Age | Commit message (Collapse) | Author |
|
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>
|
|
By now, the majority of architectures have working relocation
support, so the few remaining architectures have become exceptions.
To make this more obvious, we make working relocation now the default
case, and flag the remaining cases with CONFIG_NEEDS_MANUAL_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>
|
|
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>
|
|
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>
|
|
u-boot.bin can be loaded at any 4-byte aligned memory location and directly
'jumped' to using the 'go' command using the load address as the start
address. Doing so performs a 'warm boot' which skips memory initialisation
and other low-level initialisations, relocates U-Boot to upper memory and
starts U-Boot in RAM as per normal 'cold boot'
|
|
Provides a small speed increase and prepares for fully relocatable image.
Downside is the TEXT_BASE, bss, load address etc must ALL be aligned on a
a 4-byte boundary which is not such a terrible restriction as everything
is already 4-byte aligned anyway
|
|
Tidy up the linker script and discard some sections to save space
|
|
Create more generic names for the symbols exported from the linker script
|
|
|
|
By reserving space for the Global Data immediately below the stack during
assembly level initialisation, the C declaration of the static global data
can be removed, along with the 'RAM Bootstrap' function. This results in
cleaner code, and the ability to pass boot-up flags from assembler into C
|
|
By using another register, reduce code size by one instruction
|
|
%ebx will hold low-level boot flags and must be preserved
|
|
Using %ebp as a return pointer prevents creating 'load anywhere' images
|
|
To allow for 'load anywhere' images, the %ebp return pointer 'hack' must
be removed, so we cannot have two 'calls' to get_mem_size
|
|
Progress indication is not relocation friendly so remove it in
preperation for full relocatability support
|
|
|
|
Change to:
- reparam=3
- no-from-pointer
- no-stack-protector
- preferred-stack-boundary=2
- no-top-level-reorder
These options make the code a little smaller and faster
|
|
Perform some basic code cleanups of the x86 files
|
|
Linux has C macros and code to load the GTD after switching to Protected
Mode. Using these greatly simplifies the assembler code
|
|
Follow the discussion of Charles Manning and Mike Frysinger.
Using gc_sections helps reduce image size.
|
|
The header of recent Linux Kernels includes the size of the image, and
therefore is not needed to be passed to zboot. Still process the third
parameter (size of image) in the event that an older kernel is being loaded
|
|
Relocation is not board-specific for the x86 architectrure, so
CONFIG_RELOC_FIXUP_WORKS can be defined globally in the config.h
|
|
This patch removes the completely unused CONFIG_SERIAL_SOFTWARE_FIFO
feature from U-Boot. It has only been implemented for PPC4xx and was not
used at all. So let's remove it and make the code smaller and cleaner.
Signed-off-by: Stefan Roese <sr@denx.de>
Acked-by: Detlev Zundel <dzu@denx.de>
|
|
bi_env is nowhere used, so delete it!
Signed-off-by: Heiko Schocher <hs@denx.de>
similar patch posted from Dirk Behme Tue Jul 27 18:36:09 CEST 2010
http://lists.denx.de/pipermail/u-boot/2010-July/074542.html
|
|
Motivation:
* Old environment code used a pessimizing implementation:
- variable lookup used linear search => slow
- changed/added variables were added at the end, i. e. most
frequently used variables had the slowest access times => slow
- each setenv() would calculate the CRC32 checksum over the whole
environment block => slow
* "redundant" envrionment was locked down to two copies
* No easy way to implement features like "reset to factory defaults",
or to select one out of several pre-defined (previously saved) sets
of environment settings ("profiles")
* No easy way to import or export environment settings
======================================================================
API Changes:
- Variable names starting with '#' are no longer allowed
I didn't find any such variable names being used; it is highly
recommended to follow standard conventions and start variable names
with an alphanumeric character
- "printenv" will now print a backslash at the end of all but the last
lines of a multi-line variable value.
Multi-line variables have never been formally defined, allthough
there is no reason not to use them. Now we define rules how to deal
with them, allowing for import and export.
- Function forceenv() and the related code in saveenv() was removed.
At the moment this is causing build problems for the only user of
this code (schmoogie - which has no entry in MAINTAINERS); may be
fixed later by implementing the "env set -f" feature.
Inconsistencies:
- "printenv" will '\\'-escape the '\n' in multi-line variables, while
"printenv var" will not do that.
======================================================================
Advantages:
- "printenv" output much better readable (sorted)
- faster!
- extendable (additional variable properties can be added)
- new, powerful features like "factory reset" or easy switching
between several different environment settings ("profiles")
Disadvantages:
- Image size grows by typically 5...7 KiB (might shrink a bit again on
systems with redundant environment with a following patch series)
======================================================================
Implemented:
- env command with subcommands:
- env print [arg ...]
same as "printenv": print environment
- env set [-f] name [arg ...]
same as "setenv": set (and delete) environment variables
["-f" - force setting even for read-only variables - not
implemented yet.]
