/* Copyright (C) 1995, 1989 Transarc Corporation - All rights reserved */ /* * For copyright information, see IPL which you accepted in order to * download this software. * */ #include "../afs/param.h" /* Should be always first */ #include "../afs/sysincludes.h" /* Standard vendor system headers */ #include "../afs/afsincludes.h" /* Afs-based standard headers */ #include "../afs/afs_stats.h" /* afs statistics */ #ifdef AFS_AIX_ENV #include /* for vm_att(), vm_det() */ #endif static char memZero; /* address of 0 bytes for kmem_alloc */ extern int afs_osicred_initialized; struct osimem { struct osimem *next; }; /* osi_Init -- do once per kernel installation initialization. * -- On Solaris this is called from modload initialization. * -- On AIX called from afs_config. * -- On HP called from afsc_link. * -- On SGI called from afs_init. */ #ifdef AFS_SGI53_ENV lock_t afs_event_lock; #endif #ifdef AFS_SGI64_ENV flid_t osi_flid; #endif void osi_Init() { static int once = 0; if (once++ > 0) /* just in case */ return; #if defined(AFS_HPUX_ENV) osi_InitGlock(); #else /* AFS_HPUX_ENV */ #if defined(AFS_GLOBAL_SUNLOCK) #if defined(AFS_SGI62_ENV) mutex_init(&afs_global_lock, MUTEX_DEFAULT, "afs_global_lock"); #elif defined(AFS_OSF_ENV) usimple_lock_init(&afs_global_lock); afs_global_owner = (thread_t)0; #elif defined(AFS_AIX41_ENV) lock_alloc((void*)&afs_global_lock, LOCK_ALLOC_PIN, 1, 1); simple_lock_init((void *)&afs_global_lock); #else #ifndef AFS_LINUX22_ENV /* Linux initialization in osi directory. Should move the others. */ mutex_init(&afs_global_lock, "afs_global_lock", MUTEX_DEFAULT, NULL); #endif #endif /* afs_rxglobal_lock is initialized in rx_Init. */ #endif #endif /* AFS_HPUX_ENV */ if ( !afs_osicred_initialized ) { bzero((char *)&afs_osi_cred, sizeof(struct AFS_UCRED)); crhold(&afs_osi_cred); /* don't let it evaporate */ afs_osicred_initialized = 1; } #ifdef AFS_SGI64_ENV osi_flid.fl_pid = osi_flid.fl_sysid = 0; #endif } osi_Active(avc) register struct vcache *avc; { AFS_STATCNT(osi_Active); #if defined(AFS_SUN_ENV) || defined(AFS_AIX_ENV) || defined(AFS_OSF_ENV) || defined(AFS_SUN5_ENV) || (AFS_LINUX20_ENV) if ((avc->opens > 0) || (avc->states & CMAPPED)) return 1; /* XXX: Warning, verify this XXX */ #else #if defined(AFS_MACH_ENV) if (avc->opens > 0 || ((avc->v.v_flag & VTEXT) && !inode_uncache_try(avc))) return 1; #else #if defined(AFS_SGI_ENV) if ((avc->opens > 0) || AFS_VN_MAPPED((struct vnode *)avc)) return 1; #else if (avc->opens > 0 || (avc->v.v_flag & VTEXT)) return(1); #endif #endif /* AFS_MACH_ENV */ #endif return 0; } /* this call, unlike osi_FlushText, is supposed to discard caches that may contain invalid information if a file is written remotely, but that may contain valid information that needs to be written back if the file is being written locally. It doesn't subsume osi_FlushText, since the latter function may be needed to flush caches that are invalidated by local writes. avc->pvnLock is already held, avc->lock is guaranteed not to be held (by us, of course). */ void osi_FlushPages(avc, credp) register struct vcache *avc; struct AFS_UCRED *credp; { afs_hyper_t origDV; ObtainReadLock(&avc->lock); /* If we've already purged this version, or if we're the ones writing this version, don't flush it (could lose the data we're writing). */ if ((hcmp((avc->m.DataVersion), (avc->mapDV)) <= 0) || ((avc->execsOrWriters > 