Commit Graph

104 Commits

Author SHA1 Message Date
Matthew Dillon
b4e36adf1c Fix warnings preparing for -Wall -Wcast-qual
Also disable one usb module in LINT due to fatal compilation errors,
    temporary.
1999-01-27 20:09:21 +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
Matt Jacob
667d00a796 put it back the way it was 1999-01-20 23:00:31 +00:00
Matt Jacob
7781f5a621 temporary fix so alpha stays working while configure code is fixed 1999-01-20 19:08:45 +00:00
Kenneth D. Merry
6dd76100e4 By popular demand, back out the CONNER CFP* quirk entry, and return it to
its original form.  (Originally, it only applied to the CFP 2107.)

Hopefully we can come to some conclusion about which Conner drives are
broken for tagged queueing.
1999-01-20 01:13:20 +00:00
Kenneth D. Merry
1f04f5abd7 The Samsung WN34324U is broken for tagged queueing.
PR:		kern/9535
Submitted by:	NAKAJI Hiroyuki <nakaji@zeisei.dpri.kyoto-u.ac.jp>
1999-01-19 16:59:36 +00:00
Kenneth D. Merry
59190eaa10 Generalize the quirk entry for the Conner CFP* drives. It did just cover
the CFP2107, but it appears (not surprisingly) that the 1 gig and 4 gig
versions of that drive have the same problem with tagged queueing.

Also, fix the problem reported in PR kern/9482.  The XPT_DEV_MATCH case in
xptioctl() wasn't putting a proper path in the CCB before it called
xpt_action().  When CAMDEBUG is defined, and CAM_DEBUG_TRACE debugging is
turned on, the CAM_DEBUG statement at the beginning of xpt_action would end
up deferencing a NULL path pointer.  That of course caused a panic.

My solution is to just stick the xpt peripheral's path in the CCB.

PR:		kern/9482
Reviewed by:	gibbs
1999-01-19 01:02:47 +00:00
Peter Wemm
6d2a8f1ca0 Call the xpt_init() hook during at the start of the configure() process
via SYSINIT().  This gets a little closer to making cam useable as a
module.
1999-01-19 00:13:05 +00:00
Matt Jacob
15da947cba Sascha Blank <blank@uni-trier.de> convinced me I was an
idiot about testing SA_QUIRK_2FM in samount. Fixed.

Removed the NORRLS quirk (to save quirk space) and left
the behaviour of being quiet about failed reserve/release
(failed due Illegal Request) the same.

Added a SF_QUIET_IR for prevent/allow for the same purposes.
1999-01-16 19:20:30 +00:00
Matt Jacob
54e609ba5c More bandaids. One important one from Sascha Blank
(<blank@fox.uni-trier.de>) about quirks being set as
arithmetic values, not as bitfields. Add HP, Kennedy
and M4 1/2" reel quirk entries.

Do a lot of gratuitous source changing.

Audit all functions that build ccbs for the tape driver
and decide whether each one can be retried or not.

Still to do is some more state management post errors.
1999-01-16 04:02:31 +00:00
John Polstra
0ec81012da Replace includes of <sys/kernel.h> with includes of
<sys/linker_set.h> in those files that use only the linker set
definitions.
1999-01-14 06:22:10 +00:00
Justin T. Gibbs
c8bead2a12 Add support for wildcard device entries in the EDT. The target mode
'Black Hole' device uses this feature to schedule itself against any
target or lun attached to a controller that receives an unwanted request
from an initiator instead of having an instance per potential target/lun
request.

Use the wildcard entries to simplify wildcard async callback storage.

Don't announce devices twice to peripheral drivers.  The devices will
be announced as soon as the AC_PATH_REGISTERED event is registered by
the peripheral driver, so no manaul push of this event is required.

Reviewed by: Kenneth Merry <ken@FreeBSD.org>
1999-01-14 06:03:59 +00:00
Justin T. Gibbs
6fda903f5b The target mode 'Black Hole' device. This peripheral driver provides inquiry
data and sense information for target mode devices for which no other
peripheral driver is attached.  This simplifies the task of dealing with
luns that are not otherwise enabled for target mode if the controller
does not have firmware that automatically deals with this case (e.g.
the aic7xxx driver).
1999-01-14 06:00:29 +00:00
Justin T. Gibbs
443af83864 Properly handle transfers that only consume partial buffers. 1999-01-14 05:57:32 +00:00
Eivind Eklund
fcfc314003 Make dsopen() and dsioctl() use cdstrategy1 instead of cdstrategy.
This silence the warnings and makes it more obvious where the bugs
have to be fixed.
1999-01-12 16:26:19 +00:00
Matt Jacob
26fd36dffa Clean up and fix quirk table (was missing necessary wildcards) and add a couple
from the old driver. Change format of quirk table to have a preferred block
size for devices that need to be QUIRK_FIXED- this is loaded into the
last_media_blocksize tag at saregister time and will be used in the first
samount case.

