/* * Copyright 2000, International Business Machines Corporation and others. * All Rights Reserved. * * This software has been released under the terms of the IBM Public * License. For details, see the LICENSE file in the top-level source * directory or online at http://www.openafs.org/dl/license10.html */ #include #include #include #include #include #include #include #include #include #include "afsd.h" #include "cm_btree.h" /*extern void afsi_log(char *pattern, ...);*/ extern osi_hyper_t hzero; /* File locks */ osi_queue_t *cm_allFileLocks; osi_queue_t *cm_freeFileLocks; unsigned long cm_lockRefreshCycle; /* lock for globals */ osi_rwlock_t cm_scacheLock; /* Dummy scache entry for use with pioctl fids */ cm_scache_t cm_fakeSCache; osi_queue_t * cm_allFreeWaiters; /* protected by cm_scacheLock */ #ifdef AFS_FREELANCE_CLIENT extern osi_mutex_t cm_Freelance_Lock; #endif /* must be called with cm_scacheLock write-locked! */ void cm_AdjustScacheLRU(cm_scache_t *scp) { if (scp == cm_data.scacheLRULastp) cm_data.scacheLRULastp = (cm_scache_t *) osi_QPrev(&scp->q); osi_QRemoveHT((osi_queue_t **) &cm_data.scacheLRUFirstp, (osi_queue_t **) &cm_data.scacheLRULastp, &scp->q); osi_QAdd((osi_queue_t **) &cm_data.scacheLRUFirstp, &scp->q); if (!cm_data.scacheLRULastp) cm_data.scacheLRULastp = scp; } /* call with scache write-locked and mutex held */ void cm_RemoveSCacheFromHashTable(cm_scache_t *scp) { cm_scache_t **lscpp; cm_scache_t *tscp; int i; if (scp->flags & CM_SCACHEFLAG_INHASH) { /* hash it out first */ i = CM_SCACHE_HASH(&scp->fid); for (lscpp = &cm_data.scacheHashTablep[i], tscp = cm_data.scacheHashTablep[i]; tscp; lscpp = &tscp->nextp, tscp = tscp->nextp) { if (tscp == scp) { *lscpp = scp->nextp; scp->flags &= ~CM_SCACHEFLAG_INHASH; break; } } } } /* called with cm_scacheLock write-locked; recycles an existing scp. * * this function ignores all of the locking hierarchy. */ long cm_RecycleSCache(cm_scache_t *scp, afs_int32 flags) { if (scp->refCount != 0) { return -1; } if (scp->flags & CM_SCACHEFLAG_SMB_FID) { osi_Log1(afsd_logp,"cm_RecycleSCache CM_SCACHEFLAG_SMB_FID detected scp 0x%p", scp); #ifdef DEBUG osi_panic("cm_RecycleSCache CM_SCACHEFLAG_SMB_FID detected",__FILE__,__LINE__); #endif return -1; } cm_RemoveSCacheFromHashTable(scp); #if 0 if (flags & CM_SCACHE_RECYCLEFLAG_DESTROY_BUFFERS) { osi_queueData_t *qdp; cm_buf_t *bufp; while(qdp = scp->bufWritesp) { bufp = osi_GetQData(qdp); osi_QRemove((osi_queue_t **) &scp->bufWritesp, &qdp->q); osi_QDFree(qdp); if (bufp) { lock_ObtainMutex(&bufp->mx); bufp->cmFlags &= ~CM_BUF_CMSTORING; bufp->flags &= ~CM_BUF_DIRTY; bufp->dirty_offset = 0; bufp->dirty_length = 0; bufp->flags |= CM_BUF_ERROR; bufp->error = VNOVNODE; bufp->dataVersion = -1; /* bad */ bufp->dirtyCounter++; if (bufp->flags & CM_BUF_WAITING) { osi_Log2(afsd_logp, "CM RecycleSCache Waking [scp 0x%x] bufp 0x%x", scp, bufp); osi_Wakeup((long) &bufp); } lock_ReleaseMutex(&bufp->mx); buf_Release(bufp); } } while(qdp = scp->bufReadsp) { bufp = osi_GetQData(qdp); osi_QRemove((osi_queue_t **) &scp->bufReadsp, &qdp->q); osi_QDFree(qdp); if (bufp) { lock_ObtainMutex(&bufp->mx); bufp->cmFlags &= ~CM_BUF_CMFETCHING; bufp->flags &= ~CM_BUF_DIRTY; bufp->dirty_offset = 0; bufp->dirty_length = 0; bufp->flags |= CM_BUF_ERROR; bufp->error = VNOVNODE; bufp->dataVersion = -1; /* bad */ bufp->dirtyCounter++; if (bufp->flags & CM_BUF_WAITING) { osi_Log2(afsd_logp, "CM RecycleSCache Waking [scp 0x%x] bufp 0x%x", scp, bufp); osi_Wakeup((long) &bufp); } lock_ReleaseMutex(&bufp->mx); buf_Release(bufp); } } buf_CleanDirtyBuffers(scp); } else { /* look for things that shouldn't still be set */ osi_assertx(scp->bufWritesp == NULL, "non-null cm_scache_t bufWritesp"); osi_assertx(scp->bufReadsp == NULL, "non-null cm_scache_t bufReadsp"); } #endif /* invalidate so next merge works fine; * also initialize some flags */ scp->fileType = 0; scp->flags &= ~(CM_SCACHEFLAG_STATD | CM_SCACHEFLAG_DELETED | CM_SCACHEFLAG_RO | CM_SCACHEFLAG_PURERO | CM_SCACHEFLAG_OVERQUOTA | CM_SCACHEFLAG_OUTOFSPACE | CM_SCACHEFLAG_EACCESS); scp->serverModTime = 0; scp->dataVersion = 0; scp->bulkStatProgress = hzero; scp->waitCount = 0; scp->waitQueueT = NULL; if (scp->cbServerp) { cm_PutServer(scp->cbServerp); scp->cbServerp = NULL; } scp->cbExpires = 0; scp->fid.vnode = 0; scp->fid.volume = 0; scp->fid.unique = 0; scp->fid.cell = 0; /* remove from dnlc */ cm_dnlcPurgedp(scp); cm_dnlcPurgevp(scp); /* discard cached status; if non-zero, Close * tried to store this to server but failed */ scp->mask = 0; /* drop held volume ref */ if (scp->volp) { cm_PutVolume(scp->volp); scp->volp = NULL; } /* discard symlink info */ scp->mountPointStringp[0] = '\0'; memset(&scp->mountRootFid, 0, sizeof(cm_fid_t)); memset(&scp->dotdotFid, 0, sizeof(cm_fid_t)); /* reset locking info */ scp->fileLocksH = NULL; scp->fileLocksT = NULL; scp->serverLock = (-1); scp->exclusiveLocks = 0; scp->sharedLocks = 0; scp->lockDataVersion = -1; /* not locked, but there can be no references to this guy * while we hold the global refcount lock. */ cm_FreeAllACLEnts(scp); #ifdef USE_BPLUS /* destroy directory Bplus Tree */ if (scp->dirBplus) { LARGE_INTEGER start, end; QueryPerformanceCounter(&start); bplus_free_tree++; freeBtree(scp->dirBplus); scp->dirBplus = NULL; QueryPerformanceCounter(&end); bplus_free_time += (end.QuadPart - start.QuadPart); } #endif return 0; } /* called with cm_scacheLock write-locked; find a vnode to recycle. * Can allocate a new one if desperate, or if below quota (cm_data.maxSCaches). */ cm_scache_t *cm_GetNewSCache(void) { cm_scache_t *scp; int retry = 0; #if 0 /* first pass - look for deleted objects */ for ( scp = cm_data.scacheLRULastp; scp; scp = (cm_scache_t *) osi_QPrev(&scp->q)) { osi_assertx(scp >= cm_data.scacheBaseAddress && scp < (cm_scache_t *)cm_data.scacheHashTablep, "invalid cm_scache_t address"); if (scp->refCount == 0) { if (scp->flags & CM_SCACHEFLAG_DELETED) { osi_Log1(afsd_logp, "GetNewSCache attempting to recycle deleted scp 0x%x", scp); if (!cm_RecycleSCache(scp, CM_SCACHE_RECYCLEFLAG_DESTROY_BUFFERS)) { /* we found an entry, so return it */ /* now remove from the LRU queue and put it back at the * head of the LRU queue. */ cm_AdjustScacheLRU(scp); /* and we're done */ return scp; } osi_Log1(afsd_logp, "GetNewSCache recycled failed scp 0x%x", scp); } else if (!