Re: Subject: [PATCH] crypto: add zbewalgo compression algorithm for zram

["Stephan Müller" <smueller@chronox.de>]

Am Dienstag, 30. Januar 2018, 20:49:00 CET schrieb Benjamin Warnke:

Hi Benjamin,

> > In general: I do not think that having larger C functions in header files
> > is a proper coding style.
> 
> How should I solve this?
> 
> Option 1:
> Move everything in the lib/zbewalgo folder into a single source file.
> This way there is no function defined in a header file.
> I separated the code into different files because the different partial
> compression algorithms are independent from each other.
> 
> Option 2:
> Move each header file in its own *.c file, while keeping the
> function-declarations in the header. If the code is scattered in multiple
> source files each of the partial algorithms would show up as an independent
> module. All these modules would load simultaneously with the module
> zbewalgo_compress The module zbewalgo_compress requires an array of all
> partial algorithms. This would spam the 'lsmod' list with unneccessary
> details.

A module may be compiled from multiple C files. So, moving the code into 
separate C files and link them into one object / KO should be considered.

> 
> > Why are there __attribute__ ((unused)) function parameters, such as in
> > compress_bitshuffle and others?
> 
> The zbewalgo algorithm is a container-algorithm for compression functions
> with the ability to concatenate multiple algorithms. To be able to call any
> compression algorithm the same way, I defined 'struct zbewalgo_alg' as the
> interface for all those independent compression algorithms. Some of the
> partial algorithms do not require all parameters. To silence compiler
> warnings (if additional warning flags are enabled) I explicitly add the
> 'unused'-parameter

Linux does not enable the compiler warning about unused parameters.

Ciao
Stephan