For file mounts, the directory vnode is not available from namei and this
prevents the use of vn_fullpath_hardlink. In this case, we can use the
vnode which was covered by the file mount with vn_fullpath.
This also disallows file mounts over files with link counts greater than
one to ensure a deterministic path to the mount point.
Reviewed by: mjg, kib
Tested by: pho
The main use-case for this is to support mounting config files and
secrets into OCI containers. My current workaround copies the files into
the container which is messy and risks secrets leaking into container
images if the cleanup fails.
This adds a VFCF flag to indicate whether the filesystem supports file
mounts and allows fspath to be either a directory or a file if the flag
is set.
Test Plan:
$ sudo mkdir -p /mnt
$ sudo touch /mnt/foo
$ sudo mount -t nullfs /COPYRIGHT /mnt/foo
Reviewed by: mjg, kib
Tested by: pho
instead of looking at SAVESTART
This is a step towards removing the flag.
Reviewed by: mckusick
Tested by: pho
Differential Revision: https://reviews.freebsd.org/D34468
This patch adds "allow.nfsd" to the jail code based on a
new kernel build option VNET_NFSD. This will not work
until future patches fix nmount(2) to allow mountd to
run in a vnet prison and the NFS server code is patched
so that global variables are in a vnet.
The jail(8) man page will be patched in a future commit.
Reviewed by: jamie
MFC after: 4 months
Differential Revision: https://reviews.freebsd.org/D37637
It appears that, prior to r158857 vfs_domount() checked
suser() when MNT_EXPORTED was specified.
r158857 appears to have broken this, since MNT_EXPORTED
was no longer set when mountd.c was converted to use nmount(2).
r164033 replaced the suser() check with
priv_check(td, PRIV_VFS_MOUNT_EXPORTED), which does the
same thing (ie. checks for effective uid == 0 assuming suses_enabled
is set).
This patch restores this check by setting MNT_EXPORTED when the
"export" mount option is specified to nmount().
I think this is reasonable since only mountd(8) should be setting
exports and I doubt any non-root mounted file system would
be setting its own exports.
Reviewed by: kib, markj
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D37718
pr_abort calls tcp_usr_abort which calls tcp_drop with ECONNABORTED.
After pr_abort returns, the so_error is then set to a more specific
error. However, a reader can observe and return the ECONNABORTED
error before so_error is set to the desired error value. This is
resulting in spurious test failures of recently added tests for
invalid conditions such as invalid headers.
To fix, refactor the code to abort a connection to call tcp_drop
directly with the desired error value. ktls_reset_send_tag already
calls tcp_drop directly when it aborts a connection due to an error.
Reviewed by: gallatin
Reported by: CI (jenkins), gallatin, olivier
Sponsored by: Chelsio Communications
Differential Revision: https://reviews.freebsd.org/D37692
Sockets have special handling for EPIPE on a write, that was spread out
into several places. Treating transient errors is also special - if
protocol is atomic, than we should ignore any changes to uio_resid, a
transient error means the write had completely failed (see d2b3a0ed31).
- Provide sousrsend() that expects a valid uio, and leave sosend() for
kernel consumers only. Do all special error handling right here.
- In dofilewrite() don't do special handling of error for DTYPE_SOCKET.
- For send(2), write(2) and aio_write(2) call into sousrsend() and remove
error handling for kern_sendit(), soo_write() and soaio_process_job().
PR: 265087
Reported by: rz-rpi03 at h-ka.de
Reviewed by: markj
Differential revision: https://reviews.freebsd.org/D35863
When VV_CROSSLOCK is present, the lock for the vnode at the current
stage of lookup must be held across the VFS_ROOT() call for the
filesystem mounted at the vnode. Since VV_CROSSLOCK implies that
the root vnode reuses the already-held lock, the possibility for
recursion should be made clear in the flags passed to VFS_ROOT().
For cases in which the lock is held exclusive, this means passing
LK_CANRECURSE. For cases in which the lock is held shared, it
means clearing LK_NODDLKTREAT to allow VFS_ROOT() to potentially
recurse on the shared lock even in the presence of an exclusive
waiter.
That the existing code works for unionfs is due to a coincidence
of the current unionfs implementation.
Reviewed by: kib
Tested by: pho
Differential Revision: https://reviews.freebsd.org/D37458
Return the value as stat(2) st_blocks.
