Original bug description from Feng (fdt_resize() bug caused "WARNING:
could not set linux, initrd-start FDT_ERR_NOSPACE."):
What I got is an error: "WARNING: could not set linux,initrd-start
FDT_ERR_NOSPACE." after loading Device Tree blob. This in turn caused
linux to miss init part.
After some digging, I found out the reason for this error, it is caused
by fdt_resize().
FDT blob got resized after filling in all board specific information of
PowerPC. (in boot_body_linux()). It reduced blob size with only extra
space for two fdt_reserve_entry, one for fdt itself, and one for initrd.
Then it's aligned to a 0x1000 page boundary. However, later in
fdt_initrd(), it could add two more properties, initrd-start AND
initrd-end, each one needs at least two fdt_reserve_entry sizes done by
_fdt_add_property() (name and value). Thus, the two fdt_reserve_entry
extra space is not sufficient.
So for some specific fdt size which is just under the page boundary
after resizing, this will cause an error of FDT_ERR_NOSPACE in
fdt_initrd() when setting those two properties, and failed to pass
initrd information to linux.
My fix is in fdt_resize(), leave at least 4 fdt_reserve_entry for
initrd. So instead of 2*sizeof(struct fdt_reserve_entry) for
actual_totalsize, use 5*sizeof(struc fdt_reserve_entry).
Stefan: I got this same error on katmai, when trying to boot with
initrd (run flash_self). This patch fixes this issue.
Signed-off-by: Feng Wang <fwang02@harris.com>
Tested-by: Stefan Roese <sr@denx.de>
Cc: Jerry Van Baren <gvb.uboot@gmail.com>
Acked-by: Gerald Van Baren <vanbaren@cideas.com>
If we are creating reference (handles) to nodes in a device tree we need
to first create a new phandle in node and this needs a new phandle
value. So we search through the whole dtb to find the max phandle value
and return the next greater value for a new phandle allocation.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Acked-by: Gerald Van Baren <vanbaren@cideas.com>
Given a compatible string and physical address try and find a node that
matches. This is useful when we want to find a specific device node to
update (for example if we have multiple PCI nodes we can use the
physical address to distinguish them when trying to update the device
tree).
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Acked-by: Gerald Van Baren <vanbaren@cideas.com>
This code is extracted out of the Linux Kernel code from
arch/powerpc/kernel/prom_parse.c.
We maintain some of the same structure to support multiple bus types even
though we only have one in the current code. In the future we might want
to translate across a PCI bus and thus it will be easier to add that
functionality back in.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Acked-by: Gerald Van Baren <vanbaren@cideas.com>
Add sec3.1 h/w geometry for fdt node fixups.
Also, technically, whilst SEC v3.3 h/w honours the tls_ssl_stream descriptor
type, it lacks the ARC4 algorithm execution unit required to be able
to execute anything meaningful with it. Change the node to agree with
the documentation that declares that the sec3.3 really doesn't have such
a descriptor type.
Reported-by: Haiying Wang <Haiying.Wang@freescale.com>
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Add a helper function that given an alias will delete both the node
the alias points to and the alias itself
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Acked-by: Gerald Van Baren <vanbaren@cideas.com>
Allow overwriting defined partitions in the device tree blob
using partition info defined in the 'mtdparts' environment
variable.
Signed-off-by: Anatolij Gustschin <agust@denx.de>
Cc: Gerald Van Baren <vanbaren@cideas.com>
Currently fdt_fixup_stdout() is using hard-coded CONFIG_CONS_INDEX
constant. With multi-serial support, the CONS_INDEX may no longer
represent actual console, so we should try to extract port number
from the current stdio device name instead of always hard-coding the
constant value.
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Acked-by: Gerald Van Baren <vanbaren@cideas.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Its possible that we end up with a device tree that happens to be a
particular size that after we call fdt_resize() we don't have any
space left for the initrd mem_rsv.
Fix this be adding a second mem_rsv into the size calculation. We
had one to cover the fdt itself and we have the potential of adding
a second for the initrd.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Acked-by: Gerald Van Baren <vanbaren@cideas.com>
This function can be used to update the size in the "reg" property
of the NOR FLASH device nodes. This is necessary for boards with
non-fixed NOR FLASH sizes.
