pci: Pass cookie through pci_find_device, pci_enumerate_bus, take 2.

New functions pci_find_device1 and pci_enumerate_bus1 have the cookie
argument. Existing symbols pci_find_device and pci_enumerate_bus are
now wrappers for the cookieless version.

This will allow pci_find_device callers to pass a cookie through to
the match function so they can keep state or pass in extra parameters
like b/d/f numbers, which will allow us to nix some horrible kludges
in the Linux PCI API emulation for drm (and, perhaps, Intel wifi).

This change drops the symbol pci_probe_device, in favour of a new
pci_probe_device1 with the cookie argument. But I don't think that
requires a revbump because it's only called by MD pci_enumerate_bus1
implementations, which don't live in modules anyway.

Take 2: Make sure to handle NULL match function.

Revert "pci: Pass cookie through pci_find_device, pci_enumerate_bus."

Evidently something is wrong with this, to be diagnosed and redone
once the builds and tests are in better shape.

pci: Pass cookie through pci_find_device, pci_enumerate_bus.

New functions pci_find_device1 and pci_enumerate_bus1 have the cookie
argument. Existing symbols pci_find_device and pci_enumerate_bus are
now wrappers for the cookieless version.

This drops the symbol pci_probe_device, in favour of a new
pci_probe_device1 with the cookie argument. But I don't think that
requires a revbump because it's only called by MD pci_enumerate_bus1
implementations, which don't live in modules anyway.

Remove unnecessary <sys/malloc.h>.

Don't try to make a tag based on node 0

Rename min/max -> uimin/uimax for better honesty.

These functions are defined on unsigned int. The generic name
min/max should not silently truncate to 32 bits on 64-bit systems.
This is purely a name change -- no functional change intended.

HOWEVER! Some subsystems have

#define min(a, b) ((a) < (b) ? (a) : (b))
#define max(a, b) ((a) > (b) ? (a) : (b))

even though our standard name for that is MIN/MAX. Although these
may invite multiple evaluation bugs, these do _not_ cause integer
truncation.

To avoid `fixing' these cases, I first changed the name in libkern,
and then compile-tested every file where min/max occurred in order to
confirm that it failed -- and thus confirm that nothing shadowed
min/max -- before changing it.

I have left a handful of bootloaders that are too annoying to
compile-test, and some dead code:

cobalt ews4800mips hp300 hppa ia64 luna68k vax
acorn32/if_ie.c (not included in any kernels)
macppc/if_gm.c (superseded by gem(4))

It should be easy to fix the fallout once identified -- this way of
doing things fails safe, and the goal here, after all, is to _avoid_
silent integer truncations, not introduce them.

Maybe one day we can reintroduce min/max as type-generic things that
never silently truncate. But we should avoid doing that for a while,
so that existing code has a chance to be detected by the compiler for
conversion to uimin/uimax without changing the semantics until we can
properly audit it all. (Who knows, maybe in some cases integer
truncation is actually intended!)

branches: 1.77.28; 1.77.30;
make pci_intr_string and eisa_intr_string take a buffer and a length
instead of relying in local static storage.

kill sprintf

branches: 1.75.2;
split device_t/softc for all remaining drivers.
replace "struct device *" with "device_t".
use device_xname(), device_unit(), etc.

branches: 1.74.2; 1.74.12;
convert some open-coded SPDB_PROBE printf calls to use DPRINTF() macro.

#include <sys/bus.h> instead of <machine/bus.h>.

when looking for a device's node we want to start with the child nodes of the
root bus, not its peers

Neither pci_dma64_available(), pci_probe_device(), pci_mapreg_map(9),
pci_find_rom(), pci_intr_map(9), pci_enumerate_bus(), nor the match
predicate passed to pciide_compat_intr_establish() should ever modify
their pci_attach_args argument, so make their pci_attach_args arguments
const and deal with the fallout throughout the kernel.

For the most part, these changes add a 'const' where there was no
'const' before, however, some drivers and MD code used to modify
pci_attach_args. Now those drivers either copy their pci_attach_args
and modify the copy, or refrain from modifying pci_attach_args:

Xen: according to Manuel Bouyer, writing to pci_attach_args in
pci_intr_map() was a leftover from Xen 2. Probably a bug. I
stopped writing it. I have not tested this change.

siside(4): sis_hostbr_match() needlessly wrote to pci_attach_args.
Probably a bug. I use a temporary variable. I have not tested this
change.

slide(4): sl82c105_chip_map() overwrote the caller's pci_attach_args.
Probably a bug. Use a local pci_attach_args. I have not tested
this change.

viaide(4): via_sata_chip_map() and via_sata_chip_map_new() overwrote the
caller's pci_attach_args. Probably a bug. Make a local copy of the
caller's pci_attach_args and modify the copy. I have not tested
this change.

While I'm here, make pci_mapreg_submap() static.

