profile/common/platform/kernel/u-boot-artik.git
9 years agopart_efi: make sure the gpt_pte is freed
Mark Langsdorf [Tue, 10 Sep 2013 20:14:56 +0000 (15:14 -0500)]
part_efi: make sure the gpt_pte is freed

the gpt_pte wasn't being freed if it was checked against an invalid
partition. The resulting memory leakage could make it impossible
to repeatedly attempt to load non-existent files in a script.

Also, downgrade the message for not finding an invalid partition
from a printf() to a debug() so as to minimize message spam in
perfectly normal situations.

Signed-off-by: Mark Langsdorf <mark.langsdorf@calxeda.com>
9 years agofs: fat: don't call disk_write with zero sector num
Wu, Josh [Wed, 24 Jul 2013 09:55:30 +0000 (17:55 +0800)]
fs: fat: don't call disk_write with zero sector num

In the set_cluster() function, it will convert the buffer size to sector
numbers. Then call disk_write() to write by sector.
For remaining buffer, the size is less than a sector, call disk_write()
again to write them in one sector.

But if the total buffer size is less then one sector, the original code
will call disk_write() with zero sector number. It is unnecessary.
So this patch fix this. Now it will not call disk_write() if total buffer size
is less than one sector.

Signed-off-by: Josh Wu <josh.wu@atmel.com>
9 years agousb:g_dnl:dfu: Download gadget and DFU function code clean up
Lukasz Majewski [Tue, 17 Sep 2013 13:58:23 +0000 (15:58 +0200)]
usb:g_dnl:dfu: Download gadget and DFU function code clean up

The download gadget code and DFU function lacks of proper declarations
for the case when a target board wants to use only one of available usb
functions.

Moreover the relevant declarations have been moved to consistent
localization (like <dfu.h>).

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Cc: Marek Vasut <marex@denx.de>
9 years agousb:gadget:Remove redundant #includes for USB composite gadget and its functions
Lukasz Majewski [Tue, 17 Sep 2013 13:58:22 +0000 (15:58 +0200)]
usb:gadget:Remove redundant #includes for USB composite gadget and its functions

Only the <linux/usb/gadget.h> requires error.h include. Hence, several
includes of error.h at USB gadget functions are not needed.

Moreover unnecessary malloc.h includes were also removed.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Cc: Marek Vasut <marex@denx.de>
9 years agousb:g_dnl:ums: Conditional compilation for mass storage function (f_mass_storage)
Lukasz Majewski [Tue, 17 Sep 2013 13:58:21 +0000 (15:58 +0200)]
usb:g_dnl:ums: Conditional compilation for mass storage function (f_mass_storage)

The mass storage composite function is now compiled in only when
CONFIG_USB_GADGET_MASS_STORAGE is defined.
Such change provides binary size reduction for boards which use USB
download gadget (like am335x_evm) with DFU, but don't use UMS.

For example at am335x_evm board reduction is more than 2KiB for
text and around 120B for data.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Cc: Marek Vasut <marex@denx.de>
9 years agodfu: ram support
Afzal Mohammed [Tue, 17 Sep 2013 19:45:24 +0000 (01:15 +0530)]
dfu: ram support

DFU spec mentions it as a method to upgrade firmware (software stored
in writable non-volatile memory). It also says other potential uses of
DFU is beyond scope of the spec.

Here such a beyond the scope use is being attempted - directly pumping
binary images from host via USB to RAM. This facility is a developer
centric one in that it gives advantage over upgrading non-volatile
memory for testing new images every time during development and/or
testing.

Directly putting image onto RAM would speed up upgrade process. This and
convenience was the initial thoughts that led to doing this, speed
improvement over MMC was only 1 second though - 6 sec on RAM as opposed
to 7 sec on MMC in beagle bone, perhaps enabling cache and/or optimizing
DFU framework to avoid multiple copy for ram (if worth) may help, and
on other platforms and other boot media like NAND maybe improvement
would be higher.

And for a platform that doesn't yet have proper DFU suppport for
non-volatile media's, DFU to RAM can be used.

Another minor advantage would be to increase life of mmc/nand as it
would be less used during development/testing.

usage: <image name> ram <start address> <size>
eg. kernel ram 0x81000000 0x1000000

Downloading images to RAM using DFU is not something new, this is
acheived in openmoko also.

DFU on RAM can be used for extracting RAM contents to host using dfu
upload. Perhaps this can be extended to io for squeezing out register
dump through usb, if it is worth.

Signed-off-by: Afzal Mohammed <afzal.mohd.ma@gmail.com>
Cc: Heiko Schocher <hs@denx.de>
Cc: Marek Vasut <marex@denx.de>
Cc: Lukasz Majewski <l.majewski@samsung.com>
Cc: Pantelis Antoniou <panto@antoniou-consulting.com>
Cc: Gerhard Sittig <gsi@denx.de>
Acked-by: Marek Vasut <marex@denx.de>
Acked-by: Lukasz Majewski <l.majewski@samsung.com>
Acked-by: Heiko Schocher <hs@denx.de>
9 years agodfu: unify mmc/nand read/write ops enum
Afzal Mohammed [Tue, 17 Sep 2013 19:44:50 +0000 (01:14 +0530)]
dfu: unify mmc/nand read/write ops enum

MMC and NAND independently defines same enumerators for read/write.
Unify them by defining enum in dfu header. RAM support that is being
added newly also can make use of it.

Signed-off-by: Afzal Mohammed <afzal.mohd.ma@gmail.com>
Cc: Heiko Schocher <hs@denx.de>
Cc: Marek Vasut <marex@denx.de>
Cc: Lukasz Majewski <l.majewski@samsung.com>
Cc: Pantelis Antoniou <panto@antoniou-consulting.com>
Acked-by: Lukasz Majewski <l.majewski@samsung.com>
Acked-by: Marek Vasut <marex@denx.de>
Acked-by: Heiko Schocher <hs@denx.de>
9 years agodfu: Extract common DFU code to handle "dfu_alt_info" environment variable
Lukasz Majewski [Wed, 11 Sep 2013 12:53:35 +0000 (14:53 +0200)]
dfu: Extract common DFU code to handle "dfu_alt_info" environment variable

New dfu_init_env_entities() function has been extracted from cmd_dfu.c and
stored at dfu core.

This is a dfu centric code, so it shall be processed in the core.

Change-Id: I756c5de922fa31399d8804eaadc004ee98844ec2
Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Tested-by: Heiko Schocher <hs@denx.de>
9 years agodfu: Make maximum DFU file size equal to default DFU data buffer
Lukasz Majewski [Tue, 10 Sep 2013 13:29:23 +0000 (15:29 +0200)]
dfu: Make maximum DFU file size equal to default DFU data buffer

Up till now the DFU maximum file size (to be written to e.g. eMMC)
was different from the DFU data buffer size. It caused errors when
one buffer was smaller than data to be written.

Now, the maximum DFU file size is equal to default DFU buffer size.
In spite of this, user is still able to manually adjust those default
values.

