locore calls to go via a locore-entry jumptable.
Use the mips/mips/mips_machdep.c code to initialize locore state,
exception vectors, and the locore entry vector, instead of doing
it all in mach_init().
Update dumpsys() to use new-style signatures for dump routines,
even though the pmax SCSI drivers don't actually support dumping kernel
memory.
locore calls to go via a locore-entry jumptable.
Cast the (int) arguments to MachTLBUpdateEntry() to avoid
warnings. Variables TLB entries are still type-punned as either structs
or ints, without any regard, when the pmax-specific VM code passes
them as arguments to functions.
locore calls to go via a locore-entry jumptable.
Use mips_btop(), mips_round_page, mips_trunc_seg() instead
of pmax_btop(), pmax_round_page, pmax_trunc_seg().
Add Per's software-readonly-bit mechanism, since the r2000 and r4000
hardware TLB entries are very different, and the r4k has no space for
software bits in TLB entries. That is, this pmap code still won't work
on r4000 machines. Some other solution, like another jump table for
clients of the pmap code, is necessary.
locore calls to go via a locore-entry jumptable.
Declare r2000- and r4000-specific exception-handler functions, to which
trap() and interrupt() dispatch exceptions. Initialize r2000- and r4000-
specific exception-handler vectors, when CPU_R4000 and CPU_R2000 are
defined.
Update the stack-traceback code (partially) to understand and print
the new low-level exception-handler code, via which machine exception-vectors
send exceptions to call trap() or interrupt(). This needs more work.
It just causes a kernel panic, and will until the rest of the pmax
autoconfiguration is redone to use a bus-specific probe and attach
argument, or until KN01 devices are no longer attached to the mainbus.
an vector (struct) of function pointers. Add prototype declarations for
each vector entry.
Add declarations for the r2000 (MIPS-I) and r4000 (MIPS-III) locore
versions of the relevant functions.
update the 4.3bsd/Ultrix-compatible qvss-style mmap'ed framebuffer/input-event
code to use mips_round_page() et.al., not pmax_round_page().
Add explicit "int" return types to functions. Add prototyped forward
declarations.
Fix the the definitions of dtop cdevsw entry point functions to match
their declarations. Add explicit "int" declarations to functions.
Add prototyped declarations for all local functions.
Return zero from dtopstop().
Include the external declarations from dtopvar.h.
to all mips ports.
So far, this consists of code to initialize a vector, or jump-table, of
pointers to locore functions that require different definitions on different
Mips CPUs (eg., r2000/3000 and r4000); a generic wrapper for setting up
CPU-specific exception vectors; and CPU and FPU identification code.
locore functions. The new names are used by C code to construct a jump-table,.
making it less infeasible to have a single kernel image work on both
r3000 and r4000 systems.
independent z8530 driver. When that driver is updated, the local copy
of those files may be removed. From Bill Studenmund.
Compiler warnings fixed by me.
boot.c: update from sun3 port, with changes for my new boot code put in.
conf.c: config from openbsd, by Theo de Raadt
dev_net.c: update from openbsd (standard dev_net with changes from theo)
dev_net.h: proto from openbsd
exec_mvme.c: from sun3 port, adapted by me. I also fixed a bug that
prevented ZMAGIC kernels from running.
libsa.h: new file by me
parse_args.c: by Theo de Raadt from OpenBSD with slight revisions from by me
to make it fit the new format.
- New delay() function, modeled after sun3/mvme68k delay,
calibrated with a model's known cpu speed.
- Use the defined constants for machineid and ectype
when determining what kind of machine we have, rather
than their expanded equivalents, for clarity.
- Move the vector table to the data segment so that it
can be patched at run-time.
two didn't seem to make much sense anyway..) to allow `esp' devices to be
attached to one of `sbus', `dma' and `espdma'.
Remove the wildcarded `espdma?' and `ledma?' attachments of `esp' and `le'
respectively, in favour of `dma?' and `lebuffer?' (but the latter is not
yet implemented), which seems to better match reality: additional SBus
SCSI/Lance boards call themselves `dma' and `lebuffer'.
on a per target basis (until the driver can sort things out on its own).
Test against "sbus" in stead of "espdma" to find out where in the
configuration tree we are: an esp can be the child of a "dma" on SBus
add-on boards.
sending of a keyboard reset code. This fixes a problem on some machine
where the reset code ends is a long loop continually trying to reset the
keyboard.
machine-independent code for more sane access to bus resources.
Defines:
bus_{io,mem}_{map,unmap}
bus_{io,mem}_{read,write}_{1,2,4,8}
bus_io_{read,write}_multi_{1,2,4,8}
functions, and several types to go with them.
Note: The bus_io_*_8 functions are not supported.
no external L2 cache.
