BSD Commit Log Search

   Put the bowels of the various VOP_PRINT routines under
   defined(DEBUG) || defined(DIAGNOSTIC) || defined(VFSLCKDEBUG)
   as they won't be used otherwise.

   Shaves a few bytes off installation kernels.

   ok kn@ semarie@

   vfs: VOP_REMOVE: move vnode unlocking and ref dropping to FS-indep part

   while here, ensure all vop_remove field are set, and always call the function.

   the change is very conservative: it only adds vnode ref drop/unlock where it was
   absent because it should be unreachable (and if it wasn't, it should fix
   things).

   ok miod@

   Move the "no (hard) linking directories" and "no cross-mount links"
   checks from all the filesystems that support hardlinks at all into
   the VFS layer.  Simplify, EPERM description in link(2).

   ok miod@ mpi@

   Remove the remnants of the leftover selinfo from vnode(9) layer. Just
   mechanical 'selinfo' to 'klist' replacement in 'vnode' structure because
   knote(9) API is already used.

   <sys/selinfo.h> headers added where is was required.

   ok bluhm

   Remove unused VOP_POLL().

   OK mpi@

   Clarify usage of __EV_POLL and __EV_SELECT

   Make __EV_POLL specific to kqueue-based poll(2), to remove overlap
   with __EV_SELECT that only select(2) uses.

   OK millert@ mpi@

   tiny little whitespace fixes

   Prohibit renames of tmpfs mount-points to fix a panic.

   From gerhard@

   Call uvm_vnp_uncache() in tmpfs_write().  We currently only call
   uvm_vnp_uncache() in tmpfs_write() when a file grows in size.  This
   is not enough.  We need to invalidate UVM's cache of the vnode every
   time the contents of the vnode are modified.  Failure to do so might
   lead to inconsistencies between read/mmap consumers.

   From Pedro Martelletto

   Fix tmpfs_lookup locking for ".." == ".".  unveil_find_cover() calls
   VFS_LOOKUP(dir, &parent) in a loop and looks up the parent directory
   ".." repeatedly. VFS_LOOKUP is expected to unlock 'dir' and return
   'parent' locked.

   So tmpfs_lookup() is called for ISDOTDOT and:
        - runs with dvp = dir, vpp = &parent
        - gets parent from tmpfs_vnode_get() and
        - re-locks dir with vn_lock(dvp)
   but skips the call to
        VOP_UNLOCK(dvp);
   on return because *vpp == dvp

   The reason for doing so is the lookup for ".".  In this case
   tmpfs_lookup() just increases the reference on dvp and copies the
   pointer:
        *vpp = dvp; vref(dvp);

   However, in our case we also have *vpp == dvp, but for a different

    [3 lines not shown]

   spelling

   Refactor klist insertion and removal

   Rename klist_{insert,remove}() to klist_{insert,remove}_locked().
   These functions assume that the caller has locked the klist. The current
   state of locking remains intact because the kernel lock is still used
   with all klists.

   Add new functions klist_insert() and klist_remove() that lock the klist
   internally. This allows some code simplification.

   OK mpi@

   Fix build of tmpfs

   The breakage was caused by the removal of #include <sys/systm.h>
   from <uvm/uvm_map.h>.

   OK deraadt@, mpi@, beck@

   tmpfs_reclaim() has to make sure the VFS cache has no more locks held
   for the vnode.

   ok beck@

   Rename poll-compatibility flag to better reflect what it is.

   While here prefix kernel-only EV flags with two underbars.

   Suggested by kettenis@, ok visa@

   Use a new EV_OLDAPI flag to match the behavior of poll(2) and select(2).

   Adapt FS kqfilters to always return true when the flag is set and bypass
   the polling mechanism of the NFS thread.

   While here implement a write filter for NFS.

   ok visa@

   Abstract the head of knote lists. This allows extending the lists,
   for example, with locking assertions.

   OK mpi@, anton@

   Replace field f_isfd with field f_flags in struct filterops to allow
   adding more filter properties without cluttering the struct.

   OK mpi@, anton@

   struct vops is not modified during runtime so use const which moves each
   into read-only data segment.
   OK deraadt@ tedu@

   Use C99 designated initializers with struct filterops. In addition,
   make the structs const so that the data are put in .rodata.

   OK mpi@, deraadt@, anton@, bluhm@

   Allow concurrent reads of the f_offset field of struct file by
   serializing both read/write operations using the existing file mutex.
   The vnode lock still grants exclusive write access to the offset; the
   mutex is only used to make the actual write atomic and prevent any
   concurrent reader from observing intermediate values.

   ok mpi@ visa@

   Revert anton@ changes about read/write unlocking
   https://marc.info/?l=openbsd-cvs&m=156277704122293&w=2

   ok anton@

   Make read/write of the f_offset field belonging to struct file MP-safe;
   as part of the effort to unlock the kernel. Instead of relying on the
   vnode lock, introduce a dedicated lock per file. Exclusive write access
   is granted using the new foffset_enter and foffset_leave API. A
   convenience function foffset_get is also available for threads that only
   need to read the current offset.

   The lock acquisition order in vn_write has been changed to match the one
   in vn_read in order to avoid a potential deadlock. This change also gets
   rid of a documented race in vn_read().

   Inspired by the FreeBSD implementation.

   With help and ok mpi@ visa@

   More "explicitely" -> "explicitly" in various comments.

   ok guenther@ tb@ deraadt@

   Make callers of VOP_CREATE(9) and VOP_MKNOD(9) responsible for
   unlocking the directory vnode.

   OK mpi@, helg@

   Call vput(dvp) in vnode operation functions instead of calling it in
   the file allocation routine. This allows stepwise changing of the vnode
   locking discipline.

   OK mpi@

   Fix build without DIAGNOSTIC, ok mikeb@

   Remove proc from the parameters of vn_lock(). The parameter is
   unnecessary because curproc always does the locking.

   OK mpi@

   Clean up the parameters of VOP_LOCK() and VOP_UNLOCK(). It is always
   curproc that does the locking or unlocking, so the proc parameter
   is pointless and can be dropped.

   OK mpi@, deraadt@

   Remove the lockmgr() API. It is only used by filesystems, where it is a
   trivial change to use rrw locks instead. All it needs is LK_* defines
   for the RW_* flags.

   tested by naddy and sthen on package building infrastructure
   input and ok jmc mpi tedu