mjw for the hint. Fixes bug #1516.
* Changed _{kern,user}_create_symlink() to no longer check the supplied
link string. BeOS seems to do that, but this is not standard
conforming. The previous implementation even used the path processed
by check_path(), which would potentially have appended a ".".
* Some style cleanup.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@22451 a95241bf-73f2-0310-859d-f6bbb57e9c96
* the page writer don't allow to block, while all other writers do. This fixes
bug #1509. The reason the page writer needs this is because it marks several
pages from different caches as busy.
* Fixed a warning about ASSERT being defined already in BFS, since
util/DoublyLinkedList.h now includes debug.h.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@22434 a95241bf-73f2-0310-859d-f6bbb57e9c96
vfs_select.h, respectively moved most of it into the new kernel
private header wait_for_objects.h.
* Added new experimental API functions wait_for_objects[_etc](). They
work pretty much like poll(), but also for semaphores, ports, and
threads.
* Removed the "ref" parameter from notify_select_events() and the
select_sync_pool functions as well as from fd_ops::fd_[de]select(). It
is no longer needed. The FS interface select() hook still has it,
though -- the VFS will always pass 0.
* de]select_fd() take a select_info* instead of a select_sync* + ref
pair, now. Added respective functions for semaphores, ports, and
threads.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@22416 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Reworked the select support:
- The io_context additionally stores a table of lists of select_infos,
which enables it to deselect events of a pending select() when
closing a FD. This prevents a race condition potentially causing a
write to stale memory.
- The opaque selectsync* passed to FSs is now actually a select_info*.
This was necessary, since the FDs deselect() hook (unlike the
select() hook) doesn't take a "ref" argument and deselecting a
single info (e.g. caused by a premature close()) was not possible.
The select() hook's "ref" argument has become superfluous.
- It should now be relatively easy to implement a poll_on_steroids()
that can also wait for objects other than FDs (e.g. semaphores,
ports, threads etc.).
* Set/reset the signal mask in common_select(). This makes pselect()
work as required.
* Reorganized vfs_resize_fd_table().
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@22391 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Removed the vm_cache/vm_store ref_count duality that besides being a bit ugly
also created the page dameon cache retrieval problem: now, only areas (and
cache consumers) retrieve a reference to the store (and therefore, the vnode).
The page daemon doesn't need to care about this at all anymore, and the pseudo
references of the vm_cache could be removed again.
* Rearranged deletion of vnodes such that its ID can be reused directly after
fs_remove_vnode() has been called.
* vm_page_allocate_page() no longer panics when it runs out of pages, but just
waits for new pages to become available using the new sFreeCondition condition
variable - to make sure this happens in an acceptable time frame, it'll
trigger a run of the low memory handlers.
* Implemented a page_thief() that steals inactive pages from caches and puts
them into the free queue. It runs as a low memory handler.
* The file cache now sets the usage count on the pages it inserts into the
cache (needs some rework though, cache_io() doesn't do it yet).
* Instead of panicking, the kernel will currently dead lock in low memory
situations, since BFS does a bit too much in bfs_release_vnode().
* Some minor cleanup.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@22315 a95241bf-73f2-0310-859d-f6bbb57e9c96
reference count to drop below 0 there.
* Added TODO describing a serious race condition between free_vnode()
and the page daemon.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@22209 a95241bf-73f2-0310-859d-f6bbb57e9c96
increment the root node vnode reference count. Otherwise it could race
with fs_unmount(). Fixes bug #1438.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@22203 a95241bf-73f2-0310-859d-f6bbb57e9c96
removed in several cases.
* vfs_get_vnode_cache() now always gives out a reference to the cache it
returns; if it needs to allocate a new one, the vnode owns one reference,
and the caller another.
* therefore, file_cache_create() now owns a reference to its vm_cache_ref, and
frees it in file_cache_delete().
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21533 a95241bf-73f2-0310-859d-f6bbb57e9c96
in the vnode hash; that's a bug in the FS that shouldn't be ignored that way.
* Minor cleanup.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@21262 a95241bf-73f2-0310-859d-f6bbb57e9c96
* read_dir() is supposed to return B_OK and and a count of 0 when
reaching the end of the directory. In case the node in question could
not be found, we were looping infinitely.
* free_dir_cookie() was not invoked.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20519 a95241bf-73f2-0310-859d-f6bbb57e9c96
short, if the FD table size wasn't a multiple of 8.
* vfs_resize_fd_table() didn't seem to know at all about the
close-on-exit bitmap. The pointer in the io_context would point to
free()d memory afterwards. This explains the sporadically closed
stdin/out/err descriptors in programs started from Tracker and
Deskbar.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20413 a95241bf-73f2-0310-859d-f6bbb57e9c96
add-ons accordingly and removed the syscall.
* Removed send_notification().
* Reimplemented notify_listener(). It used the unimplemented
send_notification(). Now it has a chance to work. Note that
notify_listener() is obsolete. I would already have removed it, if
there weren't lots of FS implementations still using it (Hint!).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20329 a95241bf-73f2-0310-859d-f6bbb57e9c96
vnode::covered_by fields. Together with sMountOpLock it allows write
access, either lock alone suffices for read access. Before
sMountOpLock had to be acquired for read (and write) access, which
meant that while mounting/unmounting a FS path resolution would have
to wait. In case of the UserlandFS this would even cause a deadlock
while mounting if the client tried to resolve the path of the device
to be mounted (e.g. by opening it).
