Index: head/sys/kern/sysv_shm.c =================================================================== --- head/sys/kern/sysv_shm.c (revision 113447) +++ head/sys/kern/sysv_shm.c (revision 113448) @@ -1,894 +1,894 @@ /* $FreeBSD$ */ /* $NetBSD: sysv_shm.c,v 1.23 1994/07/04 23:25:12 glass Exp $ */ /* * Copyright (c) 1994 Adam Glass and Charles Hannum. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by Adam Glass and Charles * Hannum. * 4. The names of the authors may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHORS ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "opt_compat.h" #include "opt_sysvipc.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include static MALLOC_DEFINE(M_SHM, "shm", "SVID compatible shared memory segments"); struct oshmctl_args; static int oshmctl(struct thread *td, struct oshmctl_args *uap); static int shmget_allocate_segment(struct thread *td, struct shmget_args *uap, int mode); static int shmget_existing(struct thread *td, struct shmget_args *uap, int mode, int segnum); /* XXX casting to (sy_call_t *) is bogus, as usual. */ static sy_call_t *shmcalls[] = { (sy_call_t *)shmat, (sy_call_t *)oshmctl, (sy_call_t *)shmdt, (sy_call_t *)shmget, (sy_call_t *)shmctl }; #define SHMSEG_FREE 0x0200 #define SHMSEG_REMOVED 0x0400 #define SHMSEG_ALLOCATED 0x0800 #define SHMSEG_WANTED 0x1000 static int shm_last_free, shm_nused, shm_committed, shmalloced; static struct shmid_ds *shmsegs; struct shm_handle { /* vm_offset_t kva; */ vm_object_t shm_object; }; struct shmmap_state { vm_offset_t va; int shmid; }; static void shm_deallocate_segment(struct shmid_ds *); static int shm_find_segment_by_key(key_t); static struct shmid_ds *shm_find_segment_by_shmid(int); static struct shmid_ds *shm_find_segment_by_shmidx(int); static int shm_delete_mapping(struct vmspace *vm, struct shmmap_state *); static void shmrealloc(void); static void shminit(void); static int sysvshm_modload(struct module *, int, void *); static int shmunload(void); static void shmexit_myhook(struct vmspace *vm); static void shmfork_myhook(struct proc *p1, struct proc *p2); static int sysctl_shmsegs(SYSCTL_HANDLER_ARGS); /* * Tuneable values. */ #ifndef SHMMAXPGS #define SHMMAXPGS 8192 /* Note: sysv shared memory is swap backed. */ #endif #ifndef SHMMAX #define SHMMAX (SHMMAXPGS*PAGE_SIZE) #endif #ifndef SHMMIN #define SHMMIN 1 #endif #ifndef SHMMNI #define SHMMNI 192 #endif #ifndef SHMSEG #define SHMSEG 128 #endif #ifndef SHMALL #define SHMALL (SHMMAXPGS) #endif struct shminfo shminfo = { SHMMAX, SHMMIN, SHMMNI, SHMSEG, SHMALL }; static int shm_use_phys; SYSCTL_DECL(_kern_ipc); SYSCTL_INT(_kern_ipc, OID_AUTO, shmmax, CTLFLAG_RW, &shminfo.shmmax, 0, ""); SYSCTL_INT(_kern_ipc, OID_AUTO, shmmin, CTLFLAG_RW, &shminfo.shmmin, 0, ""); SYSCTL_INT(_kern_ipc, OID_AUTO, shmmni, CTLFLAG_RD, &shminfo.shmmni, 0, ""); SYSCTL_INT(_kern_ipc, OID_AUTO, shmseg, CTLFLAG_RD, &shminfo.shmseg, 0, ""); SYSCTL_INT(_kern_ipc, OID_AUTO, shmall, CTLFLAG_RW, &shminfo.shmall, 0, ""); SYSCTL_INT(_kern_ipc, OID_AUTO, shm_use_phys, CTLFLAG_RW, &shm_use_phys, 0, ""); SYSCTL_PROC(_kern_ipc, OID_AUTO, shmsegs, CTLFLAG_RD, NULL, 0, sysctl_shmsegs, "", ""); static int shm_find_segment_by_key(key) key_t key; { int i; for (i = 0; i < shmalloced; i++) if ((shmsegs[i].shm_perm.mode & SHMSEG_ALLOCATED) && shmsegs[i].shm_perm.key == key) return (i); return (-1); } static struct shmid_ds * shm_find_segment_by_shmid(shmid) int shmid; { int segnum; struct shmid_ds *shmseg; segnum = IPCID_TO_IX(shmid); if (segnum < 0 || segnum >= shmalloced) return (NULL); shmseg = &shmsegs[segnum]; if ((shmseg->shm_perm.mode & (SHMSEG_ALLOCATED | SHMSEG_REMOVED)) != SHMSEG_ALLOCATED || shmseg->shm_perm.seq != IPCID_TO_SEQ(shmid)) return (NULL); return (shmseg); } static struct shmid_ds * shm_find_segment_by_shmidx(int segnum) { struct shmid_ds *shmseg; if (segnum < 0 || segnum >= shmalloced) return (NULL); shmseg = &shmsegs[segnum]; if ((shmseg->shm_perm.mode & (SHMSEG_ALLOCATED | SHMSEG_REMOVED)) != SHMSEG_ALLOCATED ) return (NULL); return (shmseg); } static void shm_deallocate_segment(shmseg) struct shmid_ds *shmseg; { struct shm_handle *shm_handle; size_t size; GIANT_REQUIRED; shm_handle = shmseg->shm_internal; vm_object_deallocate(shm_handle->shm_object); free(shm_handle, M_SHM); shmseg->shm_internal = NULL; size = round_page(shmseg->shm_segsz); shm_committed -= btoc(size); shm_nused--; shmseg->shm_perm.mode = SHMSEG_FREE; } static int shm_delete_mapping(struct vmspace *vm, struct shmmap_state *shmmap_s) { struct shmid_ds *shmseg; int segnum, result; size_t size; GIANT_REQUIRED; segnum = IPCID_TO_IX(shmmap_s->shmid); shmseg = &shmsegs[segnum]; size = round_page(shmseg->shm_segsz); result = vm_map_remove(&vm->vm_map, shmmap_s->va, shmmap_s->va + size); if (result != KERN_SUCCESS) return (EINVAL); shmmap_s->shmid = -1; shmseg->shm_dtime = time_second; if ((--shmseg->shm_nattch <= 0) && (shmseg->shm_perm.mode & SHMSEG_REMOVED)) { shm_deallocate_segment(shmseg); shm_last_free = segnum; } return (0); } #ifndef _SYS_SYSPROTO_H_ struct shmdt_args { const void *shmaddr; }; #endif /* * MPSAFE */ int shmdt(td, uap) struct thread *td; struct shmdt_args *uap; { struct proc *p = td->td_proc; struct shmmap_state *shmmap_s; int i; int error = 0; if (!jail_sysvipc_allowed && jailed(td->td_ucred)) return (ENOSYS); mtx_lock(&Giant); shmmap_s = p->p_vmspace->vm_shm; if (shmmap_s == NULL) { error = EINVAL; goto done2; } for (i = 0; i < shminfo.shmseg; i++, shmmap_s++) { if (shmmap_s->shmid != -1 && shmmap_s->va == (vm_offset_t)uap->shmaddr) { break; } } if (i == shminfo.shmseg) { error = EINVAL; goto done2; } error = shm_delete_mapping(p->p_vmspace, shmmap_s); done2: mtx_unlock(&Giant); return (error); } #ifndef _SYS_SYSPROTO_H_ struct shmat_args { int shmid; const void *shmaddr; int shmflg; }; #endif /* * MPSAFE */ int shmat(td, uap) struct thread *td; struct shmat_args *uap; { struct proc *p = td->td_proc; int i, flags; struct shmid_ds *shmseg; struct shmmap_state *shmmap_s = NULL; struct shm_handle *shm_handle; vm_offset_t attach_va; vm_prot_t prot; vm_size_t size; int rv; int error = 0; if (!jail_sysvipc_allowed && jailed(td->td_ucred)) return (ENOSYS); mtx_lock(&Giant); shmmap_s = p->p_vmspace->vm_shm; if (shmmap_s == NULL) { size = shminfo.shmseg * sizeof(struct shmmap_state); shmmap_s = malloc(size, M_SHM, M_WAITOK); for (i = 0; i < shminfo.shmseg; i++) shmmap_s[i].shmid = -1; p->p_vmspace->vm_shm = shmmap_s; } shmseg = shm_find_segment_by_shmid(uap->shmid); if (shmseg == NULL) { error = EINVAL; goto done2; } error = ipcperm(td, &shmseg->shm_perm, (uap->shmflg & SHM_RDONLY) ? IPC_R : IPC_R|IPC_W); if (error) goto done2; for (i = 0; i < shminfo.shmseg; i++) { if (shmmap_s->shmid == -1) break; shmmap_s++; } if (i >= shminfo.shmseg) { error = EMFILE; goto done2; } size = round_page(shmseg->shm_segsz); #ifdef VM_PROT_READ_IS_EXEC prot = VM_PROT_READ | VM_PROT_EXECUTE; #else prot = VM_PROT_READ; #endif if ((uap->shmflg & SHM_RDONLY) == 0) prot |= VM_PROT_WRITE; flags = MAP_ANON | MAP_SHARED; if (uap->shmaddr) { flags |= MAP_FIXED; if (uap->shmflg & SHM_RND) { attach_va = (vm_offset_t)uap->shmaddr & ~(SHMLBA-1); } else if (((vm_offset_t)uap->shmaddr & (SHMLBA-1)) == 0) { attach_va = (vm_offset_t)uap->shmaddr; } else { error = EINVAL; goto done2; } } else { /* * This is just a hint to vm_map_find() about where to * put it. */ attach_va = round_page((vm_offset_t)p->p_vmspace->vm_taddr + maxtsiz + maxdsiz); } shm_handle = shmseg->shm_internal; vm_object_reference(shm_handle->shm_object); rv = vm_map_find(&p->p_vmspace->vm_map, shm_handle->shm_object, 0, &attach_va, size, (flags & MAP_FIXED)?0:1, prot, prot, 0); if (rv != KERN_SUCCESS) { error = ENOMEM; goto done2; } vm_map_inherit(&p->p_vmspace->vm_map, attach_va, attach_va + size, VM_INHERIT_SHARE); shmmap_s->va = attach_va; shmmap_s->shmid = uap->shmid; shmseg->shm_lpid = p->p_pid; shmseg->shm_atime = time_second; shmseg->shm_nattch++; td->td_retval[0] = attach_va; done2: mtx_unlock(&Giant); return (error); } struct oshmid_ds { struct ipc_perm shm_perm; /* operation perms */ int shm_segsz; /* size of segment (bytes) */ ushort shm_cpid; /* pid, creator */ ushort shm_lpid; /* pid, last operation */ short shm_nattch; /* no. of current attaches */ time_t shm_atime; /* last attach time */ time_t shm_dtime; /* last detach time */ time_t shm_ctime; /* last change time */ void *shm_handle; /* internal handle for shm segment */ }; struct oshmctl_args { int shmid; int cmd; struct oshmid_ds *ubuf; }; /* * MPSAFE */ static int oshmctl(td, uap) struct thread *td; struct oshmctl_args *uap; { #ifdef COMPAT_43 int error = 0; struct shmid_ds *shmseg; struct oshmid_ds outbuf; if (!jail_sysvipc_allowed && jailed(td->td_ucred)) return (ENOSYS); mtx_lock(&Giant); shmseg = shm_find_segment_by_shmid(uap->shmid); if (shmseg == NULL) { error = EINVAL; goto done2; } switch (uap->cmd) { case IPC_STAT: error = ipcperm(td, &shmseg->shm_perm, IPC_R); if (error) goto done2; outbuf.shm_perm = shmseg->shm_perm; outbuf.shm_segsz = shmseg->shm_segsz; outbuf.shm_cpid = shmseg->shm_cpid; outbuf.shm_lpid = shmseg->shm_lpid; outbuf.shm_nattch = shmseg->shm_nattch; outbuf.shm_atime = shmseg->shm_atime; outbuf.shm_dtime = shmseg->shm_dtime; outbuf.shm_ctime = shmseg->shm_ctime; outbuf.shm_handle = shmseg->shm_internal; error = copyout(&outbuf, uap->ubuf, sizeof(outbuf)); if (error) goto done2; break; default: /* XXX casting to (sy_call_t *) is bogus, as usual. */ error = ((sy_call_t *)shmctl)(td, uap); break; } done2: mtx_unlock(&Giant); return (error); #else return (EINVAL); #endif } #ifndef _SYS_SYSPROTO_H_ struct shmctl_args { int shmid; int cmd; struct shmid_ds *buf; }; #endif /* * MPSAFE */ int shmctl(td, uap) struct thread *td; struct shmctl_args *uap; { int error = 0; struct shmid_ds inbuf; struct shmid_ds *shmseg; if (!jail_sysvipc_allowed && jailed(td->td_ucred)) return (ENOSYS); mtx_lock(&Giant); switch (uap->cmd) { case IPC_INFO: error = copyout(&shminfo, uap->buf, sizeof(shminfo)); if (error) goto done2; td->td_retval[0] = shmalloced; goto done2; case SHM_INFO: { struct shm_info shm_info; shm_info.used_ids = shm_nused; shm_info.shm_rss = 0; /*XXX where to get from ? */ shm_info.shm_tot = 0; /*XXX where to get from ? */ shm_info.shm_swp = 0; /*XXX where to get from ? */ shm_info.swap_attempts = 0; /*XXX where to get from ? */ shm_info.swap_successes = 0; /*XXX where to get from ? */ error = copyout(&shm_info, uap->buf, sizeof(shm_info)); if (error) goto done2; td->td_retval[0] = shmalloced; goto done2; } } if( (uap->cmd) == SHM_STAT ) shmseg = shm_find_segment_by_shmidx(uap->shmid); else shmseg = shm_find_segment_by_shmid(uap->shmid); if (shmseg == NULL) { error = EINVAL; goto done2; } switch (uap->cmd) { case SHM_STAT: case IPC_STAT: error = ipcperm(td, &shmseg->shm_perm, IPC_R); if (error) goto done2; error = copyout(shmseg, uap->buf, sizeof(inbuf)); if (error) goto done2; else if( (uap->cmd) == SHM_STAT ) td->td_retval[0] = IXSEQ_TO_IPCID( uap->shmid, shmseg->shm_perm ); break; case IPC_SET: error = ipcperm(td, &shmseg->shm_perm, IPC_M); if (error) goto done2; error = copyin(uap->buf, &inbuf, sizeof(inbuf)); if (error) goto done2; shmseg->shm_perm.uid = inbuf.shm_perm.uid; shmseg->shm_perm.gid = inbuf.shm_perm.gid; shmseg->shm_perm.mode = (shmseg->shm_perm.mode & ~ACCESSPERMS) | (inbuf.shm_perm.mode & ACCESSPERMS); shmseg->shm_ctime = time_second; break; case IPC_RMID: error = ipcperm(td, &shmseg->shm_perm, IPC_M); if (error) goto done2; shmseg->shm_perm.key = IPC_PRIVATE; shmseg->shm_perm.mode |= SHMSEG_REMOVED; if (shmseg->shm_nattch <= 0) { shm_deallocate_segment(shmseg); shm_last_free = IPCID_TO_IX(uap->shmid); } break; #if 0 case SHM_LOCK: case SHM_UNLOCK: #endif default: error = EINVAL; break; } done2: mtx_unlock(&Giant); return (error); } #ifndef _SYS_SYSPROTO_H_ struct shmget_args { key_t key; size_t size; int shmflg; }; #endif static int shmget_existing(td, uap, mode, segnum) struct thread *td; struct shmget_args *uap; int mode; int segnum; { struct