Newer boards (e.g., Vortex) take time to reset, and wedge on some machines.
Polling on S_COMMAND_IN_PROGRESS would be better, but that doesn't
work with xx_RESET commands on 3c509s.
For some reason it wouldn't get positioned right when mapped in through the
blitter memory mapped location, so switched to the register mapping, which
works.
XXX colormap handling for the cursor is still broken.
CONFIG_CNTRL bits by making the "conn" argument to epconfig()
a u_int, not a u_int16_t, and by defining 100mbit media
bits in the upper 16 bits.
A redesigned interface that fixes this properly is in the pipeline.
comments in pr port-i386/2358: "if_ep/3C590 no longer working"
* Only call Vortex media-config code on PCI-attached devices, until
we know if it works on ISA and EISA large-packet devices also.
* Fix typo in printf of MAC address
rather than unknown.
* Look for 100mbit interfaces, and for any present, set corresponding
bits in `conn' argument to epconfig(). epconfig() will now print
100Mbit media by name.
look for 100mbit Xcvrs, print their names, add to ep_connectors.
* Print 10baseT as 10baseT, since media name (utp) is no longer a unique
transceiver type.
* Redo FIFO memory size printing, also show Tx/Rx split (xxx isa 3c509?).
* Print MAC address on a separate line, with 100Mbit media and fifo-ram
info it no longer fits on the same line as media types.
* Comment the fact that the "connection" media-selection code just doesn't
grok 100Mbit media.
* add code to look for 3c59x/3c90x media-select and autoconfig bits, though
we don't yet dynamically probe media when the autoconfig bit is set
(see previous bullet).
* XXX the whole connector/media code should be redone, perhaps a' la
the BSDI ifmedia design.
For now, we really only need 8 bits of media type.
* Add bogus definitions for 100mbit transcievers, to be set in the
"conn" argument to ep_config(). Should really use the same bits
as in EP_W0_CONFIG_CTRL bits, but I don't know what bits the 3c515
&c use for 100mbit media.
* Define bitflags for 100Mbit transceiver options in ep_connectors.
* Add definitions of powerup/powerdown commands. for interfaces with
power management. Should send a powerup after APM resume?
* Add definition of fragment-busmaster (vortex) START_DMA command.
* Add definition of PCI configuration-space media-present bits.
(really just EP_W3_RESET_OPTIONS remapped into pci config space.)
* Add comments describing what the #define's above are used for.
For now, we really only need 8 bits of media type.
* Add bogus definitions for 100mbit transcievers, to be set in the
"conn" argument to ep_config(). Should really use the same bits
as in EP_W0_CONFIG_CTRL bits, but I don't know what bits the 3c515
&c use for 100mbit media.
* Define bitflags for 100Mbit transceiver options in ep_connectors.
* Add definitions of powerup/powerdown commands. for interfaces with
power management. Should send a powerup after APM resume?
* Add definition of fragment-busmaster (vortex) START_DMA command.
* Add definition of PCI configuration-space media-present bits.
(really just EP_W3_RESET_OPTIONS remapped into pci config space.)
* Add comments describing what the #define's above are used for.
getting the 'bootpreference' ourselves when it is not set to any legal
value.
- Because the relation between memtop/membot and the video base does not
hold on the Hades, skip testing on it.
* Add IPL_HIGH. Alias IPL_HIGH to IPL_RTTY
* Define splrtty (used by the scn driver).
* Add iv_level and iv_mask to struct iv. These define the IPL at
which the handler should run.
* intr_establish now takes *two* ipls. One at which the handler should run,
and one in which the interrupt should get blocked.
* Define SCN_RING_HIWAT and add sc_rbhiwat to scn_softc. When the
ringbuffer gets full up to this mark, rts gets asserted (Ian Dall).
* Add sc_rx_blocked to scn_softc. This is used to tell scnsoft that
rxinterrupts are blocked (Ian Dall).
* intr_establish now takes *two* ipls. One at which the handler should run,
and one in which the interrupt should get blocked.
