Commit Graph

365 Commits

Author SHA1 Message Date
Poul-Henning Kamp
15c672a31d Make a mutex to stop the race coming into geom_disk's done routine.
Cut up requests into smaller bits if they are longer than the drivers
disk->d_maxsize or dev->si_iosize_max.

Properly handle the race condition when using g_clone_bio() is used
without having the single-threadedness of g_down/g_up secure locking.
2003-02-11 18:32:31 +00:00
Poul-Henning Kamp
8028f16b1b Don't divide by zero if there is no stripewidth specified. 2003-02-11 15:23:41 +00:00
Poul-Henning Kamp
200c8a23cd Typo in last commit. 2003-02-11 15:20:34 +00:00
Poul-Henning Kamp
8a63edc3a1 Better names for struct disk elements: d_maxsize, d_stripeoffset
and d_stripesisze;

Introduce si_stripesize and si_stripeoffset in struct cdev so we
can make the visible to clustering code.

Add stripesize and stripeoffset to providers.

DTRT with stripesize and stripeoffset in various places in GEOM.
2003-02-11 14:57:34 +00:00
Poul-Henning Kamp
bb17c5f69f Propagate DISKFLAG_CANDELETE from struct disk to G_PF_CANDELETE on the
provider.
2003-02-11 14:12:06 +00:00
Poul-Henning Kamp
11eb149478 Wrap a long line. 2003-02-11 14:09:48 +00:00
Poul-Henning Kamp
392d56b4bd Don't short-circuit zero-length requests of they are BIO_[SG]ETATTR. 2003-02-11 13:13:10 +00:00
Poul-Henning Kamp
88b1af7730 Use the SI_CANDELETE flag on the dev_t rather than the D_CANFREE flag
on the cdevsw to determine ability to handle the BIO_DELETE request.
2003-02-11 12:49:58 +00:00
Poul-Henning Kamp
cadb987095 Unconditionally make our provider with G_PF_CANDELETE. 2003-02-11 12:37:04 +00:00
Poul-Henning Kamp
faa7b8b96e Propagate G_PF_CANDELETE to our own providers from the provider we attach to. 2003-02-11 12:36:33 +00:00
Poul-Henning Kamp
83333d4569 Introduce flag field and G_PF_CANDELETE field on providers. 2003-02-11 11:55:40 +00:00
Poul-Henning Kamp
dc803e37fb Remove another printf which does not say anything we didn't already know. 2003-02-11 11:02:27 +00:00
Poul-Henning Kamp
81e9eba33a Turn the "updating" flag (back) into two sequence number fields at
either ends of the structure so we have a way to determine if a
snapshot is consistent.
2003-02-11 11:01:26 +00:00
Poul-Henning Kamp
9358a8009a Remove a debugging printf. 2003-02-11 10:59:43 +00:00
Poul-Henning Kamp
cce7303af3 Update the statistics collection code to track busy time instead of
idle time.

Statistics now default to "on" and can be turned off with
        sysctl kern.geom.collectstats=0

Performance impact of statistics collection is on the order of
800 nsec per consumer/provider set on a 700MHz Athlon.
2003-02-09 17:04:57 +00:00
Poul-Henning Kamp
a09a56b5ad Put the name of the /dev entry in the .h file, userland will need it. 2003-02-08 14:50:27 +00:00
Poul-Henning Kamp
4ec353005c Move the g_stat struct to its own .h file, we will export it to other code.
Insted of embedding a struct g_stat in consumers and providers, merely
include a pointer.

Remove a couple of <sys/time.h> includes now unneeded.

Add a special allocator for struct g_stat.  This allocator will allocate
entire pages and hand out g_stat functions from there.  The "id" field
indicates free/used status.

Add "/dev/geom.stats" device driver whic exports the pages from the
allocator to userland with mmap(2) in read-only mode.

This mmap(2) interface should be considered a non-public interface and
the functions in libgeom (not yet committed) should be used to access
the statistics data.
2003-02-08 13:03:57 +00:00
Poul-Henning Kamp
91cd3dc6f5 Move #defines of major/minor to internal header file so other bits can
share and coordinate with geom_dev.
2003-02-08 12:30:12 +00:00
Poul-Henning Kamp
801bb689ca Commit the correct copy of the g_stat structure.
Add debug.sizeof.g_stat sysctl.

Set the id field of the g_stat when we create consumers and providers.

Remove biocount from consumer, we will use the counters in the g_stat
structure instead.  Replace one field which will need to be atomically
manipulated with two fields which will not (stat.nop and stat.nend).

Change add companion field to bio_children: bio_inbed for the exact
same reason.

Don't output the biocount in the confdot output.

Fix KASSERT in g_io_request().

Add sysctl kern.geom.collectstats defaulting to off.

