freebsd-skq/sys/dev/vinum/vinumrevive.c
Greg Lehey 458252a742 initsd: For striped and RAID-[45] plexes, don't restrict the I/O
transfer size to the stripe size.  This is a different
        situation from reviving, where this limitation is necessary.
        In initsd we're simply writing binary zeroes to the entire
        disk, so the only effect of limiting the transfer is to slow
        things down.
2003-12-15 00:45:53 +00:00

621 lines
21 KiB
C

/*-
* Copyright (c) 1997, 1998, 1999
* Nan Yang Computer Services Limited. All rights reserved.
*
* Parts copyright (c) 1997, 1998 Cybernet Corporation, NetMAX project.
*
* Written by Greg Lehey
*
* This software is distributed under the so-called ``Berkeley
* License'':
*
* 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 Nan Yang Computer
* Services Limited.
* 4. Neither the name of the Company 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 ``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 company 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.
*
* $Id: vinumrevive.c,v 1.19 2003/05/08 04:34:47 grog Exp grog $
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <dev/vinum/vinumhdr.h>
#include <dev/vinum/request.h>
/*
* Revive a block of a subdisk. Return an error
* indication. EAGAIN means successful copy, but
* that more blocks remain to be copied. EINVAL
* means that the subdisk isn't associated with a
* plex (which means a programming error if we get
* here at all; FIXME).
*/
int
revive_block(int sdno)
{
int s; /* priority level */
struct sd *sd;
struct plex *plex;
struct volume *vol;
struct buf *bp;
int error = EAGAIN;
int size; /* size of revive block, bytes */
daddr_t plexblkno; /* lblkno in plex */
int psd; /* parity subdisk number */
u_int64_t stripe; /* stripe number */
int paritysd = 0; /* set if this is the parity stripe */
struct rangelock *lock; /* for locking */
daddr_t stripeoffset; /* offset in stripe */
plexblkno = 0; /* to keep the compiler happy */
sd = &SD[sdno];
lock = NULL;
if (sd->plexno < 0) /* no plex? */
return EINVAL;
plex = &PLEX[sd->plexno]; /* point to plex */
if (plex->volno >= 0)
vol = &VOL[plex->volno];
else
vol = NULL;
if ((sd->revive_blocksize == 0) /* no block size */
||(sd->revive_blocksize & ((1 << DEV_BSHIFT) - 1))) /* or invalid block size */
sd->revive_blocksize = DEFAULT_REVIVE_BLOCKSIZE;
else if (sd->revive_blocksize > MAX_REVIVE_BLOCKSIZE)
sd->revive_blocksize = MAX_REVIVE_BLOCKSIZE;
size = min(sd->revive_blocksize >> DEV_BSHIFT, sd->sectors - sd->revived) << DEV_BSHIFT;
sd->reviver = curproc->p_pid; /* note who last had a bash at it */
/* Now decide where to read from */
switch (plex->organization) {
case plex_concat:
plexblkno = sd->revived + sd->plexoffset; /* corresponding address in plex */
break;
case plex_striped:
stripeoffset = sd->revived % plex->stripesize; /* offset from beginning of stripe */
if (stripeoffset + (size >> DEV_BSHIFT) > plex->stripesize)
size = (plex->stripesize - stripeoffset) << DEV_BSHIFT;
plexblkno = sd->plexoffset /* base */
+ (sd->revived - stripeoffset) * plex->subdisks /* offset to beginning of stripe */
+ stripeoffset; /* offset from beginning of stripe */
break;
case plex_raid4:
case plex_raid5:
stripeoffset = sd->revived % plex->stripesize; /* offset from beginning of stripe */
plexblkno = sd->plexoffset /* base */
+ (sd->revived - stripeoffset) * (plex->subdisks - 1) /* offset to beginning of stripe */
+stripeoffset; /* offset from beginning of stripe */
stripe = (sd->revived / plex->stripesize); /* stripe number */
/* Make sure we don't go beyond the end of the band. */
size = min(size, (plex->stripesize - stripeoffset) << DEV_BSHIFT);
if (plex->organization == plex_raid4)
psd = plex->subdisks - 1; /* parity subdisk for this stripe */
else
psd = plex->subdisks - 1 - stripe % plex->subdisks; /* parity subdisk for this stripe */
paritysd = plex->sdnos[psd] == sdno; /* note if it's the parity subdisk */
/*
* Now adjust for the strangenesses
* in RAID-4 and RAID-5 striping.
