB) size"
+ default 128KiB
+ help
+ This sets the UBIFS Physical erase block (PEB) size of your Chip.
+ Is your Chip (erase) "block size" without spare (spare area is reserved
+ by UBI for ECC data). Size might be specified kilobytes (KiB),
+ megabytes (MiB), gigabytes (GiB) or bytes (no modifier).
+
+config DEPLOYMENT_LEB_SIZE
+ string "ubifs logical erase block (LEB) size in Bytes"
+ default 126KiB
+ help
+ This set the UBIFS Logical Erase Block (LEB) size in bytes. This has to
+ be smaller than the PEB size because of the UBI EC and VID headers (see
+ http://www.linux-mtd.infradead.org/doc/ubi.html#L_overhead): 2 pages if
+ the flash has no sub-pages, else 1 page (nand page = minimum I/O unit).
+
+config DEPLOYMENT_MAX_LEB_COUNT
+ string "ubifs maximum logical erase block (LEB) count"
+ default 4096
+ help
+ This set maximum LEB count, the logical maximum size of the image =
+ (LEB_COUNT * LEB_SIZE). This is the maximum volume size you want to
+ support for this image to be put on. Don't exagerate, because it will
+ cost performance.
+ See http://www.linux-mtd.infradead.org/faq/ubifs.html#L_max_leb_cnt.
+
+choice
+ prompt "ubifs compression type"
+ default DEPLOYMENT_UBIFS_NONE
+ help
+ Default UBIFS build without compression.
+
+ config DEPLOYMENT_UBIFS_NONE
+ bool "no compression"
+ help
+ Choose to build ubifs image without compression.
+
+ config DEPLOYMENT_UBIFS_LZO
+ bool "LZO compression"
+ help
+ choose lzo compression ubifs image, lzo compressor is
+ generally faster than zlib but compresses worse.
+
+ config DEPLOYMENT_UBIFS_FAVOR_LZO
+ bool "try both compressions"
+ help
+ It is a method of combining "lzo" and "zlib" compressors. Namely,
+ mkfs.ubifs tries to compress data with both "lzo" and "zlib"
+ compressors, then it compares which compressor is better.
+
+ config DEPLOYMENT_UBIFS_ZLIB
+ bool "zlib compression"
+ help
+ zlib compresses better than LZO but it is slower.
+
+endchoice
+
+config DEPLOYMENT_UBIFS_COMPR_TYPE
+ string
+ default none if DEPLOYMENT_UBIFS_NONE
+ default lzo if DEPLOYMENT_UBIFS_LZO
+ default favor_lzo if DEPLOYMENT_UBIFS_FAVOR_LZO
+ default zlib if DEPLOYMENT_UBIFS_ZLIB
+
+endif
+
config DEPLOYMENT_ZIP
depends DEPLOYMENT_ROMFS
bool "Create a compressed image of the kernel/romfs image"
@@ -136,7 +250,7 @@
default n
config SYSCFG_TMPFS_SIZE
- depends SYSCFG_READONLY_FS || DEPLOYMENT_JFFS2 || DEPLOYMENT_YAFFS2
+ depends SYSCFG_READONLY_FS || DEPLOYMENT_JFFS2 || DEPLOYMENT_YAFFS2 || DEPLOYMENT_UBIFS
string "tmpfs size"
default "512k"
help
@@ -147,16 +261,16 @@
from the total available RAM to the system.
config SYSCFG_RAM_DIRS
- depends SYSCFG_READONLY_FS || DEPLOYMENT_JFFS2 || DEPLOYMENT_YAFFS2
+ depends SYSCFG_READONLY_FS || DEPLOYMENT_JFFS2 || DEPLOYMENT_YAFFS2 || DEPLOYMENT_UBIFS
string "Place these dirs in writable RAM"
default "/tmp /etc /var" if SYSCFG_READONLY_FS
- default "/tmp /var" if DEPLOYMENT_JFFS2 || DEPLOYMENT_YAFFS2
+ default "/tmp /var" if DEPLOYMENT_JFFS2 || DEPLOYMENT_YAFFS2 || DEPLOYMENT_UBIFS
help
Flash may only be erased a finite number of times (of the order
- of 100000 times for a NOR device). If you deploy to Flash,
- frequently written files should be written to RAM to avoid
- exhausting Flash. List here a space separated list of directories
- to be put in tmpfs (RAM). Changes to these directories will
+ of 100000 times for a NOR device and a NAND device with good ECC).
