- Do not specify -e twice.
- Use "which" instead of calling commands that might not exist.
- Fix bashism ("==" is C not sh)
- Carefully quote potentially empty variables
- Check command arguments before doing anything
- Rewrite argument checking to be more easily extensible
- Consistent indentation
- UNIX style error messages

fixup fastdata

- no argument => run submodule init
- "nosubmoudle" => do not run submodule
- other values => error message

The following mini6410/tiny6410 functions are available:
init_6410 - initialize clock, timer, DRAM
init_6410_flash - initializes NAND flash support
install_6410_uboot - copies u-boot image into RAM and runs it

Detect based on both manufacturer id and chip id
if manufacturer id is defined in table

- added manufacturer field
- name moved to the end for better text alignment

Signed-off-by: Stefan Mahr <stefan.mahr@sphairon.com>

Signed-off-by: Spencer Oliver <ntfreak@users.sourceforge.net>

cygwin does not define sleep, so use our internal win32 version.
caused by commit 9d4aec6bda

Signed-off-by: Spencer Oliver <ntfreak@users.sourceforge.net>

The beagleboard icepick jtag tap id's vary.

Signed-off-by: Jonas Hörberg <jhorberg@sauer-danfoss.com>

Conflicts:

	src/target/cortex_a.c

there was a check in clearing the status register that
called exit() if the target was running. target_write_memory()
has such a check and will report the error correctly.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

only set jtag global pointer if jtag->init() succeeds. Less code,
more clear what the rules are.

Fix nit that error value from init() was not propagated unmodified.

Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>

- works on Cortex-M3 with ThreadX and FreeRTOS

Compared to original patch a few nits were fixed:

- remove stricmp usage
- unsigned compare fix
- printf formatting fixes
- fixed a bug with overrunning a memory buffer allocated with malloc.

Update devices as per the latest programming manual.

We now use the full DEVID to identify the target. Previously we used
a 8bit id but that has now been changed in the manual.

Signed-off-by: Spencer Oliver <ntfreak@users.sourceforge.net>

Freescale iMX53 doesn't seem to like unaligned accesses to his memory
mapped registers.
Anyway this patch makes dump_image/load_image 4X faster for every
access through APB.

Signed-off-by: Luca Ellero <lroluk@gmail.com>

Signed-off-by: Luca Ellero <lroluk@gmail.com>

Signed-off-by: Luca Ellero <lroluk@gmail.com>