With these changes in place, I have tested the compilation of these
kernels:

alpha GENERIC
amd64 GENERIC XEN3_DOM0
arc GENERIC
atari HADES MILAN-PCIIDE
bebox GENERIC
cats GENERIC
cobalt GENERIC
evbarm-eb NSLU2
evbarm-el ADI_BRH ARMADILLO9 CP3100 GEMINI GEMINI_MASTER GEMINI_SLAVE GUMSTIX
HDL_G IMX31LITE INTEGRATOR IQ31244 IQ80310 IQ80321 IXDP425 IXM1200
KUROBOX_PRO LUBBOCK MARVELL_NAS NAPPI SHEEVAPLUG SMDK2800 TEAMASA_NPWR
TEAMASA_NPWR_FC TS7200 TWINTAIL ZAO425
evbmips-el AP30 DBAU1500 DBAU1550 MALTA MERAKI MTX-1 OMSAL400 RB153 WGT624V3
evbmips64-el XLSATX
evbppc EV64260 MPC8536DS MPC8548CDS OPENBLOCKS200 OPENBLOCKS266
OPENBLOCKS266_OPT P2020RDB PMPPC RB800 WALNUT
hp700 GENERIC
i386 ALL XEN3_DOM0 XEN3_DOMU
ibmnws GENERIC
macppc GENERIC
mvmeppc GENERIC
netwinder GENERIC
ofppc GENERIC
prep GENERIC
sandpoint GENERIC
sgimips GENERIC32_IP2x
sparc GENERIC_SUN4U KRUPS
sparc64 GENERIC

As of Sun Apr 3 15:26:26 CDT 2011, I could not compile these kernels
with or without my patches in place:

### evbmips-el GDIUM

nbmake: nbmake: don't know how to make /home/dyoung/pristine-nbsd/src/sys/arch/mips/mips/softintr.c. Stop

### evbarm-el MPCSA_GENERIC
src/sys/arch/evbarm/conf/MPCSA_GENERIC:318: ds1672rtc*: unknown device `ds1672rtc'

### ia64 GENERIC

/tmp/genassym.28085/assym.c: In function 'f111':
/tmp/genassym.28085/assym.c:67: error: invalid application of 'sizeof' to incomplete type 'struct pcb'
/tmp/genassym.28085/assym.c:76: error: dereferencing pointer to incomplete type

### sgimips GENERIC32_IP3x

crmfb.o: In function `crmfb_attach':
crmfb.c:(.text+0x2304): undefined reference to `ddc_read_edid'
crmfb.c:(.text+0x2304): relocation truncated to fit: R_MIPS_26 against `ddc_read_edid'
crmfb.c:(.text+0x234c): undefined reference to `edid_parse'
crmfb.c:(.text+0x234c): relocation truncated to fit: R_MIPS_26 against `edid_parse'
crmfb.c:(.text+0x2354): undefined reference to `edid_print'
crmfb.c:(.text+0x2354): relocation truncated to fit: R_MIPS_26 against `edid_print'

remove unnecessary iommuvar.h and psycho{reg,var}.h includes.

don't try to turn on parity or set the latency timer, or dump pci config
space for the pci bridge in sparc64_pci_enumerate_bus(). it doesn't work
on PCIe adapters. #if 0'ed for now, should move this into psycho/schizo.

branches: 1.68.2;
allow sparc64 spc_find_ino() to be NULL; schizo doesn't need it.

branches: 1.67.4;
Cope with up to four "interrupts" properties for a node.
Just use the first one for now. This should be improved.
OK: mrg@.

move the guts of pci_intr_map() into pci_machdep.c, and move the schizo
ign addition into a hook. also, don't double shift the ign in set_intr.

tested on U60 and SB2500.

convert pci_intr_map() into a chipset tag function pointer, and implement
the schizo version slightly differently.

pull out the schizo's IGN from the upaid, not the apparently broken device
ID register. from openbsd.

with this i appear to have valid working interrupts on the SB2500.

tested on U60 and SB2500.


XXX: we can probably kill (*spc_find_ino)() now that pci_intr_map() itself
XXX: is no longer a first class function.

give more debug output about mapping interrupts.

clean up some debug messages a little bit.

pci_make_tag:
- add assertion for checking spc_busnode initialization.

ofpci_make_tag():
don't write PCI_COMMAND_STATUS_REG with
PCI_COMMAND_(MEM|MASTER|IO)_ENABLE, or it in.
sparc_pci_childspace():
move psycho.c:get_childspace() into pci_machdep.c but rename it
psycho_alloc_bus_tag():
use M_ZERO

a few clean up/rearrange changes, somewhat inspired from openbsd's
schizo.c:

- rename pp_busnode{} -> spc_busnode{}, and move it to
sparc_pci_chipset{}

- move public pci functions pci_conf_read(), pci_conf_write() and
pci_intr_establish() into psycho.c, renaming them in the process,
and add pointers through sparc_pci_chipset{} to access them.
as part of this, add PDB_INTMAP and PDB_CONF debug options to
psycho.c

- move and rename sparc64 private pci_find_ino() into
sparc_pci_chipset{}, like the above

- update psycho_alloc_chipset() to set the new 4 function members


no functional change intended. XXX: not tested on cardbus.

branches: 1.59.4; 1.59.6;
pci_intr_setattr(), allows PCI interrupts to be marked MPSAFE on x86, and
other platforms if the code is added.

pci_intr_map(...)
pci_intr_setattr(pc, ih, PCI_INTR_MPSAFE, 1);
pci_intr_establish(...)

remove clause #3 from my license where there are no other
copyright holders involved.

branches: 1.57.2; 1.57.4; 1.57.6;
Change previous to use aprint_error_dev.

use device_xname()

branches: 1.55.6;
Do not panic when trying to disestablish a pci interrupt.

branches: 1.54.14; 1.54.20;
pci_find_ino(): Use interrupt pin from pci_attach_args when mapping interrupts.
Use *ihp as a pure output. Fix case where interrupt pin #D was incorrectly
mapped. Some cosmetic changes while here. Ok'd by martin.

branches: 1.53.4; 1.53.8; 1.53.10;
remove "struct cacheline" and the "cacheline" variable. (it was only set
in cpu_attach(), overwritten by multiple cpus, and only one member of the
12 or more were ever used again.) add an ecache_min_line_size to replace
this functionality.

clean up some non-SMP friendly code from cpu_attach():
- only reset the fpu state on the boot cpu
- don't bother printing "FPU version 0" always
- don't call getver() and ignore the result
- print the UPA id on the first cpu0: line

only print mp_tramp address #ifdef DEBUG.

reset the fpu state in cpu_hatch(), like cpu_attach() does for the boot cpu.


cpu_attach() is now free of SMP-unfriendly code.