*/
if (sd->plexsdno > psd) /* beyond the parity stripe, */
plexblkno -= plex->stripesize; /* one stripe less */
else if (paritysd)
plexblkno -= plex->stripesize * sd->plexsdno; /* go back to the beginning of the band */
break;
case plex_disorg: /* to keep the compiler happy */
break; /* to keep the pedants happy */
}
if (paritysd) { /* we're reviving a parity block, */
bp = parityrebuild(plex, sd->revived, size, rebuildparity, &lock, NULL); /* do the grunt work */
if (bp == NULL) /* no buffer space */
return ENOMEM; /* chicken out */
} else { /* data block */
s = splbio();
bp = geteblk(size); /* Get a buffer */
splx(s);
if (bp == NULL)
return ENOMEM;
/*
* Amount to transfer: block size, unless it
* would overlap the end.
*/
bp->b_bcount = size;
bp->b_resid = bp->b_bcount;
bp->b_blkno = plexblkno; /* start here */
if (isstriped(plex)) /* we need to lock striped plexes */
lock = lockrange(plexblkno << DEV_BSHIFT, bp, plex); /* lock it */
if (vol != NULL) /* it's part of a volume, */
/*
* First, read the data from the volume. We
* don't care which plex, that's bre's job.
*/
bp->b_dev = VOL[plex->volno].dev; /* create the device number */
else /* it's an unattached plex */
bp->b_dev = PLEX[sd->plexno].dev; /* create the device number */
bp->b_iocmd = BIO_READ; /* either way, read it */
bp->b_flags = 0;
vinumstart(bp, 1);
bufwait(bp);
}
if (bp->b_ioflags & BIO_ERROR) {
error = bp->b_error;
if (lock) /* we took a lock, */
unlockrange(sd->plexno, lock); /* give it back */
} else
/* Now write to the subdisk */
{
bp->b_dev = SD[sdno].dev; /* create the device number */
bp->b_flags &= ~B_DONE; /* no longer done */
bp->b_ioflags = 0;
bp->b_iocmd = BIO_WRITE;
bp->b_resid = bp->b_bcount;
bp->b_blkno = sd->revived; /* write it to here */
sdio(bp); /* perform the I/O */
bufwait(bp);
if (bp->b_ioflags & BIO_ERROR)
error = bp->b_error;
else {
sd->revived += bp->b_bcount >> DEV_BSHIFT; /* moved this much further down */
if (sd->revived >= sd->sectors) { /* finished */
sd->revived = 0;
set_sd_state(sdno, sd_up, setstate_force); /* bring the sd up */
log(LOG_INFO, "vinum: %s is %s\n", sd->name, sd_state(sd->state));
save_config(); /* and save the updated configuration */
error = 0; /* we're done */
}
}
if (lock) /* we took a lock, */
unlockrange(sd->plexno, lock); /* give it back */
while (sd->waitlist) { /* we have waiting requests */
#ifdef VINUMDEBUG
struct request *rq = sd->waitlist;
if (debug & DEBUG_REVIVECONFLICT)
log(LOG_DEBUG,
"Relaunch revive conflict sd %d: %p\n%s dev %d.%d, offset 0x%jx, length %ld\n",
rq->sdno,
rq,
rq->bp->b_iocmd == BIO_READ ? "Read" : "Write",
major(rq->bp->b_dev),
minor(rq->bp->b_dev),
(intmax_t) rq->bp->b_blkno,
rq->bp->b_bcount);
#endif
launch_requests(sd->waitlist, 1); /* do them now */
sd->waitlist = sd->waitlist->next; /* and move on to the next */
}
}
if (bp->b_qindex == 0) { /* not on a queue, */
bp->b_flags |= B_INVAL;
bp->b_ioflags &= ~BIO_ERROR;
brelse(bp); /* is this kosher? */
}
return error;
}
/*
* Check or rebuild the parity blocks of a RAID-4
* or RAID-5 plex.
*
* The variables plex->checkblock and
* plex->rebuildblock represent the
* subdisk-relative address of the stripe we're
* looking at, not the plex-relative address. We
* store it in the plex and not as a local
* variable because this function could be
* stopped, and we don't want to repeat the part
* we've already done. This is also the reason
* why we don't initialize it here except at the
* end. It gets initialized with the plex on
* creation.
