linux.git/fs/io_uring.c, branch v5.10

io_uring: fix file leak on error path of io ctx creation

2020-12-08T15:54:26+00:00

Put file as part of error handling when setting up io ctx to fix
memory leaks like the following one.

   BUG: memory leak
   unreferenced object 0xffff888101ea2200 (size 256):
     comm "syz-executor355", pid 8470, jiffies 4294953658 (age 32.400s)
     hex dump (first 32 bytes):
       00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
       20 59 03 01 81 88 ff ff 80 87 a8 10 81 88 ff ff   Y..............
     backtrace:
       [<000000002e0a7c5f>] kmem_cache_zalloc include/linux/slab.h:654 [inline]
       [<000000002e0a7c5f>] __alloc_file+0x1f/0x130 fs/file_table.c:101
       [<000000001a55b73a>] alloc_empty_file+0x69/0x120 fs/file_table.c:151
       [<00000000fb22349e>] alloc_file+0x33/0x1b0 fs/file_table.c:193
       [<000000006e1465bb>] alloc_file_pseudo+0xb2/0x140 fs/file_table.c:233
       [<000000007118092a>] anon_inode_getfile fs/anon_inodes.c:91 [inline]
       [<000000007118092a>] anon_inode_getfile+0xaa/0x120 fs/anon_inodes.c:74
       [<000000002ae99012>] io_uring_get_fd fs/io_uring.c:9198 [inline]
       [<000000002ae99012>] io_uring_create fs/io_uring.c:9377 [inline]
       [<000000002ae99012>] io_uring_setup+0x1125/0x1630 fs/io_uring.c:9411
       [<000000008280baad>] do_syscall_64+0x2d/0x70 arch/x86/entry/common.c:46
       [<00000000685d8cf0>] entry_SYSCALL_64_after_hwframe+0x44/0xa9

Reported-by: syzbot+71c4697e27c99fddcf17@syzkaller.appspotmail.com
Fixes: 0f2122045b94 ("io_uring: don't rely on weak ->files references")
Cc: Pavel Begunkov 
Signed-off-by: Hillf Danton 
Signed-off-by: Jens Axboe

io_uring: fix mis-seting personality's creds

2020-12-07T15:43:44+00:00

After io_identity_cow() copies an work.identity it wants to copy creds
to the new just allocated id, not the old one. Otherwise it's
akin to req->work.identity->creds = req->work.identity->creds.

Signed-off-by: Pavel Begunkov 
Signed-off-by: Jens Axboe

io_uring: fix recvmsg setup with compat buf-select

2020-11-30T18:12:03+00:00

__io_compat_recvmsg_copy_hdr() with REQ_F_BUFFER_SELECT reads out iov
len but never assigns it to iov/fast_iov, leaving sr->len with garbage.
Hopefully, following io_buffer_select() truncates it to the selected
buffer size, but the value is still may be under what was specified.

Cc:  # 5.7
Signed-off-by: Pavel Begunkov 
Signed-off-by: Jens Axboe

io_uring: fix files grab/cancel race

2020-11-26T15:50:21+00:00

When one task is in io_uring_cancel_files() and another is doing
io_prep_async_work() a race may happen. That's because after accounting
a request inflight in first call to io_grab_identity() it still may fail
and go to io_identity_cow(), which migh briefly keep dangling
work.identity and not only.

Grab files last, so io_prep_async_work() won't fail if it did get into
->inflight_list.

note: the bug shouldn't exist after making io_uring_cancel_files() not
poking into other tasks' requests.

Signed-off-by: Pavel Begunkov 
Signed-off-by: Jens Axboe

io_uring: fix ITER_BVEC check

2020-11-24T14:54:30+00:00

iov_iter::type is a bitmask that also keeps direction etc., so it
shouldn't be directly compared against ITER_*. Use proper helper.