All the following changes were provided by Ian Dall:
* Remove SCN_HSOFT. Now that the pc532's software interrupt system is running
from int14, this won't work anymore.
* Add code to detect uart type (SCN2681, SCC2692 or SCC26C92).
* Cleanup my RECOVER mess. Thank's to Ian Dall for reading the datasheet
and removing all the unnecessary RECOVERs.
* Cleanup scnrxintr.
* Implement scnhiwflow.
* Do not use automatic RTS. This and scnhiflow and some changes to the
interrupt system suggested by Ian fix the scn overrun problem at 57600
baud.
* intr_establish now takes *two* ipls. One at which the handler should run,
and one in which the interrupt should get blocked.
* Make transfersize configurable for input and output. Default the input
transfersize to 128 bytes. That way interrupts won't be disabled for more
then about 128us-256us for slow targets. This is necessary to prevent scn
fifo overruns. Cleanup the pdma code. Thank's to Ian Dall for these
changes.
* Make it work on 30mhz machines.
* Configure MSDOSFS, CD9660, KERNFS, NFSCLIENT and INET.
* Configure SCSI-cds, slip and ppp.
* Update for mi ramdisk driver.
* Initialize iv_level for all interrupts.
* Protect the call to the sir-handler with splraise.
* intr_establish now takes *two* ipls. One at which the handler should run,
and one in which the interrupt should get blocked.
* Compute iv_mask for all interrupts in intr_establish.
'const char *', and 'void *', respectively. The second arg is taken directly
from user arguments, and is const there, so must be const in the prototypes
and functions. The third arg is also taken directly from user arguments.
It doesn't have to be changed, but since it's cleaner to keep the type
the same as the user arg's type, and I'm already making the 'const char *'
change...
* Fix arguments to various copyin()/copyout() invocations, to avoid
gratuitous casts.
* Some KNF formatting fixes
* Change sockargs()'s second argument to be a const void *, to help
with dealing with the syscall argument type fixups/const poisoning.
to be a caddr_t, but since it's being passed constant user argument pointers
it needs to be const, and since it's being passed a variety of argument
pointer types it makes sense to change it to void *. That argument is
simply being handed to copyin().
'const char *', and 'void *', respectively. The second arg is taken directly
from user arguments, and is const there, so must be const in the prototypes
and functions. The third arg is also taken directly from user arguments.
It doesn't have to be changed, but since it's cleaner to keep the type
the same as the user arg's type, and I'm already making the 'const char *'
change...
__builtin_classify_type() into harmless values, so that lint doesn't
get confused, think they're real functions, and that they're being
invoked in the wrong ways.
* change in-kernel syscall prototypes to match user-land prototypes in
the following ways:
+ add 'const' where appropriate.
+ make the following "safe" type changes where appropriate:
caddr_t -> struct msghdr *
caddr_t -> struct sockaddr *
caddr_t -> void *
char * -> void *
int -> uid_t (safe because uid_t not used as index/count)
int -> gid_t (safe because gid_t not used as index/count)
u_int -> size_t
+ change "int" to "u_long" in flags arguments to chflags() and
fchflags(). This is safe because the arguments are used as
flag bits and there's nothing that would cause the top bit
of the int to be set yet, and because the user-land definitions
already specified u_long, so a u_long's worth of argument was
already being passed in.
wrong for a bunch of functions:
void: sys_exit, sys_sync
ssize_t: sys_read, sys_write, sys_recvmsg, sys_sendmsg,
sys_recvfrom, sys_readv, sys_writev, sys_sendto
long: sys_pathconf, sys_fpathconf
void *: sys_shmat
* Note that sys_open, sys_ioctl, and sys_fcntl are defined such that their
last argument is optional.
These changes should not have any real effect, because right now this
information is not actually used for anything.
* Don't output prototypes for INDIR syscalls (since they always show up as
sys_nosys() in the syscall table).