*
* Each call to this function processes at most
* one stripe. We can't loop in this function,
* because we're unstoppable, so we have to be
* called repeatedly from userland.
*/
void
parityops(struct vinum_ioctl_msg *data)
{
int plexno;
struct plex *plex;
int size; /* I/O transfer size, bytes */
int stripe; /* stripe number in plex */
int psd; /* parity subdisk number */
struct rangelock *lock; /* lock on stripe */
struct _ioctl_reply *reply;
off_t pstripe; /* pointer to our stripe counter */
struct buf *pbp;
off_t errorloc; /* offset of parity error */
enum parityop op; /* operation to perform */
plexno = data->index;
op = data->op;
pbp = NULL;
reply = (struct _ioctl_reply *) data;
reply->error = EAGAIN; /* expect to repeat this call */
plex = &PLEX[plexno];
if (!isparity(plex)) { /* not RAID-4 or RAID-5 */
reply->error = EINVAL;
return;
} else if (plex->state < plex_flaky) {
reply->error = EIO;
strcpy(reply->msg, "Plex is not completely accessible\n");
return;
}
pstripe = data->offset;
stripe = pstripe / plex->stripesize; /* stripe number */
psd = plex->subdisks - 1 - stripe % plex->subdisks; /* parity subdisk for this stripe */
size = min(DEFAULT_REVIVE_BLOCKSIZE, /* one block at a time */
plex->stripesize << DEV_BSHIFT);
pbp = parityrebuild(plex, pstripe, size, op, &lock, &errorloc); /* do the grunt work */
if (pbp == NULL) { /* no buffer space */
reply->error = ENOMEM;
return; /* chicken out */
}
/*
* Now we have a result in the data buffer of
* the parity buffer header, which we have kept.
* Decide what to do with it.
*/
reply->msg[0] = '\0'; /* until shown otherwise */
if ((pbp->b_ioflags & BIO_ERROR) == 0) { /* no error */
if ((op == rebuildparity)
|| (op == rebuildandcheckparity)) {
pbp->b_iocmd = BIO_WRITE;
pbp->b_resid = pbp->b_bcount;
sdio(pbp); /* write the parity block */
bufwait(pbp);
}
if (((op == checkparity)
|| (op == rebuildandcheckparity))
&& (errorloc != -1)) {
if (op == checkparity)
reply->error = EIO;
sprintf(reply->msg,
"Parity incorrect at offset 0x%jx\n",
(intmax_t) errorloc);
}
if (reply->error == EAGAIN) { /* still OK, */
plex->checkblock = pstripe + (pbp->b_bcount >> DEV_BSHIFT); /* moved this much further down */
if (plex->checkblock >= SD[plex->sdnos[0]].sectors) { /* finished */
plex->checkblock = 0;
reply->error = 0;
}
}
}
if (pbp->b_ioflags & BIO_ERROR)
reply->error = pbp->b_error;
pbp->b_flags |= B_INVAL;
pbp->b_ioflags &= ~BIO_ERROR;
brelse(pbp);
unlockrange(plexno, lock);
}
/*
* Rebuild a parity stripe. Return pointer to
* parity bp. On return,
*
* 1. The band is locked. The caller must unlock
* the band and release the buffer header.
*
* 2. All buffer headers except php have been
* released. The caller must release pbp.
*
* 3. For checkparity and rebuildandcheckparity,
* the parity is compared with the current
* parity block. If it's different, the
* offset of the error is returned to
* errorloc. The caller can set the value of
* the pointer to NULL if this is called for
* rebuilding parity.
*
* pstripe is the subdisk-relative base address of
* the data to be reconstructed, size is the size
* of the transfer in bytes.
*/
struct buf *
parityrebuild(struct plex *plex,
u_int64_t pstripe,
int size,
enum parityop op,
struct rangelock **lockp,
off_t * errorloc)
{
int error;
int s;
int sdno;
u_int64_t stripe; /* stripe number */
int *parity_buf; /* buffer address for current parity block */
int *newparity_buf; /* and for new parity block */
int mysize; /* I/O transfer size for this transfer */
int isize; /* mysize in ints */
int i;
int psd; /* parity subdisk number */
int newpsd; /* and "subdisk number" of new parity */
struct buf **bpp; /* pointers to our bps */
struct buf *pbp; /* buffer header for parity stripe */
int *sbuf;
int bufcount; /* number of buffers we need */
stripe = pstripe / plex->stripesize; /* stripe number */
psd = plex->subdisks - 1 - stripe % plex->subdisks; /* parity subdisk for this stripe */
parity_buf = NULL; /* to keep the compiler happy */
error = 0;
/*
* It's possible that the default transfer size
* we chose is not a factor of the stripe size.