Change-Id: Ied75d0f7b59588ebd79dae9a22af801d36622216
Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
9 years agodfu:cosmetic: Fix printf text for buffer overflow condition
Lukasz Majewski [Tue, 10 Sep 2013 13:29:22 +0000 (15:29 +0200)]
dfu:cosmetic: Fix printf text for buffer overflow condition

Correct error message if overflow is detected.

Change-Id: I8a915c7353d49822c046fbc36241237b370e6c98
Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
9 years agodfu, nand, ubi: add partubi alt settings for updating ubi partition
Heiko Schocher [Thu, 25 Jul 2013 04:43:11 +0000 (06:43 +0200)]
dfu, nand, ubi: add partubi alt settings for updating ubi partition

updating an ubi partition needs a completely erased mtd partition,
see:
http://lists.infradead.org/pipermail/linux-mtd/2011-May/035416.html

So, add partubi alt setting for the dfu_alt_info environment
variable to mark this partition as an ubi partition. In case we
update an ubi partition, we erase after flashing the image into the
partition, the remaining sektors.

Signed-off-by: Heiko Schocher <hs@denx.de>
Cc: Pantelis Antoniou <panto@antoniou-consulting.com>
Cc: Tom Rini <trini@ti.com>
Cc: Lukasz Majewski <l.majewski@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Marek Vasut <marex@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Scott Wood <scottwood@freescale.com>
9 years agousb:dfu:g_dnl: Refactoring the string definition code for g_dnl driver
Lukasz Majewski [Mon, 19 Aug 2013 15:17:19 +0000 (17:17 +0200)]
usb:dfu:g_dnl: Refactoring the string definition code for g_dnl driver

The manufacturer and product IDs are dynamically assigned when gadget is
bind.
Now the IDs aren't assigned at struct g_dnl_string_defs definition.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Cc: Marek Vasut <marex@denx.de>
Cc: "Egli, Samuel" <samuel.egli@siemens.com>
9 years agousb:dfu:g_dnl: Change number of exported configurations at composite gadget
Lukasz Majewski [Mon, 19 Aug 2013 15:17:18 +0000 (17:17 +0200)]
usb:dfu:g_dnl: Change number of exported configurations at composite gadget

USB composite gadget (g_dnl) supports only one configuration. Due to that
the corresponding field - bConfigurationValue has been changed.

Moreover more descriptive names were chosen for relevant fields.

Windows XP setup:
- Thesyscon USB Descriptor Dumper
- zadig_xp program for WinUSB installation (which is required by dfu-util)
- dfu-util for windows (version 0.6)
- TRATS target connected via USB hub to test Win XP machine.

Tested at: Trats - Exynos4210

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Cc: Marek Vasut <marex@denx.de>
Cc: "Egli, Samuel" <samuel.egli@siemens.com>
9 years agodfu: Implementation of target reset after communication with dfu-util's -R switch
Lukasz Majewski [Thu, 18 Jul 2013 11:19:14 +0000 (13:19 +0200)]
dfu: Implementation of target reset after communication with dfu-util's -R switch

This patch extends dfu code to support transmission with -R switch
specified at dfu-util.

When -R is specified, the extra USB_REQ_DFU_DETACH request is sent after
successful data transmission. Then dfu resources are released and reset
command is issued.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
9 years agodfu: Update DFU's authorship history
Łukasz Majewski [Fri, 5 Jul 2013 09:40:47 +0000 (11:40 +0200)]
dfu: Update DFU's authorship history

The DFU's state machine original implementation author and copyright were
missing.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Acked-by: Stefan Schmidt <stefan@datenfreihafen.org>
9 years agodfu:function: Fix number of allocated DFU function pointers
Lukasz Majewski [Wed, 26 Jun 2013 09:46:13 +0000 (11:46 +0200)]
dfu:function: Fix number of allocated DFU function pointers

This subtle change fix problem with too small amount of allocated
memory to store DFU function pointers.

One needs to allocate extra space for sentinel NULL pointer in this array
of function pointers.

With the previous code, the NULL value overwrites malloc internal data
and afterwards free(f_dfu->function) crashes.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Marek Vasut <marex@denx.de>
9 years agousb, g_dnl: make possibility to fixup the device_desc board specific
Heiko Schocher [Tue, 4 Jun 2013 09:19:50 +0000 (11:19 +0200)]
usb, g_dnl: make possibility to fixup the device_desc board specific

add a weak dummy function g_dnl_fixup to add the possibility to update
the device_desc board specific. Used on the upcoming siemens board
support, where idVendor and idProduct is stored in an eeprom.

Signed-off-by: Heiko Schocher <hs@denx.de>
Cc: Marek Vasut <marek.vasut@gmail.com>
Cc: Lukasz Majewski <l.majewski@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Reviewed-by: Lukasz Majewski <l.majewski@samsung.com>
9 years agousb, dfu gadget: switch to dfu mode in dfu_bind
Heiko Schocher [Tue, 25 Jun 2013 11:59:29 +0000 (13:59 +0200)]
usb, dfu gadget: switch to dfu mode in dfu_bind

- set in to_dfu_mode() f_dfu->dfu_state = DFU_STATE_dfuIDLE
  as after every to_dfu_mode call this is done, so move
  this into to_dfu_mode
- switch in dfu_bind() into dfu mode:
  This fixes wrong "dfu-util -l" output, when calling
  "dfu-util -l" after a board reset, without doing a
  download before. See also discussion here:

http://lists.denx.de/pipermail/u-boot/2013-June/157272.html

Signed-off-by: Heiko Schocher <hs@denx.de>
Cc: Lukasz Majewski <l.majewski@samsung.com>
Cc: Pantelis Antoniou <panto@antoniou-consulting.com>
Cc: Tom Rini <trini@ti.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Marek Vasut <marex@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Samuel Egli <samuel.egli@siemens.com>
Acked-by: Lukasz Majewski <l.majewski@samsung.com>
Tested-by: Lukasz Majewski <l.majewski@samsung.com>
9 years agousb:gadget: USB Mass Storage Gadget support
Lukasz Majewski [Tue, 5 Mar 2013 11:10:17 +0000 (12:10 +0100)]
usb:gadget: USB Mass Storage Gadget support

This patch adds the USB Mass Storage Gadget to u-boot
New command called "ums" is implemented to provide access
to on-device embedded persistent memory.

USB Mass Storage is supposed to work on top of the USB
Gadget framework

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Piotr Wilczek <p.wilczek@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
CC: Marek Vasut <marek.vasut@gmail.com>
Conflicts:
common/Makefile

9 years agodfu: Properly zero out timeout value
Pantelis Antoniou [Fri, 30 Nov 2012 08:01:10 +0000 (08:01 +0000)]
dfu: Properly zero out timeout value

Zero out timeout value; letting it filled with undefined values
ends up with the dfu host hanging.

Signed-off-by: Pantelis Antoniou <panto@antoniou-consulting.com>
9 years agog_dnl: Properly terminate string list.
Pantelis Antoniou [Fri, 30 Nov 2012 08:01:06 +0000 (08:01 +0000)]
g_dnl: Properly terminate string list.

Well, not terminating the list causes very interesting crashes.
As in changing the vendor & product ID crashes. Fun.

