U-Boot crashed on the last instruction:
int parse_stream_outer(struct in_str *inp, int flag)
{
effa4784: 94 21 ff 38 stwu r1,-200(r1)
effa4788: 7c 08 02 a6 mflr r0
effa478c: 42 9f 00 05 bcl- 20,4*cr7+so,effa4790 <parse_stream_outer+0xc>
effa4790: 7d 80 00 26 mfcr r12
effa4794: 13 c1 b3 21 evstdd r30,176(r1)
...which is a SPE instruction, although -mno-spe was used.
tmp/cross/ppce500v2/bin/powerpc-angstrom-linux-gnuspe-gcc --version
powerpc-angstrom-linux-gnuspe-gcc (GCC) 4.3.3
Seems to be a known issue (since 2008-04?!)
Googled some, turns out this patch/workaround works for me on MPC8536DS.
See http://gcc.gnu.org/ml/gcc-patches/2008-04/msg00311.html for more info
Signed-off-by: Leon Woestenberg <leon@sidebranch.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
According the user manual, we need loop-check the L2 enable bit set.
Signed-off-by: Dave Liu <daveliu@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
U-boot for Marvell Kirkwood boards no longer work after the EABI changes
introduced in commit f772acf8a5. This
turns out to be caused by a stack alignment issue. The armv5te
instructions ldrd/strd instructions require 8-byte alignment to work
properly (otherwise undefined behavior).
Tested on an OpenRD base board, where both printouts and ubifs stuff now
works.
Signed-off-by: Simon Kagstrom <simon.kagstrom@netinsight.net>
gpmc_config should not be a variant as it is board specific
hence make it a const parameter
Fixes issues identified by Dirk:
- build issue for zoom2
- warnings for all other OMAP3 platforms using nand/onenand etc
Signed-off-by: Nishanth Menon <nm@ti.com>
Some 4xx variants (e.g. 440EP(x)/GR(x)) have an internal
synchronous PCI clock. Knowledge about the currently configured
value might be helpful. So let's print it out upon bootup.
Signed-off-by: Stefan Roese <sr@denx.de>
PPC440EP(x)/PPC440GR(x):
In asynchronous PCI mode, the synchronous PCI clock must meet
certain requirements. The following equation describes the
relationship that must be maintained between the asynchronous PCI
clock and synchronous PCI clock. Select an appropriate PCI:PLB
ratio to maintain the relationship:
AsyncPCIClk - 1MHz <= SyncPCIclock <= (2 * AsyncPCIClk) - 1MHz
This patch now adds a function to check and reconfigure the sync
PCI clock to meet this requirement. This is in preparation for
some AMCC boards (Sequoia/Rainier and Yosemite/Yellowstone) using this
function to not violate the PCI clocking rules.
Signed-off-by: Stefan Roese <sr@denx.de>
Till now only the ranges in the ebc node are updated with the values
currently configured in the PPC4xx EBC controller. With this patch now
the NOR flash size is updated in the device tree blob as well. This is
done by scanning the compatible nodes "cfi-flash" and "jedec-flash"
for the correct chip select number.
This size fixup is enabled for all AMCC eval board right now. Other
4xx boards may want to enable it as well, if this problem with multiple
NOR FLASH sizes exists.
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Wolfgang Denk <wd@denx.de>
strcpy() was iused with the target address being a pointer to a
constant string, which potentially is read-only. Use a (writable)
array of characters instead.
Signed-off-by: Wolfgang Denk <wd@denx.de>
Signed-off-by: Stefan Roese <sr@denx.de>
mpc8313e erratum IPIC1 swapped TSEC interrupt ID numbers on rev. 1
h/w (see AN3545). The base device tree in use has rev. 1 ID numbers,
so if on Rev. 2 (and higher) h/w, we fix them up here.
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Reviewed-by: Roland Lezuo <roland.lezuo@chello.at>
When setting up the LAWs for the DDR, if there was an error,
you got the not-so-helpful error text "ERROR" and nothing
else. Not only is it non-informative, but it is also
pretty frustrating trying to grep for "ERROR" in the source.
