summaryrefslogtreecommitdiffstats
path: root/sys/fs/cd9660/TODO.hibler
diff options
context:
space:
mode:
Diffstat (limited to 'sys/fs/cd9660/TODO.hibler')
-rw-r--r--sys/fs/cd9660/TODO.hibler22
1 files changed, 22 insertions, 0 deletions
diff --git a/sys/fs/cd9660/TODO.hibler b/sys/fs/cd9660/TODO.hibler
new file mode 100644
index 0000000..3501aa2
--- /dev/null
+++ b/sys/fs/cd9660/TODO.hibler
@@ -0,0 +1,22 @@
+1. Investiate making ISOFS another UFS shared filesystem (ala FFS/MFS/LFS).
+ Since it was modelled after the inode code, we might be able to merge
+ them back. It looks like a seperate (but very similar) lookup routine
+ will be needed due to the associated file stuff.
+
+2. Make filesystem exportable. This comes for free if stacked with UFS.
+ Otherwise, the ufs_export routines need to be elevated to vfs_* routines.
+ [ DONE - hibler ]
+
+3. If it can't be merged with UFS, at least get them in sync. For example,
+ it could use the same style hashing routines as in ufs/ufs_ihash.c
+
+4. It would be nice to be able to use the vfs_cluster code.
+ Unfortunately, if the logical block size is smaller than the page size,
+ it won't work. Also, if throughtput is relatively constant for any
+ block size (as it is for the HP drive--150kbs) then clustering may not
+ buy much (or may even hurt when vfs_cluster comes up with a large sync
+ cluster).
+
+5. Seems like there should be a "notrans" or some such mount option to show
+ filenames as they really are without lower-casing, stripping of version
+ numbers, etc. Does this make sense?
OpenPOWER on IntegriCloud