Signed-off-by: Pantelis Antoniou <panto@antoniou-consulting.com>
9 years agog_dnl: Issue connect/disconnect as appropriate
Pantelis Antoniou [Fri, 30 Nov 2012 08:01:05 +0000 (08:01 +0000)]
g_dnl: Issue connect/disconnect as appropriate

Call usb_gadget_connect/usb_gadget_disconnect in g_dnl_bind/g_dnl_unbind.

Signed-off-by: Pantelis Antoniou <panto@antoniou-consulting.com>
9 years agoRemove obsolete header file
Pantelis Antoniou [Wed, 28 Nov 2012 02:43:54 +0000 (02:43 +0000)]
Remove obsolete header file

usbdescriptors.h conflicts with linux/usb/ch9.h
Remove it.

Signed-off-by: Pantelis Antoniou <panto@antoniou-consulting.com>
9 years agodfu:usb: DFU USB function (f_dfu) support for g_dnl composite gadget
Lukasz Majewski [Mon, 6 Aug 2012 12:41:06 +0000 (14:41 +0200)]
dfu:usb: DFU USB function (f_dfu) support for g_dnl composite gadget

Support for f_dfu USB function.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Marek Vasut <marex@denx.de>
9 years agodfu:usb: Support for g_dnl composite download gadget.
Lukasz Majewski [Mon, 6 Aug 2012 12:41:05 +0000 (14:41 +0200)]
dfu:usb: Support for g_dnl composite download gadget.

Composite USB download gadget support (g_dnl) for download functions.
This code works on top of composite gadget.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Marek Vasut <marex@denx.de>
9 years agoext4fs: le32_to_cpu() used on a 16-bit field
Rommel Custodio [Sun, 21 Jul 2013 08:53:25 +0000 (10:53 +0200)]
ext4fs: le32_to_cpu() used on a 16-bit field

Fix reading ext4_extent_header struture on BE machines.  Some 16 bit
fields where converted to 32 bit fields, due to the byte swap on BE
machines the containing value was corrupted. Therefore reading ext4
filesystems on BE machines where broken before.

Signed-off-by: Rommel Custodio <sessyargc+uboot@gmail.com>
[sent via git-send-email; rework commit message]
Signed-off-by: Andreas Bießmann <andreas.devel@googlemail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Simon Glass <sjg@chromium.org>
Tested-by: Lukasz Majewski <l.majewski@samsung.com>
9 years agodfu:ext4:fix Fix DFU upload functionality
Łukasz Majewski [Fri, 28 Jun 2013 16:41:50 +0000 (18:41 +0200)]
dfu:ext4:fix Fix DFU upload functionality

For the first eMMC read of data for upload, use the "large" dfu_buf (now
configurable) instead of usb request buffer allocated at composite layer
(which is 4KiB) [*].

For eMMC the whole file is read, which usually is larger than the buffer [*]
provided with usb request.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Cc: Tom Rini <trini@ti.com>
Cc: Pantelis Antoniou <panto@antoniou-consulting.com>
Cc: Marek Vasut <marex@denx.de>
Cc: Heiko Schocher <hs@denx.de>
9 years agodfu:ext4:fix Fix ext4{read|write} command formatting
Łukasz Majewski [Fri, 28 Jun 2013 16:41:49 +0000 (18:41 +0200)]
dfu:ext4:fix Fix ext4{read|write} command formatting

In the following commit:
"dfu: Support larger than memory transfers."
SHA1: ea2453d56b8860dbd18a3c517531ffc8dcb5c839

The ext4{read|write} command formatting has been changed. It removed
a write mandatory [sizebytes] parameter.

It extents DFU_FS_EXT4 case at mmc_file_op to provide mandatory
parameter for DFU write.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Cc: Tom Rini <trini@ti.com>
Cc: Pantelis Antoniou <panto@antoniou-consulting.com>
Cc: Marek Vasut <marex@denx.de>
Cc: Heiko Schocher <hs@denx.de>
9 years agodfu: make data buffer size configurable
Heiko Schocher [Wed, 12 Jun 2013 04:05:51 +0000 (06:05 +0200)]
dfu: make data buffer size configurable

Dfu transfer uses a buffer before writing data to the
raw storage device. Make the size (in bytes) of this buffer
configurable through environment variable "dfu_bufsiz".
Defaut value is configurable through CONFIG_SYS_DFU_DATA_BUF_SIZE

Signed-off-by: Heiko Schocher <hs@denx.de>
Cc: Pantelis Antoniou <panto@antoniou-consulting.com>
Cc: Tom Rini <trini@ti.com>
Cc: Lukasz Majewski <l.majewski@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Marek Vasut <marex@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
Acked-by: Tom Rini <trini@ti.com>
9 years agodfu: Change indentation of defines in <dfu.h>
Tom Rini [Thu, 14 Mar 2013 05:32:49 +0000 (05:32 +0000)]
dfu: Change indentation of defines in <dfu.h>

Signed-off-by: Tom Rini <trini@ti.com>
9 years agodfu, nand: before write a buffer to nand, erase the nand sectors
Heiko Schocher [Mon, 24 Jun 2013 16:50:40 +0000 (18:50 +0200)]
dfu, nand: before write a buffer to nand, erase the nand sectors

before writing the received buffer to nand, erase the nand
sectors. If not doing this, nand write fails. See for
more info here:

http://lists.denx.de/pipermail/u-boot/2013-June/156361.html

Using the nand erase option "spread", maybe overwrite
blocks on, for example another mtd partition, if the
erasing range contains bad blocks.
So a limit option is added to nand_erase_opts()

Signed-off-by: Heiko Schocher <hs@denx.de>
Cc: Scott Wood <scottwood@freescale.com>
Cc: Pantelis Antoniou <panto@antoniou-consulting.com>
Cc: Lukasz Majewski <l.majewski@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Marek Vasut <marex@denx.de>
Cc: Tom Rini <trini@ti.com>
Signed-off-by: Scott Wood <scottwood@freescale.com>
9 years agodisk: Fix possible out-of-bounds access in part_efi.c
Marek Vasut [Sun, 19 May 2013 12:53:34 +0000 (12:53 +0000)]
disk: Fix possible out-of-bounds access in part_efi.c

Make sure to never access beyond bounds of either EFI partition name
or DOS partition name. This situation is happening:

part.h:     disk_partition_t->name is 32-byte long
part_efi.h: gpt_entry->partition_name is 36-bytes long

The loop in part_efi.c copies over 36 bytes and thus accesses beyond
the disk_partition_t->name .

Fix this by picking the shortest of source and destination arrays and
make sure the destination array is cleared so the trailing bytes are
zeroed-out and don't cause issues with string manipulation.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Tom Rini <trini@ti.com>
Cc: Simon Glass <sjg@chromium.org>
9 years agodisk/part_dos: check harder for partition table
Egbert Eich [Tue, 9 Apr 2013 05:46:14 +0000 (05:46 +0000)]
disk/part_dos: check harder for partition table

Devices that used to have a whole disk FAT filesystem but got then
partitioned will most likely still have a FAT or FAT32 signature
in the first sector as this sector does not get overwritten by
a partitioning tool (otherwise the tool would risk to kill the mbr).

The current partition search algorithm will erronously detects such
a device as a raw FAT device.

