to look up cookies properly, at least for standard controllers.
Cookies are used so that we don't have to pass around lots of args.
All of the dmainit functions use the unit number so it is essential
that we pass them a cookie with the correct unit number.
This may break working configurations if there are bugs in the
dmainit functions like the ones I just fixed for VIA chipsets.
Broken in: rev 1.4 of ide_pci.c and rev.1.139 of wd.c.
IDE hardare. The attempted fix in rev.1.182 was a no-op except for
adding dozens of style bugs. The undocumented options ALI_V and
DISABLE_PCI_IDE go away as a side effect. ALI_V was a no-op because
rev.1.182 was a no-op. DISABLE_PCI_IDE didn't actually disable
PCI IDE. It disabled the buggy code in wdprobe() at a cost of
completely breaking support for Promise controllers.
Broken in: rev.1.139
buffer had to be left on the head of the queue for [bufq]disksort()
to sort against. This isn't right for devices that can support multiple
active i/o's, and only the fd driver did it. "Fixing" this in rev.1.36
of ufs_disksubr.c broke the fd driver in much the same way as rev.1.52
of <sys/buf.h> broke it (see rev.1.119).
Bug reported and fix tested by: dt
- Don't try to set typematic rate if there is not a keyboard.
- Fix wrong test on error code.
- Don't try to claim the keyboard twice. The second call will fail.
keyboard and video card drivers.
Because of the changes, you are required to update your kernel
configuration file now!
The files in sys/dev/syscons are still i386-specific (but less so than
before), and won't compile for alpha and PC98 yet.
syscons still directly accesses the video card registers here and
there; this will be rectified in the later stages.
+ ECP parallel port chipset FIFO detection
+ DMA+FIFO parallel I/O handled as chipset specific
+ nlpt updated in order to use the above enhanced parallel I/O.
Use 'lptcontrol -e' to use enhanced I/O
+ Various options documented in LINT
+ Full IEEE1284 NIBBLE and BYTE modes support. See ppbus(4) for
an overview of the IEEE1284 standard
+ Detection of PnP parallel devices at boot
+ Read capability added to nlpt driver to get IEEE1284 compliant
printer status with a simple 'cat /dev/lpt0'
+ IEEE1284 peripheral emulation added to BYTE mode. Two computers
may dialog according to IEEE1284 signaling method.
See PERIPH_1284 option and /sys/dev/ppbus/ppi.c
All this code is supposed to provide basic functions for IEEE1284 programming.
ppi.c and nlpt.c may act as examples.
version and the asm version are inlined, and everything is cached,
the asm version is 1.75 times slower than the C version on P5's.
On K6's, it is only 1.25 times slower.
fixing it. See rev.1.22 of ../sound/audio.c for fixes. When both
the C version and the asm version are inlined, and everything is cached,
the asm version is 1.75 times slower than the C version on P5's. On
K6's, it is only 1.25 times slower.
versions of gcc and broken for current versions of egcs. The asm
here (for translate_bytes()) is now an interesting example of one
that needs to be volatile to work.
Fixed missing "memory" in the clobber list for translate_bytes().
Submitted by: "John S. Dyson" <dyson@iquest.net> but rewritten by me
from sc, vt and sio drivers. Use instead a linker_set to collect them.
Staticize ??cngetc(), ??cnputc(), etc functions in sc and vt drivers.
We must still have siocngetc() and siocnputc() as globals because they
are directly referred to by i386-gdbstub.c :-(
Oked by: bde
- Special registers of IO-DATA device's RSA series are defined in
ic/rsa.h (new file).
Pointed out by: Bruce Evans <bde@zeta.org.au>
Submitted by: Takahashi Yoshihiro <nyan@wyvern.cc.kogakuin.ac.jp>
deltas, but it is possible since I had a few merge conflicts over the last
few days while this has been sitting ready to go.
(Part 1 was committed to the config files, but cvs aborted grrr..)
Approved by: core
in cyopen() were done in a different order than in sioopen(), partly
to (ab)use a side effect of comparam() and partly because I didn't
understand what the reset was doing (it flushes the fifos). This
turned out to be more than a cosmetic problem. Flushing the fifos
quite late is good for discarding input that arrived while the line
state was being initialized, and in the cy driver it also seems to
reduce a problem with input that arrived long ago during the previous
close (the UART loses sync too easily and for too long).
change the original code but add an extra option "ALI_V" to check the
precise IDE port.(especially, secondary) Use the same option "ALI_V" on
the kernel config file to prevent generic DMA check causes wrong result.