(scp->flags & CM_SCACHEFLAG_INHASH)) { /* we found an entry, so return it */ /* now remove from the LRU queue and put it back at the * head of the LRU queue. */ cm_AdjustScacheLRU(scp); /* and we're done */ return scp; } } } osi_Log0(afsd_logp, "GetNewSCache no deleted or recycled entries available for reuse"); #endif if (cm_data.currentSCaches >= cm_data.maxSCaches) { /* There were no deleted scache objects that we could use. Try to find * one that simply hasn't been used in a while. */ for ( scp = cm_data.scacheLRULastp; scp; scp = (cm_scache_t *) osi_QPrev(&scp->q)) { /* It is possible for the refCount to be zero and for there still * to be outstanding dirty buffers. If there are dirty buffers, * we must not recycle the scp. */ if (scp->refCount == 0 && scp->bufReadsp == NULL && scp->bufWritesp == NULL) { if (!buf_DirtyBuffersExist(&scp->fid)) { if (!cm_RecycleSCache(scp, 0)) { /* we found an entry, so return it */ /* now remove from the LRU queue and put it back at the * head of the LRU queue. */ cm_AdjustScacheLRU(scp); /* and we're done */ return scp; } } else { osi_Log1(afsd_logp,"GetNewSCache dirty buffers exist scp 0x%x", scp); } } } osi_Log1(afsd_logp, "GetNewSCache all scache entries in use (retry = %d)", retry); return NULL; } /* if we get here, we should allocate a new scache entry. We either are below * quota or we have a leak and need to allocate a new one to avoid panicing. */ scp = cm_data.scacheBaseAddress + cm_data.currentSCaches; osi_assertx(scp >= cm_data.scacheBaseAddress && scp < (cm_scache_t *)cm_data.scacheHashTablep, "invalid cm_scache_t address"); memset(scp, 0, sizeof(cm_scache_t)); scp->magic = CM_SCACHE_MAGIC; lock_InitializeMutex(&scp->mx, "cm_scache_t mutex"); lock_InitializeRWLock(&scp->bufCreateLock, "cm_scache_t bufCreateLock"); #ifdef USE_BPLUS lock_InitializeRWLock(&scp->dirlock, "cm_scache_t dirlock"); #endif scp->serverLock = -1; /* and put it in the LRU queue */ osi_QAdd((osi_queue_t **) &cm_data.scacheLRUFirstp, &scp->q); if (!cm_data.scacheLRULastp) cm_data.scacheLRULastp = scp; cm_data.currentSCaches++; cm_dnlcPurgedp(scp); /* make doubly sure that this is not in dnlc */ cm_dnlcPurgevp(scp); scp->allNextp = cm_data.allSCachesp; cm_data.allSCachesp = scp; return scp; } /* like strcmp, only for fids */ int cm_FidCmp(cm_fid_t *ap, cm_fid_t *bp) { if (ap->vnode != bp->vnode) return 1; if (ap->volume != bp->volume) return 1; if (ap->unique != bp->unique) return 1; if (ap->cell != bp->cell) return 1; return 0; } void cm_fakeSCacheInit(int newFile) { if ( newFile ) { memset(&cm_data.fakeSCache, 0, sizeof(cm_scache_t)); cm_data.fakeSCache.cbServerp = (struct cm_server *)(-1); /* can leave clientModTime at 0 */ cm_data.fakeSCache.fileType = CM_SCACHETYPE_FILE; cm_data.fakeSCache.unixModeBits = 0777; cm_data.fakeSCache.length.LowPart = 1000; cm_data.fakeSCache.linkCount = 1; cm_data.fakeSCache.refCount = 1; } lock_InitializeMutex(&cm_data.fakeSCache.mx, "cm_scache_t mutex"); } long cm_ValidateSCache(void) { cm_scache_t * scp, *lscp; long i; if ( cm_data.scacheLRUFirstp == NULL && cm_data.scacheLRULastp != NULL || cm_data.scacheLRUFirstp != NULL && cm_data.scacheLRULastp == NULL) { afsi_log("cm_ValidateSCache failure: inconsistent LRU pointers"); fprintf(stderr, "cm_ValidateSCache failure: inconsistent LRU pointers\n"); return -17; } for ( scp = cm_data.scacheLRUFirstp, lscp = NULL, i = 0; scp; lscp = scp, scp = (cm_scache_t *) osi_QNext(&scp->q), i++ ) { if (scp->magic != CM_SCACHE_MAGIC) { afsi_log("cm_ValidateSCache failure: scp->magic != CM_SCACHE_MAGIC"); fprintf(stderr, "cm_ValidateSCache failure: scp->magic != CM_SCACHE_MAGIC\n"); return -1; } if (scp->nextp && scp->nextp->magic != CM_SCACHE_MAGIC) { afsi_log("cm_ValidateSCache failure: scp->nextp->magic != CM_SCACHE_MAGIC"); fprintf(stderr, "cm_ValidateSCache failure: scp->nextp->magic != CM_SCACHE_MAGIC\n"); return -2; } if (scp->randomACLp && scp->randomACLp->magic != CM_ACLENT_MAGIC) { afsi_log("cm_ValidateSCache failure: scp->randomACLp->magic != CM_ACLENT_MAGIC"); fprintf(stderr, "cm_ValidateSCache failure: scp->randomACLp->magic != CM_ACLENT_MAGIC\n"); return -3; } if (scp->volp && scp->volp->magic != CM_VOLUME_MAGIC) { afsi_log("cm_ValidateSCache failure: scp->volp->magic != CM_VOLUME_MAGIC"); fprintf(stderr, "cm_ValidateSCache failure: scp->volp->magic != CM_VOLUME_MAGIC\n"); return -4; } if (i > cm_data.currentSCaches ) { afsi_log("cm_ValidateSCache failure: LRU First queue loops"); fprintf(stderr, "cm_ValidateSCache failure: LUR First queue loops\n"); return -13; } if (lscp != (cm_scache_t *) osi_QPrev(&scp->q)) { afsi_log("cm_ValidateSCache failure: QPrev(scp) != previous"); fprintf(stderr, "cm_ValidateSCache failure: QPrev(scp) != previous\n"); return -15; } } for ( scp = cm_data.scacheLRULastp, lscp = NULL, i = 0; scp; lscp = scp, scp = (cm_scache_t *) osi_QPrev(&scp->q), i++ ) { if (scp->magic != CM_SCACHE_MAGIC) { afsi_log("cm_ValidateSCache failure: scp->magic != CM_SCACHE_MAGIC"); fprintf(stderr, "cm_ValidateSCache failure: scp->magic != CM_SCACHE_MAGIC\n"); return -5; } if (scp->nextp && scp->nextp->magic != CM_SCACHE_MAGIC) { afsi_log("cm_ValidateSCache failure: scp->nextp->magic != CM_SCACHE_MAGIC"); fprintf(stderr, "cm_ValidateSCache failure: scp->nextp->magic != CM_SCACHE_MAGIC\n"); return -6; } if (scp->randomACLp && scp->randomACLp->magic != CM_ACLENT_MAGIC) { afsi_log("cm_ValidateSCache failure: scp->randomACLp->magic != CM_ACLENT_MAGIC"); fprintf(stderr, "cm_ValidateSCache failure: scp->randomACLp->magic != CM_ACLENT_MAGIC\n"); return -7; } if (scp->volp && scp->volp->magic != CM_VOLUME_MAGIC) { afsi_log("cm_ValidateSCache failure: scp->volp->magic != CM_VOLUME_MAGIC"); fprintf(stderr, "cm_ValidateSCache failure: scp->volp->magic != CM_VOLUME_MAGIC\n"); return -8; } if (i > cm_data.currentSCaches ) { afsi_log("cm_ValidateSCache failure: LRU Last queue loops"); fprintf(stderr, "cm_ValidateSCache failure: LUR Last queue loops\n"); return -14; } if (lscp != (cm_scache_t *) osi_QNext(&scp->q)) { afsi_log("cm_ValidateSCache failure: QNext(scp) != next"); fprintf(stderr, "cm_ValidateSCache failure: QNext(scp) != next\n"); return -16; } } for ( i=0; i < cm_data.scacheHashTableSize; i++ ) { for ( scp = cm_data.scacheHashTablep[i]; scp; scp = scp->nextp ) { if (scp->magic != CM_SCACHE_MAGIC) { afsi_log("cm_ValidateSCache failure: scp->magic != CM_SCACHE_MAGIC"); fprintf(stderr, "cm_ValidateSCache failure: scp->magic != CM_SCACHE_MAGIC\n"); return -9; } if (scp->nextp && scp->nextp->magic != CM_SCACHE_MAGIC) { afsi_log("cm_ValidateSCache failure: scp->nextp->magic != CM_SCACHE_MAGIC"); fprintf(stderr, "cm_ValidateSCache failure: scp->nextp->magic != CM_SCACHE_MAGIC\n"); return -10; } if (scp->randomACLp && scp->randomACLp->magic != CM_ACLENT_MAGIC) { afsi_log("cm_ValidateSCache failure: scp->randomACLp->magic != CM_ACLENT_MAGIC"); fprintf(stderr, "cm_ValidateSCache failure: scp->randomACLp->magic != CM_ACLENT_MAGIC\n"); return -11; } if (scp->volp && scp->volp->magic != CM_VOLUME_MAGIC) { afsi_log("cm_ValidateSCache failure: scp->volp->magic != CM_VOLUME_MAGIC"); fprintf(stderr, "cm_ValidateSCache failure: scp->volp->magic != CM_VOLUME_MAGIC\n"); return -12; } } } return cm_dnlcValidate(); } void cm_SuspendSCache(void) { cm_scache_t * scp; time_t now; cm_GiveUpAllCallbacksAllServers(TRUE); /* * After this