freebsd-nq/sys/kern/subr_disk.c

602 lines
14 KiB
C
Raw Normal View History

/*
* ----------------------------------------------------------------------------
* "THE BEER-WARE LICENSE" (Revision 42):
* <phk@FreeBSD.ORG> wrote this file. As long as you retain this notice you
* can do whatever you want with this stuff. If we meet some day, and you think
* this stuff is worth it, you can buy me a beer in return. Poul-Henning Kamp
* ----------------------------------------------------------------------------
*
* $FreeBSD$
*
*/
#include "opt_geom.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/stdint.h>
#include <sys/bio.h>
#include <sys/conf.h>
#include <sys/disk.h>
#include <sys/diskslice.h>
#include <sys/disklabel.h>
#ifdef NO_GEOM
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/sysctl.h>
#include <machine/md_var.h>
#include <sys/ctype.h>
2000-12-08 20:09:00 +00:00
static MALLOC_DEFINE(M_DISK, "disk", "disk data");
static d_strategy_t diskstrategy;
static d_open_t diskopen;
static d_close_t diskclose;
static d_ioctl_t diskioctl;
static d_psize_t diskpsize;
static LIST_HEAD(, disk) disklist = LIST_HEAD_INITIALIZER(&disklist);
void disk_dev_synth(dev_t dev);
void
disk_dev_synth(dev_t dev)
{
struct disk *dp;
int u, s, p;
dev_t pdev;
if (dksparebits(dev))
return;
LIST_FOREACH(dp, &disklist, d_list) {
if (major(dev) != dp->d_devsw->d_maj)
continue;
u = dkunit(dev);
p = RAW_PART;
s = WHOLE_DISK_SLICE;
pdev = makedev(dp->d_devsw->d_maj, dkmakeminor(u, s, p));
if (pdev->si_devsw == NULL)
return; /* Probably a unit we don't have */
s = dkslice(dev);
p = dkpart(dev);
if (s == WHOLE_DISK_SLICE && p == RAW_PART) {
/* XXX: actually should not happen */
dev = make_dev(pdev->si_devsw, dkmakeminor(u, s, p),
UID_ROOT, GID_OPERATOR, 0640, "%s%d",
dp->d_devsw->d_name, u);
dev_depends(pdev, dev);
return;
}
if (s == COMPATIBILITY_SLICE) {
dev = make_dev(pdev->si_devsw, dkmakeminor(u, s, p),
UID_ROOT, GID_OPERATOR, 0640, "%s%d%c",
dp->d_devsw->d_name, u, 'a' + p);
dev_depends(pdev, dev);
return;
}
if (p != RAW_PART) {
dev = make_dev(pdev->si_devsw, dkmakeminor(u, s, p),
UID_ROOT, GID_OPERATOR, 0640, "%s%ds%d%c",
dp->d_devsw->d_name, u, s - BASE_SLICE + 1,
'a' + p);
} else {
dev = make_dev(pdev->si_devsw, dkmakeminor(u, s, p),
UID_ROOT, GID_OPERATOR, 0640, "%s%ds%d",
dp->d_devsw->d_name, u, s - BASE_SLICE + 1);
make_dev_alias(dev, "%s%ds%dc",
dp->d_devsw->d_name, u, s - BASE_SLICE + 1);
}
dev_depends(pdev, dev);
return;
}
}
static void
disk_clone(void *arg, char *name, int namelen, dev_t *dev)
{
struct disk *dp;
char const *d;
char *e;
int j, u, s, p;
dev_t pdev;
if (*dev != NODEV)
return;
LIST_FOREACH(dp, &disklist, d_list) {
d = dp->d_devsw->d_name;
j = dev_stdclone(name, &e, d, &u);
if (j == 0)
continue;
if (u > DKMAXUNIT)
continue;
p = RAW_PART;
s = WHOLE_DISK_SLICE;
pdev = makedev(dp->d_devsw->d_maj, dkmakeminor(u, s, p));
if (pdev->si_disk == NULL)
continue;
if (*e != '\0') {
j = dev_stdclone(e, &e, "s", &s);
if (j == 0)
s = COMPATIBILITY_SLICE;
else if (j == 1 || j == 2)
s += BASE_SLICE - 1;
if (!*e)
; /* ad0s1 case */
else if (e[1] != '\0')
return; /* can never be a disk name */
else if (*e < 'a' || *e > 'h')
return; /* can never be a disk name */
else
p = *e - 'a';
}
if (s == WHOLE_DISK_SLICE && p == RAW_PART) {
return;
} else if (s >= BASE_SLICE && p != RAW_PART) {
*dev = make_dev(pdev->si_devsw, dkmakeminor(u, s, p),
UID_ROOT, GID_OPERATOR, 0640, "%s%ds%d%c",
pdev->si_devsw->d_name, u, s - BASE_SLICE + 1,
p + 'a');
} else if (s >= BASE_SLICE) {
*dev = make_dev(pdev->si_devsw, dkmakeminor(u, s, p),
UID_ROOT, GID_OPERATOR, 0640, "%s%ds%d",
pdev->si_devsw->d_name, u, s - BASE_SLICE + 1);
make_dev_alias(*dev, "%s%ds%dc",
pdev->si_devsw->d_name, u, s - BASE_SLICE + 1);
} else {
*dev = make_dev(pdev->si_devsw, dkmakeminor(u, s, p),
UID_ROOT, GID_OPERATOR, 0640, "%s%d%c",
pdev->si_devsw->d_name, u, p + 'a');
}
dev_depends(pdev, *dev);
return;
}
}
End two weeks of on and off debugging. Fix the crash on the Nth insertion of a CF card, for random values of N > 1. With these fixes, I've been able to do 100 insert/remove of the cards w/o a crash with lots of system activity going on that in the past would help trigger the crash. The problem: FreeBSD creates dev_t's on the fly as they are needed and never destroys them. These dev_t's point to a struct disk that is used for housekeeping on the disk. When a device goes away, the struct disk pointer becomes a dangling pointer. Sometimes when the device comes back, the pointer will point to the new struct disk (in which case the insertion will work). Other times it won't (especially if any length of time has passed, since it is dependent on memory returned from malloc). The Fix: There is one of these dev_t's that is always correct. The device for the WHOLE_DISK_SLICE is always right. It gets set at create_disk() time. So, the fix is to spend a little CPU time and lookup the WHOLE_DISK_SLICE dev_t and use the si_disk from that in preference to the one that's in the device asking to do the I/O. In addition, we change the test of si_disk == NULL meaning that the dev needed to inherit properties from the pdev to dev->si_disk != pdev->si_disk. This test is a little stronger than the previous test, but can sometimes be fooled into not inheriting. However, the results of this fooling are that the old values will be used, which will generally always be the same as before. si_drv[12] are the only values that are copied that might pose a problem. They tend to change as the si_disk field would change, so it is a hole, but it is a small hole. One could correctly argue that one should replace much of this code with something much much better. I would be on the pro side of that argument. Reviewed by: phk (who also ported the original patch to current) Sponsored by: Timing Solutions
