Simon Glass [Wed, 8 May 2013 08:06:06 +0000 (08:06 +0000)]
sparc: Use image_setup_linux() instead of local code
Sparc only really sets up the ramdisk, but we should still use
image_setup_linux() so that setup is common across all architectures
that use the FDT.
Cover-letter
Introduce a common image_setup_linux() function
This series continues the work to tidy up the image code. Each
architecture has its own code for setting up ready for booting linux.
An attempt is made here to unify these in a single image_setup_linux()
function.
The part of the image code that deals with FDT is split into image-fdt.c
and a few tweaks are added to make FIT images more viable in SPL.
END Signed-off-by: Simon Glass <sjg@chromium.org>
Simon Glass [Wed, 8 May 2013 08:06:03 +0000 (08:06 +0000)]
arm: Use image_setup_linux() instead of local code
Use the common FDT setup function that is now available in image. Move
the FDT-specific code to a new bootm-fdt.c and remove unused headers
from bootm.c.
Simon Glass [Wed, 8 May 2013 08:06:02 +0000 (08:06 +0000)]
arm: Refactor bootm to reduce #ifdefs
With fewer #ifdefs the code is more readable and more of the code is
compiled for all boards. Add defines in the header file to control
what features are enabled, and then use if() instead of #ifdef.
Simon Glass [Wed, 8 May 2013 08:06:01 +0000 (08:06 +0000)]
image: Add device tree setup to image library
This seems to be a common function for several architectures, so create
a common function rather than duplicating the code in each arch.
Also make an attempt to avoid introducing #ifdefs in the new code, partly
by removing useless #ifdefs around function declarations in the image.h
header.
Simon Glass [Wed, 8 May 2013 08:05:59 +0000 (08:05 +0000)]
image: Add CONFIG_FIT_SPL_PRINT to control FIT image printing in SPL
This code is very large, and in SPL it isn't always useful to print
out image information (in fact there might not even be a console
active). So disable this feature unless this option is set.
Simon Glass [Wed, 8 May 2013 08:05:57 +0000 (08:05 +0000)]
mkimage: Put FIT loading in function and tidy error handling
The fit_handle_file() function is quite long - split out the part that
loads and checks a FIT into its own function. We will use this
function for storing public keys into a destination FDT file.
The error handling is currently a bit repetitive - tidy it.
Simon Glass [Tue, 7 May 2013 06:12:01 +0000 (06:12 +0000)]
image: Rename fit_add_hashes() to fit_add_verification_data()
We intend to add signatures to FITs also, so rename this function so that
it is not specific to hashing. Also rename fit_image_set_hashes() and
make it static since it is not used outside this file.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Marek Vasut <marex@denx.de>
Simon Glass [Tue, 7 May 2013 06:11:57 +0000 (06:11 +0000)]
image: Rename fit_image_check_hashes() to fit_image_verify()
This is the main entry point to the FIT image verification code. We will
be using it to handle image verification with signatures, so rename the
function.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Marek Vasut <marex@denx.de>
Simon Glass [Tue, 7 May 2013 06:11:51 +0000 (06:11 +0000)]
image: Move timestamp #ifdefs to header file
Rather than repeat the line
#if defined(CONFIG_TIMESTAMP) || defined(CONFIG_CMD_DATE) || \
defined(USE_HOSTCC)
everywhere, put this in a header file and #define IMAGE_ENABLE_TIMESTAMP
to either 1 or 0. Then we can use a plain if() in most code and avoid
the #ifdefs.
The compiler's dead code elimination ensures that the result is the same.
Signed-off-by: Simon Glass <sjg@chromium.org> Acked-by: Marek Vasut <marex@denx.de>
Simon Glass [Tue, 7 May 2013 06:11:48 +0000 (06:11 +0000)]
bootstage: Don't build for HOSTCC
We don't measure boot timing on the host, or with SPL, so use both
conditions in the bootstage header. This allows us to avoid using
conditional compilation around bootstage_...() calls. (#ifdef)
Simon Glass [Tue, 7 May 2013 06:11:47 +0000 (06:11 +0000)]
hash: Add a way to calculate a hash for any algortihm
Rather than needing to call one of many hashing algorithms in U-Boot,
provide a function hash_block() which handles this, and can support all
available hash algorithms.
Once we have md5 supported within hashing, we can use this function in
the FIT image code.