call all servers are marked down. * Do not clear the callbacks, instead change the * expiration time so that the callbacks will be expired * when the servers are marked back up. However, we * want the callbacks to be preserved as long as the * servers are down. That way if the machine resumes * without network, the stat cache item will still be * considered valid. */ now = osi_Time(); lock_ObtainWrite(&cm_scacheLock); for ( scp = cm_data.allSCachesp; scp; scp = scp->allNextp ) { if (scp->cbServerp) { if (scp->flags & CM_SCACHEFLAG_PURERO && scp->volp) { if (scp->volp->cbExpiresRO == scp->cbExpires) { scp->volp->cbExpiresRO = now+1; } } scp->cbExpires = now+1; } } lock_ReleaseWrite(&cm_scacheLock); } long cm_ShutdownSCache(void) { cm_scache_t * scp; lock_ObtainWrite(&cm_scacheLock); for ( scp = cm_data.allSCachesp; scp; scp = scp->allNextp ) { if (scp->randomACLp) { lock_ObtainMutex(&scp->mx); cm_FreeAllACLEnts(scp); lock_ReleaseMutex(&scp->mx); } if (scp->cbServerp) { cm_PutServer(scp->cbServerp); scp->cbServerp = NULL; } scp->cbExpires = 0; scp->flags &= ~CM_SCACHEFLAG_CALLBACK; #ifdef USE_BPLUS if (scp->dirBplus) freeBtree(scp->dirBplus); scp->dirBplus = NULL; scp->dirDataVersion = -1; lock_FinalizeRWLock(&scp->dirlock); #endif lock_FinalizeMutex(&scp->mx); lock_FinalizeRWLock(&scp->bufCreateLock); } lock_ReleaseWrite(&cm_scacheLock); cm_GiveUpAllCallbacksAllServers(FALSE); return cm_dnlcShutdown(); } void cm_InitSCache(int newFile, long maxSCaches) { static osi_once_t once; if (osi_Once(&once)) { lock_InitializeRWLock(&cm_scacheLock, "cm_scacheLock"); if ( newFile ) { memset(cm_data.scacheHashTablep, 0, sizeof(cm_scache_t *) * cm_data.scacheHashTableSize); cm_data.allSCachesp = NULL; cm_data.currentSCaches = 0; cm_data.maxSCaches = maxSCaches; cm_data.scacheLRUFirstp = cm_data.scacheLRULastp = NULL; } else { cm_scache_t * scp; for ( scp = cm_data.allSCachesp; scp; scp = scp->allNextp ) { lock_InitializeMutex(&scp->mx, "cm_scache_t mutex"); lock_InitializeRWLock(&scp->bufCreateLock, "cm_scache_t bufCreateLock"); #ifdef USE_BPLUS lock_InitializeRWLock(&scp->dirlock, "cm_scache_t dirlock"); #endif scp->cbServerp = NULL; scp->cbExpires = 0; scp->fileLocksH = NULL; scp->fileLocksT = NULL; scp->serverLock = (-1); scp->lastRefreshCycle = 0; scp->exclusiveLocks = 0; scp->sharedLocks = 0; scp->openReads = 0; scp->openWrites = 0; scp->openShares = 0; scp->openExcls = 0; scp->waitCount = 0; #ifdef USE_BPLUS scp->dirBplus = NULL; scp->dirDataVersion = -1; #endif scp->waitQueueT = NULL; scp->flags &= ~CM_SCACHEFLAG_WAITING; } } cm_allFileLocks = NULL; cm_freeFileLocks = NULL; cm_lockRefreshCycle = 0; cm_fakeSCacheInit(newFile); cm_allFreeWaiters = NULL; cm_dnlcInit(newFile); osi_EndOnce(&once); } } /* version that doesn't bother creating the entry if we don't find it */ cm_scache_t *cm_FindSCache(cm_fid_t *fidp) { long hash; cm_scache_t *scp; hash = CM_SCACHE_HASH(fidp); if (fidp->cell == 0) { return NULL; } lock_ObtainWrite(&cm_scacheLock); for (scp=cm_data.scacheHashTablep[hash]; scp; scp=scp->nextp) { if (cm_FidCmp(fidp, &scp->fid) == 0) { cm_HoldSCacheNoLock(scp); cm_AdjustScacheLRU(scp); lock_ReleaseWrite(&cm_scacheLock); return scp; } } lock_ReleaseWrite(&cm_scacheLock); return NULL; } #ifdef DEBUG_REFCOUNT long cm_GetSCacheDbg(cm_fid_t *fidp, cm_scache_t **outScpp, cm_user_t *userp, cm_req_t *reqp, char * file, long line) #else long cm_GetSCache(cm_fid_t *fidp, cm_scache_t **outScpp, cm_user_t *userp, cm_req_t *reqp) #endif { long hash; cm_scache_t *scp; long code; cm_volume_t *volp = NULL; cm_cell_t *cellp; char* mp = NULL; int special; // yj: boolean variable to test if file is on root.afs int isRoot; extern cm_fid_t cm_rootFid; hash = CM_SCACHE_HASH(fidp); osi_assertx(fidp->cell != 0, "unassigned cell value"); if (fidp->cell== cm_data.rootFid.cell && fidp->volume==cm_data.rootFid.volume && fidp->vnode==0x0 && fidp->unique==0x0) { osi_Log0(afsd_logp,"cm_GetSCache called with root cell/volume and vnode=0 and unique=0"); } // yj: check if we have the scp, if so, we don't need // to do anything else lock_ObtainWrite(&cm_scacheLock); for (scp=cm_data.scacheHashTablep[hash]; scp; scp=scp->nextp) { if (cm_FidCmp(fidp, &scp->fid) == 0) { #ifdef DEBUG_REFCOUNT afsi_log("%s:%d cm_GetSCache (1) outScpp 0x%p ref %d", file, line, scp, scp->refCount); osi_Log1(afsd_logp,"cm_GetSCache (1) outScpp 0x%p", scp); #endif cm_HoldSCacheNoLock(scp); *outScpp = scp; cm_AdjustScacheLRU(scp); lock_ReleaseWrite(&cm_scacheLock); return 0; } } // yj: when we get here, it means we don't have an scp // so we need to either load it or fake it, depending // on whether the file is "special", see below. // yj: if we're trying to get an scp for a file that's // on root.afs of homecell, we want to handle it specially // because we have to fill in the status stuff 'coz we // don't want trybulkstat to fill it in for us #ifdef AFS_FREELANCE_CLIENT special = (fidp->cell==AFS_FAKE_ROOT_CELL_ID && fidp->volume==AFS_FAKE_ROOT_VOL_ID && !(fidp->vnode==0x1 && fidp->unique==0x1)); isRoot = (fidp->cell==AFS_FAKE_ROOT_CELL_ID && fidp->volume==AFS_FAKE_ROOT_VOL_ID && fidp->vnode==0x1 && fidp->unique==0x1); if (cm_freelanceEnabled && isRoot) { osi_Log0(afsd_logp,"cm_GetSCache Freelance and isRoot"); /* freelance: if we are trying to get the root scp for the first * time, we will just put in a place holder entry. */ volp = NULL; } if (cm_freelanceEnabled && special) { osi_Log0(afsd_logp,"cm_GetSCache Freelance and special"); if (fidp->vnode > 1 && fidp->vnode <= cm_noLocalMountPoints + 2) { lock_ObtainMutex(&cm_Freelance_Lock); mp =(cm_localMountPoints+fidp->vnode-2)->mountPointStringp; lock_ReleaseMutex(&cm_Freelance_Lock); } else { mp = ""; } scp = cm_GetNewSCache(); if (scp == NULL) { osi_Log0(afsd_logp,"cm_GetSCache unable to obtain *new* scache entry"); lock_ReleaseWrite(&cm_scacheLock); return CM_ERROR_WOULDBLOCK; } #if not_too_dangerous /* dropping the cm_scacheLock allows more than one thread * to obtain the same cm_scache_t from the LRU list. Since * the refCount is known to be zero at this point we have to * assume that no one else is using the one this is returned. */ lock_ReleaseWrite(&cm_scacheLock); lock_ObtainMutex(&scp->mx); lock_ObtainWrite(&cm_scacheLock); #endif scp->fid = *fidp; scp->volp = cm_data.rootSCachep->volp; cm_GetVolume(scp->volp); /* grab an additional reference */ scp->dotdotFid.cell=AFS_FAKE_ROOT_CELL_ID; scp->dotdotFid.volume=AFS_FAKE_ROOT_VOL_ID; scp->dotdotFid.unique=1; scp->dotdotFid.vnode=1; scp->flags |= (CM_SCACHEFLAG_PURERO | CM_SCACHEFLAG_RO); scp->nextp=cm_data.scacheHashTablep[hash]; cm_data.scacheHashTablep[hash]=scp; scp->flags |= CM_SCACHEFLAG_INHASH; scp->refCount = 1; osi_Log1(afsd_logp,"cm_GetSCache (freelance) sets refCount to 1 scp 0x%x", scp); if (fidp->vnode > 1 && fidp->vnode <= cm_noLocalMountPoints + 2) scp->fileType = (cm_localMountPoints+fidp->vnode-2)->fileType; else scp->fileType = CM_SCACHETYPE_INVALID; lock_ObtainMutex(&cm_Freelance_Lock); scp->length.LowPart = (DWORD)strlen(mp)+4; scp->length.HighPart = 0; strncpy(scp->mountPointStringp,mp,MOUNTPOINTLEN); scp->mountPointStringp[MOUNTPOINTLEN-1] = '\0'; lock_ReleaseMutex(&cm_Freelance_Lock); scp->owner=0x0; scp->unixModeBits=0777; scp->clientModTime=FakeFreelanceModTime; scp->serverModTime=FakeFreelanceModTime; scp->parentUnique = 0x1; scp->parentVnode=0x1; scp->group=0; scp->dataVersion=cm_data.fakeDirVersion; scp->lockDataVersion=-1; /* no lock yet */ #if not_too_dangerous lock_ReleaseMutex(&scp->mx); #endif *outScpp = scp; lock_ReleaseWrite(&cm_scacheLock); #ifdef DEBUG_REFCOUNT afsi_log("%s:%d cm_GetSCache (2) outScpp 0x%p ref %d", file, line, scp, scp->refCount); osi_Log1(afsd_logp,"cm_GetSCache (2) outScpp 0x%p", scp); #endif return 0; } // end of yj code #endif /* AFS_FREELANCE_CLIENT */ /* otherwise, we need to find the volume */ if (!cm_freelanceEnabled || !isRoot) { lock_ReleaseWrite(&cm_scacheLock); /* for perf. reasons */ cellp = cm_FindCellByID(fidp->cell); if (!cellp) return CM_ERROR_NOSUCHCELL; code = cm_GetVolumeByID(cellp, fidp->volume, userp, reqp, CM_GETVOL_FLAG_CREATE, &volp); if (code) return code; lock_ObtainWrite(&cm_scacheLock); } /* otherwise, we have the volume, now reverify that the scp doesn't * exist, and proceed. */ for (scp=cm_data.scacheHashTablep[hash]; scp; scp=scp->nextp) { if (cm_FidCmp(fidp, &scp->fid) == 0) { #ifdef DEBUG_REFCOUNT afsi_log("%s:%d cm_GetSCache (3) outScpp 0x%p ref %d", file, line, scp, scp->refCount); osi_Log1(afsd_logp,"cm_GetSCache (3) outScpp 0x%p", scp); #endif cm_HoldSCacheNoLock(scp); osi_assertx(scp->volp == volp, "cm_scache_t volume has unexpected value"); cm_AdjustScacheLRU(scp); lock_ReleaseWrite(&cm_scacheLock); if (volp) cm_PutVolume(volp); *outScpp = scp; return 0; } } /* now, if we don't have the fid, recycle something */ scp = cm_GetNewSCache(); if (scp == NULL) { osi_Log0(afsd_logp,"cm_GetNewSCache unable to obtain *new* scache entry"); lock_ReleaseWrite(&cm_scacheLock); if (volp) cm_PutVolume(volp); return CM_ERROR_WOULDBLOCK; } osi_Log2(afsd_logp,"cm_GetNewSCache returns scp 0x%x flags 0x%x", scp, scp->flags); osi_assertx(!(scp->flags & CM_SCACHEFLAG_INHASH), "CM_SCACHEFLAG_INHASH set"); #if not_too_dangerous /* dropping the cm_scacheLock allows more than one thread * to obtain the same cm_scache_t from the LRU list. Since * the refCount is known to be zero at this point we have to * assume that no one else is using the one this is returned. */ lock_ReleaseWrite(&cm_scacheLock); lock_ObtainMutex(&scp->mx); lock_ObtainWrite(&cm_scacheLock); #endif scp->fid = *fidp; scp->volp = volp; /* a held reference */ if (!cm_freelanceEnabled || !isRoot) { /* if this scache entry represents a volume root then we need * to copy the dotdotFipd from the volume structure where the * "master" copy is stored (defect 11489) */ if (scp->fid.vnode == 1 && scp->fid.unique == 1) { scp->dotdotFid = volp->dotdotFid; } if (volp->ro.ID == fidp->volume) scp->flags |= (CM_SCACHEFLAG_PURERO | CM_SCACHEFLAG_RO); else if (volp->bk.ID == fidp->volume) scp->flags |= CM_SCACHEFLAG_RO; } scp->nextp = cm_data.scacheHashTablep[hash]; cm_data.scacheHashTablep[hash] = scp; scp->flags |= CM_SCACHEFLAG_INHASH; scp->refCount = 1; osi_Log1(afsd_logp,"cm_GetSCache sets refCount to 1 scp 0x%x", scp); #if not_too_dangerous lock_ReleaseMutex(&scp->mx); #endif /* XXX - The following fields in the cm_scache are * uninitialized: * fileType * parentVnode * parentUnique */ lock_ReleaseWrite(&cm_scacheLock); /* now we have a held scache entry; just return it */ *outScpp = scp; #ifdef DEBUG_REFCOUNT afsi_log("%s:%d cm_GetSCache (4) outScpp 0x%p ref %d", file, line, scp, scp->refCount); osi_Log1(afsd_logp,"cm_GetSCache (4) outScpp 0x%p", scp); #endif return 0; } /* Returns a held reference to the scache's parent * if it exists */ cm_scache_t * cm_FindSCacheParent(cm_scache_t * scp) { long code = 0; int i; cm_fid_t parent_fid; cm_scache_t * pscp = NULL; lock_ObtainRead(&cm_scacheLock); parent_fid = scp->fid; parent_fid.vnode = scp->parentVnode; parent_fid.unique = scp->parentUnique; if (cm_FidCmp(&scp->fid, &parent_fid)) { i = CM_SCACHE_HASH(&parent_fid); for (pscp = cm_data.scacheHashTablep[i]; pscp; pscp = pscp->nextp) { if (!cm_FidCmp(&pscp->fid, &parent_fid)) { cm_HoldSCacheNoLock(pscp); break; } } } lock_ReleaseRead(&cm_scacheLock); return pscp; } void cm_SyncOpAddToWaitQueue(cm_scache_t * scp, afs_int32 flags, cm_buf_t * bufp) { cm_scache_waiter_t * w; lock_ObtainWrite(&cm_scacheLock); if (cm_allFreeWaiters == NULL) { w = malloc(sizeof(*w)); memset(w, 0, sizeof(*w)); } else { w = (cm_scache_waiter_t *) cm_allFreeWaiters; osi_QRemove(&cm_allFreeWaiters, (osi_queue_t *) w); } w->threadId = thrd_Current(); w->scp = scp; cm_HoldSCacheNoLock(scp); w->flags = flags; w->bufp = bufp; osi_QAddT(&scp->waitQueueH, &scp->waitQueueT, (osi_queue_t *) w); lock_ReleaseWrite(&cm_scacheLock); osi_Log2(afsd_logp, "cm_SyncOpAddToWaitQueue : Adding thread to wait queue scp 0x%p w 0x%p", scp, w); } int cm_SyncOpCheckContinue(cm_scache_t * scp, afs_int32 flags, cm_buf_t * bufp) { cm_scache_waiter_t * w; int this_is_me; osi_Log0(afsd_logp, "cm_SyncOpCheckContinue checking for continuation"); lock_ObtainRead(&cm_scacheLock); for (w = (cm_scache_waiter_t *)scp->waitQueueH; w; w = (cm_scache_waiter_t *)osi_QNext((osi_queue_t *) w)) { if (w->flags == flags && w->bufp == bufp) { break; } } osi_assertx(w != NULL, "null cm_scache_waiter_t"); this_is_me = (w->threadId == thrd_Current()); lock_ReleaseRead(&cm_scacheLock); if (!this_is_me) { osi_Log1(afsd_logp, "cm_SyncOpCheckContinue MISS: Waiter 0x%p", w); return 0; } osi_Log1(afsd_logp, "cm_SyncOpCheckContinue HIT: Waiter 0x%p", w); lock_ObtainWrite(&cm_scacheLock); osi_QRemoveHT(&scp->waitQueueH, &scp->waitQueueT, (osi_queue_t *) w); cm_ReleaseSCacheNoLock(scp); memset(w, 0, sizeof(*w)); osi_QAdd(&cm_allFreeWaiters, (osi_queue_t *) w); lock_ReleaseWrite(&cm_scacheLock); return 1; } /* synchronize a fetch, store, read, write, fetch status or store status. * Called with scache mutex held, and returns with it held, but temporarily * drops it during the fetch. * * At most one flag can be on in flags, if this is an RPC request. * * Also, if we're fetching or storing data, we must ensure that we have a buffer. * * There are a lot of weird restrictions here; here's an attempt to explain the * rationale for the concurrency restrictions implemented in this function. * * First, although the file server will break callbacks when *another* machine * modifies a file or status block, the client itself is responsible for * concurrency control on its own requests. Callback breaking events are rare, * and simply invalidate any concurrent new status