* Add "indir" to the comment for INDIR syscalls in the syscalls table, so
it's more obvious why they call sys_nosys().
* Deal with multi-word system call return types (i.e. foo *, or
struct foo *, or struct foo, etc.).
* Add a new class of system calls "INDIR" (for "indirect"), which
is to be used to represent indirect syscalls like syscall() and
__syscall() which are implemented in MD code and which don't want
args structures defined. (The old way of declaring this type of
syscalls still works.)
* Allow system calls to be marked as having a variable number of
arguments, by inserting "..." (no trailing comma) before the
first hf the optional arguments in the syscall definition. Because
of the way syscall arguments are handled by MI code, _ALL_ syscall
arguments must actually be included in the definition, i.e.
"optional" arguments are either "are there or aren't," i.e. these
aren't really varargs functions. Therefore, for normal syscalls,
there _must_ be arguments listed after the "...". For INDIR
syscalls, which really do have a variable number of arguments and
which aren't handled via the normal mechanism, that requirement is
not in force.
* output primitive (machine-parsable) syscall descriptions as comments
in <sys/syscall.h>. These can be used to easily build real function
prototypes, or to build stub functions for use by lint.
- Always allow RAW_PART to be opened, regardless of the partition table.
- Never do partition translation on RAW_PART; it's always offset 0.
- Always make sure the disklabel is read from RAW_PART.
- Make *strategy() return values consistent.
This fixes a condition where a bogus partition table could be written
to the disk from SYS_INST, and the user had no way to rewrite with
a correct partition table, since RAW_PART would be invalid.
bit from the extended setup inquiry and key off it to:
- Force synchronous negotiation on targs > 7.
- Read additional information returned by wide cards on inquire setup.
- Read sync period on targs > 7.
- Display sync period/offset on targs > 7.
- Set scsi_link.max_target to 15.
cvs: ----------------------------------------------------------------------
input operands. This corrects an uninitialized variable warning, and
may also explain some erratic behavior of the MRG ADB; gcc could throw
away the value of the `ticks' variable (but I haven't traced through
any ROMs, myself).
Do this in both machdep.c:setregs() and trap.c:(end of syscall)
for now. Need more analysis to see what is the best way...
Fixes PR#3034 (Thanks very much to Ian Dall for the fix!)
configuration. This way, the delay loop is calibrated before graphics and
serial hardware is touched.
This change should smooth pr 2890 by Thorsten Frueauf (also privately
reported by Laurent Badoukh). While the real problem with those is the
paranoically high delay() calls in the grf_cl initialization, it was made
even more visible by the miscalibrated (to the save side) new style delay
loop.
even if you "really didn't change anything dangerous" :-)
- While we're here, save a few bytes and clock cycles during kernel startup:
cinva ic clears the branch cache on the 68060, no need to do it explicitly.
We're about 75% there. SCSI and HP-IB are not yet supported in a new
config kernel; some autoconfiguration hackery has to be done there, yet.
These changes are enough to network boot a diskless kernel.
New config glue is enabled with the "NEWCONFIG" kernel option. If that
option is not present, an old config kernel will be built. Any kernel
configured with config(8) will automatically pick up the NEWCONFIG
option from std.hp300.
for branch prediction errors (could be used to detect strange binaries),
integer instruction, FP instruction, FP data type and FP effective address
emulations. The latter can be used to diagnose binaries which should be
recompiled with -m68060.
XXX Maybe these diagnostics should be switchable by sysctl or
XXX options DIAGNOSTIC.
indirect bus code framework) from Jason Thorpe. This allows us to dump
bus_scan() and bus_print(), and move bus_peek() and bus_mapin() to the
NuBus code (since they'll eventually go away, anyway).
Because there is no way yet to tell init(1) to execute commands before
single user-mode is entered, the inittodr() function is still partly
functional. It retrieves the value of the RTC and sets the time as if
the RTC was running at UTC. The resettodr() function is a No-Op. The
RTC can only be changed by a write to /dev/rtc.
