Re: [PATCH v4 00/17] xfs: flush related error handling cleanups

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Mon, May 04, 2020 at 10:11:37AM -0400, Brian Foster wrote:
> Hi all,
> 
> I think everything has been reviewed to this point. Only minor changes
> noted below in this release. A git repo is available here[1].
> 
> The only outstanding feedback that I'm aware of is Dave's comment on
> patch 7 of v3 [2] regarding the shutdown assert check. I'm not aware of
> any means to get through xfs_wait_buftarg() with a dirty buffer that
> hasn't undergone the permanant error sequence and thus shut down the fs.

# echo 0 > /sys/fs/xfs/<dev>/fail_at_unmount

And now any error with a "retry forever" config (the default) will
be collected by xfs_buftarg_wait() without a preceeding shutdown as
xfs_buf_iodone_callback_error() will not treat it as a permanent
error during unmount. i.e. this doesn't trigger:

        /* At unmount we may treat errors differently */
        if ((mp->m_flags & XFS_MOUNT_UNMOUNTING) && mp->m_fail_unmount)
                goto permanent_error;

and so the error handling just marks it with a write error and lets
it go for a write retry in future. These are then collected in
xfs_buftarg_wait() as nothing is going to retry them once unmount
gets to this point...

Cheers,

Dave.
-- 
Dave Chinner
david@xxxxxxxxxxxxx



[Index of Archives]     [XFS Filesystem Development (older mail)]     [Linux Filesystem Development]     [Linux Audio Users]     [Yosemite Trails]     [Linux Kernel]     [Linux RAID]     [Linux SCSI]


  Powered by Linux