- end delete [-f] name
not implemented yet
["-f" - force delete even for read-only variables]
- env save
same as "saveenv": save environment
- env export [-t | -b | -c] addr [size]
export internal representation (hash table) in formats usable for
persistent storage or processing:
-t: export as text format; if size is given, data will be
padded with '\0' bytes; if not, one terminating '\0'
will be added (which is included in the "filesize"
setting so you can for exmple copy this to flash and
keep the termination).
-b: export as binary format (name=value pairs separated by
'\0', list end marked by double "\0\0")
-c: export as checksum protected environment format as
used for example by "saveenv" command
addr: memory address where environment gets stored
size: size of output buffer
With "-c" and size is NOT given, then the export command will
format the data as currently used for the persistent storage,
i. e. it will use CONFIG_ENV_SECT_SIZE as output block size and
prepend a valid CRC32 checksum and, in case of resundant
environment, a "current" redundancy flag. If size is given, this
value will be used instead of CONFIG_ENV_SECT_SIZE; again, CRC32
checksum and redundancy flag will be inserted.
With "-b" and "-t", always only the real data (including a
terminating '\0' byte) will be written; here the optional size
argument will be used to make sure not to overflow the user
provided buffer; the command will abort if the size is not
sufficient. Any remainign space will be '\0' padded.
On successful return, the variable "filesize" will be set.
Note that filesize includes the trailing/terminating '\0'
byte(s).
Usage szenario: create a text snapshot/backup of the current
settings:
=> env export -t 100000
=> era ${backup_addr} +${filesize}
=> cp.b 100000 ${backup_addr} ${filesize}
Re-import this snapshot, deleting all other settings:
=> env import -d -t ${backup_addr}
- env import [-d] [-t | -b | -c] addr [size]
import external format (text or binary) into hash table,
optionally deleting existing values:
-d: delete existing environment before importing;
otherwise overwrite / append to existion definitions
-t: assume text format; either "size" must be given or the
text data must be '\0' terminated
-b: assume binary format ('\0' separated, "\0\0" terminated)
-c: assume checksum protected environment format
addr: memory address to read from
size: length of input data; if missing, proper '\0'
termination is mandatory
- env default -f
reset default environment: drop all environment settings and load
default environment
- env ask name [message] [size]
same as "askenv": ask for environment variable
- env edit name
same as "editenv": edit environment variable
- env run
same as "run": run commands in an environment variable
======================================================================
TODO:
- drop default env as implemented now; provide a text file based
initialization instead (eventually using several text files to
incrementally build it from common blocks) and a tool to convert it
into a binary blob / object file.
- It would be nice if we could add wildcard support for environment
variables; this is needed for variable name auto-completion,
but it would also be nice to be able to say "printenv ip*" or
"printenv *addr*"
- Some boards don't link any more due to the grown code size:
DU405, canyonlands, sequoia, socrates.
=> cc: Matthias Fuchs <matthias.fuchs@esd-electronics.com>,
Stefan Roese <sr@denx.de>,
Heiko Schocher <hs@denx.de>
- Dropping forceenv() causes build problems on schmoogie
=> cc: Sergey Kubushyn <ksi@koi8.net>
- Build tested on PPC and ARM only; runtime tested with NOR and NAND
flash only => needs testing!!
Signed-off-by: Wolfgang Denk <wd@denx.de>
Cc: Matthias Fuchs <matthias.fuchs@esd-electronics.com>,
Cc: Stefan Roese <sr@denx.de>,
Cc: Heiko Schocher <hs@denx.de>
Cc: Sergey Kubushyn <ksi@koi8.net>
|
|
So far, getenv() would work before relocation is most cases, even
though it was not intended to be used that way. When switching to a
hash table based implementation, this would break a number of boards.
For convenience, we make getenv() check if it's running before
relocation and, if so, use getenv_f() internally.
Note that this is limited to simple cases, as we use a small static
buffer (32 bytes) in the global data for this purpose.
For this reason, it is also not a good idea to convert all current
uses of getenv_f() into getenv() - some of the existing use cases need
to be able to deal with longer variable values, so getenv_f() is still
needed and recommended for use before relocation.
Signed-off-by: Wolfgang Denk <wd@denx.de>
|
|
Commit 077e1958ca4afe12d88043b123ded058c51b89f7 broke the ability of the
x86 port to boot from a cold-reset by removing the initial IDT. Re-
instate the initial IDT to allow cold-booting of x86 boards
|
|
Commit 54841ab50c20d6fa6c9cc3eb826989da3a22d934 made the argv parameter
to do_go_exec() const but did not allow for the fact that argv[-1] is
set to point to the global data structure and relies on argv being non-
const.
