Ext3 FS driver being removed from Linux Kernel
 

This was reported on Phoronix, however, here is the link from the kernel mailing list: https://lkml.org/lkml/2015/7/15/438

Maybe. https://lkml.org/lkml/2015/8/31/474

And just in case the somewhat alarmist subject alarms you (and you didn't read all of the LKML threads), no ext3 isn't going away.
It's been handled by the ext4 codebase for years.

Despite the fluidity of the discussion, it looks like Linus has been convinced after all.

When I first saw this a few days ago it caused me some concern as I still rely on ext3 (and ext2).

But after reading the lkml I see no concerns at all.

ext3 support to the user will remain unaffected as far as I can tell, except that the identical function (regression tested, guaranteed identical function) will be supplied by the ext4 code base.

So removal of redundant code is about the sum total of the changes. Sounds OK to me!

Slackware's kernel config was still using the standalone ext[23] modules rather than using the ext4 module for ext2/3.

This is one of the changes I've been making to my local kernels for a good while now:
Funnily enough, a week or so ago, before all this broke, I contemplated posting on the subject to ask whether there was any reason Slackware was still using the separate EXT2/3 modules, but decided not to bother as I knew what the answer would be -- "because we haven't needed to change it". Looks like the issue is going to be forced now.

The latest from Linus on it:

I wonder, how it will work for me? I use ext2 for my partitions.

That same ext4 driver.

His final word on it:
"On the filesystem side, the bulk of the changes (in lines of code) is
the removal of the ext3 filesystem (with ext4 remaining to support
ext3 layouts - but the separate ext3 codebase is gone)."


https://lkml.org/lkml/2015/9/12/243

So it seems that existing kernels are fine and will continue to work as always and only new kernels will require a change to config. Does this affect all branches or only the 4x branch and at what point(s) will this change take affect?

I assume that the merge window for 4.3 is closed so it will be effective in 4.4 (unless Linus decides otherwise). Of course older kernels are not affected and I see no rationale for this change to be backported. As for the configuration settings, they will probably change accordingly in the new kernel. Wait and see.

Gah, I know I sound like a curmudgeonly old geezer saying something like this, but I feel like these days they change too much too quickly in the newer mainline kernels. I preferred the old odd/even system prior to 2.6.

Hi,

Is 'backported' the correct term here?;)

Anyway, I can't recall any situation when the code was removed from older version at the time when it was removed in the newer version. Basically it's not possible because of compatibility.

Not that fast;) I'd say there are more additions than changes due to the hardware popping up all the time. But you are at least partially right, there are changes being introduced and from time to time the drivers (and other code) have to be updated. That might be a pain, for example if you learn based on LDD3, be prepared for your code to fail to compile.

--
Best regards,
Andrzej Telszewski