[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: new year, new test kernel.



On Sun, Jan 02, 2005 at 02:59:54AM +0100, Daniel Roesen wrote:

 > > The bulk of the remainder are bugfixes, and a handful
 > > of 'distro' patches that wont go upstream.
 > 
 > Thanks for that. So just to be sure I understand this correctly:
 > only ~8 patches for features

Some of those features are split up into multiple patches.
Ie, Execshield is 3-4, modsign is ~6, etc..
This makes upstream merging easier. Some of the smaller bits
of exec-shield already found their way upstream. It makes it
so much easier to rebase when you can just drop a patch instead
of having to edit a diff/hand merge rejects.

 > over 200 bug fixes, and some "distro"
 > patches? What kind of patches are those "distro" patches?

Things like the make oldconfig_nonint target, things fiddling
with default gcc options (-Os by default, and tune for P4 on the 686 kernel)
Some debugging patches (like linux-2.6.3-printopen.patch
which prints files that are opened, useful for generating
file lists for readahead), things that alter some upstream
decisions (like changing some printk's to panic's and vice versa)
WARN_ON -> BUG_ON). Upstream kernel will allow you to sometimes limp
along after something catastrophic happens, Fedora kernel
will panic as soon as possible to prevent possible data corruption.
Removal of export's of a bunch of things like sys_call_table
which only cause pain, and prevent a bunch of rootkits from running.
(Whilst there are rootkits that can still patch the system without
 this exported, it raises the bar slightly).
Finally, we deprecate a bunch of stuff that will ultimately go
away upstream.

		Dave


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]