| 1 |
ATTACH TODO |
|---|
| 2 |
|
|---|
| 3 |
* CHECK FOR ANY CONCEIVABLE ERROR |
|---|
| 4 |
|
|---|
| 5 |
fist_ioctl +a /mnt/mini_fo abc /n/fist/mini_fo/zadok |
|---|
| 6 |
oops if /n/fist/mini_fo/zadok doesn't exist |
|---|
| 7 |
CPW: Fixed, hidden_dentry not correctly validated. |
|---|
| 8 |
|
|---|
| 9 |
ensure that only file owners are allowed to attach to their own directories |
|---|
| 10 |
what about the privs of mount point? |
|---|
| 11 |
CPW: Should it be owners or people with READ/EXEC perms? |
|---|
| 12 |
Check that MAY_EXEC and MAY_READ are permitted on to. |
|---|
| 13 |
|
|---|
| 14 |
|
|---|
| 15 |
ensure that you can only attach to directories. |
|---|
| 16 |
CPW: S_ISDIR added. |
|---|
| 17 |
|
|---|
| 18 |
* works |
|---|
| 19 |
|
|---|
| 20 |
- stat /mnt/mini_fo/nonexistent IS OK |
|---|
| 21 |
|
|---|
| 22 |
- stat /mnt/mini_fo/abd/nonexistent IS OK |
|---|
| 23 |
|
|---|
| 24 |
* bugs |
|---|
| 25 |
|
|---|
| 26 |
stat /mnt/mini_fo/abd/date.txt (exists) |
|---|
| 27 |
oops1 in mini_fo_interpose. |
|---|
| 28 |
fixed: had to do w/ using static inode numbers. |
|---|
| 29 |
|
|---|
| 30 |
Problem: attach node and hidden one may have different names! |
|---|
| 31 |
is that a problem in our expected code? |
|---|
| 32 |
check if |
|---|
| 33 |
fist_ioctl +a /mnt/mini_fo zadok /n/fist/mini_fo/zadok |
|---|
| 34 |
makes a different wrt oops1 |
|---|
| 35 |
Nope: we still get an oops |
|---|
| 36 |
|
|---|
| 37 |
- busy inodes after umount (kernel message) |
|---|
| 38 |
probably b/c we haven't implemented detach yet |
|---|
| 39 |
|
|---|
| 40 |
- at some point I got an oops from "ls /mnt/mini_fo/abc" |
|---|
| 41 |
can't reproduce it. |
|---|
| 42 |
I think it was in file ops (release) |
|---|
| 43 |
|
|---|
| 44 |
- CPW: The attached name changed from "compile2" to "compile2ZZZZ" |
|---|
| 45 |
Then it disappeared while I was doing 2 concurrent gcc compiles. |
|---|