* We *must* limit this operation to a single
* stripe, at least for RAID-5 rebuild, since
* the parity subdisk changes between stripes,
* so in this case we need to perform a short
* transfer. Set variable mysize to reflect
* this.
*/
mysize = min(size, (plex->stripesize * (stripe + 1) - pstripe) << DEV_BSHIFT);
isize = mysize / (sizeof(int)); /* number of ints in the buffer */
bufcount = plex->subdisks + 1; /* sd buffers plus result buffer */
newpsd = plex->subdisks;
bpp = (struct buf **) Malloc(bufcount * sizeof(struct buf *)); /* array of pointers to bps */
/* First, build requests for all subdisks */
for (sdno = 0; sdno < bufcount; sdno++) { /* for each subdisk */
if ((sdno != psd) || (op != rebuildparity)) {
/* Get a buffer header and initialize it. */
s = splbio();
bpp[sdno] = geteblk(mysize); /* Get a buffer */
if (bpp[sdno] == NULL) {
while (sdno-- > 0) { /* release the ones we got */
bpp[sdno]->b_flags |= B_INVAL;
brelse(bpp[sdno]); /* give back our resources */
}
splx(s);
printf("vinum: can't allocate buffer space for parity op.\n");
return NULL; /* no bpps */
}
splx(s);
if (sdno == psd)
parity_buf = (int *) bpp[sdno]->b_data;
if (sdno == newpsd) /* the new one? */
bpp[sdno]->b_dev = SD[plex->sdnos[psd]].dev; /* write back to the parity SD */
else
bpp[sdno]->b_dev = SD[plex->sdnos[sdno]].dev; /* device number */
bpp[sdno]->b_iocmd = BIO_READ; /* either way, read it */
bpp[sdno]->b_flags = 0;
bpp[sdno]->b_bcount = mysize;
bpp[sdno]->b_resid = bpp[sdno]->b_bcount;
bpp[sdno]->b_blkno = pstripe; /* transfer from here */
}
}
/* Initialize result buffer */
pbp = bpp[newpsd];
newparity_buf = (int *) bpp[newpsd]->b_data;
bzero(newparity_buf, mysize);
/*
* Now lock the stripe with the first non-parity
* bp as locking bp.
*/
*lockp = lockrange(pstripe * plex->stripesize * (plex->subdisks - 1),
bpp[psd ? 0 : 1],
plex);
/*
* Then issue requests for all subdisks in
* parallel. Don't transfer the parity stripe
* if we're rebuilding parity, unless we also
* want to check it.
*/
for (sdno = 0; sdno < plex->subdisks; sdno++) { /* for each real subdisk */
if ((sdno != psd) || (op != rebuildparity)) {
sdio(bpp[sdno]);
}
}
/*
* Next, wait for the requests to complete.
* We wait in the order in which they were
* issued, which isn't necessarily the order in
* which they complete, but we don't have a
* convenient way of doing the latter, and the
* delay is minimal.
*/
for (sdno = 0; sdno < plex->subdisks; sdno++) { /* for each subdisk */
if ((sdno != psd) || (op != rebuildparity)) {
bufwait(bpp[sdno]);
if (bpp[sdno]->b_ioflags & BIO_ERROR) /* can't read, */
error = bpp[sdno]->b_error;
else if (sdno != psd) { /* update parity */
sbuf = (int *) bpp[sdno]->b_data;
for (i = 0; i < isize; i++)
((int *) newparity_buf)[i] ^= sbuf[i]; /* xor in the buffer */
}
}
if (sdno != psd) { /* release all bps except parity */
bpp[sdno]->b_flags |= B_INVAL;
brelse(bpp[sdno]); /* give back our resources */
}
}
/*
* If we're checking, compare the calculated
* and the read parity block. If they're
* different, return the plex-relative offset;
* otherwise return -1.
