Commit Graph

77 Commits

Author SHA1 Message Date
Alexander Motin
869fd29a7b Rewrite physio() to not allocate pbufs for unmapped I/O.
pbufs is a limited resource, and their allocator is not SMP-scalable.
So instead of always allocating pbuf to immediately convert it to bio,
allocate bio just here.  If buffer needs kernel mapping, then pbuf is
still allocated, but used only as a source of KVA and storage for a list
of held pages.

On 40-core system doing many 512-byte reads from user level to array of
raw SSDs this change removes huge lock congestion inside pbuf allocator.
It improves peak performance from ~300K to ~1.2M IOPS.  On my previous
24-core system this problem also existed, but was less serious.

Reviewed by:	kib
MFC after:	2 weeks
2015-04-21 10:55:53 +00:00
Kenneth D. Merry
880e57b635 Fix some issues in change 254760 pointed out by Bruce Evans:
- Remove excessive parenthesis
 - Use KNF continuation indentation
 - Cut down on excessive continuation lines
 - More consistent style in messages
 - Use uprintf() instead of printf()

Submitted by:	bde
2013-08-29 16:41:40 +00:00
Kenneth D. Merry
aaea33e51f Fix a printf format warning on 32-bit mips and powerpc.
Reported by:	bde, gjb
Pointy hat to:	ken
2013-08-24 19:02:36 +00:00
Kenneth D. Merry
93729c1796 Add support to physio(9) for devices that don't want I/O split and
configure sa(4) to request no I/O splitting by default.

For tape devices, the user needs to be able to clearly understand
what blocksize is actually being used when writing to a tape
device.  The previous behavior of physio(9) was that it would split
up any I/O that was too large for the device, or too large to fit
into MAXPHYS.  This means that if, for instance, the user wrote a
1MB block to a tape device, and MAXPHYS was 128KB, the 1MB write
would be split into 8 128K chunks.  This would be done without
informing the user.

This has suboptimal effects, especially when trying to communicate
status to the user.  In the event of an error writing to a tape
(e.g. physical end of tape) in the middle of a 1MB block that has
been split into 8 pieces, the user could have the first two 128K
pieces written successfully, the third returned with an error, and
the last 5 returned with 0 bytes written.  If the user is using
a standard write(2) system call, all he will see is the ENOSPC
error.  He won't have a clue how much actually got written.  (With
a writev(2) system call, he should be able to determine how much
got written in addition to the error.)

The solution is to prevent physio(9) from splitting the I/O.  The
new cdev flag, SI_NOSPLIT, tells physio that the driver does not
want I/O to be split beforehand.

Although the sa(4) driver now enables SI_NOSPLIT by default,
that can be disabled by two loader tunables for now.  It will not
be configurable starting in FreeBSD 11.0.  kern.cam.sa.allow_io_split
allows the user to configure I/O splitting for all sa(4) driver
instances.  kern.cam.sa.%d.allow_io_split allows the user to
configure I/O splitting for a specific sa(4) instance.

There are also now three sa(4) driver sysctl variables that let the
users see some sa(4) driver values.  kern.cam.sa.%d.allow_io_split
shows whether I/O splitting is turned on.  kern.cam.sa.%d.maxio shows
the maximum I/O size allowed by kernel configuration parameters
(e.g. MAXPHYS, DFLTPHYS) and the capabilities of the controller.
kern.cam.sa.%d.cpi_maxio shows the maximum I/O size supported by
the controller.

Note that a better long term solution would be to implement support
for chaining buffers, so that that MAXPHYS is no longer a limiting
factor for I/O size to tape and disk devices.  At that point, the
controller and the tape drive would become the limiting factors.

sys/conf.h:	Add a new cdev flag, SI_NOSPLIT, that allows a
		driver to tell physio not to split up I/O.

sys/param.h:	Bump __FreeBSD_version to 1000049 for the addition
		of the SI_NOSPLIT cdev flag.

kern_physio.c:	If the SI_NOSPLIT flag is set on the cdev, return
		any I/O that is larger than si_iosize_max or
		MAXPHYS, has more than one segment, or would have
		to be split because of misalignment with EFBIG.
		(File too large).