info. * * In the absence of callback breaking messages, we need to know how to * synchronize incoming responses describing updates to files. We synchronize * operations that update the data version by comparing the data versions. * However, updates that do not update the data, but only the status, can't be * synchronized with fetches or stores, since there's nothing to compare * to tell which operation executed first at the server. * * Thus, we can allow multiple ops that change file data, or dir data, and * fetches. However, status storing ops have to be done serially. * * Furthermore, certain data-changing ops are incompatible: we can't read or * write a buffer while doing a truncate. We can't read and write the same * buffer at the same time, or write while fetching or storing, or read while * fetching a buffer (this may change). We can't fetch and store at the same * time, either. * * With respect to status, we can't read and write at the same time, read while * fetching, write while fetching or storing, or fetch and store at the same time. * * We can't allow a get callback RPC to run in concurrently with something that * will return updated status, since we could start a call, have the server * return status, have another machine make an update to the status (which * doesn't change serverModTime), have the original machine get a new callback, * and then have the original machine merge in the early, old info from the * first call. At this point, the easiest way to avoid this problem is to have * getcallback calls conflict with all others for the same vnode. Other calls * to cm_MergeStatus that aren't associated with calls to cm_SyncOp on the same * vnode must be careful not to merge in their status unless they have obtained * a callback from the start of their call. * * Note added 1/23/96 * Concurrent StoreData RPC's can cause trouble if the file is being extended. * Each such RPC passes a FileLength parameter, which the server uses to do * pre-truncation if necessary. So if two RPC's are processed out of order at * the server, the one with the smaller FileLength will be processed last, * possibly resulting in a bogus truncation. The simplest way to avoid this * is to serialize all StoreData RPC's. This is the reason we defined * CM_SCACHESYNC_STOREDATA_EXCL and CM_SCACHEFLAG_DATASTORING. */ long cm_SyncOp(cm_scache_t *scp, cm_buf_t *bufp, cm_user_t *userp, cm_req_t *reqp, afs_uint32 rights, afs_uint32 flags) { osi_queueData_t *qdp; long code; cm_buf_t *tbufp; afs_uint32 outRights; int bufLocked; afs_uint32 sleep_scp_flags = 0; afs_uint32 sleep_buf_cmflags = 0; afs_uint32 sleep_scp_bufs = 0; int wakeupCycle; /* lookup this first */ bufLocked = flags & CM_SCACHESYNC_BUFLOCKED; if (bufp) osi_assertx(bufp->refCount > 0, "cm_buf_t refCount 0"); /* Do the access check. Now we don't really do the access check * atomically, since the caller doesn't expect the parent dir to be * returned locked, and that is what we'd have to do to prevent a * callback breaking message on the parent due to a setacl call from * being processed while we're running. So, instead, we check things * here, and if things look fine with the access, we proceed to finish * the rest of this check. Sort of a hack, but probably good enough. */ while (1) { if (flags & CM_SCACHESYNC_FETCHSTATUS) { /* if we're bringing in a new status block, ensure that * we aren't already doing so, and that no one is * changing the status concurrently, either. We need * to do this, even if the status is of a different * type, since we don't have the ability to figure out, * in the AFS 3 protocols, which status-changing * operation ran first, or even which order a read and * a write occurred in. */ if (scp->flags & (CM_SCACHEFLAG_FETCHING | CM_SCACHEFLAG_STORING | CM_SCACHEFLAG_SIZESTORING | CM_SCACHEFLAG_GETCALLBACK)) { osi_Log1(afsd_logp, "CM SyncOp scp 0x%p is FETCHING|STORING|SIZESTORING|GETCALLBACK want FETCHSTATUS", scp); goto sleep; } } if (flags & (CM_SCACHESYNC_STORESIZE | CM_SCACHESYNC_STORESTATUS | CM_SCACHESYNC_SETSIZE | CM_SCACHESYNC_GETCALLBACK)) { /* if we're going to make an RPC to change the status, make sure * that no one is bringing in or sending out the status. */ if (scp->flags & (CM_SCACHEFLAG_FETCHING | CM_SCACHEFLAG_STORING | CM_SCACHEFLAG_SIZESTORING | CM_SCACHEFLAG_GETCALLBACK)) { osi_Log1(afsd_logp, "CM SyncOp scp 0x%p is FETCHING|STORING|SIZESTORING|GETCALLBACK want STORESIZE|STORESTATUS|SETSIZE|GETCALLBACK", scp); goto sleep; } if (scp->bufReadsp || scp->bufWritesp) { osi_Log1(afsd_logp, "CM SyncOp scp 0x%p is bufRead|bufWrite want STORESIZE|STORESTATUS|SETSIZE|GETCALLBACK", scp); goto sleep; } } if (flags & CM_SCACHESYNC_FETCHDATA) { /* if we're bringing in a new chunk of data, make sure that * nothing is happening to that chunk, and that we aren't * changing the basic file status info, either. */ if (scp->flags & (CM_SCACHEFLAG_FETCHING | CM_SCACHEFLAG_STORING | CM_SCACHEFLAG_SIZESTORING | CM_SCACHEFLAG_GETCALLBACK)) { osi_Log1(afsd_logp, "CM SyncOp scp 0x%p is FETCHING|STORING|SIZESTORING|GETCALLBACK want FETCHDATA", scp); goto sleep; } if (bufp && (bufp->cmFlags & (CM_BUF_CMFETCHING | CM_BUF_CMSTORING | CM_BUF_CMWRITING))) { osi_Log2(afsd_logp, "CM SyncOp scp 0x%p bufp 0x%p is BUF_CMFETCHING|BUF_CMSTORING|BUF_CMWRITING want FETCHDATA", scp, bufp); goto sleep; } } if (flags & CM_SCACHESYNC_STOREDATA) { /* same as fetch data */ if (scp->flags & (CM_SCACHEFLAG_FETCHING | CM_SCACHEFLAG_STORING | CM_SCACHEFLAG_SIZESTORING | CM_SCACHEFLAG_GETCALLBACK)) { osi_Log1(afsd_logp, "CM SyncOp scp 0x%p is FETCHING|STORING|SIZESTORING|GETCALLBACK want STOREDATA", scp); goto sleep; } if (bufp && (bufp->cmFlags & (CM_BUF_CMFETCHING | CM_BUF_CMSTORING | CM_BUF_CMWRITING))) { osi_Log2(afsd_logp, "CM SyncOp scp 0x%p bufp 0x%p is BUF_CMFETCHING|BUF_CMSTORING|BUF_CMWRITING want STOREDATA", scp, bufp); goto sleep; } } if (flags & CM_SCACHESYNC_STOREDATA_EXCL) { /* Don't allow concurrent StoreData RPC's */ if (scp->flags & CM_SCACHEFLAG_DATASTORING) { osi_Log1(afsd_logp, "CM SyncOp scp 0x%p is DATASTORING want STOREDATA_EXCL", scp); goto sleep; } } if (flags & CM_SCACHESYNC_ASYNCSTORE) { /* Don't allow more than one BKG store request */ if (scp->flags & CM_SCACHEFLAG_ASYNCSTORING) { osi_Log1(afsd_logp, "CM SyncOp scp 0x%p is ASYNCSTORING want ASYNCSTORE", scp); goto sleep; } } if (flags & CM_SCACHESYNC_LOCK) { /* Don't allow concurrent fiddling with lock lists */ if (scp->flags & CM_SCACHEFLAG_LOCKING) { osi_Log1(afsd_logp, "CM SyncOp scp 0x%p is LOCKING want LOCK", scp); goto sleep; } } /* now the operations that don't correspond to making RPCs */ if (flags & CM_SCACHESYNC_GETSTATUS) { /* we can use the status that's here, if we're not * bringing in new status. */ if (scp->flags & (CM_SCACHEFLAG_FETCHING)) { osi_Log1(afsd_logp, "CM SyncOp scp 0x%p is FETCHING want