uClibc CLFS style

Jeremy Huntwork jhuntwork at linuxfromscratch.org
Tue Feb 7 19:57:58 PST 2006

Hello Everyone,

I wanted to apologize for my abrupt entrance into this list the other 
day. It didn't occur to me that my manner may have been offensive. I 
have some ideas I'd like to present, so please take them only as an 
effort to be helpful.

When looking over the uClibc book, it occurred to me that, as we're 
cross-compiling, some of the methods from Cross-LFS might be useful. So, 
mostly out of curiosity, I applied their full technique to the specifics 
of the uClibc HLFS book. The biggest differences to the current build 
method are:

* No adjusting of the toolchain in chapter 5
* Unecessary to keep the binutils dir around
* Once you get the cross-tools built the rest of the temporary tools (in 
chapter 5) are all cross-compiled.
* Temp tools aren't built with PIE or SSP (it might be possible to do, 
but I didn't look into it fully)
* One extra build of GCC (since the CLFS method builds first a static 
cross-compiler with only C support, and then a Shared Final 
cross-compiler with C and C++ support.)

I did my best to list the above only as differences to the current 
method without endorsing one way over the other. I'm not trying to push 
this method into HLFS, I'm just bringing it up as something to consider. 
Mostly, I'd like to know if there is interest in this. If so, I'll be 
happy to share the notes I've started to collect - If not, I'll probably 
just delete them.


More information about the hlfs-dev mailing list