Change sasetparams to take a sense_flags argument so that probe time testing
can be quieter (e.g. with SF_NO_PRINT).

Fix a couple of silly bugs in the fixed/variable determination in samount- one
was where there was a check against 'guessing' AND the density code being
default density- *SMACK* - you're only guessing if you find the media code
to be *other* than default density. Second bug was a test against current
blocksize being zero- should be a test against whether current blocksize
is not equal to the last blocksize if you had wanted to be fixed (suppose
you came up in fixed, but not the preferred size?). And if you don't
know what the fixed size should be, select 512 as the starting point,
not BLKDEV_IOSIZE (reality wins). Finally, in doing the test set to variable
mode, make sasetparams non-chatty.
1999-01-12 08:15:47 +00:00
Matt Jacob
7439f1066e Make HP T4000S quirk to FIXED mode 1999-01-11 18:26:25 +00:00
Eivind Eklund
cd651e8119 Add 'static' to the declaration, too. 1999-01-11 17:45:22 +00:00
Matt Jacob
f55ca80bcf Like the problems just fixed in scsi_da.c, make sure
to release the probe ccb before taking down the periph.
Also, don't do cdscheduling if you're not going to
attach the device after all.
Reviewed by:	ken@freebsd.org
1999-01-07 20:20:57 +00:00
Matt Jacob
f403e6edea A better fix to avoid race conditions between failed probes
and peripheral removal.
Obtained from:gibbs@freebsd.org
1999-01-07 20:19:09 +00:00
Kenneth D. Merry
4cdd022168 The Quantum Atlas III evidently has an identical problem to the Atlas II.
It keeps returning queue full until we have reduced the number of tagged
openings to the minimum.

So, put in a quirk entry with the same work-around.  This quirk entry is
only for the 9G Atlas III, once someone comes up with inquiry information
for the 18G version of that drive, we can quirk it as well.

Submitted by:	"Johan Granlund" <johan@granlund.nu>
1999-01-07 01:11:24 +00:00
Kenneth D. Merry
bfc0eb0f5a The Conner CFP2107 is a fixed-media drive, not removable media. This fixes
the quirk that disables tagged queueing for those drives.

Also, silence a warning by disabling xpt_for_all_targets() and
xpt_for_all_periphs().  These two functions are not currently used, but
they should not be removed.  They're part of a set of functions that
provide a way to execute a function for every {bus,target,device,periph} in
the system.

If anyone needs to use either function in the future, they can be
un-#ifdefed.
1999-01-05 21:37:07 +00:00
Matt Jacob
2e56e285fd Add a quirk that disables SYNCHRONIZE CACHE
PR:		8882
Obtained from:	Hellmuth Michaelis hm@kts.org
1999-01-05 20:43:41 +00:00
Matt Jacob
918d0cf6b7 Temporary workaround (bandaid) for case where you have READ
CAPACITY fail for a non-removable media device. There's a race
condition where the device entry is removed and then
xpt_release_ccb is called which attempts to give back the ccb
to a device that's now gone. In this bandaid release the ccb
early and then remember to not call xpt_release_ccb later.
1999-01-03 22:57:54 +00:00
Matt Jacob
656d1af535 Force ARCHIVE Vipers to be FIXED 1998-12-28 19:21:12 +00:00
Matt Jacob
5d754af79f a more correct change that should meet the sniff test 1998-12-24 06:01:15 +00:00
Matt Jacob
93f8eaab7d Unstaticize swi_camnet && swi_cambio so Alpha kernels can build again. 1998-12-24 02:43:41 +00:00
Matt Jacob
a1d461b7dd you can retry SYNC CACHE on UA errors 1998-12-23 16:48:17 +00:00
Eivind Eklund
3c68dd1aec Staticize.
Reviewed by:	gibbs
1998-12-22 20:05:23 +00:00
Matt Jacob
5a4f8fa5a4 Add a quirk NORRLS (no reserve/release) which can (and
will) get set for the devices that don't actually support
reserve/release (so we don't keep trying it).