2000-07-05 06:01:33 +00:00
static void
inherit_raw(dev_t pdev, dev_t dev)
{
dev->si_disk = pdev->si_disk;
dev->si_drv1 = pdev->si_drv1;
dev->si_drv2 = pdev->si_drv2;
dev->si_iosize_max = pdev->si_iosize_max;
dev->si_bsize_phys = pdev->si_bsize_phys;
dev->si_bsize_best = pdev->si_bsize_best;
}
dev_t
disk_create(int unit, struct disk *dp, int flags, struct cdevsw *cdevsw, struct cdevsw *proto)
{
static int once;
dev_t dev;
if (!once) {
EVENTHANDLER_REGISTER(dev_clone, disk_clone, 0, 1000);
once++;
}
bzero(dp, sizeof(*dp));
dp->d_label = malloc(sizeof *dp->d_label, M_DEVBUF, M_WAITOK|M_ZERO);
if (proto->d_open != diskopen) {
*proto = *cdevsw;
proto->d_open = diskopen;
proto->d_close = diskclose;
proto->d_ioctl = diskioctl;
proto->d_strategy = diskstrategy;
proto->d_psize = diskpsize;
}
1999-11-19 23:34:01 +00:00
if (bootverbose)
printf("Creating DISK %s%d\n", cdevsw->d_name, unit);
dev = make_dev(proto, dkmakeminor(unit, WHOLE_DISK_SLICE, RAW_PART),
UID_ROOT, GID_OPERATOR, 0640, "%s%d", cdevsw->d_name, unit);
dev->si_disk = dp;
dp->d_dev = dev;
dp->d_dsflags = flags;
dp->d_devsw = cdevsw;
LIST_INSERT_HEAD(&disklist, dp, d_list);
return (dev);
}
Here follows the new kernel dumping infrastructure. Caveats: The new savecore program is not complete in the sense that it emulates enough of the old savecores features to do the job, but implements none of the options yet. I would appreciate if a userland hacker could help me out getting savecore to do what we want it to do from a users point of view, compression, email-notification, space reservation etc etc. (send me email if you are interested). Currently, savecore will scan all devices marked as "swap" or "dump" in /etc/fstab _or_ any devices specified on the command-line. All architectures but i386 lack an implementation of dumpsys(), but looking at the i386 version it should be trivial for anybody familiar with the platform(s) to provide this function. Documentation is quite sparse at this time, more to come. Details: ATA and SCSI drivers should work as the dump formatting code has been removed. The IDA, TWE and AAC have not yet been converted. Dumpon now opens the device and uses ioctl(DIOCGKERNELDUMP) to set the device as dumpdev. To implement the "off" argument, /dev/null is used as the device. Savecore will fail if handed any options since they are not (yet) implemented. All devices marked "dump" or "swap" in /etc/fstab will be scanned and dumps found will be saved to diskfiles named from the MD5 hash of the header record. The header record is dumped in readable format in the .info file. The kernel is not saved. Only complete dumps will be saved. All maintainer rights for this code are disclaimed: feel free to improve and extend. Sponsored by: DARPA, NAI Labs
2002-03-31 22:37:00 +00:00
static int
diskdumpconf(u_int onoff, dev_t dev, struct disk *dp)
{
Here follows the new kernel dumping infrastructure. Caveats: The new savecore program is not complete in the sense that it emulates enough of the old savecores features to do the job, but implements none of the options yet. I would appreciate if a userland hacker could help me out getting savecore to do what we want it to do from a users point of view, compression, email-notification, space reservation etc etc. (send me email if you are interested). Currently, savecore will scan all devices marked as "swap" or "dump" in /etc/fstab _or_ any devices specified on the command-line. All architectures but i386 lack an implementation of dumpsys(), but looking at the i386 version it should be trivial for anybody familiar with the platform(s) to provide this function. Documentation is quite sparse at this time, more to come. Details: ATA and SCSI drivers should work as the dump formatting code has been removed. The IDA, TWE and AAC have not yet been converted. Dumpon now opens the device and uses ioctl(DIOCGKERNELDUMP) to set the device as dumpdev. To implement the "off" argument, /dev/null is used as the device. Savecore will fail if handed any options since they are not (yet) implemented. All devices marked "dump" or "swap" in /etc/fstab will be scanned and dumps found will be saved to diskfiles named from the MD5 hash of the header record. The header record is dumped in readable format in the .info file. The kernel is not saved. Only complete dumps will be saved. All maintainer rights for this code are disclaimed: feel free to improve and extend. Sponsored by: DARPA, NAI Labs
2002-03-31 22:37:00 +00:00
struct dumperinfo di;
struct disklabel *dl;
