Re: [Y2038] [PATCH] generic/390: Add tests for inode timestamp policy

On Thursday, November 24, 2016 6:40:55 PM CET Eryu Guan wrote:

On Wed, Nov 23, 2016 at 04:52:02PM -0800, Deepa Dinamani wrote:

...

+_filesystem_timestamp_range() +{

• device=${1:-$TEST_DEV}
• case $FSTYP in
• ext4) #dumpe2fs

• if [ $(dumpe2fs -h $device 2>/dev/null | grep "Inode size:" | cut -d: -f2) -gt 128 ]; then
  

• 	echo "-2147483648 15032385535"
  

• else
  

• 	echo "-2147483648 2147483647"
  

• fi
  

Do ext3 and ext2 follow the same config as ext4?

Those two only support the second case with 128 byte inodes, but the same check should work on all three.

I have an overview of the limits on https://kernelnewbies.org/y2038/vfs, though I'd probably check all of them again, as some of them turned out to be wrong. In particular, identifying whether the on-disk timestamps are meant to be signed or unsigned can be a matter of interpretation and there may be a specification that disagrees with the implementation.

Arnd