Signed-off-by: Stefan Roese <sr@denx.de>
Acked-by: Gerald Van Baren <vanbaren@cideas.com>
Acked-by: Wolfgang Denk <wd@denx.de>
If the size of a region equal to 4G it can't be represnted in a 32-bit
BAR so we should have marked that case as MEM64.
Additionally bump the number of inbound windows up to 4 to handle the
fact that Freescale PPCs that have an implicit window for CCSRBAR.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Use the standard lowercase "xx" capitalization that other Freescale
architectures use for CPU defines to prevent confusion and errors
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
The PCI_REGION_MEMORY and PCI_REGION_MEM are a bit to similar and
can be confusing when reading the code.
Rename PCI_REGION_MEMORY to PCI_REGION_SYS_MEMORY to clarify its used
for system memory mapping purposes.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
The code in fdt_resize() to extend the fdt size to end on a page boundary
is wrong for fdt's not located at an address aligned on a page boundary.
What's even worse, the code would make actualsize shrink rather than grow
if (blob & 0xfff) was bigger than the amount of padding added by ALIGN(),
causing fdt_add_mem_rsv to fail.
Fix it by aligning end address (blob + size) to a page boundary instead.
For aligned fdt's this is equivalent to what we had before.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
fdt_add_mem_rsv() requires space for a struct fdt_reserve_entry
(16 bytes), so make sure that fdt_resize at least adds that much
padding, no matter what the location or size of the fdt is.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Acked-by: Andy Fleming <afleming@freescale.com>
Added fdt_pci_dma_ranges() that parses the pci_region info from the
struct pci_controller and populates the dma-ranges based on it.
The max # of windws/dma-ranges we support is 3 since on embedded
PowerPC based systems this is the max number of windows.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Andrew Fleming-AFLEMING <afleming@freescale.com>
Add helper functions to return find a node and return it's property
or a default value.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Andrew Fleming-AFLEMING <afleming@freescale.com>
Acked-by: Gerald Van Baren <vanbaren@cideas.com>
Commit 2a1a2cb6 didnt remove the dummy mem reservation in fdt_chosen,
and this stopped Linux from booting with a Ramdisk. This patch fixes
this, by deleting the useless dummy mem reservation.
When booting with a Ramdisk, a fix offset FDT_RAMDISK_OVERHEAD is now
added to of_size, so we dont need anymore a dummy mem reservation.
I measured the value of FDT_RAMDISK_OVERHEAD on a MPC8270 based
system (=0x44 bytes) and rounded it up to 0x80).
Signed-off-by: Heiko Schocher <hs@denx.de>
Acked-by: Kumar Gala <galak@kernel.crashing.org>
Created a new fdt_initrd() to deal with setting the initrd properties
in the device tree and fixing up the mem reserve. We can use this
both in the choosen node handling and lets us remove some duplicated
code when we fixup the initrd info in bootm on PPC.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Move to using the environment variables 'ethaddr', 'eth1addr', etc..
instead of bd->bi_enetaddr, bi_enet1addr, etc.
This makes the code a bit more flexible to the number of ethernet
interfaces.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Currently U-Boot can only fixup the usb dr_mode, but some boards (namely
MPC8315E-RDB) can use two PHY types: ULPI (stand-alone OTG port) or UTMI
(connected to the four-ports hub, usb host only).
This patch implements support for passing Dual-Role USB controller's
device tree property phy_type through the usb_phy_type environment
variable.
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Acked-by: Gerald Van Baren <vanbaren@cideas.com>
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Delete the crypto node if not on an E-processor. If on 8360 or 834x family,
check rev and up-rev crypto node (to SEC rev. 2.4 property values)
if on an 'EA' processor, e.g. MPC8349EA.
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
The working_fdt pointer was declared in common/fdt_support.c but was
not used there. Move it to common/cmd_fdt.c where it is used (it is
also used in lib_ppc/bootm.c).
Signed-off-by: Gerald Van Baren <vanbaren@cideas.com>
differentiate with local variables of the same name by renaming the
global 'fdt' variable 'working_fdt'.
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Linux understands "host" (default), "peripheral" and "otg" (broken).