The TIMEZONE and DST options are no longer needed and removed from the
config files.
using __weak_alias in libc, __indr_reference should probably be defined
to do nothing (but defined, so that duplicate functions/variables aren't
compiled). However, when not using __weak_alias, __indr_reference causes
them to be omitted completely, which is a lose. So, until weak aliases
are going to be used, don't define __weak_alias or __indr_reference.
making me think that the Blizzard-IV and the Blizzard-2060 scsi
options have nearly identical DMA engines (just with a different
address offset). Alas, this isn't true.
Herewith I replace the "bznsc" (all-new-Blizzard-models) driver with the
"bztzsc" (Blizzard Two Zero).
second argument. The NuBus autoconfig code had to be reorganized as a
result of this, and looks much more like a directly-attached bus now.
These changes eliminate __BROKEN_INDIRECT_CONFIG.
host adapter drivers, indicating the highest SCSI target they can
address. Use this value to dynamically allocate data structures, rather
than hard-coding 8 targets.
These changes allow targets > 7 to be addressed on wide SCSI busses.
Fixes PRs #1674 and #2892.
dmacomputeipl(), which is called by drivers which use DMA once they've
hooked up their interrupts. This new function computes the appropriate
ipl to use for the DMA controller and (re-)establishes it's interrupt.
for by crt0. Actually clear a1 and a2, because unless a shared loader (i.e.
_not_ the kernel) set them when invoking crt0, they should be zero, and
the kernel DEBUG code wouldn't necessarily have them be. Pass a pointer
to the proc's ps_strings structure in a3.
Add newline to `Interface disabled' printf and clean up formatting.
Change packet read/write code to align all reads and writes into
memory to an appropriate boundary for the I/O size. This also fixes
a bug where the trailing unaligned bytes (1-3 of them) would be
put in an additional mbuf even if there was enough space for them
in the one we were currently filling.
This has been tested on a 3c905 in an Alpha (32-bit I/O, alignment
required), a 3c905 in an i386 (32-bit I/O, alignment not required)
and a 3c509 in an i386 (16-bit I/O, alignment not required). It
should be tested on a 3c509 in an Alpha with an ISA bus, if somone
can manage to get the two items together.
There are still some bugs in the driver relating to initialisation;
on my Alpha with a 3c905 you need to do an `ifconfig ep0 down;
ifconfig ep0 up' after the initial ifconfig to make it start working.
Apparently there are also some problems with a 3c579 in an i386,
although a 3c905 in an i386 works just fine.
>- Optional systems calls are "UNIMPL" if the support is not being
> compiled into the kernel.
It had implications that didn't occurr to me at the time. *sigh*
>- Optional systems calls are "UNIMPL" if the support is not being
> compiled into the kernel.
It had implications that didn't occur to me at the time. *sigh*
align 32bit integers. Use explicit sized typing at some other places.
XXX This still won't fix lfs for 64bit machines, as we have some
assumptions about sizeof(pointer)=sizeof(u_int32_t) in here, and (if I
looked right) a misaligned u_int64_t. The right fix (to cite cgd) will
be to seperate on-disk-representation from in-core, but I don't have
the time (at the moment) to do this.
on indirect-config busses a (permanent) softc that they could share
between 'match' and 'attach' routines:
Define __BROKEN_INDIRECT_CONFIG so that old autoconfiguration
interfaces are used, until drivers are converted to use the new
interfaces (actually, converted back to use the _older_ interfaces)
which prohibit indirect configuration devices from receiving a softc
in their match routine that they can share with their attach routine.
defined:
define match functions to take a struct cfdata * as their second
argument, config_search() to take a struct cfdata * as its second
argument, and config_{root,}search() to return struct cfdata *.
remove 'cd_indirect' cfdriver element.
remove config_scan().
remove config_make_softc() as a seperate function, reintegrating
its functionality into config_attach().