Here follows the new kernel dumping infrastructure. Caveats: The new savecore program is not complete in the sense that it emulates enough of the old savecores features to do the job, but implements none of the options yet. I would appreciate if a userland hacker could help me out getting savecore to do what we want it to do from a users point of view, compression, email-notification, space reservation etc etc. (send me email if you are interested). Currently, savecore will scan all devices marked as "swap" or "dump" in /etc/fstab _or_ any devices specified on the command-line. All architectures but i386 lack an implementation of dumpsys(), but looking at the i386 version it should be trivial for anybody familiar with the platform(s) to provide this function. Documentation is quite sparse at this time, more to come. Details: ATA and SCSI drivers should work as the dump formatting code has been removed. The IDA, TWE and AAC have not yet been converted. Dumpon now opens the device and uses ioctl(DIOCGKERNELDUMP) to set the device as dumpdev. To implement the "off" argument, /dev/null is used as the device. Savecore will fail if handed any options since they are not (yet) implemented. All devices marked "dump" or "swap" in /etc/fstab will be scanned and dumps found will be saved to diskfiles named from the MD5 hash of the header record. The header record is dumped in readable format in the .info file. The kernel is not saved. Only complete dumps will be saved. All maintainer rights for this code are disclaimed: feel free to improve and extend. Sponsored by: DARPA, NAI Labs
2002-03-31 22:37:00 +00:00
if (!onoff)
return(set_dumper(NULL));
dl = dsgetlabel(dev, dp->d_slice);
if (!dl)
return (ENXIO);
Here follows the new kernel dumping infrastructure. Caveats: The new savecore program is not complete in the sense that it emulates enough of the old savecores features to do the job, but implements none of the options yet. I would appreciate if a userland hacker could help me out getting savecore to do what we want it to do from a users point of view, compression, email-notification, space reservation etc etc. (send me email if you are interested). Currently, savecore will scan all devices marked as "swap" or "dump" in /etc/fstab _or_ any devices specified on the command-line. All architectures but i386 lack an implementation of dumpsys(), but looking at the i386 version it should be trivial for anybody familiar with the platform(s) to provide this function. Documentation is quite sparse at this time, more to come. Details: ATA and SCSI drivers should work as the dump formatting code has been removed. The IDA, TWE and AAC have not yet been converted. Dumpon now opens the device and uses ioctl(DIOCGKERNELDUMP) to set the device as dumpdev. To implement the "off" argument, /dev/null is used as the device. Savecore will fail if handed any options since they are not (yet) implemented. All devices marked "dump" or "swap" in /etc/fstab will be scanned and dumps found will be saved to diskfiles named from the MD5 hash of the header record. The header record is dumped in readable format in the .info file. The kernel is not saved. Only complete dumps will be saved. All maintainer rights for this code are disclaimed: feel free to improve and extend. Sponsored by: DARPA, NAI Labs
2002-03-31 22:37:00 +00:00
bzero(&di, sizeof di);
di.dumper = (dumper_t *)dp->d_devsw->d_dump;
di.priv = dp->d_dev;
di.blocksize = dl->d_secsize;
di.mediaoffset = (off_t)(dl->d_partitions[dkpart(dev)].p_offset +
dp->d_slice->dss_slices[dkslice(dev)].ds_offset) * DEV_BSIZE;
di.mediasize =
(off_t)(dl->d_partitions[dkpart(dev)].p_size) * DEV_BSIZE;
if (di.mediasize == 0)
return (EINVAL);
Here follows the new kernel dumping infrastructure. Caveats: The new savecore program is not complete in the sense that it emulates enough of the old savecores features to do the job, but implements none of the options yet. I would appreciate if a userland hacker could help me out getting savecore to do what we want it to do from a users point of view, compression, email-notification, space reservation etc etc. (send me email if you are interested). Currently, savecore will scan all devices marked as "swap" or "dump" in /etc/fstab _or_ any devices specified on the command-line. All architectures but i386 lack an implementation of dumpsys(), but looking at the i386 version it should be trivial for anybody familiar with the platform(s) to provide this function. Documentation is quite sparse at this time, more to come. Details: ATA and SCSI drivers should work as the dump formatting code has been removed. The IDA, TWE and AAC have not yet been converted. Dumpon now opens the device and uses ioctl(DIOCGKERNELDUMP) to set the device as dumpdev. To implement the "off" argument, /dev/null is used as the device. Savecore will fail if handed any options since they are not (yet) implemented. All devices marked "dump" or "swap" in /etc/fstab will be scanned and dumps found will be saved to diskfiles named from the MD5 hash of the header record. The header record is dumped in readable format in the .info file. The kernel is not saved. Only complete dumps will be saved. All maintainer rights for this code are disclaimed: feel free to improve and extend. Sponsored by: DARPA, NAI Labs
2002-03-31 22:37:00 +00:00
return(set_dumper(&di));
}
void
disk_invalidate (struct disk *disk)
{
if (disk->d_slice)
dsgone(&disk->d_slice);
}
void
disk_destroy(dev_t dev)
{
LIST_REMOVE(dev->si_disk, d_list);
free(dev->si_disk->d_label, M_DEVBUF);
bzero(dev->si_disk, sizeof(*dev->si_disk));
dev->si_disk = NULL;
destroy_dev(dev);
return;
}
struct disk *
disk_enumerate(struct disk *disk)
{
if (!disk)
return (LIST_FIRST(&disklist));
else
return (LIST_NEXT(disk, d_list));
}
static int
sysctl_disks(SYSCTL_HANDLER_ARGS)
{
struct disk *disk;
int error, first;
disk = NULL;
first = 1;
while ((disk = disk_enumerate(disk))) {
if (!first) {
error = SYSCTL_OUT(req, " ", 1);
if (error)
return error;
} else {
first = 0;
}
error = SYSCTL_OUT(req, disk->d_dev->si_name, strlen(disk->d_dev->si_name));
if (error)
return error;
}
error = SYSCTL_OUT(req, "", 1);
return error;
}
SYSCTL_PROC(_kern, OID_AUTO, disks, CTLTYPE_STRING | CTLFLAG_RD, 0, 0,
sysctl_disks, "A", "names of available disks");
/*
* The cdevsw functions
*/
static int
