Hi Marek,
On Fri, 29 Oct 2021 at 02:58, Marek Behún kabel@kernel.org wrote:
On Thu, 28 Oct 2021 21:17:41 -0600 Simon Glass sjg@chromium.org wrote:
Hi Marek,
On Wed, 27 Oct 2021 at 21:28, Marek Behún kabel@kernel.org wrote:
From: Marek Behún marek.behun@nic.cz
The default_environment[] buffer is built at compile time, but sometimes it makes sense for some default environment variables to be determined at runtime, for example:
- one board code may support different boards, and needs that fdtfile, board, board_name are set appropriately when command env default -a is executed
- some boards may want to prohibit the env default -a command to remove device MAC addresses stored in ethaddr, ethNaddr. This is the case for the ESPRESSObin board code, for example, where currently the board_late_init() function rewrites the default environment array to achieve this.
Add a new board specific function,
const char *board_special_default_env(unsigned i, const char **name);
which returns the value of i-th board special default environemnt variable, while storing it's name to *name.
Add default weak implementation of this functions returning NULL.
Add code to default environemnt handlers in env/common.c, which iterate these special board default environment variables and get it's values in precedence to values in the default_environment[] buffer.
Signed-off-by: Marek Behún marek.behun@nic.cz
env/common.c | 99 ++++++++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 96 insertions(+), 3 deletions(-)
What do you think about using a sysinfo driver for this? There are already a lot of weak functions in U-Boot.
I confess I am looking at sysinfo API for the first time just now. It seems like a reasonable thing, but how to combine with default environment for this?
The thing is that Pali's code for ESPRESSObin needs to prohibit deleting the ethaddrs from env with "env default -a", because there were issues with the mac address being lost...
The way you have done it seems reasonable to me, i.e. calling an iterator function to get the strings. Could it be done with sysinfo's get_str()? Perhaps we should add get_str_list() as a new method that takes and id and also an index?
Regards, Simon