Suggested and reviewed by: markj (previous version)
Tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D37097
This makes tmpfs size accounting correct for the sparce files. Also
correct report st_blocks/va_bytes. Previously the reported value did not
accounted for the swapped out pages.
PR: 223015
Reviewed by: markj
Tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D37097
This allows the use of chroot and/or jail environments which depend on
interpreters registed with imgact_binmisc to use emulator binaries from
the host to emulate programs inside the chroot.
Reviewed by: imp
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D37432
Early versions of this code had a free, but this one doesn't need
it. Remove the forgotten free(vv); from earlier versions.
Fixes: ed56dcfc6b
Noticed by: Michael Butler
Sponsored by: Netflix
Copy the arg that sets a variable to maximize the reuse of this
routine. There are places we call it from that are const char * and it
might not be safe to cast that away.
Sponsored by: Netflix
This reverts commit 68c3f03021. There are
some weird crashes when KVMs switch caused by this, so revert this
commit until they are sorted out.
Reported by: cy@
Sponsored by: Netflix
In the rare case that we succeed in probing, but fail to attach, flip
the default to be to disable the
device. hw.bus.disable_failed_devices=false is no required to restore
the old behavior. The old behavior dates form a time when dynamic
control of devices wasn't yet present (devctl didn't exist). Now that
one can retry probe/attach the device with devctl, the default doesn't
make sense: The more desirable behaivor is to have stable device numbers
when one has several instances of the same device in a system (common
for NICs or HBAs).
Reviewed by: jhb (verbal)
Sponsored by: Netflix
Normally, when a device fails to attach, we tear down the newbus state
for that device so that future driver loads can try again (maybe with a
different driver, or maybe with a re-loaded and fixed kld).
Sometimes, however, it is desirable to have the device fail
permanantly. We do this by calling device_disable() on a failed
attached, as well as keeping the device in DS_ATTACHING forever. This
prevents retries on that device. This is enabled via
hw.bus.disable_failed_devices=1 in either a hint via the loader, or at
runtime with a sysctl setting. Setting from 1 -> 0 at runtime will not
affect previously disabled devices, however: they remain disabled.
They can be re-enabled manually with devctl enable, however.
Sponsored by: Netflix
Reviewed by: gallatin, hselasky, jhb
Differential Revision: https://reviews.freebsd.org/D37517
One year ago, I deprecated 'kern' in favor of 'kernel' for the system
name for some power events. I'm about to remove it from the kernel, but
realized there's been no warning generated for users. Preserve POLA by
converting on the fly here and issuing a warning for 14.x, and an fatal
error after we branch 15. Make compiling it an error on 16 to remove
the gross hack after we branch.
Sponsored by: Netflix
Reviewed by: bapt
Differential Revision: https://reviews.freebsd.org/D37584
The new "kernel" system name is the one that's documented and has
been generated for a year now. Remove the old one now that 14.0
is getting close.
Sponsored by: Netflix
Reviewed by: bapt
Differential Revision: https://reviews.freebsd.org/D37582
Since ZFS reports _PC_MIN_HOLE_SIZE as 512 (although it
appears that an unwritten region must be at least f_iosize
to remain unallocated), vn_generic_copy_file_range()
uses 4096 for the copy blksize for ZFS, reulting in slow copies.
For most other file systems, _PC_MIN_HOLE_SIZE and f_iosize
are the same value, so this patch modifies the code to
use f_iosize for most cases. It also documents in comments
why the blksize is being set a certain way, so that the code
does not appear to be doing "magic math".
Reported by: allanjude
Reviewed by: allanjude, asomers
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D37076
This is mainly intended to provide a fallback for TOE TLS which may
need to use software decryption for an initial record at the start
of a connection.
Reviewed by: markj
Sponsored by: Chelsio Communications
Differential Revision: https://reviews.freebsd.org/D37370
The option is not even recognized and with that patched it does not
compile. Even if it did work, it would be prohibitively expensive to
use.
Interested parties can use pmcstat or dtrace instead.
ktls_cleanup() does not free ktls session objects, it merely
cleans (and frees) members of the object.
Change callers to use ktls_free() instead.
Reviewed by: gallatin, markj
Sponsored by: Chelsio Communications
Differential Revision: https://reviews.freebsd.org/D37352
The break-before-make requirement poses a problem when promoting or
demoting mappings containing thread structures: a CPU may raise a
translation fault while accessing curthread, and data_abort() accesses
the thread again before pmap_fault() can translate the address and
return.