diskopen(dev_t dev, int oflags, int devtype, struct thread *td)
{
dev_t pdev;
struct disk *dp;
int error;
error = 0;
pdev = dkmodpart(dkmodslice(dev, WHOLE_DISK_SLICE), RAW_PART);
dp = pdev->si_disk;
if (!dp)
return (ENXIO);
while (dp->d_flags & DISKFLAG_LOCK) {
dp->d_flags |= DISKFLAG_WANTED;
error = tsleep(dp, PRIBIO | PCATCH, "diskopen", hz);
if (error)
return (error);
}
dp->d_flags |= DISKFLAG_LOCK;
if (!dsisopen(dp->d_slice)) {
if (!pdev->si_iosize_max)
pdev->si_iosize_max = dev->si_iosize_max;
error = dp->d_devsw->d_open(pdev, oflags, devtype, td);
dp->d_label->d_secsize = dp->d_sectorsize;
dp->d_label->d_secperunit = dp->d_mediasize / dp->d_sectorsize;
dp->d_label->d_nsectors = dp->d_fwsectors;
dp->d_label->d_ntracks = dp->d_fwheads;
}
/* Inherit properties from the whole/raw dev_t */
End two weeks of on and off debugging. Fix the crash on the Nth insertion of a CF card, for random values of N > 1. With these fixes, I've been able to do 100 insert/remove of the cards w/o a crash with lots of system activity going on that in the past would help trigger the crash. The problem: FreeBSD creates dev_t's on the fly as they are needed and never destroys them. These dev_t's point to a struct disk that is used for housekeeping on the disk. When a device goes away, the struct disk pointer becomes a dangling pointer. Sometimes when the device comes back, the pointer will point to the new struct disk (in which case the insertion will work). Other times it won't (especially if any length of time has passed, since it is dependent on memory returned from malloc). The Fix: There is one of these dev_t's that is always correct. The device for the WHOLE_DISK_SLICE is always right. It gets set at create_disk() time. So, the fix is to spend a little CPU time and lookup the WHOLE_DISK_SLICE dev_t and use the si_disk from that in preference to the one that's in the device asking to do the I/O. In addition, we change the test of si_disk == NULL meaning that the dev needed to inherit properties from the pdev to dev->si_disk != pdev->si_disk. This test is a little stronger than the previous test, but can sometimes be fooled into not inheriting. However, the results of this fooling are that the old values will be used, which will generally always be the same as before. si_drv[12] are the only values that are copied that might pose a problem. They tend to change as the si_disk field would change, so it is a hole, but it is a small hole. One could correctly argue that one should replace much of this code with something much much better. I would be on the pro side of that argument. Reviewed by: phk (who also ported the original patch to current) Sponsored by: Timing Solutions
2000-07-05 06:01:33 +00:00
inherit_raw(pdev, dev);
if (error)
goto out;
error = dsopen(dev, devtype, dp->d_dsflags, &dp->d_slice, dp->d_label);
if (!dsisopen(dp->d_slice))
dp->d_devsw->d_close(pdev, oflags, devtype, td);
out:
dp->d_flags &= ~DISKFLAG_LOCK;
if (dp->d_flags & DISKFLAG_WANTED) {
dp->d_flags &= ~DISKFLAG_WANTED;
wakeup(dp);
}
return(error);
}
static int
diskclose(dev_t dev, int fflag, int devtype, struct thread *td)
{
struct disk *dp;
int error;
End two weeks of on and off debugging. Fix the crash on the Nth insertion of a CF card, for random values of N > 1. With these fixes, I've been able to do 100 insert/remove of the cards w/o a crash with lots of system activity going on that in the past would help trigger the crash. The problem: FreeBSD creates dev_t's on the fly as they are needed and never destroys them. These dev_t's point to a struct disk that is used for housekeeping on the disk. When a device goes away, the struct disk pointer becomes a dangling pointer. Sometimes when the device comes back, the pointer will point to the new struct disk (in which case the insertion will work). Other times it won't (especially if any length of time has passed, since it is dependent on memory returned from malloc). The Fix: There is one of these dev_t's that is always correct. The device for the WHOLE_DISK_SLICE is always right. It gets set at create_disk() time. So, the fix is to spend a little CPU time and lookup the WHOLE_DISK_SLICE dev_t and use the si_disk from that in preference to the one that's in the device asking to do the I/O. In addition, we change the test of si_disk == NULL meaning that the dev needed to inherit properties from the pdev to dev->si_disk != pdev->si_disk. This test is a little stronger than the previous test, but can sometimes be fooled into not inheriting. However, the results of this fooling are that the old values will be used, which will generally always be the same as before. si_drv[12] are the only values that are copied that might pose a problem. They tend to change as the si_disk field would change, so it is a hole, but it is a small hole. One could correctly argue that one should replace much of this code with something much much better. I would be on the pro side of that argument. Reviewed by: phk (who also ported the original patch to current) Sponsored by: Timing Solutions
2000-07-05 06:01:33 +00:00
dev_t pdev;
error = 0;