shmid_ds *shmseg; int error; shmseg = &shmsegs[segnum]; if (shmseg->shm_perm.mode & SHMSEG_REMOVED) { /* * This segment is in the process of being allocated. Wait * until it's done, and look the key up again (in case the * allocation failed or it was freed). */ shmseg->shm_perm.mode |= SHMSEG_WANTED; error = tsleep(shmseg, PLOCK | PCATCH, "shmget", 0); if (error) return (error); return (EAGAIN); } if ((uap->shmflg & (IPC_CREAT | IPC_EXCL)) == (IPC_CREAT | IPC_EXCL)) return (EEXIST); error = ipcperm(td, &shmseg->shm_perm, mode); if (error) return (error); if (uap->size && uap->size > shmseg->shm_segsz) return (EINVAL); td->td_retval[0] = IXSEQ_TO_IPCID(segnum, shmseg->shm_perm); return (0); } static int shmget_allocate_segment(td, uap, mode) struct thread *td; struct shmget_args *uap; int mode; { int i, segnum, shmid, size; struct ucred *cred = td->td_ucred; struct shmid_ds *shmseg; struct shm_handle *shm_handle; GIANT_REQUIRED; if (uap->size < shminfo.shmmin || uap->size > shminfo.shmmax) return (EINVAL); if (shm_nused >= shminfo.shmmni) /* Any shmids left? */ return (ENOSPC); size = round_page(uap->size); if (shm_committed + btoc(size) > shminfo.shmall) return (ENOMEM); if (shm_last_free < 0) { shmrealloc(); /* Maybe expand the shmsegs[] array. */ for (i = 0; i < shmalloced; i++) if (shmsegs[i].shm_perm.mode & SHMSEG_FREE) break; if (i == shmalloced) return (ENOSPC); segnum = i; } else { segnum = shm_last_free; shm_last_free = -1; } shmseg = &shmsegs[segnum]; /* * In case we sleep in malloc(), mark the segment present but deleted * so that noone else tries to create the same key. */ shmseg->shm_perm.mode = SHMSEG_ALLOCATED | SHMSEG_REMOVED; shmseg->shm_perm.key = uap->key; shmseg->shm_perm.seq = (shmseg->shm_perm.seq + 1) & 0x7fff; shm_handle = (struct shm_handle *) malloc(sizeof(struct shm_handle), M_SHM, M_WAITOK); shmid = IXSEQ_TO_IPCID(segnum, shmseg->shm_perm); /* * We make sure that we have allocated a pager before we need * to. */ if (shm_use_phys) { shm_handle->shm_object = vm_pager_allocate(OBJT_PHYS, 0, size, VM_PROT_DEFAULT, 0); } else { shm_handle->shm_object = vm_pager_allocate(OBJT_SWAP, 0, size, VM_PROT_DEFAULT, 0); } - vm_object_lock(shm_handle->shm_object); + VM_OBJECT_LOCK(shm_handle->shm_object); vm_object_clear_flag(shm_handle->shm_object, OBJ_ONEMAPPING); vm_object_set_flag(shm_handle->shm_object, OBJ_NOSPLIT); - vm_object_unlock(shm_handle->shm_object); + VM_OBJECT_UNLOCK(shm_handle->shm_object); shmseg->shm_internal = shm_handle; shmseg->shm_perm.cuid = shmseg->shm_perm.uid = cred->cr_uid; shmseg->shm_perm.cgid = shmseg->shm_perm.gid = cred->cr_gid; shmseg->shm_perm.mode = (shmseg->shm_perm.mode & SHMSEG_WANTED) | (mode & ACCESSPERMS) | SHMSEG_ALLOCATED; shmseg->shm_segsz = uap->size; shmseg->shm_cpid = td->td_proc->p_pid; shmseg->shm_lpid = shmseg->shm_nattch = 0; shmseg->shm_atime = shmseg->shm_dtime = 0; shmseg->shm_ctime = time_second; shm_committed += btoc(size); shm_nused++; if (shmseg->shm_perm.mode & SHMSEG_WANTED) { /* * Somebody else wanted this key while we were asleep. Wake * them up now. */ shmseg->shm_perm.mode &= ~SHMSEG_WANTED; wakeup(shmseg); } td->td_retval[0] = shmid; return (0); } /* * MPSAFE */ int shmget(td, uap) struct thread *td; struct shmget_args *uap; { int segnum, mode; int error; if (!jail_sysvipc_allowed && jailed(td->td_ucred)) return (ENOSYS); mtx_lock(&Giant); mode = uap->shmflg & ACCESSPERMS; if (uap->key != IPC_PRIVATE) { again: segnum = shm_find_segment_by_key(uap->key); if (segnum >= 0) { error = shmget_existing(td, uap, mode, segnum); if (error == EAGAIN) goto again; goto done2; } if ((uap->shmflg & IPC_CREAT) == 0) { error = ENOENT; goto done2; } } error = shmget_allocate_segment(td, uap, mode); done2: mtx_unlock(&Giant); return (error); } /* * MPSAFE */ int shmsys(td, uap) struct thread *td; /* XXX actually varargs. */ struct shmsys_args /* { u_int which; int a2; int a3; int a4; } */ *uap; { int error; if (!jail_sysvipc_allowed && jailed(td->td_ucred)) return (ENOSYS); if (uap->which >= sizeof(shmcalls)/sizeof(shmcalls[0])) return (EINVAL); mtx_lock(&Giant); error = (*shmcalls[uap->which])(td, &uap->a2); mtx_unlock(&Giant); return (error); } static void shmfork_myhook(p1, p2) struct proc *p1, *p2; { struct shmmap_state *shmmap_s; size_t size; int i; size = shminfo.shmseg * sizeof(struct shmmap_state); shmmap_s = malloc(size, M_SHM, M_WAITOK); bcopy(p1->p_vmspace->vm_shm, shmmap_s, size); p2->p_vmspace->vm_shm = shmmap_s; for (i = 0; i < shminfo.shmseg; i++, shmmap_s++) if (shmmap_s->shmid != -1) shmsegs[IPCID_TO_IX(shmmap_s->shmid)].shm_nattch++; } static void shmexit_myhook(struct vmspace *vm) { struct shmmap_state *base, *shm; int i; GIANT_REQUIRED; if ((base = vm->vm_shm) != NULL) { vm->vm_shm = NULL; for (i = 0, shm = base; i < shminfo.shmseg; i++, shm++) { if (shm->shmid != -1) shm_delete_mapping(vm, shm); } free(base, M_SHM); } } static void shmrealloc(void) { int i; struct shmid_ds *newsegs; if (shmalloced >= shminfo.shmmni) return; newsegs = malloc(shminfo.shmmni * sizeof(*newsegs), M_SHM, M_WAITOK); if (newsegs == NULL) return; for (i = 0; i < shmalloced; i++) bcopy(&shmsegs[i], &newsegs[i], sizeof(newsegs[0])); for (; i < shminfo.shmmni; i++) { shmsegs[i].shm_perm.mode = SHMSEG_FREE; shmsegs[i].shm_perm.seq = 0; } free(shmsegs, M_SHM); shmsegs = newsegs; shmalloced = shminfo.shmmni; } static void shminit() { int i; TUNABLE_INT_FETCH("kern.ipc.shmmaxpgs", &shminfo.shmall); for (i = PAGE_SIZE; i > 0; i--) { shminfo.shmmax = shminfo.shmall * PAGE_SIZE; if (shminfo.shmmax >= shminfo.shmall) break; } TUNABLE_INT_FETCH("kern.ipc.shmmin", &shminfo.shmmin); TUNABLE_INT_FETCH("kern.ipc.shmmni", &shminfo.shmmni); TUNABLE_INT_FETCH("kern.ipc.shmseg", &shminfo.shmseg); TUNABLE_INT_FETCH("kern.ipc.shm_use_phys", &shm_use_phys); shmalloced = shminfo.shmmni; shmsegs = malloc(shmalloced * sizeof(shmsegs[0]), M_SHM, M_WAITOK); if (shmsegs == NULL) panic("cannot allocate initial memory for sysvshm"); for (i = 0; i < shmalloced; i++) { shmsegs[i].shm_perm.mode = SHMSEG_FREE; shmsegs[i].shm_perm.seq = 0; } shm_last_free = 0; shm_nused = 0; shm_committed = 0; shmexit_hook = &shmexit_myhook; shmfork_hook = &shmfork_myhook; } static int shmunload() { if (shm_nused > 0) return (EBUSY); free(shmsegs, M_SHM); shmexit_hook = NULL; shmfork_hook = NULL; return (0); } static int sysctl_shmsegs(SYSCTL_HANDLER_ARGS) { return (SYSCTL_OUT(req, shmsegs, shmalloced * sizeof(shmsegs[0]))); } static int sysvshm_modload(struct module *module, int cmd, void *arg) { int error = 0; switch (cmd) { case MOD_LOAD: shminit(); break; case MOD_UNLOAD: error = shmunload(); break; case MOD_SHUTDOWN: break; default: error = EINVAL; break; } return (error); } static moduledata_t sysvshm_mod = { "sysvshm", &sysvshm_modload, NULL }; SYSCALL_MODULE_HELPER(shmsys); SYSCALL_MODULE_HELPER(shmat); SYSCALL_MODULE_HELPER(shmctl); SYSCALL_MODULE_HELPER(shmdt); SYSCALL_MODULE_HELPER(shmget); DECLARE_MODULE(sysvshm, sysvshm_mod, SI_SUB_SYSV_SHM, SI_ORDER_FIRST); MODULE_VERSION(sysvshm, 1); Index: head/sys/vm/vm_meter.c =================================================================== --- head/sys/vm/vm_meter.c (revision 113447) +++ head/sys/vm/vm_meter.c (revision 113448) @@ -1,356 +1,356 @@ /* * Copyright (c) 1982, 1986, 1989, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)vm_meter.c 8.4 (Berkeley) 1/4/94 * $FreeBSD$ */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include struct vmmeter cnt; int maxslp = MAXSLP; SYSCTL_UINT(_vm, VM_V_FREE_MIN, v_free_min, CTLFLAG_RW, &cnt.v_free_min, 0, ""); SYSCTL_UINT(_vm, VM_V_FREE_TARGET, v_free_target, CTLFLAG_RW, &cnt.v_free_target, 0, ""); SYSCTL_UINT(_vm, VM_V_FREE_RESERVED, v_free_reserved, CTLFLAG_RW, &cnt.v_free_reserved, 0, ""); SYSCTL_UINT(_vm, VM_V_INACTIVE_TARGET, v_inactive_target, CTLFLAG_RW, &cnt.v_inactive_target, 0, ""); SYSCTL_UINT(_vm, VM_V_CACHE_MIN, v_cache_min, CTLFLAG_RW, &cnt.v_cache_min, 0, ""); SYSCTL_UINT(_vm, VM_V_CACHE_MAX, v_cache_max, CTLFLAG_RW, &cnt.v_cache_max, 