0) && afs_DirtyPages(avc))) { ReleaseReadLock(&avc->lock); return; } ReleaseReadLock(&avc->lock); ObtainWriteLock(&avc->lock,10); /* Check again */ if ((hcmp((avc->m.DataVersion), (avc->mapDV)) <= 0) || ((avc->execsOrWriters > 0) && afs_DirtyPages(avc))) { ReleaseWriteLock(&avc->lock); return; } if (hiszero(avc->mapDV)) { hset(avc->mapDV, avc->m.DataVersion); ReleaseWriteLock(&avc->lock); return; } AFS_STATCNT(osi_FlushPages); hset(origDV, avc->m.DataVersion); afs_Trace3(afs_iclSetp, CM_TRACE_FLUSHPAGES, ICL_TYPE_POINTER, avc, ICL_TYPE_INT32, origDV.low, ICL_TYPE_INT32, avc->m.Length); ReleaseWriteLock(&avc->lock); AFS_GUNLOCK(); osi_VM_FlushPages(avc, credp); AFS_GLOCK(); ObtainWriteLock(&avc->lock,88); /* do this last, and to original version, since stores may occur while executing above PUTPAGE call */ hset(avc->mapDV, origDV); ReleaseWriteLock(&avc->lock); } afs_lock_t afs_ftf; /* flush text lock */ #ifdef AFS_TEXT_ENV /* This call is supposed to flush all caches that might be invalidated * by either a local write operation or a write operation done on * another client. This call may be called repeatedly on the same * version of a file, even while a file is being written, so it * shouldn't do anything that would discard newly written data before * it is written to the file system. */ void osi_FlushText_really(vp) register struct vcache *vp; { afs_hyper_t fdv; /* version before which we'll flush */ AFS_STATCNT(osi_FlushText); /* see if we've already flushed this data version */ if (hcmp(vp->m.DataVersion, vp->flushDV) <= 0) return; #ifdef AFS_DEC_ENV { void afs_gfs_FlushText(); afs_gfs_FlushText(vp); return; } #else MObtainWriteLock(&afs_ftf,317); hset(fdv, vp->m.DataVersion); /* why this disgusting code below? * xuntext, called by xrele, doesn't notice when it is called * with a freed text object. Sun continually calls xrele or xuntext * without any locking, as long as VTEXT is set on the * corresponding vnode. * But, if the text object is locked when you check the VTEXT * flag, several processes can wait in xuntext, waiting for the * text lock; when the second one finally enters xuntext's * critical region, the text object is already free, but the check * was already done by xuntext's caller. * Even worse, it turns out that xalloc locks the text object * before reading or stating a file via the vnode layer. Thus, we * could end up in getdcache, being asked to bring in a new * version of a file, but the corresponding text object could be * locked. We can't flush the text object without causing * deadlock, so now we just don't try to lock the text object * unless it is guaranteed to work. And we try to flush the text * when we need to a bit more often at the vnode layer. Sun * really blew the vm-cache flushing interface. */ #if defined (AFS_HPUX_ENV) if (vp->v.v_flag & VTEXT) { xrele(vp); if (vp->v.v_flag & VTEXT) { /* still has a text object? */ MReleaseWriteLock(&afs_ftf); return; } } #endif /* next do the stuff that need not check for deadlock problems */ mpurge(vp); /* finally, record that we've done it */ hset(vp->flushDV, fdv); MReleaseWriteLock(&afs_ftf); #endif /* AFS_DEC_ENV */ } #ifdef AFS_DEC_ENV /* I don't really like using xinval() here, because it kills processes * a bit aggressively. Previous incarnations of this functionality * used to use xrele() instead of xinval, and didn't invoke * cacheinval(). But they would panic. So it might be worth looking * into some middle ground... */ static void afs_gfs_FlushText(vp) register