r8 is used for global_data and should therefore be left alone!
For C code the compiler flag --fixed-r8 does the job, but in assembler
we need to be aware of that fact.
Signed-off-by: Andreas Biemann <andreas.devel@googlemail.com>
In case we are still at relocation target address before relocation we
do not need to load the registers needed for relocation. We should
instead skip the whole relocation part and jump over to clear_bss
immediately.
Signed-off-by: Andreas Biemann <andreas.devel@googlemail.com>
This patch uses r1 as scratch register for copy_loop(). Therefore we do
not longer need r7 for the storage of relocate_code()'s 'addr_moni' (the
destination address of relocation).
Therefore r7 can be used later on for other purposes.
Signed-off-by: Andreas Biemann <andreas.devel@googlemail.com>
suggested from Daniel Hobi<daniel.hobi@schmid-telecom.ch>
Tested on following boards:
arm1136: qong
armv7: omap3_beagle
arm926ejs: magnesium, tx25
Signed-off-by: Heiko Schocher <hs@denx.de>
cc: Daniel Hobi <daniel.hobi@schmid-telecom.ch>
cc: Albert ARIBAUD <albert.aribaud@free.fr>
Before this commit, weak symbols were not overridden by non-weak symbols
found in archive libraries when linking with recent versions of
binutils. As stated in the System V ABI, "the link editor does not
extract archive members to resolve undefined weak symbols".
This commit changes all Makefiles to use partial linking (ld -r) instead
of creating library archives, which forces all symbols to participate in
linking, allowing non-weak symbols to override weak symbols as intended.
This approach is also used by Linux, from which the gmake function
cmd_link_o_target (defined in config.mk and used in all Makefiles) is
inspired.
The name of each former library archive is preserved except for
extensions which change from ".a" to ".o". This commit updates
references accordingly where needed, in particular in some linker
scripts.
This commit reveals board configurations that exclude some features but
include source files that depend these disabled features in the build,
resulting in undefined symbols. Known such cases include:
- disabling CMD_NET but not CMD_NFS;
- enabling CONFIG_OF_LIBFDT but not CONFIG_QE.
Signed-off-by: Sebastien Carlier <sebastien.carlier@gmail.com>
older ld emitted all ELF relocations in input sections named
.rel.dyn, whereas newer ld uses names of the form .rel*. The
linker script only collected .rel.dyn input sections. Rewrite
to collect all .rel* input sections.
Signed-off-by: Albert Aribaud <albert.aribaud@free.fr>
For ARM systems, before ELF relocation was introduced,
CONFIG_SKIP_RELOCATE_UBOOT coul be used to prevent *COPYING* the
U-Boot image from whereever it was loaded to it's link address
(CONFIG_SYS_TEXT_BASE). The name was badly chosen, as no relocation
was performed at all, it was just a memcpy().
With ELF relocation, this does not work like that any more, and
related boards need to be fixed anyway. So don't keep this relict any
longer.
Signed-off-by: Wolfgang Denk <wd@denx.de>
Tested-by: Heiko Schocher <hs@denx.de>
Tested-by: Reinhard Meyer <u-boot@emk-elektronik.de>
When this define was introduced, the idea was to provide a soft
migration path for ARM boards to get adapted to the new relocation
support. However, other recent changes led to a different
implementation (ELF relocation), where this no longer works. By now
CONFIG_SYS_ARM_WITHOUT_RELOC does not only not help any more, but it
actually hurts because it obfuscates the actual code by sprinkling it
with lots of dead and non-working debris.
So let's make a clean cut and drop CONFIG_SYS_ARM_WITHOUT_RELOC.
Signed-off-by: Wolfgang Denk <wd@denx.de>
Tested-by: Heiko Schocher <hs@denx.de>
Tested-by: Reinhard Meyer <u-boot@emk-elektronik.de>
For the "fixloop" implementation in start.S a number of different
instructions was used. Unify code so all architectures use "blo"
here because it is more robust in case of incorrect alignments.