End two weeks of on and off debugging. Fix the crash on the Nth insertion of a CF card, for random values of N > 1. With these fixes, I've been able to do 100 insert/remove of the cards w/o a crash with lots of system activity going on that in the past would help trigger the crash. The problem: FreeBSD creates dev_t's on the fly as they are needed and never destroys them. These dev_t's point to a struct disk that is used for housekeeping on the disk. When a device goes away, the struct disk pointer becomes a dangling pointer. Sometimes when the device comes back, the pointer will point to the new struct disk (in which case the insertion will work). Other times it won't (especially if any length of time has passed, since it is dependent on memory returned from malloc). The Fix: There is one of these dev_t's that is always correct. The device for the WHOLE_DISK_SLICE is always right. It gets set at create_disk() time. So, the fix is to spend a little CPU time and lookup the WHOLE_DISK_SLICE dev_t and use the si_disk from that in preference to the one that's in the device asking to do the I/O. In addition, we change the test of si_disk == NULL meaning that the dev needed to inherit properties from the pdev to dev->si_disk != pdev->si_disk. This test is a little stronger than the previous test, but can sometimes be fooled into not inheriting. However, the results of this fooling are that the old values will be used, which will generally always be the same as before. si_drv[12] are the only values that are copied that might pose a problem. They tend to change as the si_disk field would change, so it is a hole, but it is a small hole. One could correctly argue that one should replace much of this code with something much much better. I would be on the pro side of that argument. Reviewed by: phk (who also ported the original patch to current) Sponsored by: Timing Solutions
2000-07-05 06:01:33 +00:00
pdev = dkmodpart(dkmodslice(dev, WHOLE_DISK_SLICE), RAW_PART);
dp = pdev->si_disk;
if (!dp)
return (ENXIO);
dsclose(dev, devtype, dp->d_slice);
if (!dsisopen(dp->d_slice))
error = dp->d_devsw->d_close(dp->d_dev, fflag, devtype, td);
return (error);
}
static void
diskstrategy(struct bio *bp)
{
dev_t pdev;
struct disk *dp;
End two weeks of on and off debugging. Fix the crash on the Nth insertion of a CF card, for random values of N > 1. With these fixes, I've been able to do 100 insert/remove of the cards w/o a crash with lots of system activity going on that in the past would help trigger the crash. The problem: FreeBSD creates dev_t's on the fly as they are needed and never destroys them. These dev_t's point to a struct disk that is used for housekeeping on the disk. When a device goes away, the struct disk pointer becomes a dangling pointer. Sometimes when the device comes back, the pointer will point to the new struct disk (in which case the insertion will work). Other times it won't (especially if any length of time has passed, since it is dependent on memory returned from malloc). The Fix: There is one of these dev_t's that is always correct. The device for the WHOLE_DISK_SLICE is always right. It gets set at create_disk() time. So, the fix is to spend a little CPU time and lookup the WHOLE_DISK_SLICE dev_t and use the si_disk from that in preference to the one that's in the device asking to do the I/O. In addition, we change the test of si_disk == NULL meaning that the dev needed to inherit properties from the pdev to dev->si_disk != pdev->si_disk. This test is a little stronger than the previous test, but can sometimes be fooled into not inheriting. However, the results of this fooling are that the old values will be used, which will generally always be the same as before. si_drv[12] are the only values that are copied that might pose a problem. They tend to change as the si_disk field would change, so it is a hole, but it is a small hole. One could correctly argue that one should replace much of this code with something much much better. I would be on the pro side of that argument. Reviewed by: phk (who also ported the original patch to current) Sponsored by: Timing Solutions
2000-07-05 06:01:33 +00:00
pdev = dkmodpart(dkmodslice(bp->bio_dev, WHOLE_DISK_SLICE), RAW_PART);
dp = pdev->si_disk;
bp->bio_resid = bp->bio_bcount;
End two weeks of on and off debugging. Fix the crash on the Nth insertion of a CF card, for random values of N > 1. With these fixes, I've been able to do 100 insert/remove of the cards w/o a crash with lots of system activity going on that in the past would help trigger the crash. The problem: FreeBSD creates dev_t's on the fly as they are needed and never destroys them. These dev_t's point to a struct disk that is used for housekeeping on the disk. When a device goes away, the struct disk pointer becomes a dangling pointer. Sometimes when the device comes back, the pointer will point to the new struct disk (in which case the insertion will work). Other times it won't (especially if any length of time has passed, since it is dependent on memory returned from malloc). The Fix: There is one of these dev_t's that is always correct. The device for the WHOLE_DISK_SLICE is always right. It gets set at create_disk() time. So, the fix is to spend a little CPU time and lookup the WHOLE_DISK_SLICE dev_t and use the si_disk from that in preference to the one that's in the device asking to do the I/O. In addition, we change the test of si_disk == NULL meaning that the dev needed to inherit properties from the pdev to dev->si_disk != pdev->si_disk. This test is a little stronger than the previous test, but can sometimes be fooled into not inheriting. However, the results of this fooling are that the old values will be used, which will generally always be the same as before. si_drv[12] are the only values that are copied that might pose a problem. They tend to change as the si_disk field would change, so it is a hole, but it is a small hole. One could correctly argue that one should replace much of this code with something much much better. I would be on the pro side of that argument. Reviewed by: phk (who also ported the original patch to current) Sponsored by: Timing Solutions