With this patch, do_go_exec() creates a new copy of the argv array with
an extra element to store global data pointer rather than simply
clobbering an arbitrary memory location.
|
|
While running from flash, i. e. before relocation, we have only a
limited C runtime environment without writable data segment. In this
phase, some configurations (for example with environment in EEPROM)
must not use the normal getenv(), but a special function. This
function had been called getenv_r(), with the idea that the "_r"
suffix would mean the same as in the _r_eentrant versions of some of
the C library functions (for example getdate vs. getdate_r, getgrent
vs. getgrent_r, etc.).
Unfortunately this was a misleading name, as in U-Boot the "_r"
generally means "running from RAM", i. e. _after_ relocation.
To avoid confusion, rename into getenv_f() [as "running from flash"]
Signed-off-by: Wolfgang Denk <wd@denx.de>
Acked-by: Detlev Zundel <dzu@denx.de>
|
|
The hush shell dynamically allocates (and re-allocates) memory for the
argument strings in the "char *argv[]" argument vector passed to
commands. Any code that modifies these pointers will cause serious
corruption of the malloc data structures and crash U-Boot, so make
sure the compiler can check that no such modifications are being done
by changing the code into "char * const argv[]".
This modification is the result of debugging a strange crash caused
after adding a new command, which used the following argument
processing code which has been working perfectly fine in all Unix
systems since version 6 - but not so in U-Boot:
int main (int argc, char **argv)
{
while (--argc > 0 && **++argv == '-') {
/* ====> */ while (*++*argv) {
switch (**argv) {
case 'd':
debug++;
break;
...
default:
usage ();
}
}
}
...
}
The line marked "====>" will corrupt the malloc data structures and
usually cause U-Boot to crash when the next command gets executed by
the shell. With the modification, the compiler will prevent this with
an
error: increment of read-only location '*argv'
N.B.: The code above can be trivially rewritten like this:
while (--argc > 0 && **++argv == '-') {
char *arg = *argv;
while (*++arg) {
switch (*arg) {
...
Signed-off-by: Wolfgang Denk <wd@denx.de>
Acked-by: Mike Frysinger <vapier@gentoo.org>
|
|
Signed-off-by: Wolfgang Denk <wd@denx.de>
|
|
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
AMD recently changed the licensing of the RAM sizing code to the
GPLv2 (or at your option any later version)
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
If the board has a high precision mico-second timer, it maked sense to use
it instead of the on-chip one
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
Add support for newer (up to 2.6.33) kernels
Add zboot command which takes the address of a bzImage as its first
argument and (optionally) the size of the bzImage as the second argument
(the second argument is needed for older kernels which do not include
the bzImage size in the header)
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
It is possibly to setup x86 boards to use non-PC/AT configurations. For
example, the sc520 is an x86 CPU with PC/AT and non-PC/AT peripherals.
This function allows the board to set itself up for maximum PC/AT
compatibility just before booting the Linux kernel (the Linux kernel
'just works' if everything is PC/AT compliant)
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
In order to locate the 16-bit BIOS code, we need to know the reloaction
offset.
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
Currently, the GDT is either located in FLASH or in the non-relocated
U-Boot image in RAM. Both of these locations are unsafe as those
locations can be erased during a U-Boot update. Move the GDT into the
highest available memory location and relocate U-Boot to just below it
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
Add a parameter to the 32-bit entry to indicate if entry is from Real
Mode or not. If entry is from Real Mode, execute the destructive 'sizer'
routine to determine memory size as we are booting cold and running in
Flash. If not entering from Real Mode, we are executing a U-Boot image
from RAM and therefore the memory size is already known (and running
'sizer' will destroy the running image)
There are now two 32-bit entry points. The first is the 'in RAM' entry
point which exists at the start of the U-Boot binary image. As such,
you can load u-boot.bin in RAM and jump directly to the load address
without needing to calculate any offsets. The second entry point is
used by the real-to-protected mode switch
This patch also changes TEXT_BASE to 0x6000000 (in RAM). You can load
the resulting image at 0x6000000 and simple go 0x6000000 from the u-boot
prompt
Hopefully a later patch will completely elliminate any dependency on
TEXT_BASE like a relocatable linux kernel (perfect world)
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
This patch allows the low-level assembler boot-strap to obtain the RAM
size without calling the destructive 'sizer' routine. This allows
boot-strapping from a U-Boot image loaded in RAM
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
There is an error in how the assembler version of the sc520 memory size
reporting code works. As a result, it will only ever report at most the
size of one bank of RAM
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
This was broken a long time ago by a49864593e083a5d0779fb9ca98e5a0f2053183d
which munged the NIOS and x86 do_go_exec()
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
Shamelessly steal the Linux x86 crash handling code and shove it into
U-Boot (cool - it fits). Be sure to include suitable attribution to
Linus
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
Change sc520 MMCR Access to use memory accessor functions
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|
|
Signed-off-by: Graeme Russ <graeme.russ@gmail.com>
|