Missing initializers

branches: 1.51.14; 1.51.16;
ANSIfication: u_intN_t -> uintN_t, use ANSI function declarations/definitions
instead of K&R ones.

branches: 1.50.2; 1.50.4; 1.50.6;
merge ktrace-lwp.

branches: 1.49.12;
make PCI devices attachable/detachable (as far as the particular drivers
allow), and allow to rescan a bus selectively (ie only the device/
function I'm looking at)

give the special PCI bus enumerator a sparc64_ prefix and make the
PCI_MACHDEP_ENUMERATE_BUS macro point to it

pci_find_ino: support pci-pci bridges.

Revert a part of rev. 1.44.

OF_getprop can get only the first "reg" property with a statically
allocated buffer, but prom_getprop cannot.
This should fix a "kernel trap 30: data access exception" when you
invoke a pcictl(8).

branches: 1.45.2;
Set prom_getprop args correctly.

Some more promlib interface conversions.

Two commas went missing in last commit.

Partial switch to the promlib interface.

For pci devices without ofw mapping find ino by pci bus/slot/device
information, from OpenBSD.

Nuke bcopy/bzero.

Cosmetic: move DEBUG defines.

__KERNEL_RCSID()

branches: 1.37.2;
When overwriting the cache line size configuration, do not use the magic
value 0x10, but instead calculate a line size that is properly aligned
to the IOMMU strbuf cache and our external cache line size.
From FreeBSD.

Move the bus speed query code into it's own function and make it default
to 33 MHz. Cosmetics, no functional change.

PCI-CardBus bridge support for sparc64:
- handle devices which has no OBP node.
- move PCI latency-timer initialization from pci_intr_map to
pci_enumerate_bus.
- make PCI bus free space extents for cardbus devices.
- fix PCI config space map size.
- some code integrations.

Remove the `flags' argument from bus_intr_establish().

Set the PCI latency timer for bus master devices.

Slightly different to the patch suggested by Takeshi Nakayama in PR 18451
(idea from FreeBSD).

Additional input from Jason Thorpe: do not hard code the bus frequency,
instead get it from OpenFirmware.

Turn on parity checking for the PCI bus.

branches: 1.31.2;
Make the PCI code actually work.

branches: 1.30.2;
* Add "pcitag_t *pba_bridgetag" to pci_attach_args. This is set to
NULL for root PCI busses. For busses behind a bridge, it points to
a persistent copy of the bridge's pcitag_t. This can be very useful
for machine-dependent PCI bus enumeration code.
* Implement a machine-dependent pci_enumerate_bus() for sparc64 which
uses OFW device nodes to enumerate the bus. When a PCI bus that is
behind a bridge is attached, pci_attach_hook() allocates a new PCI
chipset tag for the new bus and sets it's "curnode" to the OFW node
of the bridge. This is used as a starting point when enumerating
that bus. Root busses get the OFW node of the host bridge (psycho).
* Garbage-collect "ofpci" and "ofppb" from the sparc64 port.

Garbage-collect __PCI_DEV_FUNCORDER.

Implement pci_decompose_tag().

Add ofpci_make_tag().

Overhaul bus space.

bus_space_handle_t now holds an address and two ASIs, one for normal accesses
and one for streaming accesses. This allows to map individual handles
different ways, so some can use MMU bypass accesses and others use virtual
addresses. bus_space_map() will now create handles that use bypass accesses
unles BUS_SPACE_MAP_LINEAR is passed in. So only pass in BUS_SPACE_MAP_LINEAR
if you absolutely *need* to use bus_space_vaddr(). This removes at least one
extra level of indirection and should reduce TLB misses.

32-bit kernels have problems accessing 64-bit addresses, so they always use
virtual addresses.

branches: 1.25.4;
Fix the semantics of __PCI_DEV_FUNCORDER so that they're actually useful:
If __PCI_DEV_FUNCORDER is defined, don't do the song-and-dance to check if
a device is multi-function; machdep code is going to tell us exactly which
functions to probe.

Note this required changing how pci_func_devorder() works in the
sparc64 PCI machdep code; now the "curnode" is assumed to point
to the bus, rather than some function (typically 0) on the device,
just as pci_bus_devorder() makes that assumption.

All this should allow the PCI code to actually locate the second
HME device on a Sun Netra t1, which is at 3,1 -- previously, the
PCI code would have missed it because there is no device at 3,0.

(Sun deserves a brick to the head for this one -- this seems clearly
out of line with the PCI spec.)

in pci_conf_read() and pci_conf_write(), convert all printf()'s to only
occur if SPDB_CONF (and hence DEBUG). also convert the panic in
pci_conf_write() to a SPDB_CONF warning and a return.

this cleans up pcictl(8) support.

branches: 1.23.2;
Fix backwards compare. (From Jason Wright)

PCI overhaul.

branches: 1.21.2;
adapt to systems with out interrupt-map & interrupt-map-mask properties.

rework pci_attach_hook() to just deal with the INO, not the full
interrupt number. properly find interrupts for the E250. modify
pci_intr_map() accordingly. retire psycho_intr_map(). deal with
INO values upto 0x3f, not upto 0x32. restructure sabre_init() and
psycho_init() to be more similar, and display each psycho's IGN.
psycho_intr_establish() deals with INO upto 0x3f, values from 0x32
and higher get 0 for IPL.

tested on E250 & U5.

branches: 1.19.2;
use pci_bus_devorder() and pci_dev_funcorder()

Enable IO space mapping by default.
Fixes PR 11940.