2000-07-05 06:01:33 +00:00
if (dp != bp->bio_dev->si_disk)
inherit_raw(pdev, bp->bio_dev);
if (!dp) {
biofinish(bp, NULL, ENXIO);
return;
}
if (dscheck(bp, dp->d_slice) <= 0) {
biodone(bp);
return;
}
if (bp->bio_bcount == 0) {
biodone(bp);
return;
}
KASSERT(dp->d_devsw != NULL, ("NULL devsw"));
KASSERT(dp->d_devsw->d_strategy != NULL, ("NULL d_strategy"));
dp->d_devsw->d_strategy(bp);
return;
}
static int
diskioctl(dev_t dev, u_long cmd, caddr_t data, int fflag, struct thread *td)
{
struct disk *dp;
int error;
Here follows the new kernel dumping infrastructure. Caveats: The new savecore program is not complete in the sense that it emulates enough of the old savecores features to do the job, but implements none of the options yet. I would appreciate if a userland hacker could help me out getting savecore to do what we want it to do from a users point of view, compression, email-notification, space reservation etc etc. (send me email if you are interested). Currently, savecore will scan all devices marked as "swap" or "dump" in /etc/fstab _or_ any devices specified on the command-line. All architectures but i386 lack an implementation of dumpsys(), but looking at the i386 version it should be trivial for anybody familiar with the platform(s) to provide this function. Documentation is quite sparse at this time, more to come. Details: ATA and SCSI drivers should work as the dump formatting code has been removed. The IDA, TWE and AAC have not yet been converted. Dumpon now opens the device and uses ioctl(DIOCGKERNELDUMP) to set the device as dumpdev. To implement the "off" argument, /dev/null is used as the device. Savecore will fail if handed any options since they are not (yet) implemented. All devices marked "dump" or "swap" in /etc/fstab will be scanned and dumps found will be saved to diskfiles named from the MD5 hash of the header record. The header record is dumped in readable format in the .info file. The kernel is not saved. Only complete dumps will be saved. All maintainer rights for this code are disclaimed: feel free to improve and extend. Sponsored by: DARPA, NAI Labs
2002-03-31 22:37:00 +00:00
u_int u;
End two weeks of on and off debugging. Fix the crash on the Nth insertion of a CF card, for random values of N > 1. With these fixes, I've been able to do 100 insert/remove of the cards w/o a crash with lots of system activity going on that in the past would help trigger the crash. The problem: FreeBSD creates dev_t's on the fly as they are needed and never destroys them. These dev_t's point to a struct disk that is used for housekeeping on the disk. When a device goes away, the struct disk pointer becomes a dangling pointer. Sometimes when the device comes back, the pointer will point to the new struct disk (in which case the insertion will work). Other times it won't (especially if any length of time has passed, since it is dependent on memory returned from malloc). The Fix: There is one of these dev_t's that is always correct. The device for the WHOLE_DISK_SLICE is always right. It gets set at create_disk() time. So, the fix is to spend a little CPU time and lookup the WHOLE_DISK_SLICE dev_t and use the si_disk from that in preference to the one that's in the device asking to do the I/O. In addition, we change the test of si_disk == NULL meaning that the dev needed to inherit properties from the pdev to dev->si_disk != pdev->si_disk. This test is a little stronger than the previous test, but can sometimes be fooled into not inheriting. However, the results of this fooling are that the old values will be used, which will generally always be the same as before. si_drv[12] are the only values that are copied that might pose a problem. They tend to change as the si_disk field would change, so it is a hole, but it is a small hole. One could correctly argue that one should replace much of this code with something much much better. I would be on the pro side of that argument. Reviewed by: phk (who also ported the original patch to current) Sponsored by: Timing Solutions
2000-07-05 06:01:33 +00:00
dev_t pdev;
End two weeks of on and off debugging. Fix the crash on the Nth insertion of a CF card, for random values of N > 1. With these fixes, I've been able to do 100 insert/remove of the cards w/o a crash with lots of system activity going on that in the past would help trigger the crash. The problem: FreeBSD creates dev_t's on the fly as they are needed and never destroys them. These dev_t's point to a struct disk that is used for housekeeping on the disk. When a device goes away, the struct disk pointer becomes a dangling pointer. Sometimes when the device comes back, the pointer will point to the new struct disk (in which case the insertion will work). Other times it won't (especially if any length of time has passed, since it is dependent on memory returned from malloc). The Fix: There is one of these dev_t's that is always correct. The device for the WHOLE_DISK_SLICE is always right. It gets set at create_disk() time. So, the fix is to spend a little CPU time and lookup the WHOLE_DISK_SLICE dev_t and use the si_disk from that in preference to the one that's in the device asking to do the I/O. In addition, we change the test of si_disk == NULL meaning that the dev needed to inherit properties from the pdev to dev->si_disk != pdev->si_disk. This test is a little stronger than the previous test, but can sometimes be fooled into not inheriting. However, the results of this fooling are that the old values will be used, which will generally always be the same as before. si_drv[12] are the only values that are copied that might pose a problem. They tend to change as the si_disk field would change, so it is a hole, but it is a small hole. One could correctly argue that one should replace much of this code with something much much better. I would be on the pro side of that argument. Reviewed by: phk (who also ported the original patch to current) Sponsored by: Timing Solutions