Though, U-Boot doesn't restrict dr_mode variable to these values (think
of renames in future).
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
These defines embedded the u-boot env variables and/or the bd_t structure
in the fdt blob. The conclusion of discussion on the u-boot email list
was that embedding these in the fdt blob is not useful: there are better
ways of passing the data (in fact, the fdt blob itself replaces the
bd_t struct).
The only board that enables these is the stxxtc and they don't appear
to be used by linux.
Signed-off-by: Gerald Van Baren <vanbaren@cideas.com>
Acked-by: Kim Phillips <kim.phillips@freescale.com>
Move the flat device tree setup for QE related devices into
a common file shared between 83xx & 85xx platforms that have QE's.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Define the layout of a binary blob that contains a QE firmware and instructions
on how to upload it. Add function qe_upload_firmware() to parse the blob and
perform the actual upload. Add command-line command "qe fw" to take a firmware
blob in memory and upload it. Update ft_cpu_setup() on 85xx to create the
'firmware' device tree node if U-Boot has uploaded a firmware. Fully define
'struct rsp' in immap_qe.h to include the actual RISC Special Registers.
Signed-off-by: Timur Tabi <timur@freescale.com>
Implement a suggestion by Scott Wood to make the /chosen handling fine
grained. Don't overwrite pre-existing properties on a per-property basis,
so if /chosen exists but a necessary /chosen/property doesn't, it gets
created. If a /chosen property exists, it is NOT overwritten unless the
"force" flag is true.
Signed-off-by: Gerald Van Baren <vanbaren@cideas.com>
We use a combination of the serialN alias and CONFIG_CONS_INDEX to
determine which serial alias we should set linux,stdout-path to.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Added the following fdt fixup helpers:
* do_fixup_by_prop{_u32} - Find matching nodes by property name/value
* do_fixup_by_compat{_u32} - Find matching nodes by compat
The _u32 variants work the same only the property they are setting
is know to be a 32-bit integer instead of a byte buffer.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Update libfdt to commit 8eaf5e358366017aa2e846c5038d1aa19958314e from
the device tree compiler (dtc) project.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Moved the generic fixup handling code out of cpu/mpc5xxx and cpu/mpc8260
into common/fdt_support.c and renamed:
do_fixup() -> do_fixup_by_path()
do_fixup_u32() -> do_fixup_by_path_u32()
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Removed:
fdt_node_is_compatible
fdt_find_node_by_type
fdt_find_compatible_node
To ease merge of newer libfdt as we aren't using them anywhere at this time.
Also moved fdt_find_and_setprop out of libfdt into fdt_support.c for the same
reason.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Previously ft_board_setup() was called by fdt_chosen() which was not
really correctly structured. This splits ft_board_setup() out by creating
a new fdt boardsetup command.
Fix a bug when parsing fdt set command values which have the square
bracket form [00 11 22 33] - the length was updated incorrectly in when
parsing that form.
Signed-off-by: Gerald Van Baren <vanbaren@cideas.com>
The fdt_chosen() function was adding/seting some properties ad-hoc
improperly and duplicated (poorly) what was done in ft_board_setup()
Clean up long lines (setting properties, printing errors).
Signed-off-by: Gerald Van Baren <vanbaren@cideas.com>
The new name matches more closely the kernel's name, which is also
a much better description.
Signed-off-by: Wolfgang Grandegger <wg@grandegger.com>
Acked-by: Gerald Van Baren <vanbaren@cideas.com>
Declare the variable fdt properly as extern.
Call the "set_fn" function pointer the "short way" without the full
dereferencing syntax.
Signed-off-by: Gerald Van Baren <vanbaren@cideas.com>
Print more than the raw libfdt error message strings. This is especially
useful for cluing in the user when the bootm command aborts due to
blob problems.
Signed-off-by: Gerald Van Baren <vanbaren@cideas.com>
What was suppose to be a stack variable was declared as a pointer,
overwriting random memory.
Also moved the libfdt.a requirement into the main Makefile. That is
The U-Boot Way.
In bootm, create the "/chosen" node only if it doesn't already exist
(better matches the previous behavior).
Update for proper reserved memory map handling for initrd.