Fix format warnings to make it compile again.

Change pci_intr_map to get interrupt source information from a "struct
pci_attach_args *" instead of from four separate parameters which in
all cases were extracted from the same "struct pci_attach_args".

This both simplifies the driver api, and allows for alternate PCI
interrupt mapping schemes, such as one using the tables described in
the Intel Multiprocessor Spec which describe interrupt wirings for
devices behind pci-pci bridges based on the device's location rather
the bridge's location.

Tested on alpha and i386; welcome to 1.5Q

On-board devices on psycho machines seem to have the `interrupt' property
in the parent bus format (i.e. an INO) rather than being represented as
an PCI interrupt line. Provide a hack to work around this in pci_attach_hook().

Use probeget() in pci_config_read() for the psycho, to avoid bus
faults when probing PCI space.

Add a `device class' interrupt level argument (from machine/intr.h)
to bus_interrupt_establish().

It's currently only used in sparc64/dev/psycho.c to assign a CPU interrupt
level to devices in PCI slots.

remove include of <vm/vm.h>. <vm/vm.h> -> <uvm/uvm_extern.h>

remove/move more mach vm header files:

<vm/pglist.h> -> <uvm/uvm_pglist.h>
<vm/vm_inherit.h> -> <uvm/uvm_inherit.h>
<vm/vm_kern.h> -> into <uvm/uvm_extern.h>
<vm/vm_object.h> -> nothing
<vm/vm_pager.h> -> into <uvm/uvm_pager.h>

also includes a bunch of <vm/vm_page.h> include removals (due to redudancy
with <vm/vm.h>), and a scattering of other similar headers.

branches: 1.10.2;
kill dead code.

Turn on PCI MEM and DMA.

Implement the more flexiable `evcnt' interface as discussed (briefly) on
tech-kern and now documented in evcnt(9).

branches: 1.7.2;
Use the PCI function code to distinguish simba bus A from bus B instead
of the current hack.

add a note about US IIi and PCI_INTERRUPT_LINE register

- store the ebus's parent so we can find the IOMMU
- use generic iommu routines in the ebus code (uses the above)
- add some more comments.

clean this up some.

branches: 1.3.2; 1.3.4;
Make pbrobeget() and probeset() work for 64-bit values as well. To do this
the arguments are changed so the address is first and the ASI second so we
can have the address in %o0:%o1 and not worry about unused registers.

Also a bit of copyright cleanup.

clean up a bit, implement pci_conf_{read,write}() with probe[gs]et() (not yet used).

PCI driver for the UltraSPARC. this only works on the Ultra5/10 machines
(`SUNW,sabre') for now, and it doesn't really quite work there yet anyway.
the bus space/dma code is cloned from the sbus driver. the IOMMU code also
is cloned from the sbus code, but separated out into iommu.c so that we can
share it with the sbus driver. hopefully, much of the bus space/dma code
can also be re-shared with the sbus driver and the ebus driver but for now
these copies will do.

support for the real UltraSPARC PCI (`SUNW,psycho') is unwritten, though
most of this code is shared with it.

we can probe PCI config space and try to configue devices, but interrupts
don't work yet...

Revert "pci: Pass cookie through pci_find_device, pci_enumerate_bus."

Evidently something is wrong with this, to be diagnosed and redone
once the builds and tests are in better shape.

pci: Pass cookie through pci_find_device, pci_enumerate_bus.

New functions pci_find_device1 and pci_enumerate_bus1 have the cookie
argument. Existing symbols pci_find_device and pci_enumerate_bus are
now wrappers for the cookieless version.

This drops the symbol pci_probe_device, in favour of a new
pci_probe_device1 with the cookie argument. But I don't think that
requires a revbump because it's only called by MD pci_enumerate_bus1
implementations, which don't live in modules anyway.

Remove unnecessary <sys/malloc.h>.

Don't try to make a tag based on node 0

Rename min/max -> uimin/uimax for better honesty.

These functions are defined on unsigned int. The generic name
min/max should not silently truncate to 32 bits on 64-bit systems.
This is purely a name change -- no functional change intended.

HOWEVER! Some subsystems have

#define min(a, b) ((a) < (b) ? (a) : (b))
#define max(a, b) ((a) > (b) ? (a) : (b))

even though our standard name for that is MIN/MAX. Although these
may invite multiple evaluation bugs, these do _not_ cause integer
truncation.

To avoid `fixing' these cases, I first changed the name in libkern,
and then compile-tested every file where min/max occurred in order to
confirm that it failed -- and thus confirm that nothing shadowed
min/max -- before changing it.

I have left a handful of bootloaders that are too annoying to
compile-test, and some dead code:

cobalt ews4800mips hp300 hppa ia64 luna68k vax
acorn32/if_ie.c (not included in any kernels)
macppc/if_gm.c (superseded by gem(4))

It should be easy to fix the fallout once identified -- this way of
doing things fails safe, and the goal here, after all, is to _avoid_
silent integer truncations, not introduce them.