2000-07-05 06:01:33 +00:00
pdev = dkmodpart(dkmodslice(dev, WHOLE_DISK_SLICE), RAW_PART);
dp = pdev->si_disk;
if (!dp)
return (ENXIO);
if (cmd == DIOCSKERNELDUMP) {
Here follows the new kernel dumping infrastructure. Caveats: The new savecore program is not complete in the sense that it emulates enough of the old savecores features to do the job, but implements none of the options yet. I would appreciate if a userland hacker could help me out getting savecore to do what we want it to do from a users point of view, compression, email-notification, space reservation etc etc. (send me email if you are interested). Currently, savecore will scan all devices marked as "swap" or "dump" in /etc/fstab _or_ any devices specified on the command-line. All architectures but i386 lack an implementation of dumpsys(), but looking at the i386 version it should be trivial for anybody familiar with the platform(s) to provide this function. Documentation is quite sparse at this time, more to come. Details: ATA and SCSI drivers should work as the dump formatting code has been removed. The IDA, TWE and AAC have not yet been converted. Dumpon now opens the device and uses ioctl(DIOCGKERNELDUMP) to set the device as dumpdev. To implement the "off" argument, /dev/null is used as the device. Savecore will fail if handed any options since they are not (yet) implemented. All devices marked "dump" or "swap" in /etc/fstab will be scanned and dumps found will be saved to diskfiles named from the MD5 hash of the header record. The header record is dumped in readable format in the .info file. The kernel is not saved. Only complete dumps will be saved. All maintainer rights for this code are disclaimed: feel free to improve and extend. Sponsored by: DARPA, NAI Labs
2002-03-31 22:37:00 +00:00
u = *(u_int *)data;
return (diskdumpconf(u, dev, dp));
}
if (cmd == DIOCGFRONTSTUFF) {
*(off_t *)data = 8192; /* XXX: crude but enough) */
return (0);
}
error = dsioctl(dev, cmd, data, fflag, &dp->d_slice);
if (error == ENOIOCTL)
error = dp->d_devsw->d_ioctl(dev, cmd, data, fflag, td);
return (error);
}
static int
diskpsize(dev_t dev)
{
struct disk *dp;
dev_t pdev;
End two weeks of on and off debugging. Fix the crash on the Nth insertion of a CF card, for random values of N > 1. With these fixes, I've been able to do 100 insert/remove of the cards w/o a crash with lots of system activity going on that in the past would help trigger the crash. The problem: FreeBSD creates dev_t's on the fly as they are needed and never destroys them. These dev_t's point to a struct disk that is used for housekeeping on the disk. When a device goes away, the struct disk pointer becomes a dangling pointer. Sometimes when the device comes back, the pointer will point to the new struct disk (in which case the insertion will work). Other times it won't (especially if any length of time has passed, since it is dependent on memory returned from malloc). The Fix: There is one of these dev_t's that is always correct. The device for the WHOLE_DISK_SLICE is always right. It gets set at create_disk() time. So, the fix is to spend a little CPU time and lookup the WHOLE_DISK_SLICE dev_t and use the si_disk from that in preference to the one that's in the device asking to do the I/O. In addition, we change the test of si_disk == NULL meaning that the dev needed to inherit properties from the pdev to dev->si_disk != pdev->si_disk. This test is a little stronger than the previous test, but can sometimes be fooled into not inheriting. However, the results of this fooling are that the old values will be used, which will generally always be the same as before. si_drv[12] are the only values that are copied that might pose a problem. They tend to change as the si_disk field would change, so it is a hole, but it is a small hole. One could correctly argue that one should replace much of this code with something much much better. I would be on the pro side of that argument. Reviewed by: phk (who also ported the original patch to current) Sponsored by: Timing Solutions
2000-07-05 06:01:33 +00:00
pdev = dkmodpart(dkmodslice(dev, WHOLE_DISK_SLICE), RAW_PART);
dp = pdev->si_disk;
if (!dp)
return (-1);
if (dp != dev->si_disk) {
dev->si_drv1 = pdev->si_drv1;
dev->si_drv2 = pdev->si_drv2;
/* XXX: don't set bp->b_dev->si_disk (?) */
}
return (dssize(dev, &dp->d_slice));
}
SYSCTL_INT(_debug_sizeof, OID_AUTO, disklabel, CTLFLAG_RD,
0, sizeof(struct disklabel), "sizeof(struct disklabel)");
SYSCTL_INT(_debug_sizeof, OID_AUTO, diskslices, CTLFLAG_RD,
0, sizeof(struct diskslices), "sizeof(struct diskslices)");
SYSCTL_INT(_debug_sizeof, OID_AUTO, disk, CTLFLAG_RD,
0, sizeof(struct disk), "sizeof(struct disk)");
#endif /* NO_GEOM */
/*-
* Disk error is the preface to plaintive error messages
* about failing disk transfers. It prints messages of the form
* "hp0g: BLABLABLA cmd=read fsbn 12345 of 12344-12347"
* blkdone should be -1 if the position of the error is unknown.
