Re: BBC patch
David Woodhouse <[email protected]> Fri, 07 May 2004 12:13:57 +0100
| Newsgroups | gmane.linux.file-systems.jffs |
|---|---|
| Message-ID | <[email protected]> |
On Fri, 2004-05-07 at 12:09 +0200, Ferenc Havasi wrote: > Now I'm working on the separated patches, checked out the newest jffs2 > code and I see the changes. Now all compressor has a kernel and a user > space version. They are very small compressors - I think it's OK. > > But: what about the bigger compressors, and the shared code in our > compr.c (compressor registartion, compression mode handling, and later > model file handling) Should we (almost) duplicate these codes (in > fs/jffs2/ and util/) or apply the symlink-solution and ifdefs? I think the model file handling is going to _have_ to be different in the kernel and in mkfs.jffs2, isn't it? The latter can just mmap it. I'm not too unhappy with duplicating the registration code, since I think that will differ enough that the ifdefs make it too ugly to share. For the compressors, themselves, I think we can move them into a separate directory and aim to build them in both environments. I'd prefer to write them as kernel code and then make that work in userspace with #defines for kmalloc, printk etc. Rather than littering them with ifdefs, we can have a header file which is different for kernel and for userspace, and which makes it just work. We handle building for both 2.6 and 2.4 kernels that way -- it's not so hard. -- dwmw2 To unsubscribe from this list: send the line "unsubscribe jffs-dev" in the body of a message to [email protected]