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
This patch re-formats the code in cpu/arm920t and cpu/arm920t/23c24x0 in
preparation for changes to add support for the Embest SBC2440-II Board.
The changes are as follows:
- re-indent the code using Lindent
- make sure register layouts are defined using a C struct
- replace the upper-case typedef'ed C struct names with lower case
non-typedef'ed ones
- make sure registers are accessed using the proper accessor functions
- run checkpatch.pl and fix any error reports
It assumes the following patch has been applied first:
- [U-Boot][PATCH-ARM] CONFIG_SYS_HZ fix for ARM902T S3C24X0 Boards, 05/09/2009
Tested on an Embest SBC2440-II Board with local u-boot patches as I don't have
any s3c2400 or s3c2410 boards but need this patch applying before I can submit
patches for the SBC2440-II Board. Also, ran MAKEALL for all ARM9 targets and no
new warnings or errors were found.
Signed-off-by: Kevin Morfitt <kevin.morfitt@fearnside-systems.co.uk>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
This sets CONFIG_SYS_HZ to 1000 for all boards that use the s3c2400 and
s3c2410 cpu's which fixes various problems such as the timeouts in tftp being
too short.
Tested on an Embest SBC2440-II Board with local u-boot patches as I don't
have any s3c2400 or s3c2410 boards but need this patch applying before I can
submit patches for the SBC2440-II Board. Also, ran MAKEALL for all ARM9 targets
and no new warnings or errors were found.
It was originally submitted on 21/06/2009 but didn't get into the 2009.08
release, and Jean-Pierre made one comment on the original patch (see
http://lists.denx.de/pipermail/u-boot/2009-July/055470.html). I've made two
changes to the original patch:
- it's been re-based to the current release
- I've re-named get_timer_raw() to get_ticks() in response to Jean-Pierre's comment
This affects the sbc2410, smdk2400, smdk2410 and trab boards. I've copied it
directly to the maintainers of all except the sbc2410 which doesn't have an
entry in MAINTAINERS.
Signed-off-by: Kevin Morfitt <kmorfitt@aselaptop-1.localdomain>
Tested-by: Wolfgang Denk <wd@denx.de>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
This patch adds support for the Samsung s5pc100 and s5pc110
SoCs. The s5pc1xx SoC is an ARM Cortex A8 processor.
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
Signed-off-by: HeungJun, Kim <riverful.kim@samsung.com>
All arches apparently should reset the watchdog in their udelay loop as
noted on the mailing list recently:
> A comment in flash_status_check() suggests that udelay() is
> expected to reset the watchdog, but I can't find any architecture
> where it does.
If this is missing in other architectures, it should be fixed at the
root cause, i. e. in udelay() or in the respective support routines.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
FLAG_PARSE_SEMICOLON is not defined without hush.h, so include that.
Signed-off-by: Simon Kagstrom <simon.kagstrom@netinsight.net>
Signed-off-by: Prafulla Wadaskar <prafulla@marvell.com>
these boards are built around Atmel's AT91SAM9260/9G20 and have
up to 64MB of NOR flash, up to 128MB of SDRAM, up to 2GB of NAND
and include a 10/100 Ethernet PHY in RMII mode.
Signed-off-by: Eric Benard <eric@eukrea.com>
Signed-off-by: Tom Rix <Tom.Rix@windriver.com>
CPUAT91 is built around Atmel's AT91RM9200 and has up to 16MB of NOR
flash, up to 128MB of SDRAM, and includes a Micrel KS8721 PHY in RMII
mode.
Signed-off-by: Eric Benard <eric@eukrea.com>
Signed-off-by: Tom Rix <Tom.Rix@windriver.com>
DM646x is an SOC from TI which has both an ARM and a DSP.
There are multiple variants of the SOC mainly dealing with different
core speeds.
This patch adds the initial framework for the DM646x SOC.
Signed-off-by: Sandeep Paulraj <s-paulraj@ti.com>
v7_flush_dcache_all, because it depends on omap ROM code is not
generic. Rename the function to 'invalidate_dcache' and move it
to the omap cpu directory.
