Applying kernel patches

Beads · 03-03-2006, 12:07 PM

When applying a patch, the patching routine will often detect a previous patch, and asks if you want to assume a reversal with a default of [n]. Here's the question: Is it normal practice to assume 'no' and just proceed?

I just attempted to install the 2.6.15.5 patch (was previously 2.6.15.4), and answering 'yes' or 'no'either way doesn't allow the kernel to compile without failing.

Advice, please.

nx5000 · 03-03-2006, 12:12 PM

Perfect title, have a look at this:
/usr/src/linux/Documentation/applying-patches.txt

For you:

Code:

$ cd ~/linux-2.6.15.4                   # change into the kernel source dir
$ patch -p1 -R < ../patch-2.6.15.4      # revert the 2.6.15.4 patch
$ patch -p1 < ../patch-2.6.15.5         # apply the new 2.6.12.5 patch
$ cd ..
$ mv linux-2.6.15.4 linux-2.6.15.5      # rename the kernel source dir

You should never have a to answer such a question for official patches, they are always perfect. It always means you did an error.
In your case, 2.6.15.5 is not incremental from 2.6.15.4, it is to apply on 2.6.15!! Not Logic you would say? Imagine if you want to go from 2.6.15.3 to 2.6.15.5. kernel.org would have to handle a looot of files!

Beads · 03-03-2006, 12:18 PM

Ahh! OK, my understanding was that all patches were incremental. Thanks for the clarification! I'll follow your patch instructions, and get it right.

Thanks man.

nx5000 · 03-03-2006, 12:28 PM

You're welcome
By the way, if you don't want to think for applying patch, there is a new tool :
ketchup