#define ZFCP_LOG_AREA ZFCP_LOG_AREA_ERP
/* this drivers version (do not edit !!! generated and updated by cvs) */
-#define ZFCP_ERP_REVISION "$Revision: 1.61 $"
+#define ZFCP_ERP_REVISION "$Revision: 1.54 $"
#include "zfcp_ext.h"
u8 req_code, resp_code;
int retval = 0;
- if (send_els->status != 0) {
- ZFCP_LOG_NORMAL("ELS request timed out, force physical port "
- "reopen of port 0x%016Lx on adapter %s\n",
- port->wwpn, zfcp_get_busid_by_port(port));
- debug_text_event(port->adapter->erp_dbf, 3, "forcreop");
- retval = zfcp_erp_port_forced_reopen(port, 0);
- if (retval != 0) {
- ZFCP_LOG_NORMAL("reopen of remote port 0x%016Lx "
- "on adapter %s failed\n", port->wwpn,
- zfcp_get_busid_by_port(port));
- retval = -EPERM;
- }
+ if (send_els->status != 0)
goto skip_fsfstatus;
- }
req = (void*)((page_to_pfn(send_els->req->page) << PAGE_SHIFT) + send_els->req->offset);
resp = (void*)((page_to_pfn(send_els->resp->page) << PAGE_SHIFT) + send_els->resp->offset);
zfcp_erp_adapter_strategy(struct zfcp_erp_action *erp_action)
{
int retval;
+ unsigned long timeout;
struct zfcp_adapter *adapter = erp_action->adapter;
retval = zfcp_erp_adapter_strategy_close(erp_action);
ZFCP_LOG_INFO("Waiting to allow the adapter %s "
"to recover itself\n",
zfcp_get_busid_by_adapter(adapter));
- msleep(jiffies_to_msecs(ZFCP_TYPE2_RECOVERY_TIME));
+ /*
+ * SUGGESTION: substitute by
+ * timeout = ZFCP_TYPE2_RECOVERY_TIME;
+ * __ZFCP_WAIT_EVENT_TIMEOUT(timeout, 0);
+ */
+ timeout = ZFCP_TYPE2_RECOVERY_TIME;
+ set_current_state(TASK_UNINTERRUPTIBLE);
+ schedule_timeout(timeout);
}
return retval;
int i;
volatile struct qdio_buffer_element *sbale;
struct zfcp_adapter *adapter = erp_action->adapter;
+ int retval_cleanup = 0;
if (atomic_test_mask(ZFCP_STATUS_ADAPTER_QDIOUP, &adapter->status)) {
ZFCP_LOG_NORMAL("bug: second attempt to set up QDIO on "
zfcp_get_busid_by_adapter(adapter));
goto failed_qdio_establish;
}
- debug_text_event(adapter->erp_dbf, 3, "qdio_est");
+ ZFCP_LOG_DEBUG("queues established\n");
if (qdio_activate(adapter->ccw_device, 0) != 0) {
ZFCP_LOG_INFO("error: activation of QDIO queues failed "
zfcp_get_busid_by_adapter(adapter));
goto failed_qdio_activate;
}
- debug_text_event(adapter->erp_dbf, 3, "qdio_act");
+ ZFCP_LOG_DEBUG("queues activated\n");
/*
* put buffers into response queue,
/* NOP */
failed_qdio_activate:
- debug_text_event(adapter->erp_dbf, 3, "qdio_down1a");
- while (qdio_shutdown(adapter->ccw_device,
- QDIO_FLAG_CLEANUP_USING_CLEAR) == -EINPROGRESS)
- msleep(1000);
- debug_text_event(adapter->erp_dbf, 3, "qdio_down1b");
+ /* DEBUG */
+ //__ZFCP_WAIT_EVENT_TIMEOUT(timeout, 0);
+ /* cleanup queues previously established */
+ retval_cleanup = qdio_shutdown(adapter->ccw_device,
+ QDIO_FLAG_CLEANUP_USING_CLEAR);
+ if (retval_cleanup) {
+ ZFCP_LOG_NORMAL("bug: shutdown of QDIO queues failed "
+ "(retval=%d)\n", retval_cleanup);
+ }
failed_qdio_establish:
+ atomic_clear_mask(ZFCP_STATUS_ADAPTER_QDIOUP, &adapter->status);
+
failed_sanity:
retval = ZFCP_ERP_FAILED;
goto out;
}
+ /* cleanup queues previously established */
+
/*
- * Get queue_lock and clear QDIOUP flag. Thus it's guaranteed that
- * do_QDIO won't be called while qdio_shutdown is in progress.
