Just from a very quick look, it seems like a regression introduced with
the vcache changes: the MP-safe flag is set too late and not inherited
by the root vnode.
kmem_alloc() with KM_SLEEP
kmem_zalloc() with KM_SLEEP
percpu_alloc()
pserialize_create()
psref_class_create()
all of these paths include an assertion that the allocation has not failed,
so callers should not assert that again.
VOP_RECLAIM naturally has exclusive access to the vnode, so having it
locked on entry is not strictly necessary -- but it means if there
are any final operations that must be done on the vnode, such as
ffs_update, requiring exclusive access to it, we can now kassert that
the vnode is locked in those operations.
We can't just have the caller release the last lock because some file
systems don't use genfs_lock, and require the vnode to remain valid
for VOP_UNLOCK to work, notably unionfs.
to read-only and vice versa:
- Add an internal flag IMNT_WANTRDONLY.
- Set either IMNT_WANTRDWR or IMNT_WANTRDONLY if going from or to read-only.
- After successfull call to VFS_MOUNT() set or clear MNT_RDONLY.
Adapt tmpfs and rumpfs to the new protocol. Other file systems will be
updated when they get the IMNT_CAN_RWTORO property.
Welcome to 7.99.64
tmpfs_mount instead of crashing.
Came up in PR 51436, where kmem issues caused internal allocations to
wrongly fail. However, that could happen for real sometime (e.g.
probably if you tried to mount a new tmpfs when the system was very
low on memory, or possibly for other reasons entirely) and crashing
isn't the ticket.
(This is not a fix for PR 51436)
- tmpfs_node_get: restore (decrement) the node count on the error path.
- tmpfs_bytes_max: save the value of uvmexp.freetarg (since it is
unlocked/racy).
- Use tmpfs node address as key.
- Remove tn_vlock, field tn_vnode now protected by vcache.
- Add a hold count to tmpfs node to prevent nodes from disappearing
while tmpfs_fhtovp() trys to vcache_get() them. Last holder
destroys reclaimed nodes.
- Remove the now unneeded parent unlock/lock for lookup of '..'.
find.
The filesystem ones all call genfs_eopnotsupp - right now I am only
implementing the plumbing and we can implement fallocate and/or
fdiscard for files later.
The device ones call spec_fallocate (which is also genfs_eopnotsupp)
and spec_fdiscard, which dispatches to the device-level op.
The fifo ones all call vn_fifo_bypass, which also ends up being
EOPNOTSUPP.
by uvmexp.freetarg, as discussed on tech-kern.
Main purpose is to make tmpfs usable (as far as possible) on small memory
machines.
This is a bit experimental, but we need to give it some real world exposure
to see how well it works.
the vnode operations vector for active vnodes is unsafe because it
is not known whether deadfs or the original file system will be
called.
- Pass down LK_RETRY to the lock operation (hint for deadfs only).
- Change deadfs lock operation to return ENOENT if LK_RETRY is unset.
- Change all other lock operations to check for dead vnode once
the vnode is locked and unlock and return ENOENT in this case.
With these changes in place vnode lock operations will never succeed
after vclean() has marked the vnode as VI_XLOCK and before vclean()
has changed the operations vector.
Adresses PR kern/37706 (Forced unmount of file systems is unsafe)
Discussed on tech-kern.
Welcome to 6.99.33
maxv