struct vcache *vp; { afs_hyper_t fdv; /* version before which we'll flush */ register struct text *xp; struct gnode * gp; MObtainWriteLock(&afs_ftf,318); hset(fdv, vp->m.DataVersion); gp = afs_vntogn(vp); if (!gp) { /* this happens frequently after cores are created. */ MReleaseWriteLock(&afs_ftf); return; } if (gp->g_flag & GTEXT) { if (gp->g_textp) { xp = (struct text *) gp->g_textp ; /* if text object is locked, give up */ if (xp && (xp->x_flag & XLOCK)) { MReleaseWriteLock(&afs_ftf); return; } } else xp = (struct text *) 0; if (gp->g_flag & GTEXT) {/* still has a text object? */ xinval(gp); } } /* next do the stuff that need not check for deadlock problems */ /* maybe xinval(gp); here instead of above */ binval(NODEV, gp); cacheinval(gp); /* finally, record that we've done it */ hset(vp->flushDV, fdv); MReleaseWriteLock(&afs_ftf); } #endif /* AFS_DEC_ENV */ #endif /* AFS_TEXT_ENV */ /* procedure for making our processes as invisible as we can */ void afs_osi_Invisible() { #ifndef AFS_AIX32_ENV /* called once per "kernel" lwp to make it invisible */ #ifdef AFS_DEC_ENV u.u_procp->p_type |= SSYS; #else #if defined(AFS_SUN5_ENV) curproc->p_flag |= SSYS; #else #if defined(AFS_SGI_ENV) vrelvm(); #endif #ifdef AFS_SUN_ENV relvm(u.u_procp); /* release all the resources */ #endif #if defined(AFS_HPUX101_ENV) set_system_proc(u.u_procp); #else #if !defined(AFS_SGI64_ENV) && !defined(AFS_LINUX20_ENV) u.u_procp->p_flag |= SSYS; #endif /* AFS_SGI64_ENV */ #endif #endif #endif #endif AFS_STATCNT(osi_Invisible); } #ifndef AFS_LINUX20_ENV /* Linux version in osi_misc.c */ /* set the real time */ afs_osi_SetTime(atv) register osi_timeval_t *atv; { #ifdef AFS_AIX32_ENV struct timestruc_t t; t.tv_sec = atv->tv_sec; t.tv_nsec = atv->tv_usec * 1000; ksettimer(&t); /* Was -> settimer(TIMEOFDAY, &t); */ #else #ifdef AFS_SUN55_ENV stime(atv->tv_sec); #else #ifdef AFS_SUN5_ENV /* * To get more than second resolution we can use adjtime. The problem * is that the usecs from the server are wrong (by now) so it isn't * worth complicating the following code. */ struct stimea { time_t time; } sta; extern int stime(struct stimea *time, rval_t *rvp); sta.time = atv->tv_sec; stime(&sta, NULL); #else #if defined(AFS_SGI_ENV) struct stimea { sysarg_t time; } sta; extern int stime(struct stimea *time); AFS_GUNLOCK(); sta.time = atv->tv_sec; stime(&sta); AFS_GLOCK(); #else /* stolen from kern_time.c */ #ifndef AFS_AUX_ENV boottime.tv_sec += atv->tv_sec - time.tv_sec; #endif #ifdef AFS_HPUX_ENV { register ulong_t s; struct timeval t; t.tv_sec = atv->tv_sec; t.tv_usec = atv->tv_usec; s = spl7(); time = t; (void) splx(s); resettodr(atv); } #else { register int s; s = splclock(); time = *atv; (void) splx(s); } resettodr(); #endif #ifdef AFS_AUX_ENV logtchg(atv->tv_sec); #endif #endif /* AFS_SGI_ENV */ #endif /* AFS_SUN55_ENV */ #endif /* AFS_SUN5_ENV */ #endif /* AFS_AIX32_ENV */ AFS_STATCNT(osi_SetTime); return 0; } #endif /* AFS_LINUX20_ENV */ void *afs_osi_Alloc(size_t x) { register struct osimem *tm = NULL; register int size; AFS_STATCNT(osi_Alloc); /* 0-length allocs may return NULL ptr from AFS_KALLOC, so we special-case things so that NULL returned iff an error occurred */ if (x == 0) return &memZero; AFS_STATS(afs_stats_cmperf.OutStandingAllocs++); AFS_STATS(afs_stats_cmperf.OutStandingMemUsage += x); #ifdef AFS_LINUX20_ENV return osi_linux_alloc(x); #else size = x; tm = (struct osimem *) AFS_KALLOC(size); #ifdef AFS_SUN_ENV if (!tm) osi_Panic("osi_Alloc: Couldn't allocate %d bytes; out of memory!