Normally this isn't a problem because we have a hack to ensure that
slabs used by the thread zone are always accessed via the direct map,
where promotions and demotions are rare. However, this hack doesn't
work properly with UMA_MD_SMALL_ALLOC disabled, as is the case with
KASAN configured (since our KASAN implementation does not shadow the
direct map and so tries to force the use of the kernel map wherever
possible).
Fix the problem by modifying data_abort() to handle translation faults
in the kernel map without dereferencing "td", i.e., curthread, and
without enabling interrupts. pmap_klookup() has special handling for
translation faults which makes it safe to call in this context. Then,
revert the aforementioned hack.
Reviewed by: kevans, alc, kib, andrew
MFC after: 1 month
Sponsored by: Juniper Networks, Inc.
Sponsored by: Klara, Inc.
Differential Revision: https://reviews.freebsd.org/D37231
First, an sbuf_new() in device_get_path() shadows the sb
passed in by dev_wired_cache_add(), leaving its sb in an
unfinished state, leading to a failed KASSERT(). Fixing this
is as simple as removing the sbuf_new() from device_get_path()
Second, we cannot simply take a pointer to the sbuf memory and
store it in the device location cache, because that sbuf
is freed immediately after we add data to the cache, leading
to a use-after-free and eventually a double-free. Fixing this
requires allocating memory for the path.
After a discussion with jhb, we decided that one malloc was
better than two in dev_wired_cache_add, which is why it changed
so much.
Reviewed by: jhb
Sponsored by: Netflix
MFC after: 14 days
This commit brings back the driver from FreeBSD commit
f187d6dfbf plus subsequent fixes from
upstream.
Relative to upstream this commit includes a few other small fixes such
as additional INET and INET6 #ifdef's, #include cleanups, and updates
for recent API changes in main.
Reviewed by: pauamma, gbe, kevans, emaste
Obtained from: git@git.zx2c4.com:wireguard-freebsd @ 3cc22b2
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D36909
Refactor the symlink and mountpoint traversal logic to avoid
repeatedly checking the vnode type; a symlink cannot be a mountpoint
and vice versa. Avoid repeatedly checking cn_flags for NOCROSSMOUNT
and simplify the check which determines whether the vnode is a
mountpoint.
Suggested by: mjg
Reviewed by: kib
Tested by: pho
Differential Revision: https://reviews.freebsd.org/D35054
These are of limited use since the crossmp vnode locking ops have not
actually used a lock since commit
a2d3554542. We in fact require that
these operations are always issued with LK_SHARED. Additionally,
these directives can produce a false positive in certain VV_CROSSLOCK
cases which require upgrading of the covered vnode lock from shared
to exclusive.
While here, replace the runtime check of LK_SHARED with a KASSERT and
expand the check to include LK_NOWAIT, which all callers pass.
Reviewed by: kib
Tested by: pho
Differential Revision: https://reviews.freebsd.org/D35054
When a lookup operation crosses into a new mountpoint, the mountpoint
must first be busied before the root vnode can be locked. When a
filesystem is unmounted, the vnode covered by the mountpoint must
first be locked, and then the busy count for the mountpoint drained.
Ordinarily, these two operations work fine if executed concurrently,
but with a stacked filesystem the root vnode may in fact use the
same lock as the covered vnode. By design, this will always be
the case for unionfs (with either the upper or lower root vnode
depending on mount options), and can also be the case for nullfs
if the target and mount point are the same (which admittedly is
very unlikely in practice).
In this case, we have LOR. The lookup path holds the mountpoint
busy while waiting on what is effectively the covered vnode lock,
while a concurrent unmount holds the covered vnode lock and waits
for the mountpoint's busy count to drain.
Attempt to resolve this LOR by allowing the stacked filesystem
to specify a new flag, VV_CROSSLOCK, on a covered vnode as necessary.
Upon observing this flag, the vfs_lookup() will leave the covered
vnode lock held while crossing into the mountpoint. Employ this flag
for unionfs with the caveat that it can't be used for '-o below' mounts
until other unionfs locking issues are resolved.
Reported by: pho
Tested by: pho
Reviewed by: kib
Differential Revision: https://reviews.freebsd.org/D35054
In order to safely reuse excluded memory when it's reserved for special
purpose, we need to test whether or not the memory has been reserved
early in boot. physmem_excluded will return true when the entire range
is excluded, false otherwise.
Sponsored by: Netflix