of associated mbuf clusters) in the RX ring from 4 to 16. On my
really fast PI 400Mhz test machines, 4 descriptors (and associated
mbuf clusters) is enough to achieve decent performance without any
RX overruns. However, one person reported problems with the following
scenario:
- P90 system running FreeBSD with a 3c905B-TX adapter, slow IDE hard
disk (Quantum Bigfoot?)
- PII 266 with SCSI disks running LoseNT and also with a 3c905B-TX
- Both machines connected together via crossover cable at 100Mbps
full-duplex
- LoseNT machine writing largs amounts of data (2.5 GB work of
files each in the neighborhood of 1 to 2 MB in size) via samba to
the FreeBSD machine
In this case, the LoseNT machine is sending data very fast. Apparently
there weren't any problems initially because the user was writing to
one particular disk which was relatively fast, however after this disk
filled up and the user started writing to the second slower disk, RX
overruns would occur and sometimes the RX DMA engine would stall after
a 100 to 500MB had been transfered. The xl_rxeof() handler is supposed
to detect this condition and restart the upload engine; I'm not sure
why it doesn't, unless interrupts are being lost and the rx handler
isn't getting called.
This is still an improvement over the Linux driver, which uses 32
descriptors in its receive ring. :)
Problem reported by: Heiko Schaefer <hschaefer@fto.de>
'three-stage' bootstrap.
There are a number of caveats with the code in its current state:
- The i386 bootstrap only supports booting from a floppy.
- The kernel and kld do not yet know how to deal with the extended
information and module summary passed in.
- PnP-based autodetection and demand loading of modules is not implemented.
- i386 ELF kernel loading is not ready yet.
- The i386 bootstrap is loaded via an ugly blockmap.
On the alpha, both net- and disk-booting (SRM console machines only) is
supported. No blockmaps are used by this code.
Obtained from: Parts from the NetBSD/i386 standalone bootstrap.
They shouldn't be used there either. They should have gone away
about 3 years ago when the statically initialized devswitches went
away, but su.c unfortunately still frobs the cdevswitch in the old
way.
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>
If I'm reading the manual correctly, the 3c905B actually loses its
PCI configuration during the transition from D3(hot) back to D0, not
during the transition from D0 to D3(hot). This means it should be possible
to save the existing PCI settings, restet the power state, then restore
the PCI settings afterwards. Changed xl_attach() to attempt this first
thing before the normal PCI setup. I'm not certain this will work correctly,
but it shouldn't hurt.
If xl_init() is called while an autoneg session is in progress, the
autoneg timeout and chip state will get clobbered. Try to avoid this
by checking sc->xl_autoneg at the start of xl_init() and defer
the initialization until later if it's set. (xl_init() is always called
at the end of an autoneg session by xl_autoneg_mii().)
Problem pointed out by: Larry Baird <lab@gta.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.
since (hardware) ttys have too low a bandwidth to benefit significantly
from large buffers. Use twice the old limit for the new-default case
and 8 times the old limit for the driver-specifies-watermark case.
Nothing uses these cases yet.
Removed related debugging code.
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>
in a SMP system. Unexpected things could happen if each cpu
has a different ldt setting and one cpu tries to use value
of currentldt set by another cpu.
The fix is to move currentldt to the per-cpu area. It includes
patches I filed in PR i386/6219 which are also user ldt related.
PR: i386/7591, i386/6219
Submitted by: Luoqi Chen <luoqi@watermarkgroup.com>
clustering is obsolescent technology so hardly anyone noticed. On
a DORS 32160 SCSI drive with 4 tags, read clustering makes very
little difference even for huge sequential reads. However, on a
ZIP SCSI drive with 0 tags, the minimum overhead per block is about
40 msec, so very large clusters must be used to get anywhere near
the maximum transfer rate. Using clusters consisting of 1 8K block
reduces the transfer rate to about 250K/sec. Under msdosfs, missing
read clustering is normal and a cluster size of 1 512 byte block
reduces the transfer rate to about 25K/sec.