+ If you deploy to Flash, frequently written files should be written to
+ RAM to avoid exhausting Flash. List here a space separated list of
+ directories to be put in tmpfs (RAM). Changes to these directories will
be lost on reboot. You may also want to use this option if
you choose to deploy a read-only root filesystem. Note that
if you have chosen a read-only root filesystem, the default
Index: bin/Ltibutils.pm
===================================================================
RCS file: /sources/ltib/ltib/bin/Ltibutils.pm,v
retrieving revision 1.38
diff -U 3 -r1.38 Ltibutils.pm
--- bin/Ltibutils.pm 7 Aug 2011 15:05:53 -0000 1.38
+++ bin/Ltibutils.pm 3 Oct 2011 13:05:38 -0000
@@ -697,6 +688,10 @@
then
rm -f $tdir/rootfs.yaffs2
fi
+if [ "$pcf->{DEPLOYMENT_UBIFS}" = "y" ]
+then
+ rm -f $tdir/rootfs.ubifs.img
+fi
if [ "$pcf->{DEPLOYMENT_CRAMFS}" = "y" ]
then
rm -f $tdir/cramfs.*
@@ -856,6 +851,9 @@
# genext2fs seems to use 1k block size
my $blocks = $fs_size;
my $inodes = $fs_count + 400;
+ my $ubinize_config_tmpl = "bin/ubifs_ubinize_tmpl.ini";
+ my $ubinize_config_file = "$tdir/ubinize.ini";
+ my $ubinize_image_location_str = "UBI_IMAGE_LOCATION";
# try to sanely guess the endian of the target
my $endian = '-l';
@@ -890,6 +888,7 @@
fi
if [ "$pcf->{DEPLOYMENT_JFFS2}" = "y" ]
then
+ echo "creating jffs2 image: $tdir/rootfs.jffs2"
mkfs.jffs2 -n $pad_opt -D $dev_tab -U $endian -e $pcf->{DEPLOYMENT_ERASE_BLOCK_SIZE} -d $stage -o $tdir/rootfs.jffs2
ln -sf $tdir/rootfs.jffs2 $tdir/rootfs_image
fi
@@ -898,6 +897,25 @@
mkfs.yaffs2 -r -p $stage/etc/passwd -N -D $dev_tab $endian $stage $tdir/rootfs.yaffs2
ln -sf $tdir/rootfs.yaffs2 $tdir/rootfs_image
fi
+if [ "$pcf->{DEPLOYMENT_UBIFS}" = "y" ]
+then
+ echo "creating ubifs image: $tdir/rootfs.ubifs.img"
+ mkfs.ubifs -U -D $dev_tab -d $stage -o $tdir/rootfs.ubifs \\
+ -m $pcf->{DEPLOYMENT_MIN_I0_UNIT_SIZE} \\
+ -e $pcf->{DEPLOYMENT_LEB_SIZE} \\
+ -c $pcf->{DEPLOYMENT_MAX_LEB_COUNT} \\
+ -x $pcf->{DEPLOYMENT_UBIFS_COMPR_TYPE}
+ # genrate the ini file from .tmpl file.
+ sed -e 's:$ubinize_image_location_str:$tdir/rootfs.ubifs:g' \\
+ $ubinize_config_tmpl > $ubinize_config_file
+ ubinize -o $tdir/rootfs.ubifs.img \\
+ -m $pcf->{DEPLOYMENT_MIN_I0_UNIT_SIZE} \\
+ -s $pcf->{DEPLOYMENT_SUB_PAGE_SIZE} \\
+ -p $pcf->{DEPLOYMENT_PEB_SIZE} $ubinize_config_file
+ rm $ubinize_config_file
+ rm $tdir/rootfs.ubifs
+ ln -sf $tdir/rootfs.ubifs.img $tdir/rootfs_image
+fi
if [ "$pcf->{DEPLOYMENT_CRAMFS}" = "y" ]
then
mkfs.cramfs -q -D $dev_tab $endian $stage $tdir/rootfs.cramfs
[rootfs]
mode=ubi
image=UBI_IMAGE_LOCATION
vol_id=0
vol_type=dynamic
vol_name=root
vol_alignment=1
vol_flags=autoresize
_______________________________________________
LTIB home page: http://ltib.org
Ltib mailing list
[email protected]
https://lists.nongnu.org/mailman/listinfo/ltib
