fs: replace use of system_unbound_wq with system_dfl_wq

Currently if a user enqueue a work item using schedule_delayed_work() the
used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
schedule_work() that is using system_wq and queue_work(), that makes use
again of WORK_CPU_UNBOUND.

This lack of consistentcy cannot be addressed without refactoring the API.

system_unbound_wq should be the default workqueue so as not to enforce
locality constraints for random work whenever it's not required.

Adding system_dfl_wq to encourage its use when unbound work should be used.

The old system_unbound_wq will be kept for a few release cycles.

Suggested-by: Tejun Heo <[e-mail]>
Signed-off-by: Marco Crivellari <[e-mail]>
Link: https://lore.kernel.org/[e-mail]
Signed-off-by: Christian Brauner <[e-mail]>

fs/afs/callback.c

···
 	list_for_each_entry(vnode, &volume->open_mmaps, cb_mmap_link) {
 		if (vnode->cb_v_check != atomic_read(&volume->cb_v_break)) {
 			afs_clear_cb_promise(vnode, afs_cb_promise_clear_vol_init_cb);
-			queue_work(system_unbound_wq, &vnode->cb_work);
+			queue_work(system_dfl_wq, &vnode->cb_work);
 		}
 	}
 
···
 		if (reason != afs_cb_break_for_deleted &&
 		    vnode->status.type == AFS_FTYPE_FILE &&
 		    atomic_read(&vnode->cb_nr_mmap))
-			queue_work(system_unbound_wq, &vnode->cb_work);
+			queue_work(system_dfl_wq, &vnode->cb_work);
 
 		trace_afs_cb_break(&vnode->fid, vnode->cb_break, reason, true);
 	} else {

fs/afs/write.c

···
 void afs_issue_write(struct netfs_io_subrequest *subreq)
 {
 	subreq->work.func = afs_issue_write_worker;
-	if (!queue_work(system_unbound_wq, &subreq->work))
+	if (!queue_work(system_dfl_wq, &subreq->work))
 		WARN_ON_ONCE(1);
 }
 

fs/bcachefs/btree_write_buffer.c

···
 
 	if (bch2_btree_write_buffer_should_flush(c) &&
 	    __enumerated_ref_tryget(&c->writes, BCH_WRITE_REF_btree_write_buffer) &&
-	    !queue_work(system_unbound_wq, &c->btree_write_buffer.flush_work))
+	    !queue_work(system_dfl_wq, &c->btree_write_buffer.flush_work))
 		enumerated_ref_put(&c->writes, BCH_WRITE_REF_btree_write_buffer);
 
 	if (dst->wb == &wb->flushing)

fs/bcachefs/io_read.c

···
 
 	if (bch2_err_matches(ret, BCH_ERR_data_read_retry)) {
 		bch2_rbio_punt(rbio, bch2_rbio_retry,
-			       RBIO_CONTEXT_UNBOUND, system_unbound_wq);
+			       RBIO_CONTEXT_UNBOUND, system_dfl_wq);
 	} else {
 		rbio = bch2_rbio_free(rbio);
 
···
 	bch2_rbio_error(rbio, -BCH_ERR_data_read_retry_csum_err, BLK_STS_IOERR);
 	goto out;
 decompression_err:
-	bch2_rbio_punt(rbio, bch2_read_decompress_err, RBIO_CONTEXT_UNBOUND, system_unbound_wq);
+	bch2_rbio_punt(rbio, bch2_read_decompress_err, RBIO_CONTEXT_UNBOUND, system_dfl_wq);
 	goto out;
 decrypt_err:
-	bch2_rbio_punt(rbio, bch2_read_decrypt_err, RBIO_CONTEXT_UNBOUND, system_unbound_wq);
+	bch2_rbio_punt(rbio, bch2_read_decrypt_err, RBIO_CONTEXT_UNBOUND, system_dfl_wq);
 	goto out;
 }
 
···
 	    rbio->promote ||
 	    crc_is_compressed(rbio->pick.crc) ||
 	    bch2_csum_type_is_encryption(rbio->pick.crc.csum_type))
-		context = RBIO_CONTEXT_UNBOUND,	wq = system_unbound_wq;
+		context = RBIO_CONTEXT_UNBOUND,	wq = system_dfl_wq;
 	else if (rbio->pick.crc.csum_type)
 		context = RBIO_CONTEXT_HIGHPRI,	wq = system_highpri_wq;
 

fs/bcachefs/journal_io.c

···
 					  BCH_DEV_READ_REF_journal_read))
 			closure_call(&ca->journal.read,
 				     bch2_journal_read_device,
-				     system_unbound_wq,
+				     system_dfl_wq,
 				     &jlist.cl);
 		else
 			degraded = true;

fs/btrfs/block-group.c

···
 	btrfs_reclaim_sweep(fs_info);
 	spin_lock(&fs_info->unused_bgs_lock);
 	if (!list_empty(&fs_info->reclaim_bgs))
-		queue_work(system_unbound_wq, &fs_info->reclaim_bgs_work);
+		queue_work(system_dfl_wq, &fs_info->reclaim_bgs_work);
 	spin_unlock(&fs_info->unused_bgs_lock);
 }
 

fs/btrfs/extent_map.c

···
 	if (atomic64_cmpxchg(&fs_info->em_shrinker_nr_to_scan, 0, nr_to_scan) != 0)
 		return;
 
