Name already in use

Add support for the new ATOM type 5 which specifies the bank 1 GPIO
settings.

Signed-off-by: Lino Sanfilippo <[email protected]>

Add support for optional configuration of GPIOs on bank 1:
Create a second gpio_map_d structure and instead of using a global variable
for the GPIO map use a local structure pointer that points to the correct
map depending on the GPIO number.

Also support drive, hysteresis and slew for bank 1. Use the new ATOM of
type 5 to store the settings in the EEPROM image file.

Finally extend eeprom_settings.txt by the new configuration settings for
bank 1.

Signed-off-by: Lino Sanfilippo <[email protected]>

The GPIO bank 1 is available on CM1/CM3/CM3+ and CM4S devices. On these
devices an ID EEPROM can be used to get all available GPIOs in a defined
state before the operating system takes over.

Add an optional GPIO map section for GPIO bank 1.

Signed-off-by: Philipp Rosenberger <[email protected]>

If a custom file is added with the -c option no additional memory is
reserved. Thus read_custom() writes over the reserved memory.
To reproduce the issue the following eepmake call is sufficient:
./eepmake eeprom_settings.txt out.eep custom.bin

Signed-off-by: Philipp Rosenberger <[email protected]>

Signed-off-by: Phil Elwell <[email protected]>

The hexadecimal data parsing is fundamentally broken. Not only does it
insert NULs to limit sscanf()s input (which is unnecessary), it writes
the output as ints rather than chars, occupying four times more memory
than it should, corrupting the data, and possibly also corrupting the
heap.

See: #88

Signed-off-by: Phil Elwell <[email protected]>

See: #87

Signed-off-by: Phil Elwell <[email protected]>

A whitespace-only change.

Signed-off-by: Phil Elwell <[email protected]>

Change the serial_<n> structure members into a 4-element array, so that
we can read into all of it without triggering warnings about
overwriting the destination.

Signed-off-by: Phil Elwell <[email protected]>

Ensure that eepflash can be used in a non-interactive (scripting) mode
by adding a "-y" option, that will disable user confirmation.

Signed-off-by: Robin Getz <[email protected]>

Do parameter parsing and validation before testing for root privilege
in order to display help without requiring sudo.

Signed-off-by: Phil Elwell <[email protected]>

Shell expansion of ~ to $HOME (e.g. ~/filename), and TAB completion
on the command line, doesn't work in the middle of a string. Accept
-f filename and --file filename in addition to the usual -f=filename
and --file=filename as a way around this problem.

See: #76

Signed-off-by: Phil Elwell <[email protected]>

Instead of just printing that an error occurred, pass the error up the
call-chain to make the program fail gracefully and avoid producing
corrupted output.

Exit with error code 1 if anything goes wrong. This allows passing
errors in the tool up to scripts or makefiles and avoid silent failures
resulting in corrupted outputs.

BusyBox's dd does not necessarily support 'status' option,
so adapt the detection/handling slightly to be compatible.

Signed-off-by: Michael Heimpold <[email protected]>

Signed-off-by: Michael Heimpold <[email protected]>

If the file name given with -f contains a whitespace the file is not
found. Just use double quote to make sure the file name is not split up.

Signed-off-by: Philipp Rosenberger <[email protected]>

The usage function prints the script name as ./eepflash.sh in the example
it is called just ./eepflash. Use the actual script name instead of
hardcoding the name.

Signed-off-by: Philipp Rosenberger <[email protected]>

Signed-off-by: Philipp Rosenberger <[email protected]>

- fix indention
- move brace of the function to a new line
- move tailing comment over the code line

Signed-off-by: Philipp Rosenberger <[email protected]>

Pi4B also does not have a ZVD

Signed-off-by: Robin Getz <[email protected]>

Don't force people to always add it by hand, or use use the full path.

Signed-off-by: Robin Getz <[email protected]>

by not segfaulting when pointing this to random files (like other
EEPROMs in the system, which might have other data in different formats
in them).

This now checks signature and verison, and keeps track of file length
to make sure things are correct.

tested on files created with eepmake, on disk and pointing to EEPROMs.

Signed-off-by: Robin Getz <[email protected]>

We store the signature in host format, but need to write it to the disk
in big endian format (so it can be read by the Pi properly).

Tested on powerpc64le, 686 and arm.

Signed-off-by: Robin Getz <[email protected]>

This cleans up compiler warnings, to try to keep things cleaner going
forward.

Tested on i686 (gcc 8.3.0, Debian 8.3.0-6); ARMv8a (gcc version 8.3.0,
Raspbian 8.3.0-6+rpi1) and ARMv7 (gcc version 4.8.2, Ubuntu/Linaro 4.8.2-19ubuntu1).

Signed-off-by: Robin Getz <[email protected]>

* Remove references to rpi-update

* Fix up usage: capitalise EEPROM, adjust alignment

Co-authored-by: andrum99 <[email protected]>

The custom_data command is not well documented.  Through a reading
of eepmake.c, it was clear that it was possible to include custom
data in the eeprom_settings.txt file, but the format was not clear.
Once I figured out the format, I found that eepmake would segfault:

```
hats/eepromutils$ ./eepmake eeprom_settings.txt /dev/null
Opening file eeprom_settings.txt for read
UUID=71b07c8e-10fe-472c-b24a-906f5301fb4e
Done reading
Writing out...
eepmake(99370,0x7fffacf01380) malloc: *** error for object 0x7fe3c6c02818: incorrect checksum for freed object - object was probably modified after being freed.
*** set a breakpoint in malloc_error_break to debug
Abort trap: 6
```

This is because the data_receive flag is never set back to false.
There is a final call to finish_data() in read_text(), which I think
is there to allow long custom data or device tree blobs to be tacked
onto the end of the file without an "end" command.  However, since
data_receive was not being set to false, finish_data() is run fully
twice.  This results in a count of two custom atoms when there is
only one.  eepmake crashes when trying to write out the second,
non-existant custom atom.

I've added a small example to the eeprom_settings.txt to show
the custom_data command.

Signed-off-by: Phil Elwell <[email protected]>

See #68.

Explain that overlays can be embedded directly or by name. Give some
hints about submitting code and overlays to our kernel repo.

Signed-off-by: Phil Elwell <[email protected]>

Signed-off-by: Phil Elwell <[email protected]>