+ * MUST NOT LOCK - qdio_cleanup might call schedule
+ * FIXME: need another way to make cleanup safe
*/
-
- write_lock_irq(&adapter->request_queue.queue_lock);
- atomic_clear_mask(ZFCP_STATUS_ADAPTER_QDIOUP, &adapter->status);
- write_unlock_irq(&adapter->request_queue.queue_lock);
-
- debug_text_event(adapter->erp_dbf, 3, "qdio_down2a");
- while (qdio_shutdown(adapter->ccw_device,
- QDIO_FLAG_CLEANUP_USING_CLEAR) == -EINPROGRESS)
- msleep(1000);
- debug_text_event(adapter->erp_dbf, 3, "qdio_down2b");
+ /* Note:
+ * We need the request_queue lock here, otherwise there exists the
+ * following race:
+ *
+ * queuecommand calls create_fcp_commmand_task...calls req_create,
+ * gets sbal x to x+y - meanwhile adapter reopen is called, completes
+ * - req_send calls do_QDIO for sbal x to x+y, i.e. wrong indices.
+ *
+ * with lock:
+ * queuecommand calls create_fcp_commmand_task...calls req_create,
+ * gets sbal x to x+y - meanwhile adapter reopen is called, waits
+ * - req_send calls do_QDIO for sbal x to x+y, i.e. wrong indices
+ * but do_QDIO fails as adapter_reopen is still waiting for the lock
+ * OR
+ * queuecommand calls create_fcp_commmand_task...calls req_create
+ * - meanwhile adapter reopen is called...completes,
+ * - gets sbal 0 to 0+y, - req_send calls do_QDIO for sbal 0 to 0+y,
+ * i.e. correct indices...though an fcp command is called before
+ * exchange config data...that should be fine, however
+ */
+ if (qdio_shutdown(adapter->ccw_device, QDIO_FLAG_CLEANUP_USING_CLEAR)) {
+ /*
+ * FIXME(design):
+ * What went wrong? What to do best? Proper retval?
+ */
+ ZFCP_LOG_NORMAL("bug: shutdown of QDIO queues failed on "
+ "adapter %s\n",
+ zfcp_get_busid_by_adapter(adapter));
+ } else
+ ZFCP_LOG_DEBUG("queues cleaned up\n");
/*
* First we had to stop QDIO operation.
adapter->request_queue.free_index = 0;
atomic_set(&adapter->request_queue.free_count, 0);
adapter->request_queue.distance_from_int = 0;
+
+ atomic_clear_mask(ZFCP_STATUS_ADAPTER_QDIOUP, &adapter->status);
out:
return retval;
}
ZFCP_LOG_DEBUG("host connection still initialising... "
"waiting and retrying...\n");
/* sleep a little bit before retry */
- msleep(jiffies_to_msecs(ZFCP_EXCHANGE_CONFIG_DATA_SLEEP));
+ set_current_state(TASK_INTERRUPTIBLE);
+ schedule_timeout(ZFCP_EXCHANGE_CONFIG_DATA_SLEEP);
}
} while ((retries--) &&
atomic_test_mask(ZFCP_STATUS_ADAPTER_HOST_CON_INIT,
git://git.onelab.eu / linux-2.6.git / blobdiff