Collect the other omap cache routines l2_cache_enable and
l2_cache_disable with invalide_dcache into cache.S. This
means removing the old cache.c file that contained l2_cache_enable
and l2_cache_disable.
The conversion from cache.c to cache.S was done most through
disassembling the uboot binary. The only significant change was
to change the comparision for the return of get_cpu_rev from
cmp r0, #0
beq earlier_than_label
Which was lost information to
cmp r0, #CPU_3XX_ES20
blt earlier_than_label
The paths through the enable routine were verified by
adding an infinite loop and seeing the hang. Then
removing the infinite loop and seeing it continue.
The disable routine is similar enough that it was not
tested with this method.
Run tested by cold booting from nand on beagle and zoom1.
Compile tested on MAKEALL arm.
Signed-off-by: Tom Rix <Tom.Rix@windriver.com>
NULL is an absolute value and should not be relocated.
After this correction code like:
void weak_fun(void) __attribute__((weak));
printf("weak_fun:%p\n", weak_fun);
will still print null after relocation.
Signed-off-by: Joakim Tjernlund <Joakim.Tjernlund@transmode.se>
When U-Boot is relocated from flash to RAM pointers are modified
accordingly. However, pointers initialzed with NULL values should not
be modified so that they maintain their intended NULL value. If the
BSS segment is linked at address 0 its address will not be
updated as necessary during relocation.
This is a temporary workaround. The end goal is to add support to
U-Boot to dynamically locate the BSS at an arbitrary address at
runtime. When the ability to fixup the BSS inteligently is
added, this workaround can be removed and the 85xx link script
can put the BSS at a fixed address at link time.
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Commit 054197ba and later fixes used an array to initialize some of
the MDDRC parameters; however, the use of an array turned out to be a
bad idea as it was not possible to correlate structure entries to
array indices in readable and reliable way. Now we use a struct
instead, which makes this self-explanatory.
Signed-off-by: Wolfgang Denk <wd@denx.de>
The command "reginfo" got an overhaul for the ppc4xx. It dumps all the
relevant HW configuration registers (address, symbolic name, content).
This allows to easily detect errors in *.h files and changes in the HW
configuration.
Signed-off-by: Niklaus Giger <niklaus.giger@member.fsf.org>
Signed-off-by: Stefan Roese <sr@denx.de>
Modify all existing *.c files to use the new register names
as seen in the AMCC manuals.
Signed-off-by: Niklaus Giger <niklaus.giger@member.fsf.org>
Signed-off-by: Stefan Roese <sr@denx.de>
Unfortunately some Rev D PPC405EX/405EXr PVR's are identical with older
405EX(r) parts. Here a list:
0x12911475 - 405EX Rev D with Security *and* 405EX Rev A/B witout Sec
0x12911473 - 405EX Rev D without Security *and* 405EXr Rev A/B with Sec
Since there are only a few older parts in the field, this patch now
changes the PVR's above to represent the new Rev D versions.
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Phong Vo" <pvo@amcc.com>
This message is printed upon PCIe bus scan, not only upon error, but also
if no PCIe device is detected at all. Since this is not an error, let's
remove this message in this case. We already have the message
"link is not up." if there is no PCIe device present.
Signed-off-by: Stefan Roese <sr@denx.de>
Acked-by: Wolfgang Denk <wd@denx.de>
The SPD detection code for the Denali memory controller used on some
ppc4xx
processors incorrectly encodes DDR0_42. With certain memory
configurations,
this can cause the bootwrapper to incorrectly calculate the installed
memory
size, because the number of row bits is wrong. This patch fixes that
encoding.
Signed-off-by: Mike Nuss <mike@terascala.com>
Signed-off-by: Stefan Roese <sr@denx.de>
The following changes allow U-Boot to fully relocate from flash to
RAM:
- Remove linker scripts' .fixup sections from the .text section
- Add -mrelocatable to PLATFORM_RELFLAGS for all boards
- Define CONFIG_RELOC_FIXUP_WORKS for all boards
Previously, U-Boot would partially relocate, but statically initialized
pointers needed to be manually relocated.