Fixes: ff6165b2d7f6 ("io_uring: retain iov_iter state over io_read/io_write calls")
Reported-by: David Howells 
Signed-off-by: Pavel Begunkov 
Cc:  # 5.9
Signed-off-by: Jens Axboe

io_uring: fix shift-out-of-bounds when round up cq size

2020-11-24T14:54:30+00:00

Abaci Fuzz reported a shift-out-of-bounds BUG in io_uring_create():

[ 59.598207] UBSAN: shift-out-of-bounds in ./include/linux/log2.h:57:13
[ 59.599665] shift exponent 64 is too large for 64-bit type 'long unsigned int'
[ 59.601230] CPU: 0 PID: 963 Comm: a.out Not tainted 5.10.0-rc4+ #3
[ 59.602502] Hardware name: Red Hat KVM, BIOS 0.5.1 01/01/2011
[ 59.603673] Call Trace:
[ 59.604286] dump_stack+0x107/0x163
[ 59.605237] ubsan_epilogue+0xb/0x5a
[ 59.606094] __ubsan_handle_shift_out_of_bounds.cold+0xb2/0x20e
[ 59.607335] ? lock_downgrade+0x6c0/0x6c0
[ 59.608182] ? rcu_read_lock_sched_held+0xaf/0xe0
[ 59.609166] io_uring_create.cold+0x99/0x149
[ 59.610114] io_uring_setup+0xd6/0x140
[ 59.610975] ? io_uring_create+0x2510/0x2510
[ 59.611945] ? lockdep_hardirqs_on_prepare+0x286/0x400
[ 59.613007] ? syscall_enter_from_user_mode+0x27/0x80
[ 59.614038] ? trace_hardirqs_on+0x5b/0x180
[ 59.615056] do_syscall_64+0x2d/0x40
[ 59.615940] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 59.617007] RIP: 0033:0x7f2bb8a0b239

This is caused by roundup_pow_of_two() if the input entries larger
enough, e.g. 2^32-1. For sq_entries, it will check first and we allow
at most IORING_MAX_ENTRIES, so it is okay. But for cq_entries, we do
round up first, that may overflow and truncate it to 0, which is not
the expected behavior. So check the cq size first and then do round up.

Fixes: 88ec3211e463 ("io_uring: round-up cq size before comparing with rounded sq size")
Reported-by: Abaci Fuzz 
Signed-off-by: Joseph Qi 
Reviewed-by: Stefano Garzarella 
Signed-off-by: Jens Axboe

Merge tag 'io_uring-5.10-2020-11-20' of git://git.kernel.dk/linux-block

2020-11-20T19:47:22+00:00

Pull io_uring fixes from Jens Axboe:
 "Mostly regression or stable fodder:

   - Disallow async path resolution of /proc/self

   - Tighten constraints for segmented async buffered reads

   - Fix double completion for a retry error case

   - Fix for fixed file life times (Pavel)"

* tag 'io_uring-5.10-2020-11-20' of git://git.kernel.dk/linux-block:
  io_uring: order refnode recycling
  io_uring: get an active ref_node from files_data
  io_uring: don't double complete failed reissue request
  mm: never attempt async page lock if we've transferred data already
  io_uring: handle -EOPNOTSUPP on path resolution
  proc: don't allow async path resolution of /proc/self components

io_uring: order refnode recycling

2020-11-18T15:02:10+00:00

Don't recycle a refnode until we're done with all requests of nodes
ejected before.

Signed-off-by: Pavel Begunkov 
Cc: stable@vger.kernel.org # v5.7+
Signed-off-by: Jens Axboe

io_uring: get an active ref_node from files_data

2020-11-18T15:02:10+00:00

An active ref_node always can be found in ctx->files_data, it's much
safer to get it this way instead of poking into files_data->ref_list.

Signed-off-by: Pavel Begunkov 
Cc: stable@vger.kernel.org # v5.7+
Signed-off-by: Jens Axboe

io_uring: don't double complete failed reissue request

2020-11-17T22:17:29+00:00

Zorro reports that an xfstest test case is failing, and it turns out that
for the reissue path we can potentially issue a double completion on the
request for the failure path. There's an issue around the retry as well,
but for now, at least just make sure that we handle the error path
correctly.

Cc: stable@vger.kernel.org
Fixes: b63534c41e20 ("io_uring: re-issue block requests that failed because of resources")
Reported-by: Zorro Lang 
Signed-off-by: Jens Axboe