XXX this is ugly and will go away when cpu_softc gets done for the 1.3 release
Also, copyright police (s/Harvard University/Harvard College/).
do some re-arrangement (cleanup), and document which devices exist on
which machines, so that people can be better informed when they trim
down their kernels. There are a LOT of comments in this file now!
kmem_map (the VA space is allocated from kernel_map).
In both bus_mem_map() and bus_mem_unmap(), adjust some arithmetic to
avoid wasting a page of VA space per call to bus_mem_map().
This should reduce the possibility of a panic when unable to allocate
a new pv_entry under low memory conditions.
Changed a couple of kmem_free() calls to kmem_free_wakeup() as in the
hp300 pmap.c.
-Wall cleanup: unused variable when M68060 not defined.
documentation I have calls them, and update for new definitions in
sys/exec_ecoff.h. Also, fix ECOFF_SEGMENT_ALIGNMENT macro so that it
actually returns the right alignment.
* for the M68060 part: store buffer and branch target cache aren't
enabled yet (this needs cleanup of lots of locore.s code which is a
maze of little passages, all a little different) (and it wasn't yet
tested in an accellerated Amiga, only in the DraCo).
I've included a workaround for 2 of the CPU bugs in chips with Masks
1F43G and earlier, but didn't bother to deal with the can of worms
in the [0-3]D11W chips. Be sure to get "68060 rev. 1" or more
reported at kernel startup time, or at least mention it (or the mask
revision, if available) when reporting problems.
* for the DraCo: only machines with a CIA timer.
I assigned machine id 32000+nn (0x7Dnn), where n is the machine
readable Quicklogic custom chip revision (also printed at boot
time). "Guaranteed to work" up to rev. 3, newer DraCo's aren't
guaranteed to have any CIA (we don't have a driver for the new timer
yet).
Supported are:
- MF-II keyboards on the native interface and A3000 keyboards via
the CIA.
- builtin SCSI interface (yet another instance of siop)
- CIA timer.
- Zorro II devices which don't do DMA (don't get mapped to Zorro II
address space in the DraCo)
- "local bus" devices which are autoconfigured by the boot rom
(should be all); only an Altais driver is there (looks like a Retina Z3)
Not yet supported are:
- native timer of newer machines.
- Real Time Clock.
- serial, parallel + floppy on the SuperIO chip (that is also: no mouse)
XXX You need an enhanced boot loader, which will committed in a few days.
XXX std.draco should and will go away.
Changing types from unsigned {long,short} to u_int{32,16}_t causes problems
in a lot of userlevel code because those types aren't defined when machine/
endian.h is included. And IMO including sys/types.h in machine/endian.h
is clearly wrong.
XPG4.2 defines new types in_addr_t and in_port_t for the {h,n}to{n,h}{l,s}
functions. But we don't have time to figure out the implementation issues
given the current release schedule.
visible screen memory. This results in a considerable performance win
at all depths.
- Correct clear_screen() behavior when clearing to the top and bottom of
screen; start at current cursor position, not current cursor row.
- Add local prototypes, and other KNF.
device and a printable "external name" (name + unit number), thus eliminating
if_name and if_unit. Updated interface to (*if_watchdog)() and (*if_reset)()
to take a struct ifnet *, rather than a unit number.
device and a printable "external name" (name + unit number), thus eliminating
if_name and if_unit. Updated interface to (*if_watchdog)() and (*if_start)()
to take a struct ifnet *, rather than a unit number.
naming conflicts between bus attachments on ports that can have
multiple instances of the LANCE.
Changed struct ifnet to have a pointer to the softc of the underlying
device and a printable "external name" (name + unit number), thus eliminating
if_name and if_unit. Updated interface to (*if_watchdog)() and (*if_start)()
to take a struct ifnet *, rather than a unit number.
device and a printable "external name" (name + unit number), thus eliminating
if_name and if_unit. Updated interface to (*if_watchdog)() and (*if_start)()
to take a struct ifnet *, rather than a unit number.
naming conflicts between bus attachments on ports that can have
multiple instances of the LANCE.
Changed struct ifnet to have a pointer to the softc of the underlying
device and a printable "external name" (name + unit number), thus eliminating
if_name and if_unit. Updated interface to (*if_watchdog)() and (*if_start)()
to take a struct ifnet *, rather than a unit number.
naming conflicts between bus attachments on ports that can have
multiple instances of the LANCE.
Changed struct ifnet to have a pointer to the softc of the underlying
device and a printable "external name" (name + unit number), thus eliminating
if_name and if_unit. Updated interface to (*if_watchdog)() and (*if_start)()
to take a struct ifnet *, rather than a unit number.
device and a printable "external name" (name + unit number), thus eliminating
if_name and if_unit. Updated interface to (*if_watchdog)() and (*if_start)()
to take a struct ifnet *, rather than a unit number.
naming conflicts between bus attachments on ports that can have
multiple instances of the LANCE.
