nfs_bio.c

早期freebsd实现

/*
 * Copyright (c) 1989, 1993
 *	The Regents of the University of California.  All rights reserved.
 *
 * This code is derived from software contributed to Berkeley by
 * Rick Macklem at The University of Guelph.
 *
 * 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.
 *
 *	@(#)nfs_bio.c	8.5 (Berkeley) 1/4/94
 */

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/resourcevar.h>
#include <sys/proc.h>
#include <sys/buf.h>
#include <sys/vnode.h>
#include <sys/trace.h>
#include <sys/mount.h>
#include <sys/kernel.h>

#include <vm/vm.h>

#include <nfs/nfsnode.h>
#include <nfs/rpcv2.h>
#include <nfs/nfsv2.h>
#include <nfs/nfs.h>
#include <nfs/nfsmount.h>
#include <nfs/nqnfs.h>

struct buf *incore(), *nfs_getcacheblk();
extern struct proc *nfs_iodwant[NFS_MAXASYNCDAEMON];
extern int nfs_numasync;

/*
 * Vnode op for read using bio
 * Any similarity to readip() is purely coincidental
 */
nfs_bioread(vp, uio, ioflag, cred)
	register struct vnode *vp;
	register struct uio *uio;
	int ioflag;
	struct ucred *cred;
{
	register struct nfsnode *np = VTONFS(vp);
	register int biosize, diff;
	struct buf *bp, *rabp;
	struct vattr vattr;
	struct proc *p;
	struct nfsmount *nmp;
	daddr_t lbn, bn, rabn;
	caddr_t baddr;
	int got_buf, nra, error = 0, n, on, not_readin;

#ifdef lint
	ioflag = ioflag;
#endif /* lint */
#ifdef DIAGNOSTIC
	if (uio->uio_rw != UIO_READ)
		panic("nfs_read mode");
#endif
	if (uio->uio_resid == 0)
		return (0);
	if (uio->uio_offset < 0 && vp->v_type != VDIR)
		return (EINVAL);
	nmp = VFSTONFS(vp->v_mount);
	biosize = nmp->nm_rsize;
	p = uio->uio_procp;
	/*
	 * For nfs, cache consistency can only be maintained approximately.
	 * Although RFC1094 does not specify the criteria, the following is
	 * believed to be compatible with the reference port.
	 * For nqnfs, full cache consistency is maintained within the loop.
	 * For nfs:
	 * If the file's modify time on the server has changed since the
	 * last read rpc or you have written to the file,
	 * you may have lost data cache consistency with the
	 * server, so flush all of the file's data out of the cache.
	 * Then force a getattr rpc to ensure that you have up to date
	 * attributes.
	 * The mount flag NFSMNT_MYWRITE says "Assume that my writes are
	 * the ones changing the modify time.
	 * NB: This implies that cache data can be read when up to
	 * NFS_ATTRTIMEO seconds out of date. If you find that you need current
	 * attributes this could be forced by setting n_attrstamp to 0 before
	 * the VOP_GETATTR() call.
	 */
	if ((nmp->nm_flag & NFSMNT_NQNFS) == 0 && vp->v_type != VLNK) {
		if (np->n_flag & NMODIFIED) {
			if ((nmp->nm_flag & NFSMNT_MYWRITE) == 0 ||
			     vp->v_type != VREG) {
				if (error = nfs_vinvalbuf(vp, V_SAVE, cred, p, 1))
					return (error);
			}
			np->n_attrstamp = 0;
			np->n_direofoffset = 0;
			if (error = VOP_GETATTR(vp, &vattr, cred, p))
				return (error);
			np->n_mtime = vattr.va_mtime.ts_sec;
		} else {
			if (error = VOP_GETATTR(vp, &vattr, cred, p))
				return (error);
			if (np->n_mtime != vattr.va_mtime.ts_sec) {
				np->n_direofoffset = 0;
				if (error = nfs_vinvalbuf(vp, V_SAVE, cred, p, 1))
					return (error);
				np->n_mtime = vattr.va_mtime.ts_sec;
			}
		}
	}
	do {

