From: Jeffrey Altman <jaltman@your-file-system.com>
Date: Sat, 5 May 2012 23:11:07 +0000 (-0400)
Subject: Windows: cm_GetNewSCache drop lock to permit change
X-Git-Tag: openafs-stable-1_8_0pre1~2443
X-Git-Url: https://git.openafs.org/?p=openafs.git;a=commitdiff_plain;h=46c1f1391b6e1b95798e183b8f34cba5c074e0ad

Windows: cm_GetNewSCache drop lock to permit change

In cm_GetNewSCache the entire LRU queue is searched for a
cm_scache_t object that is safe to recycle.  If none are the LRU
queue was immediately searched again without dropping the
cm_scacheLock or taking a pause.  As a result it is quite possible
that a thread about to release a cm_scache_t was blocked from
doing so.

This patchset factors some of the logic a bit differently to
improve readability and adds new log messages to help diagnose
the cause of a problem if no cm_scache_t ever becomes available.

Change-Id: Ica6ebee0ce0456e879ae7188d9c8cdc935a92e5b
Reviewed-on: http://gerrit.openafs.org/7352
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
---

diff --git a/src/WINNT/afsd/cm_scache.c b/src/WINNT/afsd/cm_scache.c
index 9b65dae..31afbb0 100644
--- a/src/WINNT/afsd/cm_scache.c
+++ b/src/WINNT/afsd/cm_scache.c
@@ -317,22 +317,21 @@ cm_GetNewSCache(afs_uint32 locked)
                     if (!buf_dirty && !buf_rdr) {
                         cm_fid_t   fid;
                         afs_uint32 fileType;
+                        int        success;
 
-                        if (!lock_TryWrite(&scp->rw)) {
-                            lock_ObtainWrite(&cm_scacheLock);
-                            if (scp_prev != (cm_scache_t *) osi_QPrev(&scp->q) &&
-                                scp_next != (cm_scache_t *) osi_QNext(&scp->q))
-                                break;
-                            else
-                                continue;
-                        }
+                        success = lock_TryWrite(&scp->rw);
 
                         lock_ObtainWrite(&cm_scacheLock);
                         if (scp_prev != (cm_scache_t *) osi_QPrev(&scp->q) &&
                             scp_next != (cm_scache_t *) osi_QNext(&scp->q))
                         {
-                            lock_ReleaseWrite(&scp->rw);
+                            osi_Log1(afsd_logp, "GetNewSCache scp 0x%p; LRU order changed", scp);
+                            if (success)
+                                lock_ReleaseWrite(&scp->rw);
                             break;
+                        } else if (!success) {
+                                osi_Log1(afsd_logp, "GetNewSCache failed to obtain lock scp 0x%p", scp);
+                                continue;
                         }
 
                         /* Found a likely candidate.  Save type and fid in case we succeed */
@@ -351,23 +350,34 @@ cm_GetNewSCache(afs_uint32 locked)
                             goto done;
                         }
                         lock_ReleaseWrite(&scp->rw);
-                    } else if (!buf_rdr) {
-                        osi_Log1(afsd_logp, "GetNewSCache dirty buffers scp 0x%p", scp);
-                        lock_ObtainWrite(&cm_scacheLock);
-                        if (scp_prev != (cm_scache_t *) osi_QPrev(&scp->q) &&
-                            scp_next != (cm_scache_t *) osi_QNext(&scp->q))
-                            break;
                     } else {
-                        osi_Log1(afsd_logp,"GetNewSCache redirector is holding extents scp 0x%p", scp);
+                        if (buf_rdr)
+                            osi_Log1(afsd_logp,"GetNewSCache redirector is holding extents scp 0x%p", scp);
+                        else
+                            osi_Log1(afsd_logp, "GetNewSCache dirty buffers scp 0x%p", scp);
+
                         lock_ObtainWrite(&cm_scacheLock);
                         if (scp_prev != (cm_scache_t *) osi_QPrev(&scp->q) &&
                             scp_next != (cm_scache_t *) osi_QNext(&scp->q))
+                        {
+                            osi_Log1(afsd_logp, "GetNewSCache scp 0x%p; LRU order changed", scp);
                             break;
+                        }
                     }
                 }
             } /* for */
 
             osi_Log2(afsd_logp, "GetNewSCache all scache entries in use (attempt = %d, count = %u)", attempt, count);
+            if (scp == NULL) {
+                /*
+                * The entire LRU queue was walked and no available cm_scache_t was
+                * found.  Drop the cm_scacheLock and sleep for a moment to give a
+                * chance for cm_scache_t objects to be released.
+                */
+                lock_ReleaseWrite(&cm_scacheLock);
+                Sleep(50);
+                lock_ObtainWrite(&cm_scacheLock);
+            }
         }
         /* FAILURE */
         scp = NULL;