		In the event of an error, print a console message to
		give the user a clue about what happened.

scsi_sa.c:	Set the SI_NOSPLIT cdev flag on the devices created
		for the sa(4) driver by default.

		Add tunables to control whether we allow I/O splitting
		in physio(9).

		Explain in the comments that allowing I/O splitting
		will be deprecated for the sa(4) driver in FreeBSD
		11.0.

		Add sysctl variables to display the maximum I/O
		size we can do (which could be further limited by
		read block limits) and the maximum I/O size that
		the controller can do.

		Limit our maximum I/O size (recorded in the cdev's
		si_iosize_max) by MAXPHYS.  This isn't strictly
		necessary, because physio(9) will limit it to
		MAXPHYS, but it will provide some clarity for the
		application.

		Record the controller's maximum I/O size reported
		in the Path Inquiry CCB.

sa.4:		Document the block size behavior, and explain that
		the option of allowing physio(9) to split the I/O
		will disappear in FreeBSD 11.0.

Sponsored by:	Spectra Logic
2013-08-24 04:52:22 +00:00
Kenneth D. Merry
ce625ec719 Change the way that unmapped I/O capability is advertised.
The previous method was to set the D_UNMAPPED_IO flag in the cdevsw
for the driver.  The problem with this is that in many cases (e.g.
sa(4)) there may be some instances of the driver that can handle
unmapped I/O and some that can't.  The isp(4) driver can handle
unmapped I/O, but the esp(4) driver currently cannot.  The cdevsw
is shared among all driver instances.

So instead of setting a flag on the cdevsw, set a flag on the cdev.
This allows drivers to indicate support for unmapped I/O on a
per-instance basis.

sys/conf.h:	Remove the D_UNMAPPED_IO cdevsw flag and replace it
		with an SI_UNMAPPED cdev flag.

kern_physio.c:	Look at the cdev SI_UNMAPPED flag to determine
		whether or not a particular driver can handle
		unmapped I/O.

geom_dev.c:	Set the SI_UNMAPPED flag for all GEOM cdevs.
		Since GEOM will create a temporary mapping when
		needed, setting SI_UNMAPPED unconditionally will
		work.

		Remove the D_UNMAPPED_IO flag.

nvme_ns.c:	Set the SI_UNMAPPED flag on cdevs created here
		if NVME_UNMAPPED_BIO_SUPPORT is enabled.

vfs_aio.c:	In aio_qphysio(), check the SI_UNMAPPED flag on a
		cdev instead of the D_UNMAPPED_IO flag on the cdevsw.

sys/param.h:	Bump __FreeBSD_version to 1000045 for the switch from
		setting the D_UNMAPPED_IO flag in the cdevsw to setting
		SI_UNMAPPED in the cdev.

Reviewed by:	kib, jimharris
MFC after:	1 week
Sponsored by:	Spectra Logic
2013-08-15 22:52:39 +00:00
Konstantin Belousov
d1e99f43ed Add dev_strategy_csw() function, which is similar to dev_strategy()
but assumes that a thread reference was already obtained on the passed
device.  Use the function from physio(), to avoid two extra dev_mtx
lock and unlock.  Note that physio() is always used as the cdevsw
method, or is called from a cdevsw method, and the caller already owns
the reference.

dev_strategy() is left to keep KPI intact, but now it is implemented
as a wrapper around dev_strategy_csw().

Do some style cleanup in physio().

Requested and reviewed by:	kan (previous version)
Sponsored by:	The FreeBSD Foundation
MFC after:	2 weeks
2013-03-27 11:34:27 +00:00
Alexander Kabaev
31932fae1e Do not pass unmapped buffers to drivers that cannot handle them
In physio, check if device can handle unmapped IO and pass an
appropriately mapped buffer to the driver strategy routine. The
only driver in the tree that can handle unmapped buffers is one
exposed by GEOM, so mark it as such with the new flag in the
driver cdevsw structure.

