Re: new module/init structure test first results

Hi.

Look on memory usage. Storage contains compressed code, linux image in 
RAM will be uncompressed after loading. I think that built-in modules 
will require much more RAM.

15.12.2015 17:38, Erich Titl пишет:
> Hi Folks
>
> first results from my test with module handling
>
> I compiled kernels with all the modules from initmod compiled into the
> kernel for each of the three i386 architectures
>
> SALT# ls -l linux* initrd*
> -rwxr-xr-x    1 root     root       1313792 Dec 15 08:37 initrd.lrp
> -rwxr-xr-x    1 root     root        546639 Dec 15 15:19 initrd1.lrp
> -rwxr-xr-x    1 root     root       1786576 Dec 15 08:37 linux
> -rwxr-xr-x    1 root     root       2214912 Dec 15 15:21 linux1
>
> Adding another entry to boot linux1 with initrd1 to the grub menu is, of
> course, extremely simple. My test system is a WRAP running with a non
> optimized i486 variant of the kernel
>
> We can save about roughly 800 KB on initrd which does not need any
> modules anymore. We have to spend about 450 KB on the kernel, which
> definitely holds too many drivers for my type of architecture. As a net
> result it takes about 350 KB less space on storage than the traditional
> initmod approach.
>
> In this setup there is no moddb involved.
> I still need to test modifications on init to just load the modules from
> sqfs instead of copying them to /lib/modules and getting rid of the
> calls to depmod as this information is written to the modules.sqfs anyway.
>
> cheers
>
> ET
>
> ------------------------------------------------------------------------------
>
> _______________________________________________
> leaf-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/leaf-devel


------------------------------------------------------------------------------

_______________________________________________
leaf-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/leaf-devel