Instead of looking for the FAT or FAT32 signatures immediately we
use the same algorithm as used by the Linux kernel and first check
for a valid boot indicator flag on each of the 4 partitions.
If the value of this flag is invalid for the first entry we then
do the raw partition check.
If the flag for any higher partition is wrong we assume the device
is neiter a MBR nor PBR device.

Signed-off-by: Egbert Eich <eich@suse.com>
9 years agofs: Add support for saving data to filesystems
Simon Glass [Sat, 20 Apr 2013 08:42:50 +0000 (08:42 +0000)]
fs: Add support for saving data to filesystems

Add a new method for saving that filesystems can implement. This mirrors the
existing load method.

Signed-off-by: Simon Glass <sjg@chromium.org>
9 years agodfu: NAND specific routines for DFU operation
Pantelis Antoniou [Thu, 14 Mar 2013 05:32:52 +0000 (05:32 +0000)]
dfu: NAND specific routines for DFU operation

Support for NAND storage devices to work with the DFU framework.

Signed-off-by: Pantelis Antoniou <panto@antoniou-consulting.com>
Signed-off-by: Tom Rini <trini@ti.com>
Acked-by: Scott Wood <scottwood@freescale.com>
9 years agodfu: Support larger than memory transfers.
Pantelis Antoniou [Thu, 14 Mar 2013 05:32:48 +0000 (05:32 +0000)]
dfu: Support larger than memory transfers.

Previously we didn't support upload/download larger than available
memory.  This is pretty bad when you have to update your root filesystem
for example.

This patch removes that limitation (and the crashes when you transfered
any file larger than 4MB) by making raw image writes be done in chunks
and making file maximum size be configurable.

The sequence number is a 16 bit counter; make sure we handle rollover
correctly. This fixes the wrong transfers for large (> 256MB) images.

Also utilize a variable to handle initialization, so that we don't rely
on just the counter sent by the host.

Signed-off-by: Pantelis Antoniou <panto@antoniou-consulting.com>
Signed-off-by: Tom Rini <trini@ti.com>
Conflicts:
README

9 years agodfu: Increase DFU buffer size from 4MiB to 8MiB
Łukasz Majewski [Thu, 4 Apr 2013 04:32:58 +0000 (04:32 +0000)]
dfu: Increase DFU buffer size from 4MiB to 8MiB

Increase size of DMA buffer from 4MiB to 8MiB. This is necessary due to
uImage size increase.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
9 years agodisk: fix unaligned access in efi partitions
Marc Dietrich [Fri, 29 Mar 2013 07:57:10 +0000 (07:57 +0000)]
disk: fix unaligned access in efi partitions

start_sect is not aligned to a 4 byte boundary thus causing exceptions
on ARM platforms. Access this field via the get_unaligned_le32 macro.

Signed-off-by: Marc Dietrich <marvin24@gmx.de>
9 years agocmd_ext4: BREAK and correct ext4write parameter order
Tom Rini [Wed, 20 Mar 2013 04:21:38 +0000 (04:21 +0000)]
cmd_ext4: BREAK and correct ext4write parameter order

The ext4write command was taking the in-memory address and filename path
in reverse order from the rest of the filesystem read and write
commands.  This corrects the order to be the same as fatload, etc.

Signed-off-by: Tom Rini <trini@ti.com>
9 years agodfu:ext4:fix: Change ext4write command order of parameters
Łukasz Majewski [Thu, 4 Apr 2013 04:32:57 +0000 (04:32 +0000)]
dfu:ext4:fix: Change ext4write command order of parameters

Following commit:
"cmd_ext4: BREAK and correct ext4write parameter order"
SHA1:0171d52c410cbaa9290b1b214e695697c835bfe5

introduced cleanup of ext4write semantics to be consistent with other
filesystem's writing commands (e.g. fatwrite).
This commit provides correct ext4write command generation at DFU eMMC
code.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
9 years agodfu: Add a partition type target
Pantelis Antoniou [Fri, 30 Nov 2012 08:01:11 +0000 (08:01 +0000)]
dfu: Add a partition type target

Dealing with raw block numbers with the dfu is very annoying.
Introduce a partition method.

Signed-off-by: Pantelis Antoniou <panto@antoniou-consulting.com>
9 years agodfu:usb:fix: Read the "filesize" environment variable only when file read
Łukasz Majewski [Thu, 23 Aug 2012 23:33:56 +0000 (23:33 +0000)]
dfu:usb:fix: Read the "filesize" environment variable only when file read

The "filesize" environment variable shall be read only when relevant
file is read.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Reviewed-by: Marek Vasut <marex@denx.de>
9 years agodfu:usb: Support for ext4
Łukasz Majewski [Thu, 23 Aug 2012 23:33:55 +0000 (23:33 +0000)]
dfu:usb: Support for ext4

Support for ext4 file system handling at DFU.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Reviewed-by: Marek Vasut <marex@denx.de>
9 years agodfu: Fix crash when wrong number of arguments given
Pantelis Antoniou [Fri, 30 Nov 2012 08:01:08 +0000 (08:01 +0000)]
dfu: Fix crash when wrong number of arguments given

Fix obvious crash when not enough arguments are given to the dfu
command.

Signed-off-by: Pantelis Antoniou <panto@antoniou-consulting.com>
9 years agodfu: Only perform DFU board_usb_init() for TRATS
Pantelis Antoniou [Fri, 30 Nov 2012 08:01:07 +0000 (08:01 +0000)]
dfu: Only perform DFU board_usb_init() for TRATS

USB initialization shouldn't happen for all the boards.

Signed-off-by: Pantelis Antoniou <panto@antoniou-consulting.com>
9 years agodisk: define HAVE_BLOCK_DEVICE in a common place
Stephen Warren [Thu, 28 Feb 2013 15:03:46 +0000 (15:03 +0000)]
disk: define HAVE_BLOCK_DEVICE in a common place

This set of ifdefs is used in a number of places. Move its definition
somewhere common so it doesn't have to be repeated.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Simon Glass <sjg@chromium.org>
Acked-by: Tom Rini <trini@ti.com>
Signed-off-by: Tom Warren <twarren@nvidia.com>
9 years agosandbox: Add host filesystem
Simon Glass [Wed, 26 Dec 2012 09:53:35 +0000 (09:53 +0000)]
sandbox: Add host filesystem

This allows reading of files from the host filesystem in sandbox.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@ti.com>
9 years agofs: Move ls and read methods into ext4, fat
Simon Glass [Wed, 26 Dec 2012 09:53:33 +0000 (09:53 +0000)]
fs: Move ls and read methods into ext4, fat

It doesn't make a lot of sense to have these methods in fs.c. They are
filesystem-specific, not generic code. Add each to the relevant
filesystem and remove the associated #ifdefs in fs.c.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@ti.com>
9 years agofs: Use map_sysmem() on read
Simon Glass [Wed, 26 Dec 2012 09:53:32 +0000 (09:53 +0000)]
fs: Use map_sysmem() on read

This allows us to use filesystems on sandbox. It has no effect on other
architectures.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@ti.com>
9 years agofs: Tell probe functions where to put their results
Simon Glass [Wed, 26 Dec 2012 09:53:31 +0000 (09:53 +0000)]
fs: Tell probe functions where to put their results