0, ""); SYSCTL_UINT(_vm, VM_V_PAGEOUT_FREE_MIN, v_pageout_free_min, CTLFLAG_RW, &cnt.v_pageout_free_min, 0, ""); SYSCTL_UINT(_vm, OID_AUTO, v_free_severe, CTLFLAG_RW, &cnt.v_free_severe, 0, ""); SYSCTL_STRUCT(_vm, VM_LOADAVG, loadavg, CTLFLAG_RD, &averunnable, loadavg, "Machine loadaverage history"); static int vmtotal(SYSCTL_HANDLER_ARGS) { /* XXXKSE almost completely broken */ struct proc *p; struct vmtotal total, *totalp; vm_map_entry_t entry; vm_object_t object; vm_map_t map; int paging; struct thread *td; totalp = &total; bzero(totalp, sizeof *totalp); /* * Mark all objects as inactive. */ GIANT_REQUIRED; mtx_lock(&vm_object_list_mtx); TAILQ_FOREACH(object, &vm_object_list, object_list) { - vm_object_lock(object); + VM_OBJECT_LOCK(object); vm_object_clear_flag(object, OBJ_ACTIVE); - vm_object_unlock(object); + VM_OBJECT_UNLOCK(object); } mtx_unlock(&vm_object_list_mtx); /* * Calculate process statistics. */ sx_slock(&allproc_lock); FOREACH_PROC_IN_SYSTEM(p) { if (p->p_flag & P_SYSTEM) continue; mtx_lock_spin(&sched_lock); switch (p->p_state) { case PRS_NEW: mtx_unlock_spin(&sched_lock); continue; break; default: FOREACH_THREAD_IN_PROC(p, td) { /* Need new statistics XXX */ switch (td->td_state) { case TDS_INHIBITED: if (TD_ON_LOCK(td) || (td->td_inhibitors == TDI_SWAPPED)) { totalp->t_sw++; } else if (TD_IS_SLEEPING(td) || TD_AWAITING_INTR(td) || TD_IS_SUSPENDED(td)) { if (td->td_priority <= PZERO) totalp->t_dw++; else totalp->t_sl++; } break; case TDS_CAN_RUN: totalp->t_sw++; break; case TDS_RUNQ: case TDS_RUNNING: totalp->t_rq++; continue; default: break; } } } mtx_unlock_spin(&sched_lock); /* * Note active objects. */ paging = 0; map = &p->p_vmspace->vm_map; vm_map_lock_read(map); for (entry = map->header.next; entry != &map->header; entry = entry->next) { if ((entry->eflags & MAP_ENTRY_IS_SUB_MAP) || (object = entry->object.vm_object) == NULL) continue; - vm_object_lock(object); + VM_OBJECT_LOCK(object); vm_object_set_flag(object, OBJ_ACTIVE); paging |= object->paging_in_progress; - vm_object_unlock(object); + VM_OBJECT_UNLOCK(object); } vm_map_unlock_read(map); if (paging) totalp->t_pw++; } sx_sunlock(&allproc_lock); /* * Calculate object memory usage statistics. */ mtx_lock(&vm_object_list_mtx); TAILQ_FOREACH(object, &vm_object_list, object_list) { - vm_object_lock(object); + VM_OBJECT_LOCK(object); /* * devices, like /dev/mem, will badly skew our totals */ if (object->type == OBJT_DEVICE) { - vm_object_unlock(object); + VM_OBJECT_UNLOCK(object); continue; } totalp->t_vm += object->size; totalp->t_rm += object->resident_page_count; if (object->flags & OBJ_ACTIVE) { totalp->t_avm += object->size; totalp->t_arm += object->resident_page_count; } if (object->shadow_count > 1) { /* shared object */ totalp->t_vmshr += object->size; totalp->t_rmshr += object->resident_page_count; if (object->flags & OBJ_ACTIVE) { totalp->t_avmshr += object->size; totalp->t_armshr += object->resident_page_count; } } - vm_object_unlock(object); + VM_OBJECT_UNLOCK(object); } mtx_unlock(&vm_object_list_mtx); totalp->t_free = cnt.v_free_count + cnt.v_cache_count; return (sysctl_handle_opaque(oidp, totalp, sizeof total, req)); } /* * vcnt() - accumulate statistics from all cpus and the global cnt * structure. * * The vmmeter structure is now per-cpu as well as global. Those * statistics which can be kept on a per-cpu basis (to avoid cache * stalls between cpus) can be moved to the per-cpu vmmeter. Remaining * statistics, such as v_free_reserved, are left in the global * structure. * * (sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req) */ static int vcnt(SYSCTL_HANDLER_ARGS) { int error = 0; int count = *(int *)arg1; #ifdef SMP int i; int offset = (char *)arg1 - (char *)&cnt; for (i = 0; i < mp_ncpus; ++i) { struct pcpu *pcpu = pcpu_find(i); count += *(int *)((char *)&pcpu->pc_cnt + offset); } #else int offset = (char *)arg1 - (char *)&cnt; count += *(int *)((char *)PCPU_PTR(cnt) + offset); #endif error = SYSCTL_OUT(req, &count, sizeof(int)); return(error); } SYSCTL_PROC(_vm, VM_TOTAL, vmtotal, CTLTYPE_OPAQUE|CTLFLAG_RD, 0, sizeof(struct vmtotal), vmtotal, "S,vmtotal", "System virtual memory statistics"); SYSCTL_NODE(_vm, OID_AUTO, stats, CTLFLAG_RW, 0, "VM meter stats"); SYSCTL_NODE(_vm_stats, OID_AUTO, sys, CTLFLAG_RW, 0, "VM meter sys stats"); SYSCTL_NODE(_vm_stats, OID_AUTO, vm, CTLFLAG_RW, 0, "VM meter vm stats"); SYSCTL_NODE(_vm_stats, OID_AUTO, misc, CTLFLAG_RW, 0, "VM meter misc stats"); SYSCTL_PROC(_vm_stats_sys, OID_AUTO, v_swtch, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_swtch, 0, vcnt, "IU", "Context switches"); SYSCTL_PROC(_vm_stats_sys, OID_AUTO, v_trap, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_trap, 0, vcnt, "IU", "Traps"); SYSCTL_PROC(_vm_stats_sys, OID_AUTO, v_syscall, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_syscall, 0, vcnt, "IU", "Syscalls"); SYSCTL_PROC(_vm_stats_sys, OID_AUTO, v_intr, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_intr, 0, vcnt, "IU", "Hardware interrupts"); SYSCTL_PROC(_vm_stats_sys, OID_AUTO, v_soft, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_soft, 0, vcnt, "IU", "Software interrupts"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_vm_faults, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_vm_faults, 0, vcnt, "IU", "VM faults"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_cow_faults, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_cow_faults, 0, vcnt, "IU", "COW faults"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_cow_optim, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_cow_optim, 0, vcnt, "IU", "Optimized COW faults"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_zfod, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_zfod, 0, vcnt, "IU", "Zero fill"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_ozfod, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_ozfod, 0, vcnt, "IU", "Optimized zero fill"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_swapin, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_swapin, 0, vcnt, "IU", "Swapin operations"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_swapout, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_swapout, 0, vcnt, "IU", "Swapout operations"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_swappgsin, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_swappgsin, 0, vcnt, "IU", "Swapin pages"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_swappgsout, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_swappgsout, 0, vcnt, "IU", "Swapout pages"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_vnodein, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_vnodein, 0, vcnt, "IU", "Vnodein operations"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_vnodeout, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_vnodeout, 0, vcnt, "IU", "Vnodeout operations"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_vnodepgsin, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_vnodepgsin, 0, vcnt, "IU", "Vnodein pages"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_vnodepgsout, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_vnodepgsout, 0, vcnt, "IU", "Vnodeout pages"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_intrans, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_intrans, 0, vcnt, "IU", "In transit page blocking"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_reactivated, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_reactivated, 0, vcnt, "IU", "Reactivated pages"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_pdwakeups, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_pdwakeups, 0, vcnt, "IU", "Pagedaemon wakeups"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_pdpages, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_pdpages, 0, vcnt, "IU", "Pagedaemon page