	    /*
	     * Get a valid lease. If cached data is stale, flush it.
	     */
	    if (nmp->nm_flag & NFSMNT_NQNFS) {
		if (NQNFS_CKINVALID(vp, np, NQL_READ)) {
		    do {
			error = nqnfs_getlease(vp, NQL_READ, cred, p);
		    } while (error == NQNFS_EXPIRED);
		    if (error)
			return (error);
		    if (np->n_lrev != np->n_brev ||
			(np->n_flag & NQNFSNONCACHE) ||
			((np->n_flag & NMODIFIED) && vp->v_type == VDIR)) {
			if (vp->v_type == VDIR) {
			    np->n_direofoffset = 0;
			    cache_purge(vp);
			}
			if (error = nfs_vinvalbuf(vp, V_SAVE, cred, p, 1))
			    return (error);
			np->n_brev = np->n_lrev;
		    }
		} else if (vp->v_type == VDIR && (np->n_flag & NMODIFIED)) {
		    np->n_direofoffset = 0;
		    cache_purge(vp);
		    if (error = nfs_vinvalbuf(vp, V_SAVE, cred, p, 1))
			return (error);
		}
	    }
	    if (np->n_flag & NQNFSNONCACHE) {
		switch (vp->v_type) {
		case VREG:
			error = nfs_readrpc(vp, uio, cred);
			break;
		case VLNK:
			error = nfs_readlinkrpc(vp, uio, cred);
			break;
		case VDIR:
			error = nfs_readdirrpc(vp, uio, cred);
			break;
		};
		return (error);
	    }
	    baddr = (caddr_t)0;
	    switch (vp->v_type) {
	    case VREG:
		nfsstats.biocache_reads++;
		lbn = uio->uio_offset / biosize;
		on = uio->uio_offset & (biosize-1);
		bn = lbn * (biosize / DEV_BSIZE);
		not_readin = 1;

		/*
		 * Start the read ahead(s), as required.
		 */
		if (nfs_numasync > 0 && nmp->nm_readahead > 0 &&
		    lbn == vp->v_lastr + 1) {
		    for (nra = 0; nra < nmp->nm_readahead &&
			(lbn + 1 + nra) * biosize < np->n_size; nra++) {
			rabn = (lbn + 1 + nra) * (biosize / DEV_BSIZE);
			if (!incore(vp, rabn)) {
			    rabp = nfs_getcacheblk(vp, rabn, biosize, p);
			    if (!rabp)
				return (EINTR);
			    if ((rabp->b_flags & (B_DELWRI | B_DONE)) == 0) {
				rabp->b_flags |= (B_READ | B_ASYNC);
				if (nfs_asyncio(rabp, cred)) {
				    rabp->b_flags |= B_INVAL;
				    brelse(rabp);
				}
			    }
			}
		    }
		}

		/*
		 * If the block is in the cache and has the required data
		 * in a valid region, just copy it out.
		 * Otherwise, get the block and write back/read in,
		 * as required.
		 */
		if ((bp = incore(vp, bn)) &&
		    (bp->b_flags & (B_BUSY | B_WRITEINPROG)) ==
		    (B_BUSY | B_WRITEINPROG))
			got_buf = 0;
		else {
again:
			bp = nfs_getcacheblk(vp, bn, biosize, p);
			if (!bp)
				return (EINTR);
			got_buf = 1;
			if ((bp->b_flags & (B_DONE | B_DELWRI)) == 0) {
				bp->b_flags |= B_READ;
				not_readin = 0;
				if (error = nfs_doio(bp, cred, p)) {
				    brelse(bp);
				    return (error);
				}
			}
		}
		n = min((unsigned)(biosize - on), uio->uio_resid);
		diff = np->n_size - uio->uio_offset;
		if (diff < n)
			n = diff;
		if (not_readin && n > 0) {
			if (on < bp->b_validoff || (on + n) > bp->b_validend) {
				if (!got_buf) {
				    bp = nfs_getcacheblk(vp, bn, biosize, p);
				    if (!bp)
					return (EINTR);
				    got_buf = 1;
				}
				bp->b_flags |= B_INVAL;
				if (bp->b_dirtyend > 0) {
				    if ((bp->b_flags & B_DELWRI) == 0)
					panic("nfsbioread");
				    if (VOP_BWRITE(bp) == EINTR)
					return (EINTR);
				} else
				    brelse(bp);
				goto again;
			}
		}
		vp->v_lastr = lbn;
		diff = (on >= bp->b_validend) ? 0 : (bp->b_validend - on);
		if (diff < n)
			n = diff;
		break;
	    case VLNK:
		nfsstats.biocache_readlinks++;
		bp = nfs_getcacheblk(vp, (daddr_t)0, NFS_MAXPATHLEN, p);
		if (!bp)
			return (EINTR);
		if ((bp->b_flags & B_DONE) == 0) {
			bp->b_flags |= B_READ;
			if (error = nfs_doio(bp, cred, p)) {
				brelse(bp);
				return (error);
			}
		}
		n = min(uio->uio_resid, NFS_MAXPATHLEN - bp->b_resid);
		got_buf = 1;
		on = 0;
		break;
	    case VDIR:
		nfsstats.biocache_readdirs++;
		bn = (daddr_t)uio->uio_offset;
		bp = nfs_getcacheblk(vp, bn, NFS_DIRBLKSIZ, p);
		if (!bp)
			return (EINTR);
		if ((bp->b_flags & B_DONE) == 0) {
			bp->b_flags |= B_READ;
			if (error = nfs_doio(bp, cred, p)) {
				brelse(bp);
				return (error);
			}
		}

