Re: XFS hole punch races

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

 



On Tue, 2016-03-22 at 16:57 +0100, Jan Kara wrote:

Hi,

similarly to ext4 also XFS had races between hole punching and page faults
which could result in data corruption. The fixes were merged in 4.1-rc1 but
it might make sense to backport them to older stable releases given the
nature of the issue.

Relevant fixes are:

de0e8c20ba3a65b0f15040aabbefdc1999876e6b
075a924d45cc69c75a35f20b4912b85aa98b180a
e8e9ad42c1f1e1bfbe0e8c32c8cac02e9ebfb7ef
0f9160b444e4de33b65dfcd3b901358a3129461a
723cac48473358939759885a18e8df113ea96138
ec56b1f1fdc69599963574ce94cc5693d535dd64


You missed the first in that sequence:

653c60b633a9 xfs: introduce mmap/truncate lock

For 3.16, I've queued up all those fixes with one further prerequisite:

812176832169 xfs: fix swapext ilock deadlock

For 3.2, I've queued up all but 723cac484733, with these additional
prerequisites:

f38996f57687 xfs: reduce ilock hold times in xfs_setattr_size
bc4010ecb8f4 xfs: use iolock on XFS_IOC_ALLOCSP calls
76ca4c238cf5 xfs: always take the iolock around xfs_setattr_size
5f8aca8b43f4 xfs: always hold the iolock when calling xfs_change_file_space
I realise I'll need to check for regressions with xfstests.

Ben.

-- 
Ben Hutchings
The most exhausting thing in life is being insincere. - Anne Morrow
Lindberg

Attachment: signature.asc
Description: This is a digitally signed message part


[Index of Archives]     [Linux Kernel]     [Kernel Development Newbies]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite Hiking]     [Linux Kernel]     [Linux SCSI]