Broken in: rev.1.18
not the necessarily the same as the seconds part of getmicrotime()
yet, and anyway, we should have used `time_second' if we only wanted
a sloppy value for the seconds part. There is no point in making
ibcs2's time(2) more efficient than FreeBSD's time(3).
description of DPT_SHUTDOWN_SLEEP in LINT. Didn't add timestamps
so that the (combined?) sleep interval can be printed as intended
in the original printf.
stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC
and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate
support for the XL adapters in the vortex driver. LAstly, add a man
page.
(Also added an MLINKS entry for the ThunderLAN man page which I forgot
previously.)
applications. Here's how it works.
The kernel should include <machine/elf.h> to get the definitions
for the native architecture. It can reference the various ELF
structures with generic names like Elf_Sym, Elf_Shdr, etc. A define
__ELF_WORD_SIZE is also available with the value 32 or 64 as
appropriate for the native architecture.
Generic applications should include <elf.h>, which is just a wrapper
for <machine/elf.h>.
Applications such as object file dumpers that need to deal with
foreign ELF files can include <sys/elf32.h> and/or <sys/elf64.h>.
Both can be included from the same source file if desired. The
structure names must be referenced using wordsize-specific names
like Elf32_Sym, Elf64_Shdr, etc.
I haven't change the alpha stuff, but I haven't broken it either.
Cast pointers to (vm_offset_t) instead of to (u_long) (as before) or to
(uintptr_t)(void *) (as would be more correct). Don't cast vm_offset_t's
to (u_long) just to do arithmetic on them.
mp_machdep.c:
Cast pointers to (uintptr_t) instead of to (u_long). Don't forget
to cast pointers to (void *) first or to recover from integral
possible integral promotions, although this is too much work for
machine-dependent code.
vm code generally avoids warnings for pointer vs long size mismatches
by using vm_offset_t to represent pointers; pmap.c often uses plain
`unsigned int' instead of vm_offset_t and didn't use u_long elsewhere,
but this style was messed up by code apparently imported from mp_machdep.c.
address constants. This fixes some warnings for conversions from
64-bit integers to 32-bit pointers on i386's with 64-bit longs.
vm86 still uses too many u_longs.
pointers. Neither is portable, but "correct" casts to integral
types are much uglier - they lead to expressions like
ptr = (struct struct_with_too_long_a_name *)(void *)(uintptr_t)
((uintptr_t)(void *)ptr + offset);
Here the cast to the struct pointer is to match the surrounding
style of this file (and not depend on C's feature of properly
converting `void *' on assignment or cast), the `void *' casts are
because uintptr_t is only specified to work on `void *' pointers
(I missed this in about 100 lines of previous changes from [u]long
to [u]intptr_t), the outer cast to a uintptr_t is in case the
addition promoted the type, and the inner cast to a uintptr_t
corresponds to the one cast to an integer in the original code.
Don't depend on gcc's feature of casting lvalues.
It can be integral or a struct in POSIX, so it is difficult to print,
but it is actually declared as unsigned long. Assume that it is
unsigned integral.
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
ioctl() routine at the end of if_delmulti() so that interfaces with
hardware multicast filtering can update their filters in a timely
manner.
If the interface doesn't support hardware multicast filtering, then
reception of multicast frames is done using 'promiscious mode' or
'capture all multicast frames' mode and software filtering in the
kernel. In this case, it doesn't matter if if_delmulti() ever does
an SCIODELMULTI on the interface or not: if MULTICAST support is
enabled, then we join the 'all hosts' group when the interface is
configured, and remain in it until the interface is brought down.
Without hardware filtering, joining one group means joining all
groups, so it makes no difference if we call the SIOCDELMULTI
routine.
If the interface does support hardware multicast filtering, then
by not reprogramming the hardware filter in if_delmulti(), we have
to wait until somebody calls if_setmulti(), during which time the
interface is receiving frames for multicast groups in which we are
no longer interested.
was broken), 1.30 (COMPAT_43 option header was missing), 1.31 (DEVFS
option header was missing), 1.33 (garbage pointers were followed
in debugging code). Cosmetic changes from 1.27, 1.32, 1.36, 1.37.
Of course, the DEVFS code didn't even compile. Fixed. Not tested.
Forgotten by: brian
This file should not exist. It is the same as dgb.c except for lots of
renamed variables, about 250 lines removed, and only about 100 lines of
real differences.