Doug Anderson [Wed, 17 Apr 2013 16:13:41 +0000 (16:13 +0000)]
bootstage: Copy bootstage strings post-relocation
Any pointers to name strings that were passed to bootstage_mark_name()
pre-relocation should be copied post-relocation so that they don't get
trashed as the original location of U-Boot is re-used for other
purposes.
This change introduces a new API call that should be called from
board_init_r() after malloc has been initted on any board that uses
bootstage.
Signed-off-by: Doug Anderson <dianders@chromium.org> Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Simon Glass <sjg@chromium.org>
Simon Glass [Wed, 17 Apr 2013 16:13:39 +0000 (16:13 +0000)]
x86: Re-enable PCAT timer 2 for beeping
While we don't want PCAT timers for timing, we want timer 2 so that we can
still make a beep. Re-purpose the PCAT driver for this, and enable it in
coreboot.
Simon Glass [Wed, 17 Apr 2013 16:13:37 +0000 (16:13 +0000)]
x86: Remove old broken timer implementation
Tidy up some old broken and unneeded implementations. These are not used
by coreboot or anything else now.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Gabe Black <gabeblack@chromium.org> Reviewed-by: Michael Spang <spang@chromium.org> Reviewed-by: Vadim Bendebury <vbendeb@chromium.org> Acked-by: Graeme Russ <graeme.russ@gmail.com>
Simon Glass [Wed, 17 Apr 2013 16:13:35 +0000 (16:13 +0000)]
x86: Rationalise kernel booting logic and bootstage
The 'Starting linux' message appears twice in the code, but both call
through the same place. Unify these and add calls to bootstage to
mark the occasion.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Michael Spang <spang@chromium.org> Acked-by: Graeme Russ <graeme.russ@gmail.com>
Simon Glass [Wed, 17 Apr 2013 16:13:33 +0000 (16:13 +0000)]
x86: Declare global_data pointer when it is used
Several files use the global_data pointer without declaring it. This works
because the declaration is currently a NOP. But still it is better to
fix this so that x86 lines up with other archs.
Commit 1865286466a5d0c7f2e3c37632da56556c838e9e (Introduce generic link
section.h symbol files) changed the __bss_end symbol type from char[] to
ulong. This led to wrong relocation parameters which ended up in a not working
u-boot. Unfortunately this is not clear to see cause due to RAM aliasing we
may get a 'half-working' u-boot then.
Fix this by dereferencing the __bss_end symbol where needed.
Signed-off-by: Andreas Bießmann <andreas.devel@googlemail.com>
Sonic Zhang [Thu, 2 May 2013 05:46:21 +0000 (13:46 +0800)]
bfin: Move gpio support for bf54x and bf60x into the generic driver folder.
The gpio spec for bf54x and bf60x differ a lot from the old gpio driver for bf5xx.
A lot of machine macros are used to accomodate both code in one gpio driver.
This patch split the old gpio driver and move new gpio2 support to the generic
gpio driver folder.
- To enable gpio2 driver, macro CONFIG_ADI_GPIO2 should be defined in the board's
config header file.
- The gpio2 driver supports bf54x, bf60x and future ADI processors, while the
older gpio driver supports bf50x, bf51x, bf52x, bf53x and bf561.
- All blackfin specific gpio function names are replaced by the generic gpio APIs.
Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
Sonic Zhang [Sun, 7 Apr 2013 11:04:14 +0000 (19:04 +0800)]
blackfin: Move blackfin serial driver out of blackfin arch folder.
- Move blackfin serial driver to the generic driver folder.
- Move blackfin serial headers to blackfin arch head folder.
- Update the include path to blackfin serial header in start up code.
Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
Sonic Zhang [Wed, 13 Mar 2013 11:06:16 +0000 (19:06 +0800)]
blackfin: Enable early print via the generic serial API.
Remove blackfin specific implementation of the generic serial API when
early print macro is defined.
In BFIN_BOOT_BYPASS mode, don't call generic serial_puts, because
early print in bypass mode is running before code binary is relocated
to the link address.
Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
Sonic Zhang [Wed, 20 Feb 2013 10:05:16 +0000 (18:05 +0800)]
blackfin: Fit u-boot image size into limited nor flash on blackfin.
- Disable NAND driver on bf537-stamp.
- Make MMC_SPI optional.
- Disable LCD driver on bf527-ezkit.
- Enlarge BF609 nor flash reserved size from 256k to 512k bytes.
Signed-off-by: Sonic Zhang <sonic.zhang@analog.com> Signed-off-by: Sonic Zhang <sonic.adi@gmail.com>
Sonic Zhang [Fri, 30 Nov 2012 09:39:32 +0000 (17:39 +0800)]
blackfin: Correct early serial mess output in BYPASS boot mode.