Rather than rely on global variables for the probe functions, pass in
the information that we need filled in. This allows us to potentially
keep the variables private to fs.c in the future, and the meaning of
the probe function is clearer.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@ti.com>
9 years agofs: Use filesystem methods instead of switch()
Simon Glass [Wed, 26 Dec 2012 09:53:30 +0000 (09:53 +0000)]
fs: Use filesystem methods instead of switch()

We can use the available methods and avoid using switch(). When the
filesystem is not supported, we fall through to the 'unsupported'
methods: fs_probe_unsupported() prints an error, so the others do
not need to.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@ti.com>
9 years agofs: Fully populate the filesystem method struct
Simon Glass [Wed, 26 Dec 2012 09:53:29 +0000 (09:53 +0000)]
fs: Fully populate the filesystem method struct

There is a structure in fs.c with just a probe method. By adding methods
for other operations, we can avoid lots of #ifdefs and switch()s. As a
first step, create the structure ready for use.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@ti.com>
9 years agoext4: Split write support into its own file
Simon Glass [Wed, 26 Dec 2012 09:53:28 +0000 (09:53 +0000)]
ext4: Split write support into its own file

This code seems to be entirely othogonal, so remove the #ifdef and put
the condition in the Makefile instead.

Signed-off-by: Simon Glass <sjg@chromium.org>
9 years agofs: Use new numeric setenv functions
Simon Glass [Sun, 24 Feb 2013 17:33:23 +0000 (17:33 +0000)]
fs: Use new numeric setenv functions

Use setenv_ulong(), setenv_hex() and setenv_addr() in fs/

Signed-off-by: Simon Glass <sjg@chromium.org>
9 years agogpt: fix partion size limit
Piotr Wilczek [Sun, 27 Jan 2013 22:59:25 +0000 (22:59 +0000)]
gpt: fix partion size limit

Currently, in gpt command, partion size is converted from string
to unsigned long type using 'ustrtol' function. That type limits
the partition size to 4GB.

This patch changes the conversion function to 'ustrtoll' to return
unsigned long long type.

Signed-off-by: Piotr Wilczek <p.wilczek@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
9 years agoFAT: remove ifdefs to make the code more readable
Richard Genoud [Thu, 13 Dec 2012 03:30:10 +0000 (03:30 +0000)]
FAT: remove ifdefs to make the code more readable

ifdefs in the code are making it harder to read.
The use of simple if(vfat_enabled) makes no more code and is cleaner.
(the code is discarded by the compiler instead of the preprocessor.)
NB: if -O0 is used, the code won't be discarded

and bonus, now the code compiles even if CONFIG_SUPPORT_VFAT is not
defined.

Signed-off-by: Richard Genoud <richard.genoud@gmail.com>
9 years agoFAT: use toupper/tolower instead of recoding them
Richard Genoud [Thu, 13 Dec 2012 00:47:36 +0000 (00:47 +0000)]
FAT: use toupper/tolower instead of recoding them

toupper/tolower function are already declared, so use them.

Signed-off-by: Richard Genoud <richard.genoud@gmail.com>
Acked-by: Marek Vasut <marex@denx.de>
Acked-by: Stefano Babic <sbabic@denx.de>
9 years agovfat: Fix mkcksum argument sizes
Marek Vasut [Fri, 11 Jan 2013 03:35:48 +0000 (03:35 +0000)]
vfat: Fix mkcksum argument sizes

In case a function argument is known/fixed size array in C, the argument is
still decoyed as pointer instead ( T f(U n[k]) ~= T fn(U *n) ) and therefore
calling sizeof on the function argument will result in the size of the pointer,
not the size of the array.

The VFAT code contains such a bug, this patch fixes it.

Reported-by: Aaron Williams <Aaron.Williams@cavium.com>
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Tom Rini <tom.rini@gmail.com>
Cc: Aaron Williams <Aaron.Williams@cavium.com>
Tested-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Joe Hershberger <joe.hershberger@ni.com>
9 years agogpt: Support for new "gpt" command
Piotr Wilczek [Tue, 11 Dec 2012 10:09:47 +0000 (11:09 +0100)]
gpt: Support for new "gpt" command

New command - "gpt" is supported. It restores the GPT partition table.
It looks into the given environment variable for partitions definition.
It can be enabled at target configuration file with CONFIG_CMD_GPT.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Piotr Wilczek <p.wilczek@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
9 years agogpt: Support for GPT (GUID Partition Table) restoration
Lukasz Majewski [Tue, 11 Dec 2012 10:09:46 +0000 (11:09 +0100)]
gpt: Support for GPT (GUID Partition Table) restoration

The restoration of GPT table (both primary and secondary) is now possible.
Function 'gpt_restore' presents example of partition restoration process.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Piotr Wilczek <p.wilczek@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
9 years agogpt: The leXX_to_int() calls replaced with ones defined at <compiler.h>
Chang Hyun Park [Tue, 11 Dec 2012 10:09:45 +0000 (11:09 +0100)]
gpt: The leXX_to_int() calls replaced with ones defined at <compiler.h>

Custom definitions of le_XX_to_int functions have been replaced with
standard ones, defined at <compiler.h>

Replacement of several GPT related structures members with ones
indicating its endianness and proper size.

Signed-off-by: Chang Hyun Park <heartinpiece@outlook.com>
Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
9 years agopart:efi: Move part_efi.h file to ./include
Lukasz Majewski [Tue, 11 Dec 2012 10:09:43 +0000 (11:09 +0100)]
part:efi: Move part_efi.h file to ./include

This move is necessary to export gpt header and GPT partition entries to be
used with other commands or subsystems.
Additionally the part_efi.h file has been cleaned-up to supress checkpatch's
warnings.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
9 years agofs:ext4:write: Initialize cache aligned filename buffer
Łukasz Majewski [Wed, 5 Dec 2012 08:06:40 +0000 (08:06 +0000)]
fs:ext4:write: Initialize cache aligned filename buffer

The filename buffer is allocated dynamically. It must be cache aligned.
Moreover, it is necessary to erase its content before we use it for
file name operations.

This prevents from corruption of written file names.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
9 years agofs:ext4:fix: Code refactoring to suppress compiler warnings
Łukasz Majewski [Wed, 5 Dec 2012 08:06:39 +0000 (08:06 +0000)]
fs:ext4:fix: Code refactoring to suppress compiler warnings

Several fixes to suppress compiler's (eldk-5.[12].x  gcc 4.6)
warning  [-Wunused-but-set-variable]

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
9 years agofs:ext4:write: Store block device descriptor in file system structure
Łukasz Majewski [Wed, 5 Dec 2012 08:06:38 +0000 (08:06 +0000)]
fs:ext4:write: Store block device descriptor in file system structure

The device block descriptor (block_dev_desc_t) )shall be stored at
ext4 early code (at ext4fs_set_blk_dev in this case) to be available
for latter use (like put_ext4()).

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
9 years agofs:ext4:write: Add lldiv and do_div to perform 64-32 bits division
Łukasz Majewski [Wed, 5 Dec 2012 08:06:37 +0000 (08:06 +0000)]
fs:ext4:write: Add lldiv and do_div to perform 64-32 bits division

The ext4write code has been using direct calls to 64-32 division
(/ and %).