GETSTATUS", scp); goto sleep; } } if (flags & CM_SCACHESYNC_SETSTATUS) { /* we can make a change to the local status, as long as * the status isn't changing now. * * If we're fetching or storing a chunk of data, we can * change the status locally, since the fetch/store * operations don't change any of the data that we're * changing here. */ if (scp->flags & (CM_SCACHEFLAG_FETCHING | CM_SCACHEFLAG_STORING | CM_SCACHEFLAG_SIZESTORING)) { osi_Log1(afsd_logp, "CM SyncOp scp 0x%p is FETCHING|STORING|SIZESTORING want SETSTATUS", scp); goto sleep; } } if (flags & CM_SCACHESYNC_READ) { /* we're going to read the data, make sure that the * status is available, and that the data is here. It * is OK to read while storing the data back. */ if (scp->flags & CM_SCACHEFLAG_FETCHING) { osi_Log1(afsd_logp, "CM SyncOp scp 0x%p is FETCHING want READ", scp); goto sleep; } if (bufp && ((bufp->cmFlags & (CM_BUF_CMFETCHING | CM_BUF_CMFULLYFETCHED)) == CM_BUF_CMFETCHING)) { osi_Log2(afsd_logp, "CM SyncOp scp 0x%p bufp 0x%p is BUF_CMFETCHING want READ", scp, bufp); goto sleep; } if (bufp && (bufp->cmFlags & CM_BUF_CMWRITING)) { osi_Log2(afsd_logp, "CM SyncOp scp 0x%p bufp 0x%p is BUF_CMWRITING want READ", scp, bufp); goto sleep; } } if (flags & CM_SCACHESYNC_WRITE) { /* don't write unless the status is stable and the chunk * is stable. */ if (scp->flags & (CM_SCACHEFLAG_FETCHING | CM_SCACHEFLAG_STORING | CM_SCACHEFLAG_SIZESTORING)) { osi_Log1(afsd_logp, "CM SyncOp scp 0x%p is FETCHING|STORING|SIZESTORING want WRITE", scp); goto sleep; } if (bufp && (bufp->cmFlags & (CM_BUF_CMFETCHING | CM_BUF_CMSTORING | CM_BUF_CMWRITING))) { osi_Log3(afsd_logp, "CM SyncOp scp 0x%p bufp 0x%p is %s want WRITE", scp, bufp, ((bufp->cmFlags & CM_BUF_CMFETCHING) ? "CM_BUF_CMFETCHING": ((bufp->cmFlags & CM_BUF_CMSTORING) ? "CM_BUF_CMSTORING" : ((bufp->cmFlags & CM_BUF_CMWRITING) ? "CM_BUF_CMWRITING" : "UNKNOWN!!!")))); goto sleep; } } // yj: modified this so that callback only checked if we're // not checking something on /afs /* fix the conditional to match the one in cm_HaveCallback */ if ((flags & CM_SCACHESYNC_NEEDCALLBACK) #ifdef AFS_FREELANCE_CLIENT && (!cm_freelanceEnabled || !(scp->fid.vnode==0x1 && scp->fid.unique==0x1) || scp->fid.cell!=AFS_FAKE_ROOT_CELL_ID || scp->fid.volume!=AFS_FAKE_ROOT_VOL_ID || cm_fakeDirCallback < 2) #endif /* AFS_FREELANCE_CLIENT */ ) { if ((flags & CM_SCACHESYNC_FORCECB) || !cm_HaveCallback(scp)) { osi_Log1(afsd_logp, "CM SyncOp getting callback on scp 0x%p", scp); if (bufLocked) lock_ReleaseMutex(&bufp->mx); code = cm_GetCallback(scp, userp, reqp, (flags & CM_SCACHESYNC_FORCECB)?1:0); if (bufLocked) { lock_ReleaseMutex(&scp->mx); lock_ObtainMutex(&bufp->mx); lock_ObtainMutex(&scp->mx); } if (code) return code; flags &= ~CM_SCACHESYNC_FORCECB; /* only force once */ continue; } } if (rights) { /* can't check access rights without a callback */ osi_assertx(flags & CM_SCACHESYNC_NEEDCALLBACK, "!CM_SCACHESYNC_NEEDCALLBACK"); if ((rights & PRSFS_WRITE) && (scp->flags & CM_SCACHEFLAG_RO)) return CM_ERROR_READONLY; if (cm_HaveAccessRights(scp, userp, rights, &outRights)) { if (~outRights & rights) return CM_ERROR_NOACCESS; } else { /* we don't know the required access rights */ if (bufLocked) lock_ReleaseMutex(&bufp->mx); code = cm_GetAccessRights(scp, userp, reqp); if (bufLocked) { lock_ReleaseMutex(&scp->mx); lock_ObtainMutex(&bufp->mx); lock_ObtainMutex(&scp->mx); } if (code) return code; continue; } } /* if we get here, we're happy */ break; sleep: /* first check if we're not supposed to wait: fail * in this case, returning with everything still locked. */ if (flags & CM_SCACHESYNC_NOWAIT) return CM_ERROR_WOULDBLOCK; /* These are used for minidump debugging */ sleep_scp_flags = scp->flags; /* so we know why we slept */ sleep_buf_cmflags = bufp ? bufp->cmFlags : 0; sleep_scp_bufs = (scp->bufReadsp ? 1 : 0) | (scp->bufWritesp ? 2 : 0); /* wait here, then try again */ osi_Log1(afsd_logp, "CM SyncOp sleeping scp 0x%p", scp); if ( scp->flags & CM_SCACHEFLAG_WAITING ) { scp->waitCount++; scp->waitRequests++; osi_Log3(afsd_logp, "CM SyncOp CM_SCACHEFLAG_WAITING already set for 0x%p; %d threads; %d requests", scp, scp->waitCount, scp->waitRequests); } else { osi_Log1(afsd_logp, "CM SyncOp CM_SCACHEFLAG_WAITING set for 0x%p", scp); scp->flags |= CM_SCACHEFLAG_WAITING; scp->waitCount = scp->waitRequests = 1; } if (bufLocked) lock_ReleaseMutex(&bufp->mx); cm_SyncOpAddToWaitQueue(scp, flags, bufp); wakeupCycle = 0; do { if (wakeupCycle++ != 0) lock_ObtainMutex(&scp->mx); osi_SleepM((LONG_PTR) &scp->flags, &scp->mx); } while (!cm_SyncOpCheckContinue(scp, flags, bufp)); smb_UpdateServerPriority(); if (bufLocked) lock_ObtainMutex(&bufp->mx); lock_ObtainMutex(&scp->mx); scp->waitCount--; osi_Log3(afsd_logp, "CM SyncOp woke! scp 0x%p; still waiting %d threads of %d requests", scp, scp->waitCount, scp->waitRequests); if (scp->waitCount == 0) { osi_Log1(afsd_logp, "CM SyncOp CM_SCACHEFLAG_WAITING reset for 0x%p", scp); scp->flags &= ~CM_SCACHEFLAG_WAITING; scp->waitRequests = 0; } } /* big while loop */ /* now, update the recorded state for RPC-type calls */ if (flags & CM_SCACHESYNC_FETCHSTATUS) scp->flags |= CM_SCACHEFLAG_FETCHING; if (flags & CM_SCACHESYNC_STORESTATUS) scp->flags |= CM_SCACHEFLAG_STORING; if (flags & CM_SCACHESYNC_STORESIZE) scp->flags |= CM_SCACHEFLAG_SIZESTORING; if (flags & CM_SCACHESYNC_GETCALLBACK) scp->flags |= CM_SCACHEFLAG_GETCALLBACK; if (flags & CM_SCACHESYNC_STOREDATA_EXCL) scp->flags |= CM_SCACHEFLAG_DATASTORING; if (flags & CM_SCACHESYNC_ASYNCSTORE) scp->flags |= CM_SCACHEFLAG_ASYNCSTORING; if (flags & CM_SCACHESYNC_LOCK) scp->flags |= CM_SCACHEFLAG_LOCKING; /* now update the buffer pointer */ if (flags & CM_SCACHESYNC_FETCHDATA) { /* ensure that the buffer isn't already in the I/O list */ if (bufp) { for(qdp = scp->bufReadsp; qdp; qdp = (osi_queueData_t *) osi_QNext(&qdp->q)) { tbufp = osi_GetQData(qdp); osi_assertx(tbufp != bufp, "unexpected cm_buf_t value"); } } /* queue a held reference to the buffer in the "reading" I/O list */ qdp = osi_QDAlloc(); osi_SetQData(qdp, bufp); if (bufp) { buf_Hold(bufp); bufp->cmFlags |= CM_BUF_CMFETCHING; } osi_QAdd((osi_queue_t **) &scp->bufReadsp, &qdp->q); } if (flags & CM_SCACHESYNC_STOREDATA) { /* ensure that the buffer isn't already in the I/O list */ if (bufp) { for(qdp = scp->bufWritesp; qdp; qdp = (osi_queueData_t *) osi_QNext(&qdp->q)) { tbufp = osi_GetQData(qdp); osi_assertx(tbufp != bufp, "unexpected cm_buf_t value"); } } /* queue a held reference to the buffer in the "writing" I/O list */ qdp = osi_QDAlloc(); osi_SetQData(qdp, bufp); if (bufp) { buf_Hold(bufp); bufp->cmFlags |= CM_BUF_CMSTORING; } osi_QAdd((osi_queue_t **) &scp->bufWritesp, &qdp->q); } if (flags & CM_SCACHESYNC_WRITE) { /* mark the buffer as being written to. */ if (bufp) { bufp->cmFlags |= CM_BUF_CMWRITING; } } return 0; } /* for those syncops that setup for RPCs. * Called with scache locked. */ void cm_SyncOpDone(cm_scache_t *scp, cm_buf_t *bufp, afs_uint32 flags) { osi_queueData_t *qdp; cm_buf_t *tbufp; lock_AssertMutex(&scp->mx); /* now, update the recorded state for RPC-type calls */ if (flags & CM_SCACHESYNC_FETCHSTATUS) scp->flags &= ~CM_SCACHEFLAG_FETCHING; if (flags & CM_SCACHESYNC_STORESTATUS) scp->flags &= ~CM_SCACHEFLAG_STORING; if (flags & CM_SCACHESYNC_STORESIZE) scp->flags &= ~CM_SCACHEFLAG_SIZESTORING; if (flags & CM_SCACHESYNC_GETCALLBACK) scp->flags &= ~CM_SCACHEFLAG_GETCALLBACK; if (flags & CM_SCACHESYNC_STOREDATA_EXCL) scp->flags &= ~CM_SCACHEFLAG_DATASTORING; if (flags & CM_SCACHESYNC_ASYNCSTORE) scp->flags &= ~CM_SCACHEFLAG_ASYNCSTORING; if (flags & CM_SCACHESYNC_LOCK) scp->flags &= ~CM_SCACHEFLAG_LOCKING; /* now update the buffer pointer */ if (flags & CM_SCACHESYNC_FETCHDATA) { int release = 0; /* ensure that the buffer isn't already in the I/O list */ for(qdp = scp->bufReadsp; qdp; qdp = (osi_queueData_t *) osi_QNext(&qdp->q)) { tbufp = osi_GetQData(qdp); if (tbufp == bufp) break; } if (qdp) { osi_QRemove((osi_queue_t **) &scp->bufReadsp, &qdp->q); osi_QDFree(qdp); release = 1; } if (bufp) { bufp->cmFlags &= ~(CM_BUF_CMFETCHING | CM_BUF_CMFULLYFETCHED); if (bufp->flags & CM_BUF_WAITING) { osi_Log2(afsd_logp, "CM SyncOpDone Waking [scp 0x%p] bufp 0x%p", scp, bufp); osi_Wakeup((LONG_PTR) &bufp); } if (release) buf_Release(bufp); } } /* now update the buffer pointer */ if (flags & CM_SCACHESYNC_STOREDATA) { int release = 0; /* ensure that the buffer isn't already in the I/O list */ for(qdp = scp->bufWritesp; qdp; qdp = (osi_queueData_t *) osi_QNext(&qdp->q)) { tbufp = osi_GetQData(qdp); if (tbufp == bufp) break; } if (qdp) { osi_QRemove((osi_queue_t **) &scp->bufWritesp, &qdp->q); osi_QDFree(qdp); release = 1; } if (bufp) { bufp->cmFlags &= ~CM_BUF_CMSTORING; if (bufp->flags & CM_BUF_WAITING) { osi_Log2(afsd_logp, "CM SyncOpDone Waking [scp 0x%p] bufp 0x%p", scp, bufp); osi_Wakeup((LONG_PTR) &bufp); } if (release) buf_Release(bufp); } } if (flags & CM_SCACHESYNC_WRITE) { if (bufp) { osi_assertx(bufp->cmFlags & CM_BUF_CMWRITING, "!CM_BUF_CMWRITING"); bufp->cmFlags &= ~CM_BUF_CMWRITING; } } /* and wakeup anyone who is waiting */ if (scp->flags & CM_SCACHEFLAG_WAITING) { osi_Log1(afsd_logp, "CM SyncOpDone Waking scp 0x%p", scp); osi_Wakeup((LONG_PTR) &scp->flags); } } /* merge in a response from an RPC. The scp must be locked, and the callback * is optional. * * Don't overwrite any status info that is dirty, since we could have a store * operation (such as store data) that merges some info in, and we don't want * to lose the local updates. Typically, there aren't many updates we do * locally, anyway, probably only mtime. * * There is probably a bug in here where a chmod (which doesn't change * serverModTime) that occurs between two fetches, both of whose responses are * handled after the callback breaking is done, but only one of whose calls * started before that, can cause old info to be merged from the first call. */ void cm_MergeStatus(cm_scache_t *dscp, cm_scache_t *scp, AFSFetchStatus *statusp, AFSVolSync *volsyncp, cm_user_t *userp, afs_uint32 flags) { // yj: i want to create some fake status for the /afs directory and the // entries under that directory #ifdef AFS_FREELANCE_CLIENT if (cm_freelanceEnabled && scp == cm_data.rootSCachep) { osi_Log0(afsd_logp,"cm_MergeStatus Freelance cm_data.rootSCachep"); statusp->InterfaceVersion = 0x1; statusp->FileType = CM_SCACHETYPE_DIRECTORY; statusp->LinkCount = scp->linkCount; statusp->Length = cm_fakeDirSize; statusp->Length_hi = 0; statusp->DataVersion = cm_data.fakeDirVersion; statusp->Author = 0x1; statusp->Owner = 0x0; statusp->CallerAccess = 0x9; statusp->AnonymousAccess = 0x9; statusp->UnixModeBits = 0777; statusp->ParentVnode = 0x1; statusp->ParentUnique = 0x1; statusp->ResidencyMask = 0; statusp->ClientModTime = FakeFreelanceModTime; statusp->ServerModTime = FakeFreelanceModTime; statusp->Group = 0; statusp->SyncCounter = 0; statusp->dataVersionHigh = 0; statusp->errorCode = 0; } #endif /* AFS_FREELANCE_CLIENT */ if (statusp->errorCode != 0) { scp->flags |= CM_SCACHEFLAG_EACCESS; osi_Log2(afsd_logp, "Merge, Failure scp %x code 0x%x", scp, statusp->errorCode); scp->fileType = 0; /* unknown */ scp->serverModTime = 0; scp->clientModTime = 0; scp->length.LowPart = 0; scp->length.HighPart = 0; scp->serverLength.LowPart = 0; scp->serverLength.HighPart = 0; scp->linkCount = 0; scp->owner = 0; scp->group = 0; scp->unixModeBits = 0; scp->anyAccess = 0; scp->dataVersion = 0; if (dscp) { scp->parentVnode = dscp->fid.vnode; scp->parentUnique = dscp->fid.unique; } else { scp->parentVnode = 0; scp->parentUnique = 0; } return; } else { scp->flags &= ~CM_SCACHEFLAG_EACCESS; } if (!(flags & CM_MERGEFLAG_FORCE) && statusp->DataVersion < (unsigned long) scp->dataVersion) { struct cm_cell *cellp; cellp = cm_FindCellByID(scp->fid.cell); if (scp->cbServerp) { struct cm_volume *volp = NULL; cm_GetVolumeByID(cellp, scp->fid.volume, userp, (cm_req_t *) NULL, CM_GETVOL_FLAG_CREATE, &volp); osi_Log2(afsd_logp, "old data from server %x volume %s", scp->cbServerp->addr.sin_addr.s_addr, volp ? volp->namep : "(unknown)"); if (volp) cm_PutVolume(volp); } osi_Log3(afsd_logp, "Bad merge, scp %x, scp dv %d, RPC dv %d", scp, scp->dataVersion, statusp->DataVersion); /* we have a number of data fetch/store operations running * concurrently, and we can tell which one executed last at the * server by its mtime. * Choose the one with the largest mtime, and ignore the rest. * * These concurrent calls are incompatible with setting the * mtime, so we won't have a locally changed mtime here. * * We could also have ACL info for a different user than usual, * in which case we have to do that part of the merge, anyway. * We won't have to worry about the info being old, since we * won't have concurrent calls * that change file status running from this machine. * * Added 3/17/98: if we see data version regression on an RO * file, it's probably due to a server holding an out-of-date * replica, rather than to concurrent RPC's. Failures to * release replicas are now flagged by the volserver, but only * since AFS 3.4 5.22, so there are plenty of clients getting * out-of-date replicas out there. * * If we discover an out-of-date replica, by this time it's too * late to go to another server and retry. Also, we can't * reject the merge, because then there is no way for * GetAccess to do its work, and the caller gets into an * infinite loop. So we just grin and bear it. */ if (!(scp->flags & CM_SCACHEFLAG_RO)) return; } scp->serverModTime = statusp->ServerModTime; if (!(scp->mask & CM_SCACHEMASK_CLIENTMODTIME)) { scp->clientModTime = statusp->ClientModTime; } if (!