'noextents' no longer recognized? [Solved: Obsoleted]

[Cyker]

I just rebooted to go from 2.6.28 to 2.6.30 and found that my RAID array won't mount on boot anymore.

To cut a long story short, I tracked the problem down to the noextents option; Apparently it is no longer recognized as a valid mount option!

Obviously I don't want extents; I want to keep the option to go back to ext3 (In fact, the SOLE reason I'm using ext4 is for the journal checksum; I'm not using any of the other options because I feel they compromise integrity for speed!)

The removal of the option seems to be completely undocumented. I found a reference to the removal of the 'extents' option in 2.6.29, but nothing about 'noextents' or what the default behavior has been changed to.

I know previously, mounting with 'extents' or nothing would convert the filesystem to use extents which was why all the FAQs and HOWTOs were saying mount with 'noextents' if you want to stay ext3 compatible.

So, my query is, if I remove the noextents mount option, what will happen? Will it irreversibly switch my filesystem to ext4 or am I safe?
And, if it is safe now, what are the chances of the kernel people reverting behavior later so that I need to put 'noextents' back?

[Hu]

The mount option was removed in v2.6.28-5772-g83982b6. Based on a reading of the code, I think that you will be fine as long as the extents feature flag is not set in the superblock. However, I am not a filesystem developer and I have not tested this. If you are concerned, I suggest making an ext3 filesystem in a loopback file, mounting it with the newer kernel, and then checking whether any extents were used for large files.

[Cyker]

Been reading around and have come to a similar conclusion, although all my data comes from stuff mentioned in passing!

Anyway, tried it and it seems okay; The dreaded extents flag has not appeared \o/

I just need to make sure I am damned careful if I touch any old kernels now