Maybe one day we can reintroduce min/max as type-generic things that
never silently truncate. But we should avoid doing that for a while,
so that existing code has a chance to be detected by the compiler for
conversion to uimin/uimax without changing the semantics until we can
properly audit it all. (Who knows, maybe in some cases integer
truncation is actually intended!)

branches: 1.77.28; 1.77.30;
make pci_intr_string and eisa_intr_string take a buffer and a length
instead of relying in local static storage.

kill sprintf

branches: 1.75.2;
split device_t/softc for all remaining drivers.
replace "struct device *" with "device_t".
use device_xname(), device_unit(), etc.

branches: 1.74.2; 1.74.12;
convert some open-coded SPDB_PROBE printf calls to use DPRINTF() macro.

#include <sys/bus.h> instead of <machine/bus.h>.

when looking for a device's node we want to start with the child nodes of the
root bus, not its peers

Neither pci_dma64_available(), pci_probe_device(), pci_mapreg_map(9),
pci_find_rom(), pci_intr_map(9), pci_enumerate_bus(), nor the match
predicate passed to pciide_compat_intr_establish() should ever modify
their pci_attach_args argument, so make their pci_attach_args arguments
const and deal with the fallout throughout the kernel.

For the most part, these changes add a 'const' where there was no
'const' before, however, some drivers and MD code used to modify
pci_attach_args. Now those drivers either copy their pci_attach_args
and modify the copy, or refrain from modifying pci_attach_args:

Xen: according to Manuel Bouyer, writing to pci_attach_args in
pci_intr_map() was a leftover from Xen 2. Probably a bug. I
stopped writing it. I have not tested this change.

siside(4): sis_hostbr_match() needlessly wrote to pci_attach_args.
Probably a bug. I use a temporary variable. I have not tested this
change.

slide(4): sl82c105_chip_map() overwrote the caller's pci_attach_args.
Probably a bug. Use a local pci_attach_args. I have not tested
this change.

viaide(4): via_sata_chip_map() and via_sata_chip_map_new() overwrote the
caller's pci_attach_args. Probably a bug. Make a local copy of the
caller's pci_attach_args and modify the copy. I have not tested
this change.

While I'm here, make pci_mapreg_submap() static.

With these changes in place, I have tested the compilation of these
kernels:

alpha GENERIC
amd64 GENERIC XEN3_DOM0
arc GENERIC
atari HADES MILAN-PCIIDE
bebox GENERIC
cats GENERIC
cobalt GENERIC
evbarm-eb NSLU2
evbarm-el ADI_BRH ARMADILLO9 CP3100 GEMINI GEMINI_MASTER GEMINI_SLAVE GUMSTIX
HDL_G IMX31LITE INTEGRATOR IQ31244 IQ80310 IQ80321 IXDP425 IXM1200
KUROBOX_PRO LUBBOCK MARVELL_NAS NAPPI SHEEVAPLUG SMDK2800 TEAMASA_NPWR
TEAMASA_NPWR_FC TS7200 TWINTAIL ZAO425
evbmips-el AP30 DBAU1500 DBAU1550 MALTA MERAKI MTX-1 OMSAL400 RB153 WGT624V3
evbmips64-el XLSATX
evbppc EV64260 MPC8536DS MPC8548CDS OPENBLOCKS200 OPENBLOCKS266
OPENBLOCKS266_OPT P2020RDB PMPPC RB800 WALNUT
hp700 GENERIC
i386 ALL XEN3_DOM0 XEN3_DOMU
ibmnws GENERIC
macppc GENERIC
mvmeppc GENERIC
netwinder GENERIC
ofppc GENERIC
prep GENERIC
sandpoint GENERIC
sgimips GENERIC32_IP2x
sparc GENERIC_SUN4U KRUPS
sparc64 GENERIC

As of Sun Apr 3 15:26:26 CDT 2011, I could not compile these kernels
with or without my patches in place:

### evbmips-el GDIUM

nbmake: nbmake: don't know how to make /home/dyoung/pristine-nbsd/src/sys/arch/mips/mips/softintr.c. Stop

### evbarm-el MPCSA_GENERIC
src/sys/arch/evbarm/conf/MPCSA_GENERIC:318: ds1672rtc*: unknown device `ds1672rtc'

### ia64 GENERIC

/tmp/genassym.28085/assym.c: In function 'f111':
/tmp/genassym.28085/assym.c:67: error: invalid application of 'sizeof' to incomplete type 'struct pcb'
/tmp/genassym.28085/assym.c:76: error: dereferencing pointer to incomplete type

### sgimips GENERIC32_IP3x

crmfb.o: In function `crmfb_attach':
crmfb.c:(.text+0x2304): undefined reference to `ddc_read_edid'
crmfb.c:(.text+0x2304): relocation truncated to fit: R_MIPS_26 against `ddc_read_edid'
crmfb.c:(.text+0x234c): undefined reference to `edid_parse'
crmfb.c:(.text+0x234c): relocation truncated to fit: R_MIPS_26 against `edid_parse'
crmfb.c:(.text+0x2354): undefined reference to `edid_print'
crmfb.c:(.text+0x2354): relocation truncated to fit: R_MIPS_26 against `edid_print'

remove unnecessary iommuvar.h and psycho{reg,var}.h includes.

don't try to turn on parity or set the latency timer, or dump pci config
space for the pci bridge in sparc64_pci_enumerate_bus(). it doesn't work
on PCIe adapters. #if 0'ed for now, should move this into psycho/schizo.

branches: 1.68.2;
allow sparc64 spc_find_ino() to be NULL; schizo doesn't need it.

branches: 1.67.4;
Cope with up to four "interrupts" properties for a node.
Just use the first one for now. This should be improved.
OK: mrg@.

move the guts of pci_intr_map() into pci_machdep.c, and move the schizo
ign addition into a hook. also, don't double shift the ign in set_intr.

tested on U60 and SB2500.

convert pci_intr_map() into a chipset tag function pointer, and implement
the schizo version slightly differently.

pull out the schizo's IGN from the upaid, not the apparently broken device
ID register. from openbsd.

with this i appear to have valid working interrupts on the SB2500.

tested on U60 and SB2500.