Collect the following raw statistics conditioned on this sysctl:

    for each consumer and provider {
        total number of operations started.
        total number of operations completed.
        time last operation completed.
        sum of idle-time.
        for each of BIO_READ, BIO_WRITE and BIO_DELETE {
            number of operations completed.
            number of bytes completed.
            number of ENOMEM errors.
            number of other errors.
            sum of transaction time.
        }
    }

API for getting hold of these statistics data not included yet.
2003-02-07 23:08:24 +00:00
Poul-Henning Kamp
886d1942e8 Fix some sleep strings to make more sense. 2003-02-07 21:53:55 +00:00
Poul-Henning Kamp
7725f875ee Add the new statistics structure, put one in consumers and providers.
include <sys/time.h> as necessary.
2003-02-07 21:43:34 +00:00
Poul-Henning Kamp
936cc4614b Rename bio_linkage to the more obvious bio_parent.
Add bio_t0 timestamp, and include <sys/time.h> where needed
2003-02-07 21:09:51 +00:00
Gordon Tetlow
ddfa3cce05 Add some comments about the deficiencies of this module. I had hoped to get
around to addressing them some more, but Real Life (tm) has gotten in the
way.
2003-02-07 17:50:33 +00:00
Poul-Henning Kamp
a42106534f Check return value of g_clone_bio(). 2003-02-06 22:00:47 +00:00
Poul-Henning Kamp
ae2afc437b Experimentally don't let go of Giant in geom_disk's done.
We may actually be increasing Giant contention doing so because the
actual stuff we do is very cheap.

Also I am not convinced there is not a tiny window for a race here.
2003-02-06 21:50:39 +00:00
Poul-Henning Kamp
e39d70d4fd Put the checks we perform on a bio before calling ::start in their
own function, handle all validation and truncation at the time we
process the bio instead of when it gets scheduled.
2003-02-06 21:01:36 +00:00
Poul-Henning Kamp
51dd8980f4 Implement the new "struct disk" centered API for device drivers.
This commit should not change anything as no device drivers use the
new API yet.
2003-02-05 21:13:08 +00:00
Poul-Henning Kamp
237d2765f9 Pave the road to removing the fixed size limit on device nodes:
Change the si_name of dev_t's to be a char * and put a private buffer for
holding the name at then end of the struct.

Initialize si_name to point to the private buffer.

Put a KASSERT in geom_disk to prevent overrun on the fake dev_t we still
have to generate for the disk_drivers.
2003-02-04 10:32:40 +00:00
Gordon Tetlow
6be2e1dd37 Correct a comment. GEOM modules do not create /dev entries. They create
providers.

Pointed out by:	phk
2003-02-03 17:21:43 +00:00
Gordon Tetlow
a6fff888c3 Add the GEOM module that makes volume labels useful. A kernel compiled with
this will cause volume labels to be exposed in /dev/vol/<volname>. Currently,
there is no conflict resolution if more than one FS has the same volume name.

Reviewed by:	phk
2003-02-03 17:14:29 +00:00
Poul-Henning Kamp
ffa918814d Add a bio_disk pointer for use between geom_disk and the device drivers. 2003-02-02 11:59:54 +00:00
Poul-Henning Kamp
806d5cfff0 Eliminate the sc_openmask, ccdopen() and ccdclose() functions, we
can use the flag maintained by geom_disk.c

Having only a strategy method to intialize, don't waste space using
a cdevsw structure to do so.
2003-02-01 11:17:22 +00:00
Poul-Henning Kamp
5f6fb2dde3 Move configuration of geom/providers into its own function in preparation
for adding on-the-fly config interface.
2003-02-01 08:28:43 +00:00
Poul-Henning Kamp
a908904a1e Remove commented out g_enc_dos_partition(). We won't be needing it. 2003-01-31 21:18:41 +00:00
Poul-Henning Kamp
83dde52716 Add a rudimentary class for slicing Apple partitioned disks.
More work is needed on this, stakeholders please contact me.

Not quite asked for by:	rwatson
2003-01-31 16:27:07 +00:00
Poul-Henning Kamp
31a1093fc2 Add some agility to the disk_create() API:
Make passing the methods in a cdevsw structure optional.

Move "CANFREE" and "NOGIANT" flags into struct disk instead of the
cdevsw which may or may not be there.

Rename CANFREE to CANDELETE to match BIO_DELETE operation.

Add "OPEN" flag so drivers don't have to provide open/close methods
just to maintain such a flag.

Add temporary stopgap include of <sys/conf.h> to <sys/disk.h> until
the files which have them in the other order are fixed.

Add KASSERTS to make sure we don't get fed too many NULL pointers.