(It seems buggy even on PIIX4 chipset, and I don't know when this bug start)
Should I add the option "ALI_V" into /sys/i386/conf/LINT ?
ethernet driver.
The BUGS section is still impressive, but the driver seems to work for
me now. Disclaimer: i haven't been able to test this under -current
so far (but it compiles, and the notebook it's intended for can now be
updated to -current more easy than before). Don't be afraid of the
many #ifdefs on __FreeBSD_version in the imported file; i want them in
the repository on the vendor-branch so other people can also manually
integrate it into older systems. I'll clean it up on the -current
branch in a followup commit. The vendor-banch version right now
supports systems back to 2.2R.
This driver should be layered upon ppc(4), but i currently have no
idea how to do this.
Eventually i'll further develop the driver to also support the more
modern RTL 8012 success, which seems to be present in a number of
cheap pocket ethernet adapters these days. Right now, i doubt it will
run with the 8012 without any changes.
Finally a big Thanks! to RealTek for promptly providing me with
documentation and with the source code for the 8012 pocket driver upon
request. I wish all vendors were that cooperative!.
The optimisation of only waiting before issuing new commands is
obviously invalid in general and it caused many errors in NIST-PCTS.
I think the errors were mostly for characters sent with the wrong
parity, etc., after a half complete tcsetattr().
Use microtime() instead of a magic loop count to limit the wait.
The wait is a busy-wait :-( and normally takes about 500 usec.
almost works properly. Unfortunately, there is no way to flush
the rx fifo without resetting the channel, which also flushes the
tx fifo. We avoid resetting even when both fifos need to be flushed,
since resetting seems to cause the rx to lose sync if it is done
while data is arriving.
Reminded by: NIST-PCTS
hardware is interrupt-driven to a fault and sending a BREAK requires
mode switching. Always running in the BREAK-capable mode as in PR 8318
would double the overhead for sending \0's.
Reminded by: PR 8318
Not tested on the if_sr, if_cx and if_ar drivers, but
expected to work just the same as it used to.
Any users of these drivers (or even better: donors
of hardware for them) please contact phk@freebsd.org
so we can test the next batch of changes to if_sppp.
as well as several functional additions.
(1) dot3 MIB support.
(2) if_media selection method support.
(3) bridge support.
(4) new boards support. Supported boards are as follows.
[PC/AT]
* Fujitsu FMV-180 series
* Allied-Telesis RE2000 series
* Allied-Telesyn AT1700 series
* Gateway Communications G/Ether series
* UB networks Access/PC ISA series
* TDK/LANX LAC-AX series
* ICL EtherTeam16i series
* RATOC REX-5586/5587
[PC-98]
* Allied-Telesis RE1000 series
* Allied-Telesis RE1000Plus/ME1500 series
* Contec C-NET(9N)E series
* Contec C-NET(98)P2 series
* UB networks Access/PC N98C+ series
* TDK/LANX LAC-98 series(not tested)
Submitted by: seki@sysrap.cs.fujitsu.co.jp (Masahiro Sekiguchi) and
chi@bd.mbn.or.jp (Chiharu Shibata)
in rev.1.30 (just before FreeBSD-1.1R) to almost match corresponding
breakage in FreeBSD-1.x's diskerr(). FreeBSD-2.x's diskerr() never
had the breakage.
floppy is used on the toshiba Libretto line of subnotebook computers.
It differs from a normal floppy in that you must use PIO rather than
DMA to transfer the data.
To enable this, you must add options "FDC_YE" to your kernel. I don't
have a machine that has a floppy and a pcmcia slot to test to make
sure that this doesn't impact normal floppy units, so I've left this as
an option.
I have ported this to -current and made an attempt to ensure that the
indentation conforms to style(9), aka the bruce filter.
Reviewed by: nate, markm
Submitted by: David Horwitt (dhorwitt@ucsd.edu)
for possible buffer overflow problems. Replaced most sprintf()'s
with snprintf(); for others cases, added terminating NUL bytes where
appropriate, replaced constants like "16" with sizeof(), etc.
These changes include several bug fixes, but most changes are for
maintainability's sake. Any instance where it wasn't "immediately
obvious" that a buffer overflow could not occur was made safer.