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.
instead of 0) was "restored" after handling a trap that occurred while
returning to user mode. This bug was most noticeable for VM86 and is
still detected and fixed up (on return from the next exception) in doreti
if VM86 is configured.
different size (on i386's with 64 bit longs). Cosmetic. Non-cosmetic
unportabilities were already hidden by using ntohl() to convert a
32-bit user DMA address to a long.
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.
* Support for AlphaStation 200, 250, 255, 400
* Untested support for UDB, Multia, AXPpci33 (Noname)
* Support for Personal Workstation 433a/433au, 500a/500au, 600a/600au (Miata)
* Some minor fixes and improvements to interrupt handling.
Submitted by: Andrew Gallatin <gallatin@cs.duke.edu> (AS200, Miata)
Obtained from: NetBSD (some code for AS200, Miata, Noname)
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>
when certain .mk files include other .mk files. This will remove the
need for multiple include protection in some other makefiles around the
tree (and helps some elf conditionals).
hopefully become a portable driver usable by all architectures. The api
support files have had to be copied to sys/alpha/include since userland
programs expect to find them in <machine/*.h>.
All the revision history of the i386 syscons has been retained by a
repository copy.
managed to avoid corruption of this variable by luck (the compiler used a
memory read-modify-write instruction which wasn't interruptable) but other
architectures cannot.
With this change, I am now able to 'make buildworld' on the alpha (sfx: the
crowd goes wild...)
Fixed nearby bugs (in linux_alarm()):
- the itimer for the alarm was relative to the epoch instead of relative
to the boot time. This was harmless because the itimer's interval is 0.
- the seconds arg was not checked for validity before converting it to a
possibly different value.
- printf format errors.
Improvements:
Don't use splclock(). splsoftclock() suffices. Don't complicate things
by micro-optimizing interrupt latency.
Minor improvements:
Various micro-optimizations to exploit the specialness of the alarm itimer
and the value 0.
interrupts which now defers them until the transmit queue if filled
up with completed buffers. This has two advantages: first, it reduces
the number of transmitter interrupts to just 1/120th of the rate
that they occured previously, and two, running down many buffers
at once has much improved cache effects.
- 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)
- probe for PHYs by checking the BMSR (phy status) register instead
of the vendor ID register.
- fix the autonegotiation routine so that it figures out the autonegotiated
modes correctly.
- add tweaks to support the Olicom OC-2326 now that I've actually had
a chance to test one
o Olicom appears to encode the ethernet address in the EEPROM
in 16-bit chunks in network byte order. If we detect an
Olicom card (based on the PCI vendor ID), byte-swap the station
address accordingly.
XXX The Linux driver does not do this. I find this odd since
the README from the Linux driver indicates that patches to
support the Olicom cards came from somebody at Olicom; you'd
think if anyone would get that right, it'd be them. Regardless,
I accepted the word of the disgnoatic program that came bundled
with the card as gospel and fixed the attach routine to make
the station address match what it says.
o The version of the 2326 card that I got for testing is a
strange beast: the card does not look like the picture on
the box in which it was packed. For one thing, the picture
shows what looks like an external NS 83840A PHY, but the
actual card doesn't have one. The card has a TNETE100APCM
chip, which appears to have not only the usual internal
tlan 10Mbps PHY at MII address 32, but also a 10/100 PHY
at MII address 0. Curiously, this PHY's vendor and device ID
registers always return 0x0000. I suspect that this is
a mutant version of the ThunderLAN chip with 100Mbps support.
This combination behaves a little strangely and required the
following changes:
- The internal PHY has to be enabled in tl_softreset().
- The internal PHY doesn't seem to come to life after
detecting the 100Mbps PHY unless it's reset twice.
- If you want to use 100Mbps modes, you have to isolate
the internal PHY.
- If you want to use 10Mbps modes, you have to un-isolate
the internal PHY.
The latter two changes are handled at the end of tl_init(): if
the PHY vendor ID is 0x0000 (which should not be possible if we
have a real external PHY), then tl_init() forces the internal
PHY's BMCR register to the proper values.
interface congestion (eg: nfs over a ppp link, etc). Don't log these
for UDP mounts, and don't cause syscalls to fail with EINTR.
This stops the 'nfs send error 55' warnings.