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
This patch merges the ECC handling (ECC parity byte writing) into one
file (ecc.c) for all PPC4xx SDRAM controllers except for PPC440EPx/GRx.
This exception is because only those PPC's use the completely different
Denali SDRAM controller core.
Previously we had two routines to generate/write the ECC parity bytes.
With this patch we now only have one core function left.
Tested on Kilauea (no ECC) and Katmai (with and without ECC).
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Felix Radensky <felix@embedded-sol.com>
Cc: Grant Erickson <gerickson@nuovations.com>
Cc: Pieter Voorthuijsen <pv@prodrive.nl>
Reorganize DDR2 ECC handling to use common code for
SPD DIMMs and soldered SDRAM. Also, use common code
to display SDRAM info (ECC, CAS latency) for SPD and
soldered SDRAM variants.
Signed-off-by: Felix Radensky <felix@embedded-sol.com>
Signed-off-by: Stefan Roese <sr@denx.de>
The latest PPC4xx register cleanup patch missed some SDRAM defines.
This patch now changes lower case UIC defines to upper case. Also
some names are changed to match the naming in the IBM/AMCC users
manuals (e.g. mem_mcopt1 -> SDRAM0_CFG).
Signed-off-by: Stefan Roese <sr@denx.de>
The latest PPC4xx register cleanup patch missed the UIC defines.
This patch now changes lower case UIC defines to upper case.
Signed-off-by: Stefan Roese <sr@denx.de>
some LCRR bits are not documented throughout the 83xx family RMs.
New board porters copying similar board configurations might omit
setting e.g., DBYP since it was not documented in their SoC's RM.
Prevent them bricking their board by retaining power on reset values
in bit fields that the board porter doesn't explicitly configure
via CONFIG_SYS_<registername>_<bitfield> assignments in the board
config file.
also move LCRR assignment to cpu_init_r[am] to help ensure no
transactions are being executed via the local bus while CLKDIV is being
modified.
also start to use i/o accessors.
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
U-Boot can detect if an IDE device is present or not.
If not, and this new config option is activated, U-Boot
removes the ATA node from the DTS before booting Linux,
so the Linux IDE driver does not probe the device and
crash. This is needed for buggy hardware (uc101) where
no pull down resistor is connected to the signal IDE5V_DD7.
Signed-off-by: Heiko Schocher <hs@denx.de>
Signed-off-by: Martha M Stan <mmarx@silicontkx.com>
Minor cleanup:
Re-ordered default_mddrc_config[] to have matching indices.
This allows to use the same index "N" for source and target fields;
before, we had code like this
out_be32(&im->mddrc.ddr_time_config2, mddrc_config[3]);
which always looked like a copy & paste error because 2 != 3.
Also, use NULL when meaning a null pointer.
Signed-off-by: Wolfgang Denk <wd@denx.de>
The means to determine the core, bus, and DDR frequencies are completely
new on CoreNet style platforms. Additionally on p4080 we can have
different frequencies for FMAN and PME IP blocks. We need to keep track
of the FMAN & PME frequencies since they are used for time stamping
capabilities inside each block.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
On CoreNet style platforms the timebase frequency is the bus frequency
defined by 16 (on PQ3 it is divide by 8). Also on the CoreNet platforms
the core not longer controls the enabling of the timebase. We now need
to enable the boot core's timebase via CCSR register writes.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
There are various locations that we have chip specific info:
* Makefile for which ddr code to build
* Added p4080 & p4040 to cpu_type_list and SVR list
* Added number of LAWs for p4080
* Set CONFIG_MAX_CPUS to 8 for p4080
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
The CoreNet platform style of bringing secondary cores out of reset is
a bit different that the PQ3 style. Mostly the registers that we use
to setup boot translation, enable time bases, and boot release the cores
have moved around.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
On CoreNet based platforms the CCSRBAR address is split between an high &
low register and we no longer shift the address.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Scott Wood <scottwood@freescale.com>