Officially supported u-boot toolchains (eldk-5.[12].x) generate calls
to __aeabi_uldivmod(), which is niether defined in the toolchain libs
nor u-boot source tree.

Due to that, when the ext4write command has been executed, "undefined
instruction" execption was generated (since the __aeabi_uldivmod()
is not provided).

To fix this error, lldiv() for division and do_div() for modulo have
been used.

Those two functions are recommended for performing 64-32 bit number
division in u-boot.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
9 years agopart: check each variable for capability calculation
Jerry Huang [Tue, 6 Nov 2012 15:33:12 +0000 (15:33 +0000)]
part: check each variable for capability calculation

In order to calculate the capability, we use the below expression to check:
((dev_desc->lba * dev_desc->blksz)>0L)
If the capability is greater than 4GB (e.g. 8GB = 8 * 1024 * 104 * 1024),
the result will overflow, the low 32bit may be zero.

Therefore, change to check each variable to fix this potential issue.

Signed-off-by: Jerry Huang <Chang-Ming.Huang@freescale.com>
9 years agocbfs: Remove mention of CREDITS files
Simon Glass [Mon, 5 Nov 2012 12:16:24 +0000 (12:16 +0000)]
cbfs: Remove mention of CREDITS files

As requested by Wolfgang, remove references to CREDITS in the CBFS
files.

Signed-off-by: Simon Glass <sjg@chromium.org>
9 years agofs/fs.c: do_fsload: measure throughput
Andreas Bießmann [Wed, 14 Nov 2012 12:32:37 +0000 (13:32 +0100)]
fs/fs.c: do_fsload: measure throughput

This patch adds time measurement and throughput calculation for
all supported load commands.

The output of ext2load changes from

---8<---
1830666 bytes read
--->8---

to

---8<---
1830666 bytes read in 237 ms (7.4 MiB/s)
--->8---

Signed-off-by: Andreas Bießmann <andreas.devel@googlemail.com>
[agust: rebased and revised commit log]
Signed-off-by: Anatolij Gustschin <agust@denx.de>
9 years agofs: zfs: fix illegal use of fp
Alejandro Mery [Wed, 31 Oct 2012 08:21:33 +0000 (08:21 +0000)]
fs: zfs: fix illegal use of fp

the upcoming sunxi (allwinner a10/a13) platform enables zfs
by default, and using linaro's hf -msoft-float makes the build
fail because this u64 division.

Signed-off-by: Alejandro Mery <amery@geeks.cl>
Acked-by: Stefan Roese <sr@denx.de>
9 years agoinclude/linux/byteorder: import latest endian definitions from linux
Kim Phillips [Mon, 29 Oct 2012 13:34:23 +0000 (13:34 +0000)]
include/linux/byteorder: import latest endian definitions from linux

u-boot's byteorder headers did not contain endianness attributions
for use with sparse, causing a lot of false positives.  Import the
kernel's latest definitions, and enable them by including compiler.h
and types.h.  They come with 'const' added for some swab functions, so
fix those up, too:

include/linux/byteorder/big_endian.h:46:2: warning: passing argument 1 of '__swab64p' discards 'const' qualifier from pointer target type [enabled by default]

Also, note: u-boot's historic __BYTE_ORDER definition has been
preserved (for the time being at least).

We also remove ad-hoc barrier() definitions, since we're including
compiler.h in files that hadn't in the past:

macb.c:54:0: warning: "barrier" redefined [enabled by default]

In addition, including compiler.h in byteorder changes the 'noinline'
definition to expand to __attribute__((noinline)).  This fixes
arch/powerpc/lib/bootm.c:

bootm.c:329:16: error: attribute '__attribute__': unknown attribute
bootm.c:329:16: error: expected ')' before '__attribute__'
bootm.c:329:25: error: expected identifier or '(' before ')' token

powerpc sparse builds yield:

include/common.h:356:22: error: marked inline, but without a definition

the unknown-reason inlining without a definition is considered obsolete
given it was part of the 2002 initial commit, and no arm version was
'fixed.'

also fixed:
ydirectenv.h:60:0: warning: "inline" redefined [enabled by default]

and:

Configuring for devconcenter - Board: intip, Options: DEVCONCENTER
make[1]: *** [4xx_ibm_ddr2_autocalib.o] Error 1
make: *** [arch/powerpc/cpu/ppc4xx/libppc4xx.o] Error 2
powerpc-fsl-linux-size: './u-boot': No such file
4xx_ibm_ddr2_autocalib.c: In function 'DQS_autocalibration':
include/asm/ppc4xx-sdram.h:1407:13: sorry, unimplemented: inlining failed in call to 'ppc4xx_ibm_ddr2_register_dump': function body not available
4xx_ibm_ddr2_autocalib.c:1243:32: sorry, unimplemented: called from here

and:

In file included from crc32.c:50:0:
crc32table.h:4:1: warning: implicit declaration of function '___constant_swab32' [-Wimplicit-function-declaration]
crc32table.h:4:1: error: initializer element is not constant
crc32table.h:4:1: error: (near initialization for 'crc32table_le[0]')

Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
[trini: Remove '#endif' in include/common.h around setenv portion]
Signed-off-by: Tom Rini <trini@ti.com>
9 years agofs: rename fsload command to load
Stephen Warren [Wed, 31 Oct 2012 11:05:07 +0000 (11:05 +0000)]
fs: rename fsload command to load

When the generic filesystem load command "fsload" was written, I felt
that "load" was too generic of a name for it, since many other similar
commands already existed. However, it turns out that there is already
an "fsload" command, so that name cannot be used. Rename the new
"fsload" to plain "load" to avoid the conflict. At least anyone who's
used a Basic interpreter should feel familiar with the name!

Signed-off-by: Stephen Warren <swarren@nvidia.com>
9 years agofs: fix number base behaviour change in fatload/ext*load
Stephen Warren [Tue, 30 Oct 2012 12:04:19 +0000 (12:04 +0000)]
fs: fix number base behaviour change in fatload/ext*load

Commit 045fa1e "fs: add filesystem switch libary, implement ls and
fsload commands" unified the implementation of fatload and ext*load
with the new command fsload. However, this altered the interpretation
of command-line numbers from always being base-16, to requiring a "0x"
prefix for base-16 numbers. Enhance do_fsload() to allow commands to
specify which base to use.

Use base 0, thus requiring a "0x" prefix for the new fsload command.
This feels much cleaner than assuming base 16.

Use base 16 for the pre-existing fatload and ext*load to prevent a
change in behaviour.

Use base 16 exclusively for the loadaddr environment variable, since
that variable is interpreted in multiple places, so we don't want the
behaviour to change.

Update command help text to make it clear where numbers are assumed to
be hex, and where an explicit "0x" prefix is required.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
9 years agocmd_ext4: remove TABs from command help text
Stephen Warren [Tue, 30 Oct 2012 12:04:18 +0000 (12:04 +0000)]
cmd_ext4: remove TABs from command help text

TABs in the help text won't line up in the same place on the console as
in a source editor. Replace them with spaces to make ensuring correct
alignment easier.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
9 years agofs: fix do_fsload() handling of optional arguments
Stephen Warren [Tue, 30 Oct 2012 12:04:17 +0000 (12:04 +0000)]
fs: fix do_fsload() handling of optional arguments

Most arguments to the shell command do_fsload() implements are optional.
Fix the minimum argc check to respect that. Cater for the situation
where argv[2] is not provided.