Changed struct ifnet to have a pointer to the softc of the underlying
device and a printable "external name" (name + unit number), thus eliminating
if_name and if_unit. Updated interface to (*if_watchdog)() and (*if_start)()
to take a struct ifnet *, rather than a unit number.
device and a printable "external name" (name + unit number), thus eliminating
if_name and if_unit. Updated interface to (*if_watchdog)() and (*if_start)()
to take a struct ifnet *, rather than a unit number.
naming conflicts between bus attachments on ports that can have
multiple instances of the LANCE.
Changed struct ifnet to have a pointer to the softc of the underlying
device and a printable "external name" (name + unit number), thus eliminating
if_name and if_unit. Updated interface to (*if_watchdog)() and (*if_start)()
to take a struct ifnet *, rather than a unit number.
device and a printable "external name" (name + unit number), thus eliminating
if_name and if_unit. Updated interface to (*if_watchdog)() and (*if_start)()
to take a struct ifnet *, rather than a unit number.
naming conflicts between bus attachments on ports that can have
multiple instances of the LANCE.
Changed struct ifnet to have a pointer to the softc of the underlying
device and a printable "external name" (name + unit number), thus eliminating
if_name and if_unit. Updated interface to (*if_watchdog)() and (*if_start)()
to take a struct ifnet *, rather than a unit number.
Added a new routine kbdreset() to conduct a full keyboard reset.
Added definitions of all keyboard command codes and responces.
Replaced kbd_set_leds() with new function that used kbdcmd().
The keyboard attach function now detects whether a keyboard is
attached or not.
irq_claim().
Clear the active flag in the irq handler when it is removed from the irq
chain in irq_release().
Checks for podule IRQ's in irq_claim() are now guarded with
#if NPODULEBUS > 0
fiq_release() now retrieves the FIQ more registers and places them
back in the fiq handler structure.
Also change the device probing scheme to use something a bit more rational.
A current side-effect is that nubus cards are double-mapped. I expect
to fix that shortly.
Also change splclock() to block everything but serial hardware interrupts.
This removes the need to have enough chip memory to load the kernel
when booting. Kernel memory is dual-mapped (VA == 0 and VA == PA)
for enabling the MMU, and the VA == PA mapping removed after start_c()
has completed. Chip memory is still required for the /dev/reload
process.
Startup modified to work with kernel loaded directly into fast memory.
This removes the need to have enough chip memory to load the kernel
when booting. Kernel memory is dual-mapped (VA == 0 and VA == PA)
for enabling the MMU, and the VA == PA mapping removed after start_c()
has completed. Chip memory is still required for the /dev/reload
process.
loading into fastmem. This removes the requirement that the kernel fit
into chipmem.
Add a -Z flag to force loading into chip memory.
Add a -s option for compatibility with my bootblock loader.
Move the test exit to just before starting the kernel; this lets the
test mode actually load the kernel and do the version checks.
handling code so that if a given interrupt is disabled (and therefore
can't have caused the actual I/O interrupt), its handler won't be
called even if its bit is set in the interrupt register.
against the MMU switch code being in different locations between
the running kernel and the new kernel. Returns ENODEV if the reload
fails.
Change the error return from EFAULT to ENOEXEC when kernel_reload
gets a short write on the exec header.
reasons: it won't attach as console, and there's currently no way to do
keyboard input on TC machines), and has no real RAMDAC (colormap, cursor,
etc.) support. Digital UNIX does not support CFB frame buffers in the
Alpha, but they appear to work OK (with an appropriate monitor) in my
3000/300.
reasons: it won't attach as console, and there's currently no way to do
keyboard input on TC machines), and has no real RAMDAC (colormap,
cursor, etc.) support.
Clean up the debugging code a bit and the warnings from -Wall.
Don't define ESDEBUG - it can now be set from the config file.
If the Ethernet chip gets reset during the copy of the transmit buffer,
requeue the current packet and reinitialize the controller. This recovers
from an apparent hardware bug when running on my A2000/Zeus system.
a condition that occurs with some slow SCSI devices when they disconnect
(e.g. the AppleCD 600), generating spurious selection interrupts.
- Reorganize the debugging code slightly.
Don't mess with the flags when entering a software interrupt.
Use IF, not VIF.
NOTE: These changes break dosemu, but are required for proper emulation.
Don't seek to track 0 before checking for diskchange: if drive steps,
the diskchange flag is cleared. Just select the drive, then test for
a diskchange.
Clean up for -Wall with FDDEBUG defined, and fix a message typo.
Only call Debugger() if DDB is defined.
If the wait for reselect is interrupted and sc_nexus was NULL, only
print the debug message if DEBUG is defined.
matchpodule() is now used to test podule_attach_args for correct id's.
Moved the probing code that used the softc into the attach function.
Fixed compiler warnings generated with -Wall.
the podulebus attach function.
For each podule found, config_found_sm() is now called. Device drivers
now only need to look at the podule_attach_args information during
the probe.
A new function matchpodule() has been added for use in the
podule device driver probe functions.