Reviewed by: Bruce Evans <bde@zeta.org.au>
Reviewed by: Matthew Dillon <dillon@apollo.backplane.com>
Reviewed by: Mike Spengler <mks@networkcs.com>
is completely empty. There is an interrupt for output completion. It
is painful to use, but polling method used in the corresponding fix in
sio.c (rev.1.152) can't be used because there is no status bit for
transmitter-empty. Now ttywait() works right.
Reminded by: NIST-PCTS
Don't call timeout() for DTR wakeup if the relevant timeout is already
active. This fixes "timeout table full" panics when sufficiently many
cyopen()s are interrupted while they are sleeping waiting for the
timeout to expire.
were half of their physical offsets for ISA and 1/4 of their physical
offsets for PCI, while all other Cyclades offsets were physical/1 for
ISA and physical/2 for PCI. Logically wrong macros were used to scale
CY16_RESET and CY_CLEAR_INTR to the correct physical offsets.
Fixed some style bugs (mostly long lines).
interrupt handlers. Instead, load and use it atomically as necessary.
This reduces mode switching overhead for "polled" mode interrupt handling
from 5 i/o's to 3 (per service type, per port) so that polled mode is only
slightly more inefficient than "interrupt" mode.
UARTs when their divisor latch registers are selected while they are
doing output. Waiting for (some) output to drain is not permitted for
the TCSANOW case of tcsetattr(). NIST-PCTS easily detects bugs like
this by testing at at speeds that no one would want to use (50 bps).
Deleted stale comments related to flushing i/o. Flushing works properly
for 16550s according to NIST-PCTS (it can't work properly for 16450s).
This finishes fixing all sio(16550)-related bugs found by NIST-PCTS.
keyboard/mouse/display switch products (console switches). Some
products claim they emulate the PS/2 mouse when the host computer
talks to the mouse while the mouse is actually routed to another
host.
flags 0x200 Do not try to identify the mouse model. All
mice will be recognized as "generic PS/2".
0x400 Do not reset the mouse. Some switches' response
to the reset command is too slow and the psm
will timeout.
0x1000 Relax error checking when probing the mouse
port.
- Added another flag for pad devices.
flags 0x800 Assume the pad behaves like ALPS GlidePoint
when the user `taps' the surface of the pad;
it will be reported as the fourth button.
devstart_start_transaction() call is misplaced - it is after the
wdustart() call that queues the transaction on the controller queue.
Normally this doesn't matter because we're running at splbio() so
nothing will look at the controller queue. However, obsolescent
code for syncing labels sometimes slept after the transaction was
started, so the transaction sometimes completed before it was
[recorded as] started. This code was misplaced even for syncing
labels. Move it to the right place. It should go away, but
something may depend on its side effects.
truncated to 32 bits.
* Change the calling convention of the device mmap entry point to
pass a vm_offset_t instead of an int for the offset allowing
devices with a larger memory map than (1<<32) to be supported
on the alpha (/dev/mem is one such).
These changes are required to allow the X server to mmap the various
I/O regions used for device port and memory access on the alpha.
minor devices.
Improve PLL/OCXO DAC dithering.
General remodeling.
Performance is now 2.5e-11 in frequency and +/- 100 nsec in time, both
of which are actually the limits of the transmitted signal.
bug sound was not played if the total amount of data written to
the device was less than one blocksize
Noticed by: NABETANI Masaki and FreeBSD-users-jp
1) The vnode pager wasn't properly tracking the file size due to
"size" being page rounded in some cases and not in others.
This sometimes resulted in corrupted files. First noticed by
Terry Lambert.
Fixed by changing the "size" pager_alloc parameter to be a 64bit
byte value (as opposed to a 32bit page index) and changing the
pagers and their callers to deal with this properly.
2) Fixed a bogus type cast in round_page() and trunc_page() that
caused some 64bit offsets and sizes to be scrambled. Removing
the cast required adding casts at a few dozen callers.
There may be problems with other bogus casts in close-by
macros. A quick check seemed to indicate that those were okay,
however.
hasseen_isadev so this will be less noisy when conflicts do exist.
Also eliminate redundant warnings about conflicts.
Requested by: bde
Reviewed by: gibbs
specified. This makes haveseen_isadev() useful for searching for a
free resource. This increases the bitrot in the pci RESOURCE_CHECK
code.
Fixed the pre-attach conflict message. The flag for distinguishing
pre-attach conflict checks from pre-probe ones was never set.
o For bt and aha only probe the one I/O range if a specific I/O is specified
in the config file.
o Don't even try to probe I/O ranges that have been seen already.
o If we conflict with an IRQ or DRQ, then fail the probe.