Enhance both do_fsload() and do_ls() to check the maximum number of
arguments too. While this check would typically be implemented via
U_BOOT_CMD()'s max_args parameter, if these functions are called
directly, then that check won't exist.

Finally, alter do_ls() to check (argc >= 4) rather than (argc == 4) so
that if the function is enhanced to allow extra arguments in the future,
this test won't need to be changed at that time.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
9 years agoZFS: Fix compile warning in fs/zfs/zfs.c (GCC 4.6.4 from ELDK 5.2.1)
Stefan Roese [Fri, 26 Oct 2012 04:24:01 +0000 (04:24 +0000)]
ZFS: Fix compile warning in fs/zfs/zfs.c (GCC 4.6.4 from ELDK 5.2.1)

This patch fixes the following compile warning:
zfs.c:2006:1: warning: 'zfs_label' defined but not used [-Wunused-function]
zfs.c:2029:1: warning: 'zfs_uuid' defined but not used [-Wunused-function]

Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Jorgen Lundman <lundman@lundman.net>
9 years agofs: handle CONFIG_NEEDS_MANUAL_RELOC
Stephen Warren [Tue, 30 Oct 2012 07:50:47 +0000 (07:50 +0000)]
fs: handle CONFIG_NEEDS_MANUAL_RELOC

Without this, fstypes[].probe points at the wrong place, so calling the
function results in undefined behaviour.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Tested-by: Andreas Bießmann <andreas.devel@googlemail.com>
9 years agoext4: Fix printf() format string error
Simon Glass [Tue, 23 Oct 2012 13:49:25 +0000 (13:49 +0000)]
ext4: Fix printf() format string error

Fix the following error in the ext4 command:

cmd_ext4.c:110:3: error: format '%lu' expects argument of type
'long unsigned int', but argument 4 has type 'int' [-Werror=format]

Signed-off-by: Simon Glass <sjg@chromium.org>
9 years agofs: add filesystem switch libary, implement ls and fsload commands
Stephen Warren [Mon, 22 Oct 2012 06:43:51 +0000 (06:43 +0000)]
fs: add filesystem switch libary, implement ls and fsload commands

Implement "ls" and "fsload" commands that act like {fat,ext2}{ls,load},
and transparently handle either file-system. This scheme could easily be
extended to other filesystem types; I only didn't do it for zfs because
I don't have any filesystems of that type to test with.

Replace the implementation of {fat,ext[24]}{ls,load} with this new code
too.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Conflicts:
Makefile

9 years agofs: separate CONFIG_FS_{FAT, EXT4} from CONFIG_CMD_{FAT, EXT*}
Stephen Warren [Mon, 22 Oct 2012 06:43:50 +0000 (06:43 +0000)]
fs: separate CONFIG_FS_{FAT, EXT4} from CONFIG_CMD_{FAT, EXT*}

This makes the FAT and ext4 filesystem implementations build if
CONFIG_FS_{FAT,EXT4} are defined, rather than basing the build on
whether CONFIG_CMD_{FAT,EXT*} are defined. This will allow the
filesystems to be built separately from the filesystem-specific commands
that use them. This paves the way for the creation of filesystem-generic
commands that used the filesystems, without requiring the filesystem-
specific commands.

Minor documentation changes are made for this change.

The new config options are automatically selected by the old config
options to retain backwards-compatibility.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
9 years agofs: delete unused Makefile
Stephen Warren [Mon, 22 Oct 2012 06:43:49 +0000 (06:43 +0000)]
fs: delete unused Makefile

fs/Makefile is unused. The top-level Makefile sets LIBS-y += fs/xxx and
hence causes make to directly descend two directory levels into each
individual filesystem, and it never descends into fs/ itself.

So, delete this useless file.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agoFAT: implement fat_set_blk_dev(), convert cmd_fat.c
Stephen Warren [Wed, 17 Oct 2012 06:44:59 +0000 (06:44 +0000)]
FAT: implement fat_set_blk_dev(), convert cmd_fat.c

This makes the FAT filesystem API more consistent with other block-based
filesystems. If in the future standard multi-filesystem commands such as
"ls" or "load" are implemented, having FAT work the same way as other
filesystems will be necessary.

Convert cmd_fat.c to the new API, so the code looks more like other files
implementing the same commands for other filesystems.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
9 years agoFAT: initialize all fields in cur_part_info, simplify init
Stephen Warren [Wed, 17 Oct 2012 06:44:58 +0000 (06:44 +0000)]
FAT: initialize all fields in cur_part_info, simplify init

cur_part_info.{name,type} are strings. So, we don't need to memset()
the entire thing, just put the NULL-termination in the first byte.

Add missing initialization of the bootable and uuid fields.

None of these fields are actually used by fat.c. However, since it
stores the entire disk_partition_t, we should make sure that all fields
are valid.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
9 years agoFAT: remove cur_part_nr
Stephen Warren [Wed, 17 Oct 2012 06:44:57 +0000 (06:44 +0000)]
FAT: remove cur_part_nr

A future patch will implement the more standard filesystem API
fat_set_blk_dev(). This API has no way to know which partition number
the partition represents. Equally, future DM rework will make the
concept of partition number harder to pass around.

So, simply remove cur_part_nr from fat.c; its only use is in a
diagnostic printf, and the context where it's printed should make it
obvious which partition is referred to anyway (since the partition ID
would come from the user command-line that caused it).

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
9 years agofs: Add a Coreboot Filesystem (CBFS) driver and commands
Gabe Black [Fri, 12 Oct 2012 14:26:11 +0000 (14:26 +0000)]
fs: Add a Coreboot Filesystem (CBFS) driver and commands

This change adds CBFS support and some commands to use it to u-boot. These
commands are:

cbfsinit - Initialize CBFS support and pull all metadata into RAM. The end of
the ROM is an optional parameter which defaults to the standard 0xffffffff and
can be used to support multiple CBFSes in a system. The last one set up with
cbfsinit is the one that will be used.

cbfsinfo - Print information from the CBFS header.

cbfsls - Print out the size, type, and name of all the files in the current
CBFS. Recognized types are translated into symbolic names.

cbfsload - Load a file from CBFS into memory. Like the similar command for fat
filesystems, you can optionally provide a maximum size.

Support for CBFS is compiled in when the CONFIG_CMD_CBFS option is specified.

The CBFS driver can also be used programmatically from within u-boot.

If u-boot needs something out of CBFS very early before the heap is
configured, it won't be able to use the normal CBFS support which caches some
information in memory it allocates from the heap. The
cbfs_file_find_uncached function searches a CBFS instance without touching
the heap.

Signed-off-by: Gabe Black <gabeblack@google.com>
Signed-off-by: Stefan Reinauer <reinauer@chromium.org>
Signed-off-by: Simon Glass <sjg@chromium.org>
Conflicts:
Makefile

9 years agodisk: Address cast and format errors
Taylor Hutt [Fri, 12 Oct 2012 14:26:09 +0000 (14:26 +0000)]
disk: Address cast and format errors

This change addresses a few printf-formatting errors, and a typecast
error.

