On Tue, Sep 3, 2019 at 9:39 PM Andreas Dilger <adilger@xxxxxxxxx> wrote:
>
> On Sep 3, 2019, at 12:15 PM, Qian Cai <cai@xxxxxx> wrote:
> >
> > On Tue, 2019-09-03 at 09:36 -0700, Deepa Dinamani wrote:
> >> We might also want to consider updating the file system the LTP is
> >> being run on here.
> >
> > It simply format (mkfs.ext4) a loop back device on ext4 with the kernel.
> >
> > CONFIG_EXT4_FS=m
> > # CONFIG_EXT4_USE_FOR_EXT2 is not set
> > # CONFIG_EXT4_FS_POSIX_ACL is not set
> > # CONFIG_EXT4_FS_SECURITY is not set
> > # CONFIG_EXT4_DEBUG is not set
> >
> > using e2fsprogs-1.44.6. Do you mean people now need to update the kernel to
> > enable additional config to avoid the spam of warnings now?
>
> Strange. The defaults for mkfs.ext4 _should_ default to use options that
> allow enough space for the extra timestamps.
>
> Can you please provide "dumpe2fs -h" output for your filesystem, and the
> formatting options that you used when creating this filesystem.
According to the man page,
"The default inode size is controlled by the mke2fs.conf(5)
file. In the
mke2fs.conf file shipped with e2fsprogs, the default inode size is 256
bytes for most file systems, except for small file systems
where the inode
size will be 128 bytes."
If this (small file systems) is the problem, then I think we need to
do two things:
1. Change the per-inode warning to not warn if the inode size for the
file system is less than 256. We already get a mount-time warning
in that case.
2. Change the mkfs.ext4 defaults to never pick a 128 byte inode unless
the user really wants this (maybe not even then).
Arnd
