Lines Matching defs:from
267 * when it falls to zero, the perdm structure is removed from the list and
433 * Remove `el' from the list using `chase' and `curr' pointers and return result
526 * Remove a queue from the syncq list
568 /* Hide the definition from the header file. */
852 * Init routine run from main at boot time.
937 /* Have to hold sd_lock to prevent siglist from changing */
1043 * pidlock to prevent the membership in the process group from changing
1296 * pair of queues from the syncq.
1313 /* freeq removes us from the outer perimeter if any */
1317 /* Prevent service procedures from being called */
1348 * Remove queue from qhead/qtail if it is enabled.
1349 * Only reset QENAB if the queue was removed from the runlist.
1352 * It is removed from the service list but QENAB is still set.
1355 * Thus we can not reset QENAB unless we actually removed it from the service
1379 * The removal of queues from the runlist is not atomic with the
1394 * Try to remove queues from qhead/qtail list.
1401 * Wait till the syncqs associated with the queue disappear from the
1408 * functions remove the queue from its syncq list, so sqthread will not
1435 * Put ioctl data from userland buffer `arg' into the mblk chain `bp'.
1821 * STPLEX prevents any threads from entering the stream from above. It
1823 * from below may cause stream head I/O routines to be called and these
1825 * nothing may come from below since the pass queue syncq is blocked.
1828 * head (e.g. because of message propagation from the pass queue) stream
1831 * When STPLEX is set, nothing may come into the stream from above and
1895 * XXX Remove any "odd" messages from the queue.
2077 * lbfree() have removed it from mux_nodes[] and linkinfo_list,
2116 * Messages could be flowing from underneath. We will
2129 * XXX Messages can also flow from top into this
2131 * in putnext() called from top that has still not
2305 * Would prefer finding the dev_t from before any cloning,
2347 * Translate the device flags (from conf.h) to the corresponding
3408 * Discard any messages left on the syncq(s), remove the syncq(s) from the
3435 * Note: we are already removed from the stream so that the
3442 /* Tidy up - removeq only does a half-remove from stream */
3542 * Consolidation Private, and of course only callable from the stream head or
3585 * get estimate of available memory from kmem_avail().
3794 * If the syncq is placed on the sqhead/sqtail queue, try to remove it from the
4125 strcopyin(void *from, void *to, size_t len, int copyflag)
4129 if (copyin(from, to, len))
4133 bcopy(from, to, len);
4139 strcopyout(void *from, void *to, size_t len, int copyflag)
4142 if (copyout(from, to, len))
4146 bcopy(from, to, len);
4206 * Backenable the first queue upstream from `q' with a service procedure.
4215 * stream from popping/pushing since the caller of getq might not
4468 * passes messages from one queues put/service procedure to another queue.
4489 * the framework also has to prevent messages being drained from
4492 * mechanism to prevent qwriter(PERIM_OUTER) from running after
4501 * - qprocsoff (disable_svc) sets QWCLOSE to prevent runservice from
4504 * canputnext, backenable etc from dereferencing the q_next that will
4528 * from the stream, we do not need to worry about them accessing a stale q_next
4533 * prevent threads from entering a module in the stream (i.e. sq_count can
4819 * Given a read queue, unlink it from any neighbors.
5493 * Must never be called from driver or module put entry point.
5558 * Prevent q_next from changing in this stream by incrementing sq_count.
5614 * Prevent q_next from changing in this stream by incrementing sd_refcnt.
5692 * differently from the inner syncqs.
5695 * inner syncqs that belong to the outer perimeter from
5700 * inner syncqs are added to or removed from the
5714 * threads from having write access at the same time. SQ_WRITER has to be part
5721 * threads from entering the inner syncqs while it is waiting for sq_count to
5826 * Atomically (from the perspective of threads calling become_writer)
5902 * Remove a syncq from an outer perimeter.
5971 * This routine can only be called from put or service procedures plus
6101 * Dequeue all writer callbacks from the outer perimeter and run them.
6208 * Messages from the event queue must be taken off in
6544 * this queue from the sq_head list when it is
6616 * qdrain_syncq can be called (currently) from only one of two places:
6622 * If called from drain_syncq, we found it in the list of queues needing
6625 * If called from some putnext variation, it was because the
6676 * claim either from the sq_count, or a putcount. We could
6677 * get an erroneous value from other counts, but ours won't
6696 * go away. Note that there is nothing to prevent this case from
6812 /* Remove the q from the syncq list if all the messages are drained. */
6820 * Since the queue is removed from the list, reset its priority.
6935 * Remove all messages from a syncq (if qp is NULL) or remove all messages
6944 * NOTE: This function assumes that it is called from the close() context and
7028 * Now remove the queue from the syncq.
7080 * Propagate all messages from a syncq to the next syncq that are associated
7088 * as there is no point in propagating the messages from the driver syncq
7180 * When messages are moved from high priority queue to
7246 * This routine can only be called from put or service procedures plus
7658 * is WRONG; the next queue forward from wnew with a
7670 * module may have been pushed from other end and
7824 * This prevents any messages from arriving during the setq
7831 * After the new queue is inserted, all messages coming from below are
8289 /* if we came here from the background thread, clear the flag */
8730 * gone by the time other cleanup happens from the destroy callbacks.