This is the core of the case-insensitive support - supporting and enforcing UTF-8 (Unicode) filenames. All filename and user-level extended attribute names are checked for UTF-8 compliance and the ha
I objected to the part of the patch I've quoted (and the bitsrelated to it), not all of it. That how we do reviews in kernel land, not sure how samba handles it if you have a binary object/don't obje
It very much depends on the usage case. We have many users who have large numbers of files per directory, and not having to search these in userspace when we get a stat cache miss is helpful. Just ru
oops, look like the quote actually got deleted accidentally. sorry I'ltake that comment back. The part I object to are the various calls to xfs_unicode_validate in the namespace operations.
You can bet that one person's "invalid unicode sequence" will have been used as a valid filename on someone else's filesystem even using unicode. Meanings of "invalid unicode sequence" vary (a lot) d
I'd like to see the numbers... Simo tested an earlier version of this patch, and it was not faster.... Jeremy, what would be a representative test setup to use? Thanks, -Eric
IIRC, the last version of this patch series didn't help on the performance scale. Any updates as to whether that got fixed up? Josef 'Jeff' Sipek. -- In personal conversations with technical people,
CI filenames can work perfectly fine without adding validation of file names by treating non-conformant bytestreams as not having lower/upper case variants.
Eric Sandeen wrote: Jeremy Allison wrote: On Thu, Apr 03, 2008 at 01:14:50PM -0400, Christoph Hellwig wrote: Validating file names is not the filesystem job. In fact it's utterly stupid, a unix filen
tch starts using struct xfs_name more for the xattr code and is another step for using xfs_name in xfs_da_args. Also, the cred parameter is removed from xfs_attr_get and xfs_attr_fetch. The ne
er branch should now be fully up-to-date. The three mods listed below were in the merge. Lachlan McIlroy wrote: I pushed all available changes, at the time I did the merge, into the master br
mplements the code to store the actual filename found during a lookup in the dentry cache and to avoid multiple entries in the dcache pointing to the same inode. It also introduces a new type,
ted to the part of the patch I've quoted (and the bitsrelated to it), not all of it. That how we do reviews in kernel land, not sure how samba handles it if you have a binary object/don't obje
n IRIX remnant. On 64 bit Linux systems, xfs_db (and hence xfs_check) will be built as 64 bit binaries. Use xfs_repair -n, xfs_check does not scale (and will hopefully be replaced by xfs_repai