The early serial should not be configured again in initcode() for BYPASS
boot mode and in start() for the other LDR boot modes.
In BYPASS boot mode, the start up code is located in Nor flash address other
than the DRAM address defined in link script. The code embedded string can't
be addressed by its compile time symbol. Calculate it according to the flash
offset.
Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
Sonic Zhang [Tue, 5 Feb 2013 10:57:49 +0000 (18:57 +0800)]
blackfin: run core1 from L1 code sram start address in uboot init code on core 0
Define core 1 L1 code sram start address.
Add function to enable core 1 for BF609 and BF561.
Add config macro to allow customer to run core 1 in uboot init code on core 0.
Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
Mike Frysinger [Tue, 31 Jul 2012 09:38:56 +0000 (05:38 -0400)]
Blackfin: adjust asm constraints with NMI workaround
Newer gcc versions will sometimes use a Preg when "r" constraints, but
that'll fail if we use an Ireg in the assignment. So force the code
to always use a Dreg.
This also fixes early boot crashes for older Blackfin parts when compiled
with gcc-4.5. This version ends up selecting the same register for the
input and output variables which corrupts the output assignment triggering
an exception.
P2 = 0xffe02008; /* EVT2 */
R0 = RETS;
CALL 1f;
RTN;
1: P2 = RETS; <-- BAD
RETS = R0;
[P2] = P2; <-- BAD
Signed-off-by: Mike Frysinger <vapier@gentoo.org> Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
Albert ARIBAUD [Sat, 11 May 2013 04:29:50 +0000 (04:29 +0000)]
Revert wrong removal of nand_init and nand_deselect
The manual resolution in commit ec7023db wrongly removed functions
nand_init and nand_deselect from file drivers/mtd/nand/mxc_nand_spl.c.
Revert this removal.
Signed-off-by: Albert ARIBAUD <albert.u.boot@aribaud.net> Acked-by: Stefano Babic <sbabic@denx.de>
Justin Sobota [Fri, 15 Feb 2013 16:06:10 +0000 (11:06 -0500)]
Added license header to dtc/libfdt/fdt.h and libfdt_env.h
This commit adds a license header to fdt.h and libfdt_env.h
because the license was omitted.
U-Boot note: the u-boot libfdt_env.h header portion was not applied to
the u-boot libfdt_env.h because that file was created by Gerald Van Baren
(with a license header). - gvb
SRICHARAN R [Wed, 24 Apr 2013 00:41:25 +0000 (00:41 +0000)]
ARM: OMAP: Add arch_cpu_init function
The boot parameters passed from SPL to UBOOT
must be saved as a part of uboot's gd data
as early as possible, before we will inadvertently
overwrite it. So adding a arch_cpu_init for the required
Socs to save it.
Signed-off-by: Sricharan R <r.sricharan@ti.com>
[trini: Add igep0033 hunk] Signed-off-by: Tom Rini <trini@ti.com>
SRICHARAN R [Wed, 24 Apr 2013 00:41:24 +0000 (00:41 +0000)]
ARM: OMAP: Cleanup boot parameters usage
The boot parameters are read from individual variables
assigned for each of them. This been corrected and now
they are stored as a part of the global data 'gd'
structure. So read them from 'gd' instead.
Signed-off-by: Sricharan R <r.sricharan@ti.com>
[trini: Add igep0033 hunk] Signed-off-by: Tom Rini <trini@ti.com>
SRICHARAN R [Wed, 24 Apr 2013 00:41:23 +0000 (00:41 +0000)]
ARM: OMAP: Correct save_boot_params and replace with 'C' function
Currently save_boot_params saves the boot parameters passed
from romcode. But this is not stored in a writable location
consistently. So the current code would not work for a
'XIP' boot. Change this by saving the boot parameters in
'gd' which is always writable. Also add a 'C' function
instead of an assembly code that is more readable.
Eric Benard [Mon, 22 Apr 2013 05:55:00 +0000 (05:55 +0000)]
davinci: handle CONFIG_SYS_CLE_MASK and CONFIG_SYS_ALE_MASK
these variables are curently defined in several config files but the
driver doesn't use them and defaults to hardcoded values in
nand_defs.h
It's interesting to be able to change this hardcoded valude when the
hardware is not using the default adress signals to drive ALE and CLE
and two configuration defines already exist for this purpose so use
them.