If the error is because the system is really hosed, this is the least
of your problems...
Sparse macros have moved to <machine/swiz.h>.
Fix sparse memory access so that it actually works as intended.
Tidy up sparse configuration access slightly.
st_write_filemarks(). This means that it is possible to write a file
on the tape for which all the writes and the close return without
errors, but the all bits did not make it onto the tape.
PR: 741
Reviewed by: phk
Submitted by: Andrew Heybey <ath@niksun.com>
of some disk ioctls and uniformizes bounds checking in the strategy
routine. EOF handling got fixed as a side effect. The changes
are similar to old changes for SCSI disks, except slices and labels
are not searched for on the disk and the in-core label has a few
more details (mostly just for backwards compatibility). Bugs in
the in-core label had to be fixed to get dsopen() to accept it.
The slice interfaces had to support large sectors for all this to
work.
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.
`make world' to avoid problems with picking up (new) target or (stale)
host shared libraries.
Don't honor -static in LDFLAGS for linking LKMs. LDFLAGS is not
actually for ld, but we use it anyway, and must prevent -static being
misinterpreted as -s.
Don't hide any of the link steps.
- 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.
protection checks. Using the partition-relative blkno in some
parts broke the write protection for partitions at unusual
offsets (only for partitions at offset 1 on i386's).
several new features are added:
- support vc/vp shaping
- support pvc shadow interface
code cleanup:
- remove WMAYBE related code. ENI WMAYBE DMA doen't work.
- remove updating if_lastchange for every packet.
- BPF related code is moved to midway.c as it should be.
(bpfwrite should work if atm_pseudohdr and LLC/SNAP are
prepended.)
- BPF link type is changed to DLT_ATM_RFC1483.
BPF now understands only LLC/SNAP!! (because bpf can't
handle variable link header length.)
It is recommended to use LLC/SNAP instead of NULL
encapsulation for various reasons. (BPF, IPv6,
interoperability, etc.)
the code has been used for months in ALTQ and KAME IPv6.
OKed by phk long time ago.
best place to set it, and the wd and wfd strategy routines don't
set it (for failed transfers) because they expect dscheck() to
initialize everything necessary. dscheck() has always set B_ERROR,
but this is not quite sufficient, because b_resid is used by physio()
to decide how much of a B_ERROR'ed i/o was done.
formats and args to match. Fixed old printf format errors (all related;
most were hidden by calling printf indirectly).
This change somehow avoids compiler bugs for 64-bit longs on i386's,
although it increases the number of 64-bit calculations.
fork_trampoline() if switchtime is valid. This fixes not accounting
for the time between the previous context switch and and the current
time (when the forked child starts up here) in most cases - the time
is now counted in the child's runtime. I think it actually fixes
all cases, and switchtime is always valid here, since there must have
been a context switch just before the forked child starts up. Some
code should be removed if this is correct. The check that switchtime
is valid sometimes gives a false negative because the check isn't
correct until the after the first context switch after the system
has been up for >= 1 second.
The machine can not recognize this PD drive
as neither PD drive nor CD-ROM Drive.
So I can not use CD-ROMs and PDs from FreeBSD.
PR: 7423
Reviewed by: phk
Submitted by: Takura Koyama <takura@popweb.ne.jp>
timeslice of the exiting process was counted for both the exiting
process and the next process to run if the next process runs
immediately.
Broken in: mostly in kern_clock.c rev.1.70 (1998/05/28)
potentially re-use the stack page.
Cosmetic cleanup of the code to de-obfuscate it and make it easier
to follow. There should be no functional changes in this commit.
but isn't valid.
Reimplement pmap_remove() to be much more efficient at removing large
stretches of addresses.
As part of reimplementing pmap_remove() fix pmap_protect() so that it stands
a hope of working.
Change the port address argument to pci_map_port to pci_port_t* which is
defined as u_int on the alpha, u_short on i386. This is a stopgap with a
hopefully limited lifetime.
Discussed with: Stefan Esser <se@freebsd.org>
code still left in there. The macros it describes disapeared some-
time since 4.4BSD lite.
PR: 7246
Reviewed by: phk
Submitted by: Stefan Eggers <seggers@semyam.dinoco.de>
