Re: spin gsync patch

[Samuel Thibault]

Hello,

Svante Signell, on Thu 03 Nov 2016 09:47:26 +0100, wrote:
> Sorry to bring bad news. The filesystem corruption is back, but not as bad as
> before :( And when disks are checked the date is again set to: Tue Apr 13
> 23:07:40 2032. Going back to the version without the tg-libpthread-gsync-
> spin.diff patch.

Are you really sure the corruption is due to the patch? Also, we have
seen such kind of odd date jump in the past already.

I have updated the patch again to even more be cautious, this time
there is no reason why tg-libpthread-gsync-spin.diff should change
the existing ext2fs behavior at all: the only function that existing
ext2fs calls is _pthread_spin_lock, whose behavior is just to call
__spin_try_lock in a loop, and __spin_try_lock is not changed by this
patch.

Samuel