Add softc storage and manage storing last I/O and CTL
commands that had errors (for correlative purposes).

In saclose clear the 'MOUNTED' bit if we either rewind or
unload (yes, this shouldn't be necessary since the next open
should catch whether a tape change occurred, but I'm having
some questions about that actually working so this is
safer for the moment). Oh, forgot to mention in previous
commit messages that some of the failures particularly at
close time cause the tape to be ejected (for the sake
of safety)- all this prior to redoing the state machine
(which is in progress) which will try and handle this better.

Complete the addition of the setmark support
(from Martin.Birgmeier@aon.at).
1998-12-22 17:26:13 +00:00
Matt Jacob
990635d6e8 1) Fix some serious bugs (1 botch on my part which caused a filemark to be
written even it the tape was opened readonly- 2 botches in deferred error
handling for FIXED LENGTH mode which caused panic && hand resp.). Fixed
a memory leak in sa_mount.
2) Fixed an annoying bug when turning of compression to actually reflect
this for future status calls.
3) Implement the MTIOCERRSTAT call where latched control and I/O residuals
and sense data are returned to the application asking for them.
1998-12-19 23:33:21 +00:00
Matt Jacob
b5c6d4c555 Add in block position/block locate functions. 1998-12-18 04:31:43 +00:00
Matt Jacob
d902101e8e Add structures and function definitions pertinent for hardware locate support. 1998-12-18 04:29:16 +00:00
Justin T. Gibbs
2e66665367 Correct the definition of the changer device capabilities page. The
previous definition confused some reserved bytes for exchange capabilities.
1998-12-17 22:26:39 +00:00
Matt Jacob
303826838c Add some tape specific density codes- only enough of them here to recognize
some specific older units so we can choose 2FM@EOD or FIXED blocksize
quirks.
1998-12-17 19:04:18 +00:00
Matt Jacob
888bec0f0c Several changes having to do blocksize- mostly to force variable as the default.
Attempt to determine (at mount time if not done so already) via density code
whether a device should default to fixed mode or not. Attempts to set to
variable that fail will cause fixed to be selected.

Similarly, the '2 filemarks at EOM' quirk is now determined (or attempted to
be determined) via density code. Some as yet not entirely tested code for
coping with 2FM@EOD position is now also in place.
1998-12-17 18:56:23 +00:00
Justin T. Gibbs
e500cd9896 Correctly track allocated accept target I/O ccbs so that they can be
aborted prior to disabling our lun.  This requires a second set of
links since we use the ones in the ccb_hdr during normal operations.

Nuke some unused variables.
1998-12-17 00:03:14 +00:00
Kenneth D. Merry
ff1fe75fab At Justin's request, limit the size of buffers that can be mapped into
and out of kernel address space (via the pass(4) and xpt(4) peripheral
drivers) to 64K (DFLTPHYS).  Some controllers, like the Adaptec 1542,
don't support more than 64K transactions.

We plan on eventually having the capability of limiting this size based
on min(MAXPHYS, controller max), but since that capability isn't here yet,
limit things to the lowest common denominator.
1998-12-16 21:00:06 +00:00
Kenneth D. Merry
79d49a061b Probable fix for the "cdda2wav" panics that various people have been
reporting since this past summer.  (I think Daniel O'Conner was the first.)

The problem appears to have been something like this:

 - cdda2wav by default passes in a buffer that is close to the 128K MAXPHYS
   limit.
 - many times, the buffer is not page aligned
 - vmapbuf() truncates the address, so that it is page aligned
 - that causes the total size of the buffer to be greater than MAXPHYS,
   which of course is a bad thing.

Here's a quote from the PR (kern/9067):

==================
In particular, note bp->b_bufsize = 0x0001f950 and bp->b_data = 0xf2219960
(which does not start on a page boundary).  vunmapbuf() loops through all
the pages without any difficulty until addr reaches 0xf2239000, and then
the panic occurs.  This seems to indicate that we are exceeding MAXPHYS
since we actually started from the middle of a page (the data is being
transfered to a non page aligned location).

