linux.git/fs/afs, branch v5.11

rxrpc: Fix deadlock around release of dst cached on udp tunnel

2021-01-30T05:38:11+00:00

AF_RXRPC sockets use UDP ports in encap mode.  This causes socket and dst
from an incoming packet to get stolen and attached to the UDP socket from
whence it is leaked when that socket is closed.

When a network namespace is removed, the wait for dst records to be cleaned
up happens before the cleanup of the rxrpc and UDP socket, meaning that the
wait never finishes.

Fix this by moving the rxrpc (and, by dependence, the afs) private
per-network namespace registrations to the device group rather than subsys
group.  This allows cached rxrpc local endpoints to be cleared and their
UDP sockets closed before we try waiting for the dst records.

The symptom is that lines looking like the following:

	unregister_netdevice: waiting for lo to become free

get emitted at regular intervals after running something like the
referenced syzbot test.

Thanks to Vadim for tracking this down and work out the fix.

Reported-by: syzbot+df400f2f24a1677cd7e0@syzkaller.appspotmail.com
Reported-by: Vadim Fedorenko 
Fixes: 5271953cad31 ("rxrpc: Use the UDP encap_rcv hook")
Signed-off-by: David Howells 
Acked-by: Vadim Fedorenko 
Link: https://lore.kernel.org/r/161196443016.3868642.5577440140646403533.stgit@warthog.procyon.org.uk
Signed-off-by: Jakub Kicinski

afs: Fix directory entry size calculation

2021-01-04T12:25:19+00:00

The number of dirent records used by an AFS directory entry should be
calculated using the assumption that there is a 16-byte name field in the
first block, rather than a 20-byte name field (which is actually the case).
This miscalculation is historic and effectively standard, so we have to use
it.

The calculation we need to use is:

	1 + (((strlen(name) + 1) + 15) >> 5)

where we are adding one to the strlen() result to account for the NUL
termination.

Fix this by the following means:

 (1) Create an inline function to do the calculation for a given name
     length.

 (2) Use the function to calculate the number of records used for a dirent
     in afs_dir_iterate_block().

     Use this to move the over-end check out of the loop since it only
     needs to be done once.

     Further use this to only go through the loop for the 2nd+ records
     composing an entry.  The only test there now is for if the record is
     allocated - and we already checked the first block at the top of the
     outer loop.

 (3) Add a max name length check in afs_dir_iterate_block().

 (4) Make afs_edit_dir_add() and afs_edit_dir_remove() use the function
     from (1) to calculate the number of blocks rather than doing it
     incorrectly themselves.

Fixes: 63a4681ff39c ("afs: Locally edit directory data for mkdir/create/unlink/...")
Fixes: ^1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: David Howells 
Tested-by: Marc Dionne

afs: Work around strnlen() oops with CONFIG_FORTIFIED_SOURCE=y

2021-01-04T12:25:19+00:00

AFS has a structured layout in its directory contents (AFS dirs are
downloaded as files and parsed locally by the client for lookup/readdir).
The slots in the directory are defined by union afs_xdr_dirent. This,
however, only directly allows a name of a length that will fit into that
union. To support a longer name, the next 1-8 contiguous entries are
annexed to the first one and the name flows across these.

afs_dir_iterate_block() uses strnlen(), limited to the space to the end of
the page, to find out how long the name is. This worked fine until
6a39e62abbaf. With that commit, the compiler determines the size of the
array and asserts that the string fits inside that array. This is a
problem for AFS because we *expect* it to overflow one or more arrays.

A similar problem also occurs in afs_dir_scan_block() when a directory file
is being locally edited to avoid the need to redownload it. There strlen()
was being used safely because each page has the last byte set to 0 when the
file is downloaded and validated (in afs_dir_check_page()).

Fix this by changing the afs_xdr_dirent union name field to an
indeterminate-length array and dropping the overflow field.

(Note that whilst looking at this, I realised that the calculation of the
number of slots a dirent used is non-standard and not quite right, but I'll
address that in a separate patch.)