		/*
		 * If not eof and read aheads are enabled, start one.
		 * (You need the current block first, so that you have the
		 *  directory offset cookie of the next block.
		 */
		rabn = bp->b_blkno;
		if (nfs_numasync > 0 && nmp->nm_readahead > 0 &&
		    rabn != 0 && rabn != np->n_direofoffset &&
		    !incore(vp, rabn)) {
			rabp = nfs_getcacheblk(vp, rabn, NFS_DIRBLKSIZ, p);
			if (rabp) {
			    if ((rabp->b_flags & (B_DONE | B_DELWRI)) == 0) {
				rabp->b_flags |= (B_READ | B_ASYNC);
				if (nfs_asyncio(rabp, cred)) {
				    rabp->b_flags |= B_INVAL;
				    brelse(rabp);
				}
			    }
			}
		}
		on = 0;
		n = min(uio->uio_resid, NFS_DIRBLKSIZ - bp->b_resid);
		got_buf = 1;
		break;
	    };

	    if (n > 0) {
		if (!baddr)
			baddr = bp->b_data;
		error = uiomove(baddr + on, (int)n, uio);
	    }
	    switch (vp->v_type) {
	    case VREG:
		if (n + on == biosize || uio->uio_offset == np->n_size)
			bp->b_flags |= B_AGE;
		break;
	    case VLNK:
		n = 0;
		break;
	    case VDIR:
		uio->uio_offset = bp->b_blkno;
		break;
	    };
	    if (got_buf)
		brelse(bp);
	} while (error == 0 && uio->uio_resid > 0 && n > 0);
	return (error);
}

/*
 * Vnode op for write using bio
 */
nfs_write(ap)
	struct vop_write_args /* {
		struct vnode *a_vp;
		struct uio *a_uio;
		int  a_ioflag;
		struct ucred *a_cred;
	} */ *ap;
{
	register int biosize;
	register struct uio *uio = ap->a_uio;
	struct proc *p = uio->uio_procp;
	register struct vnode *vp = ap->a_vp;
	struct nfsnode *np = VTONFS(vp);
	register struct ucred *cred = ap->a_cred;
	int ioflag = ap->a_ioflag;
	struct buf *bp;
	struct vattr vattr;
	struct nfsmount *nmp;
	daddr_t lbn, bn;
	int n, on, error = 0;

#ifdef DIAGNOSTIC
	if (uio->uio_rw != UIO_WRITE)
		panic("nfs_write mode");
	if (uio->uio_segflg == UIO_USERSPACE && uio->uio_procp != curproc)
		panic("nfs_write proc");
#endif
	if (vp->v_type != VREG)
		return (EIO);
	if (np->n_flag & NWRITEERR) {
		np->n_flag &= ~NWRITEERR;
		return (np->n_error);
	}
	if (ioflag & (IO_APPEND | IO_SYNC)) {
		if (np->n_flag & NMODIFIED) {
			np->n_attrstamp = 0;
			if (error = nfs_vinvalbuf(vp, V_SAVE, cred, p, 1))
				return (error);
		}
		if (ioflag & IO_APPEND) {
			np->n_attrstamp = 0;
			if (error = VOP_GETATTR(vp, &vattr, cred, p))
				return (error);
			uio->uio_offset = np->n_size;
		}
	}
	nmp = VFSTONFS(vp->v_mount);
	if (uio->uio_offset < 0)
		return (EINVAL);
	if (uio->uio_resid == 0)
		return (0);
	/*