This fixes insta-panics on hosts, running dconschat, as /dev/fwmem
is an example of the driver that makes use of physio routine, but
bypasses the g_down thread, where the buffer gets mapped normally.

Discussed with: kib (earlier version)
2013-03-26 01:17:06 +00:00
Konstantin Belousov
e81ff91e62 Do not remap usermode pages into KVA for physio.
Sponsored by:	The FreeBSD Foundation
Tested by:	pho
2013-03-19 14:43:57 +00:00
Konstantin Belousov
eea7f71c81 Account i/o done on cdevs.
Reported and tested by:	Adam Vande More <amvandemore gmail com>
MFC after:	1 week
2010-11-25 20:05:11 +00:00
Warner Losh
9454b2d864 /* -> /*- for copyright notices, minor format tweaks as necessary 2005-01-06 23:35:40 +00:00
Poul-Henning Kamp
c569065139 Remove buf->b_dev field. 2004-11-04 07:59:57 +00:00
Poul-Henning Kamp
6afb3b1c37 Give dev_strategy() an explict cdev argument in preparation for removing
buf->b-dev.

Put a bio between the buf passed to dev_strategy() and the device driver
strategy routine in order to not clobber fields in the buf.

Assert copyright on vfs_bio.c and update copyright message to canonical
text.  There is no legal difference between John Dysons two-clause
abbreviated BSD license and the canonical text.
2004-10-29 07:16:37 +00:00
Poul-Henning Kamp
1a52a73d68 Eliminate DEV_STRATEGY() macro: call dev_strategy() directly.
Make dev_strategy() handle errors and departing devices properly.
2004-09-23 14:45:04 +00:00
Alan Cox
fad44deea3 Eliminate the acquisition and release of Giant within physio(). Remove
the spl calls.

Reviewed by: phk@
Discussed with: scottl@
2004-08-10 21:47:11 +00:00
Poul-Henning Kamp
89c9c53da0 Do the dreaded s/dev_t/struct cdev */
Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
Poul-Henning Kamp
00cbe31bd8 Send B_PHYS out to pasture, it no longer serves any function. 2003-11-15 09:28:09 +00:00
Poul-Henning Kamp
01758670e9 Initialize b_iooffset before calling strategy 2003-10-18 19:48:21 +00:00
Poul-Henning Kamp
f7e56e489d Grab Giant in physio() since non-giant drivers are starting to appear. 2003-08-02 09:40:53 +00:00
David E. O'Brien
677b542ea2 Use __FBSDID(). 2003-06-11 00:56:59 +00:00
Alan Cox
ef38cda165 Don't reinitialize fields that are already initialized by getpbuf(). 2003-04-05 23:02:58 +00:00
Alan Cox
cdb06eda66 Sufficient access checks are performed by vmapbuf() that calling
useracc() is pointless.  Remove the call to useracc() from physio().

Reviewed by:	tegge
2003-04-05 21:19:58 +00:00
Jeff Roberson
749ffa4ecd - Add a lock for protecting against msleep(bp, ...) wakeup(bp) races.
- Create a new function bdone() which sets B_DONE and calls wakup(bp). This
   is suitable for use as b_iodone for buf consumers who are not going
   through the buf cache.
 - Create a new function bwait() which waits for the buf to be done at a set
   priority and with a specific wmesg.
 - Replace several cases where the above functionality was implemented
   without locking with the new functions.
2003-03-13 07:31:45 +00:00
Matthew Dillon
2d5c7e4506 Close the remaining user address mapping races for physical
I/O, CAM, and AIO.  Still TODO: streamline useracc() checks.

Reviewed by:	alc, tegge
MFC after:	7 days
2003-01-20 17:46:48 +00:00
Poul-Henning Kamp
e2a3ea1c45 Remove unused second argument from DEV_STRATEGY(). 2003-01-03 05:57:35 +00:00
Mike Barcroft
2b7f24d210 Change iov_base's type from char *' to the standard void *'. All
uses of iov_base which assume its type is `char *' (in order to do
pointer arithmetic) have been updated to cast iov_base to `char *'.
2002-10-11 14:58:34 +00:00
Alfred Perlstein
7f05b0353a More caddr_t removal, make fo_ioctl take a void * instead of a caddr_t. 2002-06-29 01:50:25 +00:00
Poul-Henning Kamp
e96d018d92 Use btodb() macro.
Sponsored by: DARPA & NAI Labs.
2002-05-18 09:34:09 +00:00
Poul-Henning Kamp
9626b608de Separate the struct bio related stuff out of <sys/buf.h> into
<sys/bio.h>.