The issue can be triggered by something like:

touch /afs/example.com/thisisaveryveryverylongname

and it generates a report that looks like:

detected buffer overflow in strnlen
------------[ cut here ]------------
kernel BUG at lib/string.c:1149!
...
RIP: 0010:fortify_panic+0xf/0x11
...
Call Trace:
afs_dir_iterate_block+0x12b/0x35b
afs_dir_iterate+0x14e/0x1ce
afs_do_lookup+0x131/0x417
afs_lookup+0x24f/0x344
lookup_open.isra.0+0x1bb/0x27d
open_last_lookups+0x166/0x237
path_openat+0xe0/0x159
do_filp_open+0x48/0xa4
? kmem_cache_alloc+0xf5/0x16e
? __clear_close_on_exec+0x13/0x22
? _raw_spin_unlock+0xa/0xb
do_sys_openat2+0x72/0xde
do_sys_open+0x3b/0x58
do_syscall_64+0x2d/0x3a
entry_SYSCALL_64_after_hwframe+0x44/0xa9

Fixes: 6a39e62abbaf ("lib: string.h: detect intra-object overflow in fortified string functions")
Reported-by: Marc Dionne
Signed-off-by: David Howells
Tested-by: Marc Dionne
cc: Daniel Axtens

afs: Fix memory leak when mounting with multiple source parameters

2020-12-08T23:59:25+00:00

There's a memory leak in afs_parse_source() whereby multiple source=
parameters overwrite fc->source in the fs_context struct without freeing
the previously recorded source.

Fix this by only permitting a single source parameter and rejecting with
an error all subsequent ones.

This was caught by syzbot with the kernel memory leak detector, showing
something like the following trace:

  unreferenced object 0xffff888114375440 (size 32):
    comm "repro", pid 5168, jiffies 4294923723 (age 569.948s)
    backtrace:
      slab_post_alloc_hook+0x42/0x79
      __kmalloc_track_caller+0x125/0x16a
      kmemdup_nul+0x24/0x3c
      vfs_parse_fs_string+0x5a/0xa1
      generic_parse_monolithic+0x9d/0xc5
      do_new_mount+0x10d/0x15a
      do_mount+0x5f/0x8e
      __do_sys_mount+0xff/0x127
      do_syscall_64+0x2d/0x3a
      entry_SYSCALL_64_after_hwframe+0x44/0xa9

Fixes: 13fcc6837049 ("afs: Add fs_context support")
Reported-by: syzbot+86dc6632faaca40133ab@syzkaller.appspotmail.com
Signed-off-by: David Howells 
cc: Randy Dunlap 
Signed-off-by: Linus Torvalds

afs: Fix speculative status fetch going out of order wrt to modifications

2020-11-22T19:27:03+00:00

When doing a lookup in a directory, the afs filesystem uses a bulk
status fetch to speculatively retrieve the statuses of up to 48 other
vnodes found in the same directory and it will then either update extant
inodes or create new ones - effectively doing 'lookup ahead'.

To avoid the possibility of deadlocking itself, however, the filesystem
doesn't lock all of those inodes; rather just the directory inode is
locked (by the VFS).

When the operation completes, afs_inode_init_from_status() or
afs_apply_status() is called, depending on whether the inode already
exists, to commit the new status.

A case exists, however, where the speculative status fetch operation may
straddle a modification operation on one of those vnodes.  What can then
happen is that the speculative bulk status RPC retrieves the old status,
and whilst that is happening, the modification happens - which returns
an updated status, then the modification status is committed, then we
attempt to commit the speculative status.

This results in something like the following being seen in dmesg:

	kAFS: vnode modified {100058:861} 8->9 YFS.InlineBulkStatus

showing that for vnode 861 on volume 100058, we saw YFS.InlineBulkStatus
say that the vnode had data version 8 when we'd already recorded version
9 due to a local modification.  This was causing the cache to be
invalidated for that vnode when it shouldn't have been.  If it happens
on a data file, this might lead to local changes being lost.

Fix this by ignoring speculative status updates if the data version
doesn't match the expected value.

Note that it is possible to get a DV regression if a volume gets
restored from a backup - but we should get a callback break in such a
case that should trigger a recheck anyway.  It might be worth checking
the volume creation time in the volsync info and, if a change is
observed in that (as would happen on a restore), invalidate all caches
associated with the volume.

Fixes: 5cf9dd55a0ec ("afs: Prospectively look up extra files when doing a single lookup")
Signed-off-by: David Howells 
Signed-off-by: Linus Torvalds

afs: Fix afs_write_end() when called with copied == 0 [ver #3]

2020-11-14T19:51:18+00:00

When afs_write_end() is called with copied == 0, it tries to set the
dirty region, but there's no way to actually encode a 0-length region in
the encoding in page->private.