Clear our geom's softc pointer before we wither.
2003-01-30 20:34:23 +00:00
Poul-Henning Kamp
6e0b29aeca NO_GEOM cleanup: Remove sys/disklabel.h include. 2003-01-30 19:55:19 +00:00
Poul-Henning Kamp
1686ac18f9 NO_GEOM cleanup: retire disk_invalidate() 2003-01-30 19:43:50 +00:00
Poul-Henning Kamp
bf17219d59 NO_GEOM cleanup: Mark the last arg to disk_create() as unused. 2003-01-30 11:39:36 +00:00
Poul-Henning Kamp
5a67919a5e Add code to repsect the D_NOGIANT flag, should the disk device driver set it.
NO_GEOM cleanup:        remove ifdefs.

Still untested.
2003-01-29 19:47:25 +00:00
Poul-Henning Kamp
fc2b90eaab Sort these functions as the author instructed. 2003-01-29 18:37:29 +00:00
Poul-Henning Kamp
c516a05a66 Mark some args unused so this compiles in userland. 2003-01-28 09:48:09 +00:00
Poul-Henning Kamp
e55d9b0bd1 Use a void * to carry the private data for return-call'ed ioctl requests.
Amongst other things this avoids a complex workaround in the userland
regression bits.
2003-01-28 09:47:50 +00:00
Poul-Henning Kamp
7253da9c05 Implement DIOCBSDBB ioctl which overwrites first BBSIZE bytes of BSD
labeled disk.

This is complicated by the fact that BBSIZE is greater than the
PAGE_SIZE limit ioctl inflicts on arguments which are automatically
copied in.

As long as we don't need access to userland memory (copyin/out) we
can deal with the ioctl using g_callme() which executes it from the
GEOM event thread.

Once we need copyin/out, we need to return the bio with EDIRIOCTL
in order to make geom_dev call us back in the original process context
where copyin will work.

Unfortunately, that results in us getting called with Giant, so
we have to DROP_GIANT/PICKUP_GIANT around the code where we diddle
GEOMs internals.

Sometimes you just can't win...

... But it does make geom_bsd.c an almost complete example of the
GEOM beastiarium.
2003-01-26 21:54:36 +00:00
Alfred Perlstein
44956c9863 Remove M_TRYWAIT/M_WAITOK/M_WAIT. Callers should use 0.
Merge M_NOWAIT/M_DONTWAIT into a single flag M_NOWAIT.
2003-01-21 08:56:16 +00:00
Poul-Henning Kamp
c0805171aa disk_dev_synth() is a NO_GEOM hack. 2003-01-20 11:29:07 +00:00
Poul-Henning Kamp
30f5ffb8d9 Remove need for <sys/diskslice.h> but retain numerical compatibilty just in case. 2003-01-20 11:23:00 +00:00
Poul-Henning Kamp
0f76d6d822 Finally give CCD the disk mini-layer treatment:
CAUTION:

        Previously CCD would be different from all other disks in
        the system in that there were no "ccd0" device, only a
        "ccd0c" device.

        This is no longer so after this commit.  If you access a
        ccd device through the "/dev/ccd0c" device _and_ have not
        actually put a BSD disklabel on the device, you will have
        to use the name "/dev/ccd0".  If your CCD device contains
        a BSD disklabel there should be no difference.

        You need to recompile ccdconfig(8) using the changed
        src/sys/sys/ccdvar.h for the -g "show me" option to work.

        I have run the regression test I created before I started
        overhauling CCD and it flags no problems, but this code
        is mildly evil, so take care.  If you would cry if you lost
        what's on CCD, make a back before you upgrade.

Create separate cdevsw for the /dev/ccd.ctl device.

Remove the cloning function, the disk-minilayer will do all naming
for us.

Remove the ccdunit and ccdpart functions and carry the softc pointer
in the relevant dev_t's and structures.

Release all memory when a CCD device is unconfigured, previously
the softc would linger behind.

Remove all traces of BSD disklabel fiddling code.

Remove ccdpsize, the disk mini-layer does this for us.

Don't allocate memory with M_WAITOK in ccdstrategy().

Remove boundary checks which the disk mini-layer does for us.

Don't allocate space for more than 2 ccdbuf, RAID was never implemented.

NB: I have not tried to address any of the preexisting ailments of CCD.
2003-01-19 15:00:58 +00:00
Poul-Henning Kamp
1dccd08a34 Unifdef -UDEBUG on the CCD driver. The debugging is mostly useless
and can be added back selectively, should anybody start to interest
themselves for the internal workings of ccd.

This commit will make the diffs for the following commits much more
readable.
2003-01-19 14:35:38 +00:00
Poul-Henning Kamp
e9fe7d1f7f Inline now trivial functions getccdbuf() and putccdbuf().
Fix another trivial memory-leak.
2003-01-18 12:23:49 +00:00