diff options
| author | Thomas Gleixner <[email protected]> | 2018-09-15 20:50:42 +0200 |
|---|---|---|
| committer | Thomas Gleixner <[email protected]> | 2018-09-15 20:50:42 +0200 |
| commit | 9ac669fc01dbfef707ecaa6b618c0d03294cca16 (patch) | |
| tree | 3207995766ab5664c60026daae5da268806a3262 /net/unix/af_unix.c | |
| parent | fbfa9260085b5b578a049a90135e5c51928c5f7f (diff) | |
| parent | 67314ec7b0250290cc85eaa7a2f88a8ddb9e8547 (diff) | |
Merge tag 'y2038' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/playground into timers/core
Pull more y2038 work from Arnd Bergman:
y2038: convert more syscalls
Here is another set of system call changes to prepare the change over to
64-bit time_t. As before, the strategy is to change system calls that
take a 'struct timespec' argument over to 'struct __kernel_timespec',
which for now is defined to be the same but will get redefined to use a
64-bit time_t argument once we are ready to modify the system call tables.
The major change from previous patches is that the plan is no longer
to directly use the 'compat' system calls for providing compatibility
with the existing 32-bit time_t based entry points. Instead, we rename
the compat code to something that makes more sense on 32-bit architectures,
e.g. compat_timespec becomes old_timespec32.
With the renamed types in place, change over the 'stat' and 'utimes'
families of system calls, sched_rr_get_interval, recvmmsg and
rt_sigtimedwait. Another series for poll, select and io_pgetevents is
currently being tested.
Diffstat (limited to 'net/unix/af_unix.c')
| -rw-r--r-- | net/unix/af_unix.c | 11 |
1 files changed, 8 insertions, 3 deletions
diff --git a/net/unix/af_unix.c b/net/unix/af_unix.c index e5473c03d667..d1edfa3cad61 100644 --- a/net/unix/af_unix.c +++ b/net/unix/af_unix.c @@ -430,7 +430,12 @@ static int unix_dgram_peer_wake_me(struct sock *sk, struct sock *other) connected = unix_dgram_peer_wake_connect(sk, other); - if (unix_recvq_full(other)) + /* If other is SOCK_DEAD, we want to make sure we signal + * POLLOUT, such that a subsequent write() can get a + * -ECONNREFUSED. Otherwise, if we haven't queued any skbs + * to other and its full, we will hang waiting for POLLOUT. + */ + if (unix_recvq_full(other) && !sock_flag(other, SOCK_DEAD)) return 1; if (connected) @@ -2635,7 +2640,7 @@ static __poll_t unix_poll(struct file *file, struct socket *sock, poll_table *wa struct sock *sk = sock->sk; __poll_t mask; - sock_poll_wait(file, sk_sleep(sk), wait); + sock_poll_wait(file, wait); mask = 0; /* exceptional events? */ @@ -2672,7 +2677,7 @@ static __poll_t unix_dgram_poll(struct file *file, struct socket *sock, unsigned int writable; __poll_t mask; - sock_poll_wait(file, sk_sleep(sk), wait); + sock_poll_wait(file, wait); mask = 0; /* exceptional events? */ |