Requested by: bde, gibbs
Approved by: jkh
DEB macro). There are probably quite a few other messages that warrant
a similar treatment, and many more that should be converted to plain
log messages (e.g. "WARNING: wrintr but write DMA inactive!"). Now
that I think of it, same goes for the CAM code (e.g. the famed "tagged
openings" message)
reported bug.
At using tcpdump for cs interface, tcpdump only dump packet which
src or dst MAC-address is cs interface. cs interface can't look up
packet between others.
Submitted by: MIHIRA "Sanpei" Yoshiro <sanpei@sanpei.org>
add support for Vibra16X, OPTi925, and bring in several assorted
fixes to the code and documentation.
Also present here are apm hooks so that laptops can properly
reconfigure the hardware after suspend (tested on the Libretto50).
Reviewed by: jordan
- Express various sizes in bytes, rather than Kbytes, in the video
mode and adapter information structures.
- Fill 0 in the linear buffer size field if the linear frame buffer
is not available.
- Remove SW_VESA_USER ioctl. It is still experimetal and was not meant
to be released.
- Fix missing cast operator.
- Correctly handle pointers returned by the VESA BIOS. The pointers
may point to the area either in the BIOS ROM or in the buffer supplied
by the caller.
- Set the destructive cursor at the right moment.
- there were too many global variables (there still are :-).
- the data section was bloated by explicit initializations of static
variables to 0 (only fixed the recently changed ones).
- WRAPHIST() had silly parentheses around foo->bar.
- the comment about inline functions was stale.
- the comment about Userconfig presumes too much about the boot environment.
- `i' was reused confusingly in scioctl().
- the declaration of `butmap' used a deprecated K&R misfeature.
- the initializeation of `butmap' had an unnecessary line break.
- `unsigned char' was not consistently (mis)spelled as u_char.
- English was poor in a comment in videoio.c.
Submitted by: bde
Now supposedly less harmful way of accessing VGA sequencer registers
is default. An alternative, often troublesome, I/O access is optional.
Discussed with: sos, jkh
- Handle pixel (raster text) mode properly.
- Clear screen and paint border right.
- Paint text attribute (colors).
- Fix off-by-one errors.
- Add some sanity checks.
- Fix some function prototypes.
- Add some comment lines.
- Define generic text mode numbers so that the user can just give
"80x25", "80x60", "132x25"..., rather than "VGA_xxx", to `vidcontrol'
to change the current video mode. `vidoio.c' and `vesa.c' will map
these numbers to real video mode numbers appropriate and available
with the given video hardware. I believe this will be useful to make
syscons more portable across archtectures.
went backwards when interrupts were masked for more than one i8254
interrupt period. It sometimes went backwards when the i8254 counter
was reprogrammed. Neither of these should happen in normal operation.
Update the i8254 timecounter support variables atomically. Calling
timecounter functions from fast interrupt handlers may actually work
in all cases now.
Kazu writes:
The VESA support code requires vm86 support. Make sure your kernel
configuration file has the following line.
options "VM86"
If you want to statically link the VESA support code to the kernel,
add the following option to the kernel configuration file.
options "VESA"
The vidcontrol command now accepts the following video mode names:
VESA_132x25, VESA_132x43, VESA_132x50, VESA_132x60, VESA_800x600
The VESA_800x600 mode is a raster display mode. The 80x25 text will
be displayed on the 800x600 screen. Useful for some laptop computers.
vidcontrol accepts the new `-i <info>' option, where <info> must be
either `adapter' or `mode'. When the `-i adapter' option is given,
vidcontrol will print basic information (not much) on the video
adapter. When the `-i mode' option is specified, vidcontrol will
list video modes which are actually supported by the video adapter.
Submitted by: Kazutaka YOKOTA yokota@FreeBSD.ORG
and use this when masking/unmasking interrupts.
Maintain a mapping from (iopaic number, int pin) tuple to irq number,
and use this when configuring devices and programming the ioapics.
Previous code assumed that irq number was equal to int pin number, and
that the ioapic number was 0.
Don't let an AP enter _cpu_switch before all local apics are initialized.
Add a sysctl 'machdep.cs_recv_delay' to specify how long to wait after
receiving a packet in order to check for a subsequent (back-to-back)
packet. The cs8900 has a very small receive buffer, so this helps avoid
overflows at the cost of some extra CPU overhead.