* The message is printed with printf.
*/
void
disk_err(struct bio *bp, const char *what, int blkdone, int nl)
{
daddr_t sn;
printf("%s: %s ", devtoname(bp->bio_dev), what);
switch(bp->bio_cmd) {
case BIO_READ: printf("cmd=read "); break;
case BIO_WRITE: printf("cmd=write "); break;
case BIO_DELETE: printf("cmd=delete "); break;
case BIO_GETATTR: printf("cmd=getattr "); break;
case BIO_SETATTR: printf("cmd=setattr "); break;
default: printf("cmd=%x ", bp->bio_cmd); break;
}
sn = bp->bio_blkno;
if (bp->bio_bcount <= DEV_BSIZE) {
printf("fsbn %jd%s", (intmax_t)sn, nl ? "\n" : "");
return;
}
if (blkdone >= 0) {
sn += blkdone;
printf("fsbn %jd of ", (intmax_t)sn);
}
printf("%jd-%jd", (intmax_t)bp->bio_blkno,
(intmax_t)(bp->bio_blkno + (bp->bio_bcount - 1) / DEV_BSIZE));
if (nl)
printf("\n");
}
/*
* Seek sort for disks.
*
* The buf_queue keep two queues, sorted in ascending block order. The first
* queue holds those requests which are positioned after the current block
* (in the first request); the second, which starts at queue->switch_point,
* holds requests which came in after their block number was passed. Thus
* we implement a one way scan, retracting after reaching the end of the drive
* to the first request on the second queue, at which time it becomes the
* first queue.
*
* A one-way scan is natural because of the way UNIX read-ahead blocks are
* allocated.
*/
void
bioq_disksort(bioq, bp)
struct bio_queue_head *bioq;
struct bio *bp;
{
struct bio *bq;
struct bio *bn;
struct bio *be;
if (!atomic_cmpset_int(&bioq->busy, 0, 1))
panic("Recursing in bioq_disksort()");
be = TAILQ_LAST(&bioq->queue, bio_queue);
/*
* If the queue is empty or we are an
* ordered transaction, then it's easy.
*/
if ((bq = bioq_first(bioq)) == NULL) {
bioq_insert_tail(bioq, bp);
bioq->busy = 0;
return;
} else if (bioq->insert_point != NULL) {
/*
* A certain portion of the list is
* "locked" to preserve ordering, so
* we can only insert after the insert
* point.
*/
bq = bioq->insert_point;
} else {
/*
* If we lie before the last removed (currently active)
* request, and are not inserting ourselves into the
* "locked" portion of the list, then we must add ourselves
* to the second request list.
*/
if (bp->bio_pblkno < bioq->last_pblkno) {
bq = bioq->switch_point;
/*
* If we are starting a new secondary list,
* then it's easy.
*/
if (bq == NULL) {
bioq->switch_point = bp;
bioq_insert_tail(bioq, bp);
bioq->busy = 0;
return;
}
/*
* If we lie ahead of the current switch point,
* insert us before the switch point and move
* the switch point.
*/
if (bp->bio_pblkno < bq->bio_pblkno) {
bioq->switch_point = bp;
TAILQ_INSERT_BEFORE(bq, bp, bio_queue);
bioq->busy = 0;
return;
}
} else {
if (bioq->switch_point != NULL)
be = TAILQ_PREV(bioq->switch_point,
bio_queue, bio_queue);
/*
* If we lie between last_pblkno and bq,
* insert before bq.
*/
if (bp->bio_pblkno < bq->bio_pblkno) {
TAILQ_INSERT_BEFORE(bq, bp, bio_queue);
bioq->busy = 0;
return;
}
}
}
/*
* Request is at/after our current position in the list.
* Optimize for sequential I/O by seeing if we go at the tail.
*/
if (bp->bio_pblkno > be->bio_pblkno) {
TAILQ_INSERT_AFTER(&bioq->queue, be, bp, bio_queue);
bioq->busy = 0;
return;
}
/* Otherwise, insertion sort */
while ((bn = TAILQ_NEXT(bq, bio_queue)) != NULL) {
/*
* We want to go after the current request if it is the end
* of the first request list, or if the next request is a
* larger cylinder than our request.
*/
if (bn == bioq->switch_point
|| bp->bio_pblkno < bn->bio_pblkno)
break;
bq = bn;
}
TAILQ_INSERT_AFTER(&bioq->queue, bq, bp, bio_queue);
bioq->busy = 0;
}