<sys/bio.h> is now a prerequisite for <sys/buf.h> but it shall
not be made a nested include according to bdes teachings on the
subject of nested includes.

Diskdrivers and similar stuff below specfs::strategy() should no
longer need to include <sys/buf.> unless they need caching of data.

Still a few bogus uses of struct buf to track down.

Repocopy by:    peter
2000-05-05 09:59:14 +00:00
Poul-Henning Kamp
c244d2de43 Move B_ERROR flag to b_ioflags and call it BIO_ERROR.
(Much of this done by script)

Move B_ORDERED flag to b_ioflags and call it BIO_ORDERED.

Move b_pblkno and b_iodone_chain to struct bio while we transition, they
will be obsoleted once bio structs chain/stack.

Add bio_queue field for struct bio aware disksort.

Address a lot of stylistic issues brought up by bde.
2000-04-02 15:24:56 +00:00
Poul-Henning Kamp
b99c307a21 Rename the existing BUF_STRATEGY() to DEV_STRATEGY()
substitute BUF_WRITE(foo) for VOP_BWRITE(foo->b_vp, foo)

substitute BUF_STRATEGY(foo) for VOP_STRATEGY(foo->b_vp, foo)

This patch is machine generated except for the ccd.c and buf.h parts.
2000-03-20 11:29:10 +00:00
Poul-Henning Kamp
21144e3bf1 Remove B_READ, B_WRITE and B_FREEBUF and replace them with a new
field in struct buf: b_iocmd.  The b_iocmd is enforced to have
exactly one bit set.

B_WRITE was bogusly defined as zero giving rise to obvious coding
mistakes.

Also eliminate the redundant struct buf flag B_CALL, it can just
as efficiently be done by comparing b_iodone to NULL.

Should you get a panic or drop into the debugger, complaining about
"b_iocmd", don't continue.  It is likely to write on your disk
where it should have been reading.

This change is a step in the direction towards a stackable BIO capability.

A lot of this patch were machine generated (Thanks to style(9) compliance!)

Vinum users:  Greg has not had time to test this yet, be careful.
2000-03-20 10:44:49 +00:00
Poul-Henning Kamp
02c58685a4 Change useracc() and kernacc() to use VM_PROT_{READ|WRITE|EXECUTE} for the
"rw" argument, rather than hijacking B_{READ|WRITE}.

Fix two bugs (physio & cam) resulting by the confusion caused by this.

Submitted by:   Tor.Egge@fast.no
Reviewed by:    alc, ken (partly)
1999-10-30 06:32:05 +00:00
Poul-Henning Kamp
7179e74f18 Give physio a makeover.
- Let physio take read/write compatible args and have it use uio->uio_rw
  to determine the direction.

- physread/physwrite are now #defines for physio

- Remove the inversly named minphys(), dev->si_iosize_max takes over.

- Physio() always uses pbufs.

- Fix the check for non page-aligned transfers, now only unaligned
  transfers larger than (MAXPHYS - PAGE_SIZE) get fragmented (only
  interesting for tapes using max blocksize).

- General wash-and-clean of code.

Constructive input from: bde
1999-10-09 19:44:32 +00:00
Brian Feldman
0b5c7391ba Add a newline to "WARNING: %s maxphys = 0 ??" so it doesn't trip up
syslogd.  Note of course it's simply much more polite and correct, too :)
1999-10-09 05:17:05 +00:00
Søren Schmidt
dc722a14c7 In some drivers we use two devices to be able to boot.
So if si_iosize_max is allready set, dont mess with it..