Submitted by: Oleg Sharoiko <os@rsu.ru>, MIHIRA "Sanpei" Yoshiro <sanpei@sanpei.org>
for the Lite2 fix for always returning EIO in dead_read().
Cleaned up the cdevswitch initializers for all tty drivers.
Removed explicit calls to ttsetwater() from all (tty) drivers. ttsetwater()
is now called centrally for opens, not just for parameter changes.
The check for dropping unicast packets not sent to our ethernet
address is after the bpf tap, but not conditioned on it. All packets
received should get handed to bpf, and unicast packets not to us (mac)
should get dropped whether or not there is a bpf listener. I believe
that the common optimization that the interface is in hw promisc mode
iff there is a bpf listener is in general wrong, but more frequently
so on wavelans.
I think Max's fix makes bpf listeners not see unicast packets sent to
others, but I'm not sure.
One can argue that checking on MOD_ENAL is wrong, but the code only
drops packets that shouldn't be received. The correctness condition
is that it be run whenever unicast packets without our mac address can
be received.
PR: kern/7144
Submitted by: Greg Troxel <gdt@ir.bbn.com>
for 1 second's worth of input) and larger tty output buffers. The
interrupt-level buffers are still too small for speeds above 115200
bps (only a little too small for 230400 bps if RTS flow control is
enabled).
Don't call ttsetwater() explicitly in open(). It is now called for
the TTYDISC l_open() and should be static.
Don't attempt to register the cdevsw more than once.
I don't have access to a real VT220 to verify this against.
However, I'm committing the patch in `good faith' because
(a) getting hold of a real VT220 is going to be increasingly difficult
the longer the PR sits around,
(b) some one was troubled enough to in a PR and
(c) the fix is minor and has no other implications.
PR: 7559
Submitted by: Christian Weisgerber <naddy@mips.rhein-neckar.de>
of invariants to cyattach().
Fixed minor bugs:
- cyparam() returned without restoring the ipl in the error cases. This
was harmless because cyparam() is always called at spltty().
- one check for "rev. J or higher" actually checked for precisely rev. J.
swapped RTS/DTR). Merge the vendor's modification of the 2.2.6-release
version into -current for reference. Will be cleaned up in next commit.
Obtained from: ftp://ftp.cyclades.com/pub/cyclades/cyclom-y/freebsd/2.2.6/cyy226.tar.gz
instead of at compile time using ifdefs.
Use _swi_null instead of dummycamisr. CAM and dpt should call
register_swi() instead of hacking on ihandlers[] directly.
in ddb) which I broke by changing %8[l]x to %8p. Hacked the central
printf routine to not add an "0x" prefix for %p formats if the field
width is nonzero. The tables are still horribly misformatted on
64-bit machines.
Use %p instead of %8p to print pointers when the field width isn't
important.
chip.
It has been observed that the problem is most apparent:
a) in notebook computers,
b) and/or in the systems with C&T video chips.
Define the new configuration option SC_BAD_FLICKER in the kernel
configuration file to remove outb()/outw() calls in question.
DOS partition type 15 (Extended DOS, LBA) as a container for
DOS logical volumes, so the appropriate slices (e.g. sd1s5)
are not initialized.
PR: 7549
PR: 4120
Reviewed by: phk
Submitted by: Jim Mattson <jmattson@sonic.net>
- ppbus now supports PLIP via the if_plip driver
- ieee1284 infrastructure added, including parallel-port PnP
- port microsequencer added, for scripting the sort of port I/O
that is common with parallel devices without endless calls up and down
through the driver structure.
- improved bus ownership behaviour among the ppbus-using drivers.
- improved I/O chipset feature detection
The vpo driver is now implemented using the microsequencer, leading to
some performance improvements as well as providing an extensive example
of its use.
Reviewed by: msmith
Submitted by: Nicolas Souchu <Nicolas.Souchu@prism.uvsq.fr>
saver and splash screen can all work properly with syscons. Note that
the splash screen option (SC_SPLASH_SCREEN) does not work yet, as it
requires additional code from msmith.
- Reorganized the splash screen code to match the latest development
in this area.
- Delay screen switch in `switch_scr()' until the screen saver is
stopped, if one is running,
- Start the screen saver immediately, if any, when the `saver' key is
pressed. (There will be another commit for `kbdcontrol' to support
this keyword in the keymap file.)