Ports will define __BROKEN_INDIRECT_CONFIG until their drivers prototypes
are updated to work with the new definitions, and until it is sure that
their indirect-config drivers do not assume that they have a softc
in their match routine.
on indirect-config busses a (permanent) softc that they could share
between 'match' and 'attach' routines:
Check for a new definition, __BROKEN_INDIRECT_CONFIG, and if it is _not_
defined:
define match functions to take a struct cfdata * as their second
argument, config_search() to take a struct cfdata * as its second
argument, and config_{root,}search() to return struct cfdata *.
remove 'cd_indirect' cfdriver element.
remove config_scan().
Ports will define __BROKEN_INDIRECT_CONFIG until their drivers prototypes
are updated to work with the new definitions, and until it is sure that
their indirect-config drivers do not assume that they have a softc
in their match routine.
the client and server/shared data initialization into separate functions,
and calling the server/shared initialization directly from main().
Problem noted in PR #1308 (Kenneth Stailey) and PR #1780 (Chris Demetriou).
Fix suggested in PR #1780 by Chris Demetriou, and munged a bit by me,
and OK'd by Frank van der Linden <fvdl@netbsd.org>.
Improve the queuing algorithms used by NFS' asynchronous i/o. The
existing mechanism uses a global queue for some buffers and the
vp->b_dirtyblkhd queue for others. This turns sequential writes into
randomly ordered writes to the server, affecting both read and write
performance. The existing mechanism also copes badly with hung
servers, tending to block accesses to other servers when all the iods
are waiting for a hung server.
The new mechanism uses a queue for each mount point. All asynchronous
i/o goes through this queue which preserves the ordering of requests.
A simple mechanism ensures that the iods are shared out fairly between
active mount points.
Reviewed/integrated/approved by Frank van der Linden <fvdl@netbsd.org>
Lets users over-ride with makeoptions COPTS="..." in kernel config files.
Leave `mandatory' flags (like -msoft-float which on m68k enforces no
FP in kernel) in CFLAGS.
drisavar.h pretends to provide a few bus.h macros, hardwired to
that chip.
This should eventually be replaced by attachment code for the normal
com.c driver, once that one is split up into chip core driver and
attachment code, and once we have busxxx macros in NetBSD/Amiga.
along with toolchain enhancements.
aout2bb transforms a.out files with reloc information into bootblock format
files with a compressed relocation info.
bbstart.s relocates the bootblock using this compressed info, before jumping
to C code.
txlt changes some more of the absolute references to pc-relative ones (we know
we have a single code + data address space).
If you ever try to change this: don't even dare to change the compiler options;
they were found in weeks of trial and error as the ones producing the smallest
(not necessarily fastests) code.
installboot is just a script around dd, for now.
Some ideas by Michael Hitch, Leo Weppelman and Jason Thorpe; bugs added
by myself.
along with toolchain enhancements.
aout2bb transforms a.out files with reloc information into bootblock format
files with a compressed relocation info.
bbstart.s relocates the bootblock using this compressed info, before jumping
to C code.
txlt changes some more of the absolute references to pc-relative ones (we know
we have a single code + data address space).
If you ever try to change this: don't even dare to change the compiler options;
they were found in weeks of trial and error as the ones producing the smallest
(not necessarily fastests) code.
installboot is just a script around dd, for now.
Some ideas by Michael Hitch, Leo Weppelman and Jason Thorpe; bugs added
by myself.
Also, the munging is extended by inserting an instruction that branches
past the header, so the same boot image can be used on all supported
sun platforms.
primarily to nicely print version information about your PCI chipset
(try with "options PCIVERBOSE"). Eventually, it may be used to
enable/disable features/bugs of a given PCI chipset. In addition, this
driver uses the PCI-ISA bridge callback mechanism to logically attach the
ISA bus to the PCI-ISA bridge.
driver is a place-holder, which will nicely print version information
about your PCI chipset (try with "options PCIVERBOSE"). Eventually,
this can be used to enable/disable features/bugs of individual PCI
chipsets.
been attached to the system. If, by the time mainbus wants to attach
an ISA an ISA has not yet been attached to the system, attempt to attach
an ISA to mainbus.