-	queue_work(system_unbound_wq, &fs_info->em_shrinker_work);
+	queue_work(system_dfl_wq, &fs_info->em_shrinker_work);
 }
 
 void btrfs_init_extent_map_shrinker_work(struct btrfs_fs_info *fs_info)

fs/btrfs/space-info.c

···
 							  space_info->flags,
 							  orig_bytes, flush,
 							  "enospc");
-				queue_work(system_unbound_wq, async_work);
+				queue_work(system_dfl_wq, async_work);
 			}
 		} else {
 			list_add_tail(&ticket.list,
···
 		    need_preemptive_reclaim(fs_info, space_info)) {
 			trace_btrfs_trigger_flush(fs_info, space_info->flags,
 						  orig_bytes, flush, "preempt");
-			queue_work(system_unbound_wq,
+			queue_work(system_dfl_wq,
 				   &fs_info->preempt_reclaim_work);
 		}
 	}

fs/btrfs/zoned.c

···
 	refcount_inc(&eb->refs);
 	bg->last_eb = eb;
 	INIT_WORK(&bg->zone_finish_work, btrfs_zone_finish_endio_workfn);
-	queue_work(system_unbound_wq, &bg->zone_finish_work);
+	queue_work(system_dfl_wq, &bg->zone_finish_work);
 }
 
 void btrfs_clear_data_reloc_bg(struct btrfs_block_group *bg)

fs/coredump.c

···
 
 		/*
 		 * Usermode helpers are childen of either
-		 * system_unbound_wq or of kthreadd. So we know that
+		 * system_dfl_wq or of kthreadd. So we know that
 		 * we're starting off with a clean file descriptor
 		 * table. So we should always be able to use
 		 * COREDUMP_PIDFD_NUMBER as our file descriptor value.

fs/ext4/mballoc.c

···
 		list_splice_tail(&freed_data_list, &sbi->s_discard_list);
 		spin_unlock(&sbi->s_md_lock);
 		if (wake)
-			queue_work(system_unbound_wq, &sbi->s_discard_work);
+			queue_work(system_dfl_wq, &sbi->s_discard_work);
 	} else {
 		list_for_each_entry_safe(entry, tmp, &freed_data_list, efd_list)
 			kmem_cache_free(ext4_free_data_cachep, entry);

fs/netfs/misc.c

···
 {
 	if (test_bit(NETFS_RREQ_OFFLOAD_COLLECTION, &rreq->flags) &&
 	    !test_bit(NETFS_RREQ_RETRYING, &rreq->flags)) {
-		queue_work(system_unbound_wq, &rreq->work);
+		queue_work(system_dfl_wq, &rreq->work);
 	} else {
 		trace_netfs_rreq(rreq, netfs_rreq_trace_wake_queue);
 		wake_up(&rreq->waitq);

fs/netfs/objects.c

···
 		dead = __refcount_dec_and_test(&rreq->ref, &r);
 		trace_netfs_rreq_ref(debug_id, r - 1, what);
 		if (dead)
-			WARN_ON(!queue_work(system_unbound_wq, &rreq->cleanup_work));
+			WARN_ON(!queue_work(system_dfl_wq, &rreq->cleanup_work));
 	}
 }
 

fs/nfsd/filecache.c

···
 nfsd_file_schedule_laundrette(void)
 {
 	if (test_bit(NFSD_FILE_CACHE_UP, &nfsd_file_flags))
-		queue_delayed_work(system_unbound_wq, &nfsd_filecache_laundrette,
+		queue_delayed_work(system_dfl_wq, &nfsd_filecache_laundrette,
 				   NFSD_LAUNDRETTE_DELAY);
 }
 

fs/notify/mark.c

···
 		conn->destroy_next = connector_destroy_list;
 		connector_destroy_list = conn;
 		spin_unlock(&destroy_lock);
-		queue_work(system_unbound_wq, &connector_reaper_work);
+		queue_work(system_dfl_wq, &connector_reaper_work);
 	}
 	/*
 	 * Note that we didn't update flags telling whether inode cares about
···
 	spin_lock(&destroy_lock);
 	list_add(&mark->g_list, &destroy_list);
 	spin_unlock(&destroy_lock);
-	queue_delayed_work(system_unbound_wq, &reaper_work,
+	queue_delayed_work(system_dfl_wq, &reaper_work,
 			   FSNOTIFY_REAPER_DELAY);
 }
 EXPORT_SYMBOL_GPL(fsnotify_put_mark);

fs/quota/dquot.c

···
 	put_releasing_dquots(dquot);
 	atomic_dec(&dquot->dq_count);
 	spin_unlock(&dq_list_lock);
-	queue_delayed_work(system_unbound_wq, &quota_release_work, 1);
+	queue_delayed_work(system_dfl_wq, &quota_release_work, 1);
 }
 EXPORT_SYMBOL(dqput);