(scp->mask & CM_SCACHEMASK_LENGTH)) { scp->length.LowPart = statusp->Length; scp->length.HighPart = statusp->Length_hi; } scp->serverLength.LowPart = statusp->Length; scp->serverLength.HighPart = statusp->Length_hi; scp->linkCount = statusp->LinkCount; scp->owner = statusp->Owner; scp->group = statusp->Group; scp->unixModeBits = statusp->UnixModeBits & 07777; if (statusp->FileType == File) scp->fileType = CM_SCACHETYPE_FILE; else if (statusp->FileType == Directory) scp->fileType = CM_SCACHETYPE_DIRECTORY; else if (statusp->FileType == SymbolicLink) { if ((scp->unixModeBits & 0111) == 0) scp->fileType = CM_SCACHETYPE_MOUNTPOINT; else scp->fileType = CM_SCACHETYPE_SYMLINK; } else { osi_Log2(afsd_logp, "Merge, Invalid File Type (%d), scp %x", statusp->FileType, scp); scp->fileType = CM_SCACHETYPE_INVALID; /* invalid */ } /* and other stuff */ scp->parentVnode = statusp->ParentVnode; scp->parentUnique = statusp->ParentUnique; /* and merge in the private acl cache info, if this is more than the public * info; merge in the public stuff in any case. */ scp->anyAccess = statusp->AnonymousAccess; if (userp != NULL) { cm_AddACLCache(scp, userp, statusp->CallerAccess); } if ((flags & CM_MERGEFLAG_STOREDATA) && statusp->DataVersion - scp->dataVersion == 1) { cm_buf_t *bp; for (bp = cm_data.buf_fileHashTablepp[BUF_FILEHASH(&scp->fid)]; bp; bp=bp->fileHashp) { if (cm_FidCmp(&scp->fid, &bp->fid) == 0 && bp->dataVersion == scp->dataVersion) bp->dataVersion = statusp->DataVersion; } } scp->dataVersion = statusp->DataVersion; } /* note that our stat cache info is incorrect, so force us eventually * to stat the file again. There may be dirty data associated with * this vnode, and we want to preserve that information. * * This function works by simply simulating a loss of the callback. * * This function must be called with the scache locked. */ void cm_DiscardSCache(cm_scache_t *scp) { lock_AssertMutex(&scp->mx); if (scp->cbServerp) { cm_PutServer(scp->cbServerp); scp->cbServerp = NULL; } scp->cbExpires = 0; scp->flags &= ~CM_SCACHEFLAG_CALLBACK; cm_dnlcPurgedp(scp); cm_dnlcPurgevp(scp); cm_FreeAllACLEnts(scp); /* Force mount points and symlinks to be re-evaluated */ scp->mountPointStringp[0] = '\0'; } void cm_AFSFidFromFid(AFSFid *afsFidp, cm_fid_t *fidp) { afsFidp->Volume = fidp->volume; afsFidp->Vnode = fidp->vnode; afsFidp->Unique = fidp->unique; } #ifdef DEBUG_REFCOUNT void cm_HoldSCacheNoLockDbg(cm_scache_t *scp, char * file, long line) #else void cm_HoldSCacheNoLock(cm_scache_t *scp) #endif { osi_assertx(scp != NULL, "null cm_scache_t"); scp->refCount++; #ifdef DEBUG_REFCOUNT osi_Log2(afsd_logp,"cm_HoldSCacheNoLock scp 0x%p ref %d",scp, scp->refCount); afsi_log("%s:%d cm_HoldSCacheNoLock scp 0x%p, ref %d", file, line, scp, scp->refCount); #endif } #ifdef DEBUG_REFCOUNT void cm_HoldSCacheDbg(cm_scache_t *scp, char * file, long line) #else void cm_HoldSCache(cm_scache_t *scp) #endif { osi_assertx(scp != NULL, "null cm_scache_t"); lock_ObtainWrite(&cm_scacheLock); scp->refCount++; #ifdef DEBUG_REFCOUNT osi_Log2(afsd_logp,"cm_HoldSCache scp 0x%p ref %d",scp, scp->refCount); afsi_log("%s:%d cm_HoldSCache scp 0x%p ref %d", file, line, scp, scp->refCount); #endif lock_ReleaseWrite(&cm_scacheLock); } #ifdef DEBUG_REFCOUNT void cm_ReleaseSCacheNoLockDbg(cm_scache_t *scp, char * file, long line) #else void cm_ReleaseSCacheNoLock(cm_scache_t *scp) #endif { osi_assertx(scp != NULL, "null cm_scache_t"); if (scp->refCount == 0) osi_Log1(afsd_logp,"cm_ReleaseSCacheNoLock about to panic scp 0x%x",scp); osi_assertx(scp->refCount-- >= 0, "cm_scache_t refCount 0"); #ifdef DEBUG_REFCOUNT osi_Log2(afsd_logp,"cm_ReleaseSCacheNoLock scp 0x%p ref %d",scp,scp->refCount); afsi_log("%s:%d cm_ReleaseSCacheNoLock scp 0x%p ref %d", file, line, scp, scp->refCount); #endif } #ifdef DEBUG_REFCOUNT void cm_ReleaseSCacheDbg(cm_scache_t *scp, char * file, long line) #else void cm_ReleaseSCache(cm_scache_t *scp) #endif { osi_assertx(scp != NULL, "null cm_scache_t"); lock_ObtainWrite(&cm_scacheLock); if (scp->refCount == 0) osi_Log1(afsd_logp,"cm_ReleaseSCache about to panic scp 0x%x",scp); osi_assertx(scp->refCount != 0, "cm_scache_t refCount 0"); scp->refCount--; #ifdef DEBUG_REFCOUNT osi_Log2(afsd_logp,"cm_ReleaseSCache scp 0x%p ref %d",scp,scp->refCount); afsi_log("%s:%d cm_ReleaseSCache scp 0x%p ref %d", file, line, scp, scp->refCount); #endif lock_ReleaseWrite(&cm_scacheLock); } /* just look for the scp entry to get filetype */ /* doesn't need to be perfectly accurate, so locking doesn't matter too much */ int cm_FindFileType(cm_fid_t *fidp) { long hash; cm_scache_t *scp; hash = CM_SCACHE_HASH(fidp); osi_assertx(fidp->cell != 0, "unassigned cell value"); lock_ObtainWrite(&cm_scacheLock); for (scp=cm_data.scacheHashTablep[hash]; scp; scp=scp->nextp) { if (cm_FidCmp(fidp, &scp->fid) == 0) { lock_ReleaseWrite(&cm_scacheLock); return scp->fileType; } } lock_ReleaseWrite(&cm_scacheLock); return 0; } /* dump all scp's that have reference count > 0 to a file. * cookie is used to identify this batch for easy parsing, * and it a string provided by a caller */ int cm_DumpSCache(FILE *outputFile, char *cookie, int lock) { int zilch; cm_scache_t *scp; char output[2048]; int i; if (lock) lock_ObtainRead(&cm_scacheLock); sprintf(output, "%s - dumping all scache - cm_data.currentSCaches=%d, cm_data.maxSCaches=%d\r\n", cookie, cm_data.currentSCaches, cm_data.maxSCaches); WriteFile(outputFile, output, (DWORD)strlen(output), &zilch, NULL); for (scp = cm_data.allSCachesp; scp; scp = scp->allNextp) { sprintf(output, "%s scp=0x%p, fid (cell=%d, volume=%d, vnode=%d, unique=%d) volp=0x%p type=%d dv=%d len=0x%I64x mp='%s' flags=0x%x cb=0x%x refCount=%u\r\n", cookie, scp, scp->fid.cell, scp->fid.volume, scp->fid.vnode, scp->fid.unique, scp->volp, scp->fileType, scp->dataVersion, scp->length.QuadPart, scp->mountPointStringp, scp->flags, (unsigned long)scp->cbExpires, scp->refCount); WriteFile(outputFile, output, (DWORD)strlen(output), &zilch, NULL); } sprintf(output, "%s - Done dumping all scache.\r\n", cookie); WriteFile(outputFile, output, (DWORD)strlen(output), &zilch, NULL); sprintf(output, "%s - dumping cm_data.scacheHashTable - cm_data.scacheHashTableSize=%d\r\n", cookie, cm_data.scacheHashTableSize); WriteFile(outputFile, output, (DWORD)strlen(output), &zilch, NULL); for (i = 0; i < cm_data.scacheHashTableSize; i++) { for(scp = cm_data.scacheHashTablep[i]; scp; scp=scp->nextp) { sprintf(output, "%s scp=0x%p, hash=%d, fid (cell=%d, volume=%d, vnode=%d, unique=%d)\r\n", cookie, scp, i, scp->fid.cell, scp->fid.volume, scp->fid.vnode, scp->fid.unique); WriteFile(outputFile, output, (DWORD)strlen(output), &zilch, NULL); } } sprintf(output, "%s - Done dumping cm_data.scacheHashTable\r\n", cookie); WriteFile(outputFile, output, (DWORD)strlen(output), &zilch, NULL); if (lock) lock_ReleaseRead(&cm_scacheLock); return (0); }