>Pay attention to DMA errors as reported by DMAINTR() returning -1. If this
>happens reset everything.
>
>Re-schedule a timeout when first attempting to abort an operation.
>Cancel any queued timer events before re-scheduling a timeout, so esp_abort()
>can be called from other places besides the timeout handler.
as a "timeout", yet there's no specific delay in each iteration. Add
a small delay (10 usec... pretty arbitrary) in each iteration. This
fixes the "fdcresult: timeout" problems I've been having on my 200MHz P6.
after returning from hardclock(), rather than before. For some reason,
this fixes the 0xffffffff i used to see in the tv_secs of the used cpu
time of some processes.
XXX I don't fully understand the issue.
the chipset space init functions multiple times, since that would clobber
extent allocations made between the two calls. Also, deal with the
fact that the APECS and LCA no longer shared common chipset functions.
the same things, but the extent maps have to be managed differently,
since the two chipsets provide different memory and I/O region mapping
possibilities.
and PC-ish keyboard controller. (Actually, on alphas, the built-in PPI
(in the SIO) appears to be a lobotomized version of the original, but
i'd not call that a bad thing.) This driver should eventually handle all
speaker tone requests and keyboard commands, but for now it just maps
the relevant ports and passes them on to the keyboard and mouse drivers,
which are now its children (rather than children of ISA).
* Support for the new softint mechanism. Softints are now requested by
triggering an unused ICU hardware interrupt. The idea for this was
contributed by Phil Budne.
* Real probe code added.
* Duart info is now allocated only for devices that are present.
* Added IO-Recovery delays for 30mhz systems.
* Removed a few potential NULL-pointer references.
>One control block per target is insufficient if you have a full complement
>of targets attached and access those simultaneously (like in a ccd(4) array).
>We (now) allocate (somewhat arbitrarily) three per target.
>Noticed by Marshall Midden.
(1) after removing a shutdown hook (in shutdownhook_disestablish()),
free it. We created it, we have to free it. Without this,
shutdownhook_disestablish() leaks memory.
(2) in doshutdownhooks(), before running each hook, remove it from the
shutdown hook list. This makes sure that every hook is tried once
(because doshutdownhooks() is called from before rebooting, and
a fault in a shutdown hook will cause doshutdownhooks() to be called
again), but prevents the hooks from potentially being run infinitely
(as used to be possible, in the above-mentioned situation).
(e.g. from 0x3bc to 0x3bf, for example). Others may require more,
but until there's some special handling for them, 4 should be returned
so that on systems with I/O port accounting, the unit at 0x3bf can be
properly mapped, etc. (OK'd by mycroft.)
that their child busses can be attached after the PCI bus
autoconfiguration for their parent bus is done.
This works because:
(1) there can be at most one ISA/EISA bridge per PCI bus, and
(2) any ISA/EISA bridges must be attached to primary PCI
busses (i.e. bus zero).
That boils down to: there can only be one of these outstanding
at a time, it is cleared when configuring PCI bus 0 before any
subdevices have been found, and it is run after all subdevices
of PCI bus 0 have been found.
This (or something like it) is needed because there are some (legacy)
PCI devices which can show up as ISA/EISA devices as well (the prime
example of which are VGA controllers). If you attach ISA from a
PCI-ISA/EISA bridge, and the bridge is seen before the video board is,
the board can show up as an ISA device, and that can (bogusly)
complicate the PCI device's attach code, or make the PCI device not be
properly attached at all.
This could be done with machine-dependent code, but as more ports
add support for PCI (and PCI-ISA/EISA bridges) more will need it.