"0,0", for example, indicates a 1-byte region at offset 0.  The maths
miscalculates this and sets it incorrectly.

Fix it to just do nothing but unlock and put the page in this case.  We
don't actually need to mark the page dirty as nothing presumably
changed.

Fixes: 65dd2d6072d3 ("afs: Alter dirty range encoding in page->private")
Signed-off-by: David Howells 
Signed-off-by: Linus Torvalds

afs: Fix incorrect freeing of the ACL passed to the YFS ACL store op

2020-11-03T17:53:40+00:00

The cleanup for the yfs_store_opaque_acl2_operation calls the wrong
function to destroy the ACL content buffer.  It's an afs_acl struct, not
a yfs_acl struct - and the free function for latter may pass invalid
pointers to kfree().

Fix this by using the afs_acl_put() function.  The yfs_acl_put()
function is then no longer used and can be removed.

	general protection fault, probably for non-canonical address 0x7ebde00000000: 0000 [#1] SMP PTI
	...
	RIP: 0010:compound_head+0x0/0x11
	...
	Call Trace:
	 virt_to_cache+0x8/0x51
	 kfree+0x5d/0x79
	 yfs_free_opaque_acl+0x16/0x29
	 afs_put_operation+0x60/0x114
	 __vfs_setxattr+0x67/0x72
	 __vfs_setxattr_noperm+0x66/0xe9
	 vfs_setxattr+0x67/0xce
	 setxattr+0x14e/0x184
	 __do_sys_fsetxattr+0x66/0x8f
	 do_syscall_64+0x2d/0x3a
	 entry_SYSCALL_64_after_hwframe+0x44/0xa9

Fixes: e49c7b2f6de7 ("afs: Build an abstraction around an "operation" concept")
Signed-off-by: David Howells 
Signed-off-by: Linus Torvalds

afs: Fix warning due to unadvanced marshalling pointer

2020-11-03T17:53:40+00:00

When using the afs.yfs.acl xattr to change an AuriStor ACL, a warning
can be generated when the request is marshalled because the buffer
pointer isn't increased after adding the last element, thereby
triggering the check at the end if the ACL wasn't empty.  This just
causes something like the following warning, but doesn't stop the call
from happening successfully:

    kAFS: YFS.StoreOpaqueACL2: Request buffer underflow (36<108)

Fix this simply by increasing the count prior to the check.

Fixes: f5e4546347bc ("afs: Implement YFS ACL setting")
Signed-off-by: David Howells 
Signed-off-by: Linus Torvalds

afs: Fix dirty-region encoding on ppc32 with 64K pages

2020-10-29T13:53:04+00:00

The dirty region bounds stored in page->private on an afs page are 15 bits
on a 32-bit box and can, at most, represent a range of up to 32K within a
32K page with a resolution of 1 byte.  This is a problem for powerpc32 with
64K pages enabled.

Further, transparent huge pages may get up to 2M, which will be a problem
for the afs filesystem on all 32-bit arches in the future.

Fix this by decreasing the resolution.  For the moment, a 64K page will
have a resolution determined from PAGE_SIZE.  In the future, the page will
need to be passed in to the helper functions so that the page size can be
assessed and the resolution determined dynamically.

Note that this might not be the ideal way to handle this, since it may
allow some leakage of undirtied zero bytes to the server's copy in the case
of a 3rd-party conflict.  Fixing that would require a separately allocated
record and is a more complicated fix.

Fixes: 4343d00872e1 ("afs: Get rid of the afs_writeback record")
Reported-by: kernel test robot 
Signed-off-by: David Howells 
Reviewed-by: Matthew Wilcox (Oracle)

afs: Fix afs_invalidatepage to adjust the dirty region

2020-10-29T13:53:04+00:00

Fix afs_invalidatepage() to adjust the dirty region recorded in
page->private when truncating a page.  If the dirty region is entirely
removed, then the private data is cleared and the page dirty state is
cleared.

Without this, if the page is truncated and then expanded again by truncate,
zeros from the expanded, but no-longer dirty region may get written back to
the server if the page gets laundered due to a conflicting 3rd-party write.

It mustn't, however, shorten the dirty region of the page if that page is
still mmapped and has been marked dirty by afs_page_mkwrite(), so a flag is
stored in page->private to record this.

Fixes: 4343d00872e1 ("afs: Get rid of the afs_writeback record")
Signed-off-by: David Howells