scans"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_dfree, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_dfree, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_pfree, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_pfree, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_tfree, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_tfree, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_page_size, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_page_size, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_page_count, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_page_count, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_free_reserved, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_free_reserved, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_free_target, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_free_target, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_free_min, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_free_min, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_free_count, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_free_count, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_wire_count, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_wire_count, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_active_count, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_active_count, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_inactive_target, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_inactive_target, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_inactive_count, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_inactive_count, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_cache_count, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_cache_count, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_cache_min, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_cache_min, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_cache_max, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_cache_max, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_pageout_free_min, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_pageout_free_min, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_interrupt_free_min, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_interrupt_free_min, 0, vcnt, "IU", ""); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_forks, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_forks, 0, vcnt, "IU", "Number of fork() calls"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_vforks, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_vforks, 0, vcnt, "IU", "Number of vfork() calls"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_rforks, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_rforks, 0, vcnt, "IU", "Number of rfork() calls"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_kthreads, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_kthreads, 0, vcnt, "IU", "Number of fork() calls by kernel"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_forkpages, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_forkpages, 0, vcnt, "IU", "VM pages affected by fork()"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_vforkpages, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_vforkpages, 0, vcnt, "IU", "VM pages affected by vfork()"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_rforkpages, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_rforkpages, 0, vcnt, "IU", "VM pages affected by rfork()"); SYSCTL_PROC(_vm_stats_vm, OID_AUTO, v_kthreadpages, CTLTYPE_UINT|CTLFLAG_RD, &cnt.v_kthreadpages, 0, vcnt, "IU", "VM pages affected by fork() by kernel"); SYSCTL_INT(_vm_stats_misc, OID_AUTO, zero_page_count, CTLFLAG_RD, &vm_page_zero_count, 0, ""); #if 0 SYSCTL_INT(_vm_stats_misc, OID_AUTO, page_mask, CTLFLAG_RD, &page_mask, 0, ""); SYSCTL_INT(_vm_stats_misc, OID_AUTO, page_shift, CTLFLAG_RD, &page_shift, 0, ""); SYSCTL_INT(_vm_stats_misc, OID_AUTO, first_page, CTLFLAG_RD, &first_page, 0, ""); SYSCTL_INT(_vm_stats_misc, OID_AUTO, last_page, CTLFLAG_RD, &last_page, 0, ""); #endif