The i386 port could (perhaps should) be converted to use it as well.
devices actually do make sense on indirect-config busses, because you
might be able to have more than one of the busses! In addition, they're
useful because they don't require unit numbers to be wired down, so you
could e.g. have vga* at indirect? and vga* at direct?, and have the first
one found be unit number zero. Finally, devices which can divine their
own ports numbers, etc., actually should be cloning, even if you know you'll
only have one bus that they can live on.
don't machine check when a PCI Master Abort is signalled. This can
happen, for instance, when configuration space for a device that isn't
present is examined. When this is detected, act like we normally would
when machine checks are posted while examining nonexistant devices.
enabled (from the attach routine), and add comments as to why.
Some PALcode apparently 'saves' a clock interrupt for the kernel,
and if the clock interrupt handler is enabled at attach time, it
will be run when that interrupt hits, i.e. right after the spl0()
at the end of autoconfiguration. That would cause hardclock to be
run, but proc0's p_stats isn't set up by then, which would cause
hardclock to crash.
rather than and-ing 16G-1. That just strips the k0seg bits, rather
than making the false assumption that the physical address is going
to be in the lower 16G. That doesn't apply for CIA device-space
addresses, for instance.
even if PCI and the IDs are right), just for sanity, before declaring
success. Split the single 0x3b0 -> 0x3df allocation into three seperate
ones: 0x3b0 -> 0x3bc (leaving the 4 ports available for lpt),
0x3c0 -> 0x3cf, and 0x3d0 -> 0x3df. The former chunk has to be split
off if the lpt can exist there, and it's sort-of pretty to have each
group (based on second hex digit) have its own handle.
These alternative macros have a workaround for the STM^ bug in revision < 3
StrongARM CPU's that causes incorrect register saving if a cache line fill
is in progress during the STM.
a podulebus.
Make sure the podulebus driver conforms to the Acorn expansion card
specification:
- Probe the podule bus using sync access cycles rather than slow access
cycles.
- Read the podulebus header/ROM using sync access cycles rather than slow
access cycles
of targets attached and access those simultaneously (like in a ccd(4) array).
We (now) allocate (somewhat arbitrarily) three per target.
Noticed by Marshall Midden.
1. fix possible hang in en_txlaunch(). when attempting to extend
the length of an mbuf to avoid a flush we should extend it
by cnt [which is ((need - len) % 4)] rather than 4 - cnt.
also, add an EN_DEBUG printf() when we pad/FLUSH a buffer
to help with debugging/understanding what the driver is up to.
2. use interface packet counters
3. when turning off a recv VCI we recompute the new mode. make sure
we don't include the "in service" bit in the new mode, otherwise
a VCI may appear "hung" if you turn it off while a service
interrupt is pending.
4. when shutting down a VCI that is still receiving data, don't bother
going into "drain mode" if only the hardware in service bit is
set (otherwise the VCI may get "hung" in drain mode).
as a result of this we may get "unexpected rx interrupt" messages
which are not really an error, so put this printf in EN_DEBUG.
5. be sure to zero txspeed[lcv] when enabling a VCI (start at full
speed). (hooks for setting txspeed[] are currently not in
the driver, but we are playing with it locally).
credits:
#1: Detected by: Zdenek Salvet <salvet@horn.ics.muni.cz>, fix by me.
#2: Contributed by: Zdenek Salvet <salvet@horn.ics.muni.cz>
#3,#4,#5: Detected by: Milind M. Buddhikot <milind@dworkin.wustl.edu>,
fixed by me.
If not compiled with -D_KERNEL, include different includes and
do so macro magic so that this will fit sanely into test harnesses.
When used in user-land, this should be compiled with -D_EXTENT_TESTING.
Bug fixes:
(extent_insert_and_optimize) You can't do things like:
LIST_REMOVE(elem->...le_next, ...);
free(elem->...le_next, ...);
They just don't work (and will corrupt your list and/or malloc free list).
(extent_alloc_region_descriptor) Unless you wait, malloc can fail.
Don't accidentally deref a potentially-NULL pointer.
2841, plus some fixes to make the patches work on the Alpha. Seems to
improve the NCR driver a lot. We probably should try to incorporate
any updates that have happened since, too.
leo