XXX: we can probably kill (*spc_find_ino)() now that pci_intr_map() itself
XXX: is no longer a first class function.

give more debug output about mapping interrupts.

clean up some debug messages a little bit.

pci_make_tag:
- add assertion for checking spc_busnode initialization.

ofpci_make_tag():
don't write PCI_COMMAND_STATUS_REG with
PCI_COMMAND_(MEM|MASTER|IO)_ENABLE, or it in.
sparc_pci_childspace():
move psycho.c:get_childspace() into pci_machdep.c but rename it
psycho_alloc_bus_tag():
use M_ZERO

a few clean up/rearrange changes, somewhat inspired from openbsd's
schizo.c:

- rename pp_busnode{} -> spc_busnode{}, and move it to
sparc_pci_chipset{}

- move public pci functions pci_conf_read(), pci_conf_write() and
pci_intr_establish() into psycho.c, renaming them in the process,
and add pointers through sparc_pci_chipset{} to access them.
as part of this, add PDB_INTMAP and PDB_CONF debug options to
psycho.c

- move and rename sparc64 private pci_find_ino() into
sparc_pci_chipset{}, like the above

- update psycho_alloc_chipset() to set the new 4 function members


no functional change intended. XXX: not tested on cardbus.

branches: 1.59.4; 1.59.6;
pci_intr_setattr(), allows PCI interrupts to be marked MPSAFE on x86, and
other platforms if the code is added.

pci_intr_map(...)
pci_intr_setattr(pc, ih, PCI_INTR_MPSAFE, 1);
pci_intr_establish(...)

remove clause #3 from my license where there are no other
copyright holders involved.

branches: 1.57.2; 1.57.4; 1.57.6;
Change previous to use aprint_error_dev.

use device_xname()

branches: 1.55.6;
Do not panic when trying to disestablish a pci interrupt.

branches: 1.54.14; 1.54.20;
pci_find_ino(): Use interrupt pin from pci_attach_args when mapping interrupts.
Use *ihp as a pure output. Fix case where interrupt pin #D was incorrectly
mapped. Some cosmetic changes while here. Ok'd by martin.

branches: 1.53.4; 1.53.8; 1.53.10;
remove "struct cacheline" and the "cacheline" variable. (it was only set
in cpu_attach(), overwritten by multiple cpus, and only one member of the
12 or more were ever used again.) add an ecache_min_line_size to replace
this functionality.

clean up some non-SMP friendly code from cpu_attach():
- only reset the fpu state on the boot cpu
- don't bother printing "FPU version 0" always
- don't call getver() and ignore the result
- print the UPA id on the first cpu0: line

only print mp_tramp address #ifdef DEBUG.

reset the fpu state in cpu_hatch(), like cpu_attach() does for the boot cpu.


cpu_attach() is now free of SMP-unfriendly code.

Missing initializers

branches: 1.51.14; 1.51.16;
ANSIfication: u_intN_t -> uintN_t, use ANSI function declarations/definitions
instead of K&R ones.

branches: 1.50.2; 1.50.4; 1.50.6;
merge ktrace-lwp.

branches: 1.49.12;
make PCI devices attachable/detachable (as far as the particular drivers
allow), and allow to rescan a bus selectively (ie only the device/
function I'm looking at)

give the special PCI bus enumerator a sparc64_ prefix and make the
PCI_MACHDEP_ENUMERATE_BUS macro point to it

pci_find_ino: support pci-pci bridges.

Revert a part of rev. 1.44.

OF_getprop can get only the first "reg" property with a statically
allocated buffer, but prom_getprop cannot.
This should fix a "kernel trap 30: data access exception" when you
invoke a pcictl(8).

branches: 1.45.2;
Set prom_getprop args correctly.

Some more promlib interface conversions.

Two commas went missing in last commit.

Partial switch to the promlib interface.

For pci devices without ofw mapping find ino by pci bus/slot/device
information, from OpenBSD.

Nuke bcopy/bzero.

Cosmetic: move DEBUG defines.

__KERNEL_RCSID()

branches: 1.37.2;
When overwriting the cache line size configuration, do not use the magic
value 0x10, but instead calculate a line size that is properly aligned
to the IOMMU strbuf cache and our external cache line size.
From FreeBSD.

Move the bus speed query code into it's own function and make it default
to 33 MHz. Cosmetics, no functional change.

PCI-CardBus bridge support for sparc64:
- handle devices which has no OBP node.
- move PCI latency-timer initialization from pci_intr_map to
pci_enumerate_bus.
- make PCI bus free space extents for cardbus devices.
- fix PCI config space map size.
- some code integrations.

Remove the `flags' argument from bus_intr_establish().

Set the PCI latency timer for bus master devices.

Slightly different to the patch suggested by Takeshi Nakayama in PR 18451
(idea from FreeBSD).

Additional input from Jason Thorpe: do not hard code the bus frequency,
instead get it from OpenFirmware.