\n", size); #endif return (char *) tm; #endif } #if defined(AFS_SUN_ENV) || defined(AFS_SGI_ENV) void *afs_osi_Alloc_NoSleep(size_t x) { register struct osimem *tm; register int size; AFS_STATCNT(osi_Alloc); /* 0-length allocs may return NULL ptr from AFS_KALLOC, so we special-case things so that NULL returned iff an error occurred */ if (x == 0) return &memZero; size = x; AFS_STATS(afs_stats_cmperf.OutStandingAllocs++); AFS_STATS(afs_stats_cmperf.OutStandingMemUsage += x); tm = (struct osimem *) AFS_KALLOC_NOSLEEP(size); return (char *) tm; } #endif /* SUN || SGI */ void afs_osi_Free(void *x, size_t asize) { register struct osimem *tm, **lm, *um; AFS_STATCNT(osi_Free); if (x == &memZero) return; /* check for putting memZero back */ AFS_STATS(afs_stats_cmperf.OutStandingAllocs--); AFS_STATS(afs_stats_cmperf.OutStandingMemUsage -= asize); #ifdef AFS_LINUX20_ENV osi_linux_free(x); #else AFS_KFREE((struct osimem *)x, asize); #endif } /* ? is it moderately likely that there are dirty VM pages associated with * this vnode? * * Prereqs: avc must be write-locked * * System Dependencies: - *must* support each type of system for which * memory mapped files are supported, even if all * it does is return TRUE; * * NB: this routine should err on the side of caution for ProcessFS to work * correctly (or at least, not to introduce worse bugs than already exist) */ #ifdef notdef int osi_VMDirty_p(avc) struct vcache *avc; { int dirtyPages; if (avc->execsOrWriters <= 0) return 0; /* can't be many dirty pages here, I guess */ #if defined (AFS_AIX32_ENV) #ifdef notdef /* because of the level of hardware involvment with VM and all the * warnings about "This routine must be called at VMM interrupt * level", I thought it would be safest to disable interrupts while * looking at the software page fault table. */ /* convert vm handle into index into array: I think that stoinio is * always zero... Look into this XXX */ #define VMHASH(handle) ( \ ( ((handle) & ~vmker.stoinio) \ ^ ((((handle) & ~vmker.stoinio) & vmker.stoimask) << vmker.stoihash) \ ) & 0x000fffff) if (avc->vmh) { unsigned int pagef, pri, index, next; extern struct vmkerdata vmker; index = VMHASH(avc->vmh); if (scb_valid(index)) { /* could almost be an ASSERT */ pri = disable_ints(); for (pagef = scb_sidlist(index); pagef >= 0; pagef = next) { next = pft_sidfwd(pagef); if (pft_modbit(pagef)) { /* has page frame been modified? */ enable_ints(pri); return 1; } } enable_ints(pri); } } #undef VMHASH #endif #endif /* AFS_AIX32_ENV */ #if defined (AFS_SUN_ENV) if (avc->states & CMAPPED) { struct page * pg; for (pg = avc->v.v_s.v_Pages ; pg ; pg = pg->p_vpnext) { if (pg->p_mod) { return 1; } } } #endif return 0; } #endif /* notdef */ /* * Solaris osi_ReleaseVM should not drop and re-obtain the vcache entry lock. * This leads to bad races when osi_ReleaseVM() is called from * afs_InvalidateAllSegments(). * We can do this because Solaris osi_VM_Truncate() doesn't care whether the * vcache entry lock is held or not. * * For other platforms, in some cases osi_VM_Truncate() doesn't care, but * there may be cases where it does care. If so, it would be good to fix * them so they don't care. Until then, we assume the worst. * * Locking: the vcache entry lock is held. It is dropped and re-obtained. */ void osi_ReleaseVM(avc, acred) struct vcache *avc; struct AFS_UCRED *acred; { #ifdef AFS_SUN5_ENV AFS_GUNLOCK(); osi_VM_Truncate(avc, 0, acred); AFS_GLOCK(); #else ReleaseWriteLock(&avc->lock); AFS_GUNLOCK(); osi_VM_Truncate(avc, 0, acred); AFS_GLOCK(); ObtainWriteLock(&avc->lock, 80); #endif } void shutdown_osi() { extern int afs_cold_shutdown; AFS_STATCNT(shutdown_osi); if (afs_cold_shutdown) { LOCK_INIT(&afs_ftf, "afs_ftf"); } } afs_osi_suser(credp) void * credp; { #ifdef AFS_SUN5_ENV return afs_suser(credp); #else return afs_suser(); #endif } #if AFS_GCPAGS /* afs_osi_TraverseProcTable() - Walk through the systems process * table, calling afs_GCPAGs_perproc_func() for each process. */ #if defined(AFS_SUN5_ENV) void afs_osi_TraverseProcTable() { struct proc *prp; for (prp = practive; prp != NULL; prp = prp->p_next) { afs_GCPAGs_perproc_func(prp); } } #endif #if defined(AFS_HPUX_ENV) /* * NOTE: h/proc_private.h gives the process table locking rules * It indicates that access to p_cred must be protected by * mp_mtproc_lock(p); * mp_mtproc_unlock(p); * * The code in sys/pm_prot.c uses pcred_lock() to protect access to * the process creds, and uses mp_mtproc_lock() only for audit-related * changes. To be safe, we use both. */ void afs_osi_TraverseProcTable() { register proc_t *p; int endchain = 0; MP_SPINLOCK(activeproc_lock); MP_SPINLOCK(sched_lock); pcred_lock(); /* * Instead of iterating through all of proc[], traverse only * the list of active processes. As an example of this, * see foreach_process() in sys/vm_sched.c. * * We hold the locks for the entire scan in order to get a * consistent view of the current set of creds. */ for(p = proc; endchain == 0; p = &proc[p->p_fandx]) { if (p->p_fandx == 0) { endchain = 1; } if (system_proc(p)) continue; mp_mtproc_lock(p); afs_GCPAGs_perproc_func(p); mp_mtproc_unlock(p); } pcred_unlock(); MP_SPINUNLOCK(sched_lock); MP_SPINUNLOCK(activeproc_lock); } #endif #if defined(AFS_SGI_ENV) #ifdef AFS_SGI65_ENV /* TODO: Fix this later. */ static int SGI_ProcScanFunc(void *p, void *arg, int mode) { return 0; } #else /* AFS_SGI65_ENV */ static int SGI_ProcScanFunc(proc_t *p, void *arg, int mode) { afs_int32 (*perproc_func)(struct proc *) = arg; int code=0; /* we pass in the function pointer for arg, * mode ==0 for startup call, ==1 for each valid proc, * and ==2 for terminate call. */ if(mode == 1) { code = perproc_func(p); } return code; } #endif /* AFS_SGI65_ENV */ void afs_osi_TraverseProcTable() { procscan(SGI_ProcScanFunc, afs_GCPAGs_perproc_func); } #endif /* AFS_SGI_ENV */ #if defined(AFS_AIX_ENV) void afs_osi_TraverseProcTable() { struct proc *p; int i; /* * For binary compatibility, on AIX we need to be careful to use the * proper size of a struct proc, even if it is different from what * we were compiled with. */ if (!afs_gcpags_procsize) return; simple_lock(&proc_tbl_lock); for (p = (struct proc *)v.vb_proc, i = 0; p < max_proc; p = (struct proc *)((char *)p + afs_gcpags_procsize), i++) { if (p->p_stat == SNONE) continue; if (p->p_stat == SIDL) continue; if (p->p_stat == SEXIT) continue; /* sanity check */ if (PROCMASK(p->p_pid) != i) { afs_gcpags = AFS_GCPAGS_EPIDCHECK; break; } /* sanity check */ if ((p->p_nice < P_NICE_MIN) || (P_NICE_MAX < p->p_nice)) { afs_gcpags = AFS_GCPAGS_ENICECHECK; break; } afs_GCPAGs_perproc_func(p); } simple_unlock(&proc_tbl_lock); } #endif #if defined(AFS_OSF_ENV) void afs_osi_TraverseProcTable() { struct pid_entry *pe; PID_LOCK(); for (pe = pidtab; pe < pidNPID; ++pe) { if (pe->pe_proc != PROC_NULL) afs_GCPAGs_perproc_func(pe->pe_proc); } PID_UNLOCK(); } #endif /* return a pointer (sometimes a static copy ) to the cred for a * given AFS_PROC. * subsequent calls may overwrite the previously returned value. */ #if defined(AFS_SGI65_ENV) const struct AFS_UCRED *afs_osi_proc2cred(AFS_PROC *pr) { return NULL; } #elif defined(AFS_HPUX_ENV) const struct AFS_UCRED *afs_osi_proc2cred(proc_t *p) { if (!p) return; /* * Cannot use afs_warnuser() here, as the code path * eventually wants to grab sched_lock, which is * already held here */ return p_cred(p); } #elif defined(AFS_AIX_ENV) /* GLOBAL DECLARATIONS */ extern int xmattach(); /* fills out cross memory descriptor */ extern int xmdetach(); /* decrements reference count to segment */ /* * LOCKS: the caller must do * simple_lock(&proc_tbl_lock); * simple_unlock(&proc_tbl_lock); * around calls to this function. */ const struct AFS_UCRED *afs_osi_proc2cred(AFS_PROC *pproc) { struct AFS_UCRED *pcred = 0; /* * pointer to process user structure valid in *our* * address space * * The user structure for a process is stored in the user * address space (as distinct from the kernel address * space), and so to refer to the user structure of a * different process we must employ special measures. * * I followed the example used in the AIX getproc() system * call in bos/kernel/proc/getproc.c */ struct user *xmem_userp; struct xmem dp; /* ptr to xmem descriptor */ int xm; /* xmem result */ if (!pproc) { return pcred; } /* * The process private segment in which the user * area is located may disappear. We need to increment * its use count. Therefore we * - get the proc_tbl_lock to hold the segment. * - get the p_lock to lockout vm_cleardata. * - vm_att to load the segment register (no check) * - xmattach to bump its use count. * - release the p_lock. * - release the proc_tbl_lock. * - do whatever we need. * - xmdetach to decrement the use count. * - vm_det to free the segment register (no check) */ xmem_userp = NULL; xm = XMEM_FAIL; /* simple_lock(&proc_tbl_lock); */ if (pproc->p_adspace != NULLSEGVAL) { simple_lock(&pproc->p_lock); if (pproc->p_threadcount && pproc->p_threadlist) { /* * arbitrarily pick the first thread in pproc */ struct thread *pproc_thread = pproc->p_threadlist; /* * location of 'struct user' in pproc's * address space */ struct user *pproc_userp = pproc_thread->t_userp; /* * create a pointer valid in my own address space */ xmem_userp = (struct user *)vm_att(pproc->p_adspace, pproc_userp); dp.aspace_id = XMEM_INVAL; xm = xmattach(xmem_userp, sizeof(*xmem_userp), &dp, SYS_ADSPACE); } simple_unlock(&pproc->p_lock); } /* simple_unlock(&proc_tbl_lock); */ if (xm == XMEM_SUCC) { static struct AFS_UCRED cred; /* * What locking should we use to protect access to the user * area? If needed also change the code in AIX/osi_groups.c. */ /* copy cred to local address space */ cred = *xmem_userp->U_cred; pcred = &cred; xmdetach(&dp); } if (xmem_userp) { vm_det((void *)xmem_userp); } return pcred; } #elif defined(AFS_OSF_ENV) const struct AFS_UCRED *afs_osi_proc2cred(AFS_PROC *pr) { struct AFS_UCRED *rv=NULL; if(pr == NULL) { return NULL; } if((pr->p_stat == SSLEEP) || (pr->p_stat == SRUN) || (pr->p_stat == SSTOP)) rv = pr->p_rcred; return rv; } #else const struct AFS_UCRED *afs_osi_proc2cred(AFS_PROC *pr) { struct AFS_UCRED *rv=NULL; if(pr == NULL) { return NULL; } rv = pr->p_cred; return rv; } #endif #endif /* AFS_GCPAGS */