*/
if ((op == checkparity)
|| (op == rebuildandcheckparity)) {
*errorloc = -1; /* no error yet */
for (i = 0; i < isize; i++) {
if (parity_buf[i] != newparity_buf[i]) {
*errorloc = (off_t) (pstripe << DEV_BSHIFT) * (plex->subdisks - 1)
+ i * sizeof(int);
break;
}
}
bpp[psd]->b_flags |= B_INVAL;
brelse(bpp[psd]); /* give back our resources */
}
/* release our resources */
Free(bpp);
if (error) {
pbp->b_ioflags |= BIO_ERROR;
pbp->b_error = error;
}
return pbp;
}
/*
* Initialize a subdisk by writing zeroes to the
* complete address space. If verify is set,
* check each transfer for correctness.
*
* Each call to this function writes (and maybe
* checks) a single block.
*/
int
initsd(int sdno, int verify)
{
int s; /* priority level */
struct sd *sd;
struct plex *plex;
struct volume *vol;
struct buf *bp;
int error;
int size; /* size of init block, bytes */
daddr_t plexblkno; /* lblkno in plex */
int verified; /* set when we're happy with what we wrote */
error = 0;
plexblkno = 0; /* to keep the compiler happy */
sd = &SD[sdno];
if (sd->plexno < 0) /* no plex? */
return EINVAL;
plex = &PLEX[sd->plexno]; /* point to plex */
if (plex->volno >= 0)
vol = &VOL[plex->volno];
else
vol = NULL;
if (sd->init_blocksize == 0) {
sd->init_blocksize = DEFAULT_REVIVE_BLOCKSIZE;
} else if (sd->init_blocksize > MAX_REVIVE_BLOCKSIZE)
sd->init_blocksize = MAX_REVIVE_BLOCKSIZE;
size = min(sd->init_blocksize >> DEV_BSHIFT, sd->sectors - sd->initialized) << DEV_BSHIFT;
verified = 0;
while (!verified) { /* until we're happy with it, */
s = splbio();
bp = geteblk(size); /* Get a buffer */
splx(s);
if (bp == NULL)
return ENOMEM;
bp->b_bcount = size;
bp->b_resid = bp->b_bcount;
bp->b_blkno = sd->initialized; /* write it to here */
bzero(bp->b_data, bp->b_bcount);
bp->b_dev = SD[sdno].dev; /* create the device number */
bp->b_iocmd = BIO_WRITE;
sdio(bp); /* perform the I/O */
bufwait(bp);
if (bp->b_ioflags & BIO_ERROR)
error = bp->b_error;
if (bp->b_qindex == 0) { /* not on a queue, */
bp->b_flags |= B_INVAL;
bp->b_ioflags &= ~BIO_ERROR;
brelse(bp); /* is this kosher? */
}
if ((error == 0) && verify) { /* check that it got there */
s = splbio();
bp = geteblk(size); /* get a buffer */
if (bp == NULL) {
splx(s);
error = ENOMEM;
} else {
bp->b_bcount = size;
bp->b_resid = bp->b_bcount;
bp->b_blkno = sd->initialized; /* read from here */
bp->b_dev = SD[sdno].dev; /* create the device number */
bp->b_iocmd = BIO_READ; /* read it back */
splx(s);
sdio(bp);
bufwait(bp);
/*
* XXX Bug fix code. This is hopefully no
* longer needed (21 February 2000).
*/
if (bp->b_ioflags & BIO_ERROR)
error = bp->b_error;
else if ((*bp->b_data != 0) /* first word spammed */
||(bcmp(bp->b_data, &bp->b_data[1], bp->b_bcount - 1))) { /* or one of the others */
printf("vinum: init error on %s, offset 0x%llx sectors\n",
sd->name,
(long long) sd->initialized);
verified = 0;
} else
verified = 1;
if (bp->b_qindex == 0) { /* not on a queue, */
bp->b_flags |= B_INVAL;
bp->b_ioflags &= ~BIO_ERROR;
brelse(bp); /* is this kosher? */
}
}
} else
verified = 1;
}
if (error == 0) { /* did it, */
sd->initialized += size >> DEV_BSHIFT; /* moved this much further down */
if (sd->initialized >= sd->sectors) { /* finished */
sd->initialized = 0;
set_sd_state(sdno, sd_initialized, setstate_force); /* bring the sd up */
log(LOG_INFO, "vinum: %s is %s\n", sd->name, sd_state(sd->state));
save_config(); /* and save the updated configuration */
} else /* more to go, */
error = EAGAIN; /* ya'll come back, see? */
}
return error;
}
/* Local Variables: */
/* fill-column: 50 */
/* End: */