Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <5252DC88.20701@barfooze.de>
Date: Mon, 07 Oct 2013 18:08:40 +0200
From: John Spencer <maillist-musl@...fooze.de>
To: musl@...ts.openwall.com
CC: Christian Wiese <chris@...nsde.net>
Subject: Re: linux/netlink.h

On 10/07/2013 05:07 PM, Christian Wiese wrote:
> Hi John,
>
>> imo a good alternative to downloading the full kernel sources to compile
>
>> the headers.
>> an additional advantage is that kernel headers are often broken,
>> the version used by sabotage is known good.
>
> I am curious about what you mean by "kernel headers are often broken".

kernel headers are usually only tested against the kernel - i.e. they 
ship if the kernel compiles. however they often have problems that 
affect userspace apps using them (like including linux/types.h which 
clashes with libc headers, or similar things).


for example, you get a build error when you use kernel 3.8.6 headers to 
compile DOSBOX:

https://sourceforge.net/p/dosbox/patches/258/


a notoriously broken header is ext2_fs.h, which caused problems with 
extlinux in nearly every kernel release - so i decided to put it into 
sabotage directly:

https://github.com/rofl0r/sabotage/commits/master/KEEP/ext2_fs.h

>
> I am also building custom linux based systems, and I have no problems
> using
> the corresponding kernel headers for the kernel version I am building.
>
> It would be interesting for me to know what brokeness you experienced in
> the
> kernel headers.
>
> I already checked the documentation provided by the cross-lfs project, and
> it
> seems that they do not seem to have issues using the corresponding kernel
> headers for the kernel version they use.
>
> Cheers,
> Chris
>

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.