- Do not always stop the screen saver when mouse-related ioctls
are called. Stop it only if the mouse is moved or buttons are
clicked; don't stop it if any other mouse ioctls are called.
2. Added provision to write userland screen savers. (Contact me if you
are interested in writing one.)
- Added CONS_IDLE, CONS_SAVERMODE, and CONS_SAVERSTART ioctls to
support userland screen savers.
3. Some code clean-ups.
the screen mode is changed even if another vty has larger size.
Reallocate the buffer only when the new screen size is larger than
the current cut buffer size.
When bell is of "quiet" types, the console won't ring (or flush)
if the ringing process is in a background vty.
PR: i386/2853
- Modify the escape sequence 'ESC[=%d;%dB' so that bell pitch and
duration are set in hertz and msecs by kbdcontrol(1).
There will be a corresponding kbdcontrol patch.
PR: bin/6037
Submitted by: Kouichi Hirabayashi (kh@eve.mogami-wire.co.jp)
and DSO_NOLABELS flags prevent searching for slices and labels
respectively. Current drivers don't set these flags. When
DSO_NOLABELS is set, the in-core label for the whole disk is cloned
to create an in-core label for each slice. This gives the correct
result (a good in-core label for the compatibility slice) if
DSO_ONESLICE is set or only one slice is found, but usually gives
broken labels otherwise, so DSO_ONESLICE should be set if DSO_NOLABELS
is set.
- Call isa_dmadone() whenever necessary to stop DMA and/or free bounce
buffers. Undead DMA corrupted the malloc freelist fairly consistently
in the following configuration: SLICE kernel, 2 floppy drives, no disk
in fd0, disk in fd1.
- Don't call fdc_reset() from fd_timeout(). Doing so gave an "extra"
interrupt which was usually misinterpreted as being for completion
of the next FDC command; the interrupt for completion of the next
FDC command was then usually misinterpreted... There were further
complications for interrupts latched by the soft-spl mechanism so
that they were delivered after all the h/w interrupts went away.
This caused at least wrong head settle delays and may be why the
FreeBSD floppy driver seems to munch floppies more than most floppy
drivers. The reset was unnecessary anyway in cases that didn't have
the bug described next, since is was repeated a little later for
the IOTIMEDOUT state. The state machine has complications to handle
resets correctly, so just use it.
- Don't call retrier() from fd_timeout(). The IOTIMEDOUT state needs
to be processed next, and it isn't valid to set to that state if
retrier() has aborted the current transfer. Doing so caused null
pointer panics after the previous bug was fixed.
Improved error handling:
- If an i/o is aborted, arrange to reset in the state machine before
doing the next i/o. New fdc flag for this. This fixes spurious
warnings and lengthy busy-waiting for the next i/o.
- Split STARTRECAL into RESETCOMPLETE and STARTRECAL and only check
for the results from reset if we actually reset. This fixes spurious
warnings for other paths to STARTRECAL. [Oops, it may break reset
handling for motor-off resets.]
Cleanups in fd_timeout():
- Renamed to fd_iotimeout() to make it clearer that it is only used
for i/o.
- Don't handle the bp == 0 case. This case can't happen for i/o.
- Don't check for controller-busy. We know it must be.
- Don't print anything. retrier() already prints too much for normal
errors.
- Fudge the state differently so that the state machine advances
fdc->retry and the status is invalid (perhaps this should fudge a
valid state like the one for WP).
- Style fixes.
small part of a bug suite beginning in the SLICE probes but mostly in the
floppy driver. This is a quick fix: the auto case shouldn't be special;
DMA should also be stopped in isa_dma_release(); isa_dmastop() probably
shouldn't exist; common DMA registers should not be accessed without
locking.
controller reports a successful seek, it is very unlikely to report
seeking to a cylinder other than the one requested, but we check for
this, and botched the error handling for the requested_cylinder != 0
case. This error happened when the bug fixed in rev.1.52 of <sys/buf.h>
caused the head of buffer queue to change to one starting on a different
cylnder - the requested cylinder was found, but it wasn't what we
thought we requested. The fix is simply to arrange to reset the state
machine.
Corruption of the buffer queue seems to only have been a problem in the
floppy driver. Other drivers dequeue the head of the queue before doing
physical i/o on it, so the corruption at worse broke the elevator sort
order. Dequeueing breaks it anyway.
These asm statments are not quite as pessimal as when I complained
about them in rev.1.9 of audio.c. They seem to be only 40% slower
than the C version on P5's and the same speed on K6's.