To complete the description, note that the system call originates from
ReadCddaMMC12() (in scsi_cmds.c of cdda2wav) with a request to read 55
audio sectors of 2352 bytes (which is calculated to fall under MAXPHYS).
This in turn ends up calling scsi_send() (in scsi-bsd.c) which calls
cam_fill_csio() and cam_send_ccb().  This results in a CAMIOCOMMAND ioctl
with a ccb function code of XPT_SCSI_IO.
==================

The fix is to change the size check in cam_periph_mapmem() so that it is
like the one in minphys().  In particular, it is something like:

if ((buffer_length + (buf_ptr & PAGE_MASK)) > MAXPHYS)
	buffer is too big

My fix is based on the one in the PR, but I cleaned up a fair number of
things in cam_periph_mapmem().  The checks for each buffer to be mapped
are now in a separate loop from the actual mapping operation.  With the new
arrangement, we don't have to bother with unmapping any previously mapped
buffers if one of the checks fails.

Many thanks to James Liu for tracking this down.  I'd appreciate it if some
vm-savvy folks would look this over.  I believe this fix is correct, but I
could be wrong.

PR:		kern/9067 (also, kern/8112)
Reviewed by:	gibbs
Submitted by:	"James T. Liu" <jtliu@phlebas.rockefeller.edu>
1998-12-16 18:00:39 +00:00
Justin T. Gibbs
b6f7b14445 Enable/Disable our lun on open/close. Track resources kept at the controller
level so they can be reclaimed before attempting to disable our lun.
Correctly free descriptors.  Add periph locking and spl protection
around open and close.
1998-12-15 08:15:15 +00:00
Justin T. Gibbs
bb6087e5bb Wire up the XPT_ABORT and XPT_RESET_DEV ccb function codes so they can
be delivered to controller drivers.

Adjust for changes to the ccb_hdr list types in cam_queue.h
1998-12-15 08:13:10 +00:00
Justin T. Gibbs
6fc6a73645 Add definitions for TAILQ, LIST, and SLIST ccb_hdr queues. 1998-12-15 08:12:03 +00:00
Justin T. Gibbs
10fc9d90ce Return ENODEV instead of EINVAL when a particular exchange or move
operation exceeds the capabilities of the changer device.
1998-12-12 23:52:46 +00:00
Matt Jacob
ed744c4e51 Some fixes to handle fixed mode and variable mode more sensibly- and also
incorporate some notion of which revision the device is. If it's < SCSI2, for
example, READ BLOCK LIMITS is not a MANDATORY command.

At any rate, the initial state is to try and read block limits to get a notion
of the smallest and largest record size as well as the granularity. However,
this doesn't mean that the device should actually *in* fixed block mode should
the max && min  be equal... *That* choice is (for now) determined by whether
the device comes up with a blocksize of nonzero. If so, then it's a fixed block
preferred device, otherwise not (this will change again soon).

When actually doing I/O, and you're in fixed length mode, the block count is
*not* the byte count divided by the minimum block size- it's the byte count
divided by the current blocksize (or use shift/mask shortcuts if that worked
out...).

Then when you *change* the blocksize via an ioctl, make sure this actually
propagates to the stored notion of blocksize (and update the shift/mask
shortcuts).

Misc Other:
	When doing a mode select, only use the SCSI_SAME_DENSITY (0x7f) code if
the device is >= SCSI2- otherwise just use the saved density code.

	Recover from the ripple of ILLEGAL REQUEST not being 'retried' in that
RESERVE/RELEASE is not a mandatory command for < SCSI2 (so ignore it if it
fails).
1998-12-11 07:19:36 +00:00
Justin T. Gibbs
57c2edb48f Convert dadump to use reasonable data types so that some casting is unecessary. 1998-12-11 03:54:43 +00:00
Justin T. Gibbs
a0f37f55ee Do not attempt to retry commands that fail with ILLEGAL REQUEST status. 1998-12-11 03:53:05 +00:00
Justin T. Gibbs
4227e01d4d Convert debugging printfs to the CAM_DEBUG macro.
Allow sync transfers if the controller supports it.  Wide will follow
as soon as I get the kinks worked out of wide target transfers in the
aic7xxx driver (currently the only target mode driver in the tree).
1998-12-10 04:07:42 +00:00
Justin T. Gibbs
9819265820 Expand the hba_misc fied in the Path Inquiry ccb to allow a controller driver
to specify that it does not provide initiator services (PIM_NOINITIATOR)
and that the initial bus reset for device probing should be avoided
(PIM_NOBUSRESET).

Modify the XPT layer to honor these flags.
1998-12-10 04:05:49 +00:00
Matt Jacob
aa872be6f3 use CAM_DEBUG_XPT to track XPT; correct a misspelling 1998-12-06 00:06:48 +00:00
Matt Jacob
373524d467 print the appropriate SCSI revision (with CCS as a proper name for the announce message 1998-12-06 00:05:47 +00:00