Also just log the problem with maxphys not being set once.

designed by:	phk
tested by:	sos
1999-10-02 20:21:49 +00:00
Poul-Henning Kamp
45604de328 Fix a problem relating to si_iosize_max which broke scsi devices. 1999-10-02 11:17:54 +00:00
Poul-Henning Kamp
c428d4c048 Kill the cdevsw->d_maxio field.
d_maxio is replaced by the dev->si_iosize_max field which the driver
should be set in all calls to cdevsw->d_open if it has a better
idea than the system wide default.

The field is a generic dev_t field (ie: not disk specific) so that
tapes and other devices can use physio as well.
1999-09-22 19:56:14 +00:00
Poul-Henning Kamp
f5756ee9f5 Remove replace phygetvpbuf() with direct call to getpbuf(); 1999-09-12 08:27:41 +00:00
Peter Wemm
c3aac50f28 $Id$ -> $FreeBSD$ 1999-08-28 01:08:13 +00:00
Poul-Henning Kamp
60767bf422 Use more compiler friendly test for overflow.
Submitted by:	bde
1999-08-21 18:22:44 +00:00
Poul-Henning Kamp
3b782ee9d2 Detect if the the offset used to read from a raw device loose bits
when converted to block number.
1999-08-21 06:48:16 +00:00
Poul-Henning Kamp
49ff4debd3 Spring cleaning around strategy and disklabels/slices:
Introduce BUF_STRATEGY(struct buf *, int flag) macro, and use it throughout.
please see comment in sys/conf.h about the flag argument.

Remove strategy argument from all the diskslice/label/bad144
implementations, it should be found from the dev_t.

Remove bogus and unused strategy1 routines.

Remove open/close arguments from dssize().  Pick them up from dev_t.

Remove unused and unfinished setgeom support from diskslice/label/bad144 code.
1999-08-14 11:40:51 +00:00
Kirk McKusick
67812eacd7 Convert buffer locking from using the B_BUSY and B_WANTED flags to using
lockmgr locks. This commit should be functionally equivalent to the old
semantics. That is, all buffer locking is done with LK_EXCLUSIVE
requests. Changes to take advantage of LK_SHARED and LK_RECURSIVE will
be done in future commits.
1999-06-26 02:47:16 +00:00
Poul-Henning Kamp
4be2eb8c49 I got tired of seeing all the cdevsw[major(foo)] all over the place.
Made a new (inline) function devsw(dev_t dev) and substituted it.

Changed to the BDEV variant to this format as well: bdevsw(dev_t dev)

DEVFS will eventually benefit from this change too.
1999-05-08 06:40:31 +00:00
Poul-Henning Kamp
c48d17750f Introduce two functions: physread() and physwrite() and use these directly
in *devsw[] rather than the 46 local copies of the same functions.

(grog will do the same for vinum when he has time)
1999-05-07 07:03:47 +00:00
Poul-Henning Kamp
b0eeea2042 remove b_proc from struct buf, it's (now) unused.
Reviewed by:	dillon, bde
1999-05-06 20:00:34 +00:00
Peter Wemm
57dc594832 Use the reference counted PHOLD()/PRELE() rather than P_PHYSIO. 1999-04-06 03:04:47 +00:00
Matthew Dillon
1c7c3c6a86 This is a rather large commit that encompasses the new swapper,
changes to the VM system to support the new swapper, VM bug
    fixes, several VM optimizations, and some additional revamping of the
    VM code.  The specific bug fixes will be documented with additional
    forced commits.  This commit is somewhat rough in regards to code
    cleanup issues.

Reviewed by:	"John S. Dyson" <root@dyson.iquest.net>, "David Greenman" <dg@root.com>
1999-01-21 08:29:12 +00:00
Poul-Henning Kamp
f5ef029e92 Nitpicking and dusting performed on a train. Removes trivial warnings
about unused variables, labels and other lint.
1998-10-25 17:44:59 +00:00
Søren Schmidt
e620a1cbed Make struct buf->b_offset reflect the real byte offset which got
in via the uio struct. This enables device drivers to use != DEV_BSIZE
blocking on devices with wierd sector/block sizes (ie CDROM's).
1998-08-19 10:50:32 +00:00