Turn on parity checking for the PCI bus.

branches: 1.31.2;
Make the PCI code actually work.

branches: 1.30.2;
* Add "pcitag_t *pba_bridgetag" to pci_attach_args. This is set to
NULL for root PCI busses. For busses behind a bridge, it points to
a persistent copy of the bridge's pcitag_t. This can be very useful
for machine-dependent PCI bus enumeration code.
* Implement a machine-dependent pci_enumerate_bus() for sparc64 which
uses OFW device nodes to enumerate the bus. When a PCI bus that is
behind a bridge is attached, pci_attach_hook() allocates a new PCI
chipset tag for the new bus and sets it's "curnode" to the OFW node
of the bridge. This is used as a starting point when enumerating
that bus. Root busses get the OFW node of the host bridge (psycho).
* Garbage-collect "ofpci" and "ofppb" from the sparc64 port.

Garbage-collect __PCI_DEV_FUNCORDER.

Implement pci_decompose_tag().

Add ofpci_make_tag().

Overhaul bus space.

bus_space_handle_t now holds an address and two ASIs, one for normal accesses
and one for streaming accesses. This allows to map individual handles
different ways, so some can use MMU bypass accesses and others use virtual
addresses. bus_space_map() will now create handles that use bypass accesses
unles BUS_SPACE_MAP_LINEAR is passed in. So only pass in BUS_SPACE_MAP_LINEAR
if you absolutely *need* to use bus_space_vaddr(). This removes at least one
extra level of indirection and should reduce TLB misses.

32-bit kernels have problems accessing 64-bit addresses, so they always use
virtual addresses.

branches: 1.25.4;
Fix the semantics of __PCI_DEV_FUNCORDER so that they're actually useful:
If __PCI_DEV_FUNCORDER is defined, don't do the song-and-dance to check if
a device is multi-function; machdep code is going to tell us exactly which
functions to probe.

Note this required changing how pci_func_devorder() works in the
sparc64 PCI machdep code; now the "curnode" is assumed to point
to the bus, rather than some function (typically 0) on the device,
just as pci_bus_devorder() makes that assumption.

All this should allow the PCI code to actually locate the second
HME device on a Sun Netra t1, which is at 3,1 -- previously, the
PCI code would have missed it because there is no device at 3,0.

(Sun deserves a brick to the head for this one -- this seems clearly
out of line with the PCI spec.)

in pci_conf_read() and pci_conf_write(), convert all printf()'s to only
occur if SPDB_CONF (and hence DEBUG). also convert the panic in
pci_conf_write() to a SPDB_CONF warning and a return.

this cleans up pcictl(8) support.

branches: 1.23.2;
Fix backwards compare. (From Jason Wright)

PCI overhaul.

branches: 1.21.2;
adapt to systems with out interrupt-map & interrupt-map-mask properties.

rework pci_attach_hook() to just deal with the INO, not the full
interrupt number. properly find interrupts for the E250. modify
pci_intr_map() accordingly. retire psycho_intr_map(). deal with
INO values upto 0x3f, not upto 0x32. restructure sabre_init() and
psycho_init() to be more similar, and display each psycho's IGN.
psycho_intr_establish() deals with INO upto 0x3f, values from 0x32
and higher get 0 for IPL.

tested on E250 & U5.

branches: 1.19.2;
use pci_bus_devorder() and pci_dev_funcorder()

Enable IO space mapping by default.
Fixes PR 11940.

Fix format warnings to make it compile again.

Change pci_intr_map to get interrupt source information from a "struct
pci_attach_args *" instead of from four separate parameters which in
all cases were extracted from the same "struct pci_attach_args".

This both simplifies the driver api, and allows for alternate PCI
interrupt mapping schemes, such as one using the tables described in
the Intel Multiprocessor Spec which describe interrupt wirings for
devices behind pci-pci bridges based on the device's location rather
the bridge's location.

Tested on alpha and i386; welcome to 1.5Q

On-board devices on psycho machines seem to have the `interrupt' property
in the parent bus format (i.e. an INO) rather than being represented as
an PCI interrupt line. Provide a hack to work around this in pci_attach_hook().

Use probeget() in pci_config_read() for the psycho, to avoid bus
faults when probing PCI space.

Add a `device class' interrupt level argument (from machine/intr.h)
to bus_interrupt_establish().

It's currently only used in sparc64/dev/psycho.c to assign a CPU interrupt
level to devices in PCI slots.

remove include of <vm/vm.h>. <vm/vm.h> -> <uvm/uvm_extern.h>

remove/move more mach vm header files:

<vm/pglist.h> -> <uvm/uvm_pglist.h>
<vm/vm_inherit.h> -> <uvm/uvm_inherit.h>
<vm/vm_kern.h> -> into <uvm/uvm_extern.h>
<vm/vm_object.h> -> nothing
<vm/vm_pager.h> -> into <uvm/uvm_pager.h>

also includes a bunch of <vm/vm_page.h> include removals (due to redudancy
with <vm/vm.h>), and a scattering of other similar headers.

branches: 1.10.2;
kill dead code.

Turn on PCI MEM and DMA.

Implement the more flexiable `evcnt' interface as discussed (briefly) on
tech-kern and now documented in evcnt(9).

branches: 1.7.2;
Use the PCI function code to distinguish simba bus A from bus B instead
of the current hack.

add a note about US IIi and PCI_INTERRUPT_LINE register

- store the ebus's parent so we can find the IOMMU
- use generic iommu routines in the ebus code (uses the above)
- add some more comments.

clean this up some.

branches: 1.3.2; 1.3.4;
Make pbrobeget() and probeset() work for 64-bit values as well. To do this
the arguments are changed so the address is first and the ASI second so we
can have the address in %o0:%o1 and not worry about unused registers.

Also a bit of copyright cleanup.

clean up a bit, implement pci_conf_{read,write}() with probe[gs]et() (not yet used).