Signed-off-by: Wolfgang Denk <wd@denx.de>
Cc: Albert ARIBAUD <albert.aribaud@free.fr>
Cc: Minkyu Kang <mk7.kang@samsung.com>
Cc: Sandeep Paulraj <s-paulraj@ti.com>
Cc: Prafulla Wadaskar <prafulla@marvell.com>
Cc: Stefano Babic <sbabic@denx.de>
Cc: Marek Vasut <marek.vasut@gmail.com>
Acked-by: Heiko Schocher <hs@denx.de>
Fix address setup bug for ARM.
This bug stops u-boot booting if
CONFIG_SKIP_RELOCATE_UBOOT is defined.
Signed-off-by: Darius Augulis <augulis.darius@gmail.com>
CONFIG_SYS_GBL_DATA_SIZE has always been just a bad workarond for not
being able to use "sizeof(struct global_data)" in assembler files.
Recent experience has shown that manual synchronization is not
reliable enough. This patch renames CONFIG_SYS_GBL_DATA_SIZE into
GENERATED_GBL_DATA_SIZE which gets automatically generated by the
asm-offsets tool. In the result, all definitions of this value can be
deleted from the board config files. We have to make sure that all
files that reference such data include the new <asm-offsets.h> file.
No other changes have been done yet, but it is obvious that similar
changes / simplifications can be done for other, related macro
definitions as well.
Signed-off-by: Wolfgang Denk <wd@denx.de>
Acked-by: Kumar Gala <galak@kernel.crashing.org>
Conflicts:
arch/arm/include/asm/config.h
board/LaCie/edminiv2/config.mk
board/karo/tx25/config.mk
board/logicpd/imx27lite/config.mk
doc/README.arm-relocation
Signed-off-by: Wolfgang Denk <wd@denx.de>
The change is currently needed to be able to remove the board
configuration scripting from the top level Makefile and replace it by
a simple, table driven script.
Moving this configuration setting into the "CONFIG_*" name space is
also desirable because it is needed if we ever should move forward to
a Kconfig driven configuration system.
Signed-off-by: Wolfgang Denk <wd@denx.de>
Generalized misuse of ble within relocation and bss
initialization loops caused one iteration too many.
Instead of ble ('branch if lower or equal'), use
blo ('branch if lower').
While we're at it, fix all 'addreee' typos.
Signed-off-by: Albert Aribaud <albert.aribaud@free.fr>
Change the implementation for ARM11 to relocate the code to an
arbitrary address in RAM.
Tested on the qong board.
Portions of this work were supported by funding from
the CE Linux Forum.
Signed-off-by: Heiko Schocher <hs@denx.de>
Add data cache support for arm1136 systems.
Enable "cache" command on Qong board and test performance.
Test 1: Loading 127 MB of data from NAND flash into RAM:
Instr. Cache off on on
Data Cache off off on
--------------------------------------------------
QONG (ARM11) 177s 95s 43s = x 4.1
Test 2: uncompressing a gzipped image from RAM to RAM
(size compressed: 6.5 MiB, uncompressed: 35 MiB):
Instr. Cache off on on
Data Cache off off on
--------------------------------------------------
QONG (ARM11) 1.54s 0.95s 0.18s = x 8.6
Portions of this work were supported by funding from
the CE Linux Forum.
Signed-off-by: Heiko Schocher <hs@denx.de>
The ARM ABI requires that the stack be aligned to 8 bytes as it is noted
in Procedure Call Standard for the ARM Architecture:
http://infocenter.arm.com/help/topic/com.arm.doc.ihi0042d/index.html
Unaligned SP also causes the problem with variable-length arrays
allocation when VLA address becomes less than stack pointer during
aligning of this address, so the next 'push' in the stack overwrites
first 4 bytes of VLA.
Signed-off-by: Vitaly Kuzmichev <vkuzmichev@mvista.com>
The ARM1136 cache_flush() function uses the "mcr p15, 0, rn, c7, c7, 0"
instruction which means "Invalidate Both Caches" when in fact the intent
is to clean and invalidate all caches. So add an "mcr p15, 0, %0, c7,
c10, 0" instruction to "Clean Entire Data Cache" prior to the "Invalidate
Both Caches" instruction to insure that memory is consistent with any
dirty cache lines.
Also fix a couple of "flush v*" comments in ARM1136 cpu_init_crit() so
that they correctly describe the actual ARM1136 CP15 C7 Cache Operations
used.
Signed-off-by: George G. Davis <gdavis@mvista.com>