Signed-off-by: Taylor Hutt <thutt@chromium.org>
Signed-off-by: Simon Glass <sjg@chromium.org>
9 years agodisk: Make the disk partition code work with no specific partition types
Gabe Black [Fri, 12 Oct 2012 14:26:08 +0000 (14:26 +0000)]
disk: Make the disk partition code work with no specific partition types

Currently, if the disk partition code is compiled with all of the parition
types compiled out, it hits an #error which stops the build. This change
adjusts that file so that those functions will fall through to their defaults
in those cases instead of breaking the build. These functions are needed
because other code calls them, and that code is needed because other config
options are overly broad and bring in support we don't need along with
support we do.

Also reduce repetition of the 6-term #ifdef throughout the file.

Signed-off-by: Gabe Black <gabeblack@chromium.org>
Signed-off-by: Simon Glass <sjg@chromium.org>
9 years agodisk: initialize name/part fields when returning a whole disk
Stephen Warren [Wed, 10 Oct 2012 07:57:51 +0000 (07:57 +0000)]
disk: initialize name/part fields when returning a whole disk

When get_device_and_partition() finds a disk without a partition table,
under some conditions, it "returns" a disk_partition_t that describes
the entire raw disk. Make sure to initialize all fields in the partition
descriptor in that case.

The value chosen for name is just some arbitrary descriptive string.

The value chosen for info matches the check at the end of
get_device_and_partition(). However, it's probably not that important;
it's not obvious that the value is really used.

Reported-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
Signed-off-by: Tom Rini <trini@ti.com>
9 years agofs: fat: Fix mkcksum() function parameters
Marek Vasut [Tue, 9 Oct 2012 07:20:22 +0000 (07:20 +0000)]
fs: fat: Fix mkcksum() function parameters

The mkcksum() function now takes one parameter, the pointer to
11-byte wide character array, which it then operates on.

Currently, the function is wrongly passed (dir_entry)->name, which
is only 8-byte wide character array. Though by further inspecting
the dir_entry structure, it can be noticed that the name[8] entry
is immediatelly followed by ext[3] entry. Thus, name[8] and ext[3]
in the dir_entry structure actually work as this 11-byte wide array
since they're placed right next to each other by current compiler
behavior.

Depending on this is obviously wrong, thus fix this by correctly
passing both (dir_entry)->name and (dir_entry)->ext to the mkcksum()
function and adjust the function appropriately.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Tom Rini <trini@ti.com>
9 years agodisk: part_dos: print partition UUID in partition list
Stephen Warren [Mon, 8 Oct 2012 08:14:40 +0000 (08:14 +0000)]
disk: part_dos: print partition UUID in partition list

This information may be useful to compare against command "part uuid",
or if you want to manually paste the information into the kernel
command-line.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
[trini: print_one_part / print_part_dos output strings didn't quite
match before the changes]
Signed-off-by: Tom Rini <trini@ti.com>
9 years agodisk: part_dos: checkpatch cleanups
Stephen Warren [Mon, 8 Oct 2012 08:14:38 +0000 (08:14 +0000)]
disk: part_dos: checkpatch cleanups

Minor cleanups required so later patches don't trigger checkpatch.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
9 years agodisk: part_efi: set bootable flag in partition objects
Stephen Warren [Mon, 8 Oct 2012 08:14:37 +0000 (08:14 +0000)]
disk: part_efi: set bootable flag in partition objects

A partition is considered bootable if it either has the "legacy BIOS
bootable" flag set, or if the partition type UUID matches the standard
"system" type.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
9 years agodisk: part_efi: print raw partition attributes
Stephen Warren [Mon, 8 Oct 2012 08:14:36 +0000 (08:14 +0000)]
disk: part_efi: print raw partition attributes

When printing the EFI partition table, print the raw attributes. Convert
struct gpt_entry_attributes to a union to allow raw access.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
9 years agodisk: part_efi: add new partition attribute definitions
Stephen Warren [Mon, 8 Oct 2012 08:14:35 +0000 (08:14 +0000)]
disk: part_efi: add new partition attribute definitions

Add no_block_io_protocol and legacy_bios_bootable attribute definitions.
These are sourced from UEFI Spec 2.3, page 105, table 19. Credits to the
libparted source for the specification pointer.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
9 years agodisk: part_efi: print partition UUIDs
Stephen Warren [Mon, 8 Oct 2012 08:14:34 +0000 (08:14 +0000)]
disk: part_efi: print partition UUIDs

When printing the partition table, print the partition type UUID and the
individual partition UUID. Do this unconditionally, since partition UUIDs
are useful.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
9 years agodisk: part_efi: re-order partition list printf, change case
Stephen Warren [Mon, 8 Oct 2012 08:14:33 +0000 (08:14 +0000)]
disk: part_efi: re-order partition list printf, change case

The partition name is a long variable-length string. Move it last on
the line to ensure consistent layout and that the entries align with
the "header" line. Also, surround it in quotes, so if it's empty, it's
obvious that something is still being printed.

Also, change the case of the LBA numbers; lower-case looks nicer in my
opinion, and will be more consistent with the UUID printing that is
added later in this series.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
9 years agodisk: part_efi: remove indent level from loop
Stephen Warren [Mon, 8 Oct 2012 08:14:32 +0000 (08:14 +0000)]
disk: part_efi: remove indent level from loop

Simplify the partition printing loop in print_part_efi() to bail out
early when the first invalid partition is found, rather than indenting
the whole body of the loop. This simplifies later patches.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
9 years agodisk: get_device_and_partition() return value fixes
Stephen Warren [Mon, 8 Oct 2012 07:45:54 +0000 (07:45 +0000)]
disk: get_device_and_partition() return value fixes

When no valid partitions are found, guarantee that we return -1. This
most likely already happens, since the most recent get_partition_info()
will have returned an error. However, it's best to be explicit.

Remove an unnecessary assignment of ret=0 in the success case; this value
is over-written with the processed partition ID later.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
9 years agoARM: prevent misaligned array inits
Albert ARIBAUD [Tue, 9 Oct 2012 09:28:15 +0000 (09:28 +0000)]
ARM: prevent misaligned array inits

Under option -munaligned-access, gcc can perform local char
or 16-bit array initializations using misaligned native
accesses which will throw a data abort exception. Fix files
where these array initializations were unneeded, and for
files known to contain such initializations, enforce gcc
option -mno-unaligned-access.

Signed-off-by: Albert ARIBAUD <albert.u.boot@aribaud.net>
[trini: Switch to usign call cc-option for -mno-unaligned-access as
Albert had done previously as that's really correct]
Signed-off-by: Tom Rini <trini@ti.com>
9 years agodfu:cmd: Support for DFU u-boot command
Lukasz Majewski [Mon, 6 Aug 2012 12:41:09 +0000 (14:41 +0200)]
dfu:cmd: Support for DFU u-boot command

Support for u-boot's command line command "dfu <interface> <dev> [list]".

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Marek Vasut <marex@denx.de>