# Re: patch for 2.6.36.4 - From: Kent Overstreet \ - Date: 2011-08-24 17:57:09 - List: linux-bcache - Page: https://ratatoskr.run/linux-bcache/2011/08/7635342 ```text On 08/24/11 10:56, Justin Rush wrote: > If I could merge it in, that it would be better, since the gigantic patch is still applying to the 2.6.36.4 source. > > linux-2.6# git merge linux-bcache > fatal: 'linux-bcache' does not point to a commit Err, that should be whatever the bcache branch was named, probably just bcache. > > ? > > > - Justin > > > -----Original Message----- > From: Kent Overstreet [mailto:] > Sent: Wednesday, August 24, 2011 12:44 PM > To: Justin Rush > Cc: > Subject: Re: patch for 2.6.36.4 > > On 08/24/11 04:58, Justin Rush wrote: >> I am running 2.6.36.4 and I had to apply a few patches to get current hardware working as well as the lio backports so I prefer to stay on this kernel. Is there a 2.6.34.4 patch available? If not, can I make one? >> >> Looking at this helpful human's post, it looks fairly straightforward: >> http://comments.gmane.org/gmane.linux.kernel.bcache.devel/24 >> >> I checked out both trees, then switched to my branch: >> git checkout -b 2.6.36.4 >> >> Then created the patch: >> git diff linux-2.6 linux-bcache>patch > > You probably want something like > git diff v2.6.34 linux-bcache> patch > > Really you should just be able to do a git merge linux-bcache from the 2.6.34-stable branch though... > >> >> However, this resulted in a 236MB patch file, which is several orders of magnitude bigger than any other patch I have ever used. Assuming this is even the right way to do it, do I just install it by moving it into my running kernel's source and doing a patch -p0 patch ? >> >> Slightly un-related question: >> I want to use this as a cache for two sw raid6 arrays. The raid devices have a bunch of logical volumes carved out of them and those LVs are presented to LIO and served out as iSCSI LUNs. Will bcache serve as a write cache for these devices? My read performance is great, but my random write performance, or write performance in general, really sucks hard. > > Yeah, writeback caching has been implemented for quite awhile now, just have to flip it on :) > > echo 1> /sys/block/bcache0/bcache/writeback ```