* Added a clarifying comment about read access to the
fs_mount::covers_vnode/root_vnode field and removed locking in
resolve_volume_root_to_mount_point() which was not necessary for
explained reasons.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20308 a95241bf-73f2-0310-859d-f6bbb57e9c96
name (saves copying the name, if that has to be done anyway) and added a
wrapper version with the old interface.
* dir_vnode_to_path() was broken for file systems that didn't support
the get_vnode_name() hook. It resolved the mount point too early, so
that it was searching the mount point and not the FS root dir for the
node. It uses the get_vnode_name() function now (before resolving the
mount point).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20305 a95241bf-73f2-0310-859d-f6bbb57e9c96
will now grab a reference to the vnode as well if successful. This way, vfs_get_vnode_cache()
now actually works how it should: it will now always grab a reference to the cache and
its underlying vnode. This removes an extra reference to the vnode (and vm_cache) that
got ignored before and prevented volumes to be unmounted (or file caches to be removed).
Thanks to Korli for pointing this out.
* file_cache_create() is now aware of that extra vnode reference and releases it; unmounting
volumes is now working again as it should.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@18716 a95241bf-73f2-0310-859d-f6bbb57e9c96
vfs_get_vnode_cache() which did not acquire an extra reference to the
cache_ref when the cache had to be created.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@17693 a95241bf-73f2-0310-859d-f6bbb57e9c96
of the one supplied to fs_mount() (but not the /dev/disk/virtual/... entry, as that
wouldn't be that clear to the user).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@17335 a95241bf-73f2-0310-859d-f6bbb57e9c96
scan job in the calling thread).
* KDiskDeviceManager::InitialDeviceScan() now runs synchronously, so that
get_boot_partitions() doesn't need to do this ugly wait hack.
* KDiskDeviceManager::CreateFileDevice() can now run synchronously as well, which
fixes a deadlock in fs_mount() - note, mounting file devices still doesn't work,
though as Haiku's BFS doesn't allow this right now.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@17334 a95241bf-73f2-0310-859d-f6bbb57e9c96
dir_remove() did not normalize the path, and thus, could forward a "." as
name for the removed directory - which BFS didn't catch because it assumed
our VFS would work correctly...
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@17165 a95241bf-73f2-0310-859d-f6bbb57e9c96
vnode is not becoming unbusy (right now it even panics, but that can be removed
later on).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@17161 a95241bf-73f2-0310-859d-f6bbb57e9c96
* vfs_get_fs_node_from_path() now also work for absolute paths again (but
still for relative ones from the volume root) - it just tests if the
mount IDs fit, so it only returns successful if the path really is on
the desired mount.
* the Disk Device Manager publish functions now call devfs_publish_*()
correctly (by omitting the "/dev/" mount point).
* devfs_publish_partition() now accepts absolute device paths but relative
partition paths.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@17138 a95241bf-73f2-0310-859d-f6bbb57e9c96
(ie. it only worked for absolute paths, but it shouldn't work for those at
all).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@17127 a95241bf-73f2-0310-859d-f6bbb57e9c96
systems a chance to know if they have locked already.
This fixes a locking problem in BFS where one thread tried to acquire two read
locks (where someone else trying to acquire a write lock would have caused a
dead lock).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@17108 a95241bf-73f2-0310-859d-f6bbb57e9c96
changes made to a file instead of directly using vm_cache_write_modified() -
besides making the file system more independent from the file cache, this also
works around a possible dead lock (that is to be fixed in a later commit).
* fs_sync() no longer uses vnodes from the mount's vnode list directly to write
back the changes made to them, but gets them via ID instead - this makes sure
the vnode is in a valid state and fixes a race condition with free_vnode().
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@17059 a95241bf-73f2-0310-859d-f6bbb57e9c96
function, and added a vfs_disconnect_vnode() for other kernel components.
* devfs_unpublish_device() can now optionally make use of this call.
* Fixed the type check of devfs' unpublish_node().
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16907 a95241bf-73f2-0310-859d-f6bbb57e9c96
the root vnode was tried to be removed as well (which resulted in a crash).
Since playing with the root node reference count is a bad idea anyway and opens
a big race condition (with regards to the unused list), we no longer do that
now, until it's safe.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16665 a95241bf-73f2-0310-859d-f6bbb57e9c96
* fs_unmount() freed vnodes, but didn't remove them from the unused list if
needed.
* vfs_get_module_path() could put a vnode twice under several situations.
* vnode_path_to_vnode() now always puts the dir vnode, even if the provided
path is NULL. Documented the fact that it does eat the ref, too.
* Added a to-do item in vfs_get_vnode_cache() about a possible problem.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16662 a95241bf-73f2-0310-859d-f6bbb57e9c96
vfs_free_unused_vnodes()); the vnode mutex is now hold for much shorter times
only:
* Rewrote advisory_locking creation/maintenance to hold the vnode mutex only
for very short times.
* the vnode mutex is no longer held during file cache construction; instead,
the vnode is marked busy.
* Implemented an (incorrect for now) get_advisory_lock()
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16635 a95241bf-73f2-0310-859d-f6bbb57e9c96
of that list.
* Added a vfs_free_unused_vnodes() function that calls the low memory handler
directly.
* create_sem_etc() now calls the above function in case there are no semaphores
available anymore; this usually frees up to 2 semaphores per node (one from
the cache if there is a file cache attached, and eventually one from the
file system).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@16610 a95241bf-73f2-0310-859d-f6bbb57e9c96
