On Wed, 2025-01-29 at 13:31 +0300, Dmitry Antipov wrote:
In 'iwl_pnvm_load()', assume that the power table size is always equal to IWL_HARDCODED_REDUCE_POWER_SIZE. Compile tested only.
Signed-off-by: Dmitry Antipov dmantipov@yandex.ru
I would gently ask Johannes to review this before taking any other actions. This is intended for stable linux-6.1.y only in attempt to avoid possible use of an uninitialized 'len' without backporting
I don't see that there's uninitialized use of 'len'. Maybe some static checkers aren't seeing through this, but the code is fine:
If iwl_pnvm_get_from_fs() is successful, then 'len' is initialized. If it fails, we goto skip_parse.
There, if trans->reduce_power_loaded is false, 'len' again is either initialized or we go to skip_reduce_power and never use 'len'.
If trans->reduce_power_loaded is true, then we get to iwl_trans_pcie_ctx_info_gen3_set_reduce_power() which doesn't use 'len' in this case.
I'd rather not fix this non-problem in a very confusing way.
johannes