PCI driver for the UltraSPARC. this only works on the Ultra5/10 machines
(`SUNW,sabre') for now, and it doesn't really quite work there yet anyway.
the bus space/dma code is cloned from the sbus driver. the IOMMU code also
is cloned from the sbus code, but separated out into iommu.c so that we can
share it with the sbus driver. hopefully, much of the bus space/dma code
can also be re-shared with the sbus driver and the ebus driver but for now
these copies will do.

support for the real UltraSPARC PCI (`SUNW,psycho') is unwritten, though
most of this code is shared with it.

we can probe PCI config space and try to configue devices, but interrupts
don't work yet...

Remove unnecessary <sys/malloc.h>.

Don't try to make a tag based on node 0

Rename min/max -> uimin/uimax for better honesty.

These functions are defined on unsigned int. The generic name
min/max should not silently truncate to 32 bits on 64-bit systems.
This is purely a name change -- no functional change intended.

HOWEVER! Some subsystems have

#define min(a, b) ((a) < (b) ? (a) : (b))
#define max(a, b) ((a) > (b) ? (a) : (b))

even though our standard name for that is MIN/MAX. Although these
may invite multiple evaluation bugs, these do _not_ cause integer
truncation.

To avoid `fixing' these cases, I first changed the name in libkern,
and then compile-tested every file where min/max occurred in order to
confirm that it failed -- and thus confirm that nothing shadowed
min/max -- before changing it.

I have left a handful of bootloaders that are too annoying to
compile-test, and some dead code:

cobalt ews4800mips hp300 hppa ia64 luna68k vax
acorn32/if_ie.c (not included in any kernels)
macppc/if_gm.c (superseded by gem(4))

It should be easy to fix the fallout once identified -- this way of
doing things fails safe, and the goal here, after all, is to _avoid_
silent integer truncations, not introduce them.

Maybe one day we can reintroduce min/max as type-generic things that
never silently truncate. But we should avoid doing that for a while,
so that existing code has a chance to be detected by the compiler for
conversion to uimin/uimax without changing the semantics until we can
properly audit it all. (Who knows, maybe in some cases integer
truncation is actually intended!)

branches: 1.77.28; 1.77.30;
make pci_intr_string and eisa_intr_string take a buffer and a length
instead of relying in local static storage.

kill sprintf

branches: 1.75.2;
split device_t/softc for all remaining drivers.
replace "struct device *" with "device_t".
use device_xname(), device_unit(), etc.

branches: 1.74.2; 1.74.12;
convert some open-coded SPDB_PROBE printf calls to use DPRINTF() macro.

#include <sys/bus.h> instead of <machine/bus.h>.

when looking for a device's node we want to start with the child nodes of the
root bus, not its peers

Neither pci_dma64_available(), pci_probe_device(), pci_mapreg_map(9),
pci_find_rom(), pci_intr_map(9), pci_enumerate_bus(), nor the match
predicate passed to pciide_compat_intr_establish() should ever modify
their pci_attach_args argument, so make their pci_attach_args arguments
const and deal with the fallout throughout the kernel.

For the most part, these changes add a 'const' where there was no
'const' before, however, some drivers and MD code used to modify
pci_attach_args. Now those drivers either copy their pci_attach_args
and modify the copy, or refrain from modifying pci_attach_args:

Xen: according to Manuel Bouyer, writing to pci_attach_args in
pci_intr_map() was a leftover from Xen 2. Probably a bug. I
stopped writing it. I have not tested this change.

siside(4): sis_hostbr_match() needlessly wrote to pci_attach_args.
Probably a bug. I use a temporary variable. I have not tested this
change.

slide(4): sl82c105_chip_map() overwrote the caller's pci_attach_args.
Probably a bug. Use a local pci_attach_args. I have not tested
this change.

viaide(4): via_sata_chip_map() and via_sata_chip_map_new() overwrote the
caller's pci_attach_args. Probably a bug. Make a local copy of the
caller's pci_attach_args and modify the copy. I have not tested
this change.

While I'm here, make pci_mapreg_submap() static.

With these changes in place, I have tested the compilation of these
kernels:

alpha GENERIC
amd64 GENERIC XEN3_DOM0
arc GENERIC
atari HADES MILAN-PCIIDE
bebox GENERIC
cats GENERIC
cobalt GENERIC
evbarm-eb NSLU2
evbarm-el ADI_BRH ARMADILLO9 CP3100 GEMINI GEMINI_MASTER GEMINI_SLAVE GUMSTIX
HDL_G IMX31LITE INTEGRATOR IQ31244 IQ80310 IQ80321 IXDP425 IXM1200
KUROBOX_PRO LUBBOCK MARVELL_NAS NAPPI SHEEVAPLUG SMDK2800 TEAMASA_NPWR
TEAMASA_NPWR_FC TS7200 TWINTAIL ZAO425
evbmips-el AP30 DBAU1500 DBAU1550 MALTA MERAKI MTX-1 OMSAL400 RB153 WGT624V3
evbmips64-el XLSATX
evbppc EV64260 MPC8536DS MPC8548CDS OPENBLOCKS200 OPENBLOCKS266
OPENBLOCKS266_OPT P2020RDB PMPPC RB800 WALNUT
hp700 GENERIC
i386 ALL XEN3_DOM0 XEN3_DOMU
ibmnws GENERIC
macppc GENERIC
mvmeppc GENERIC
netwinder GENERIC
ofppc GENERIC
prep GENERIC
sandpoint GENERIC
sgimips GENERIC32_IP2x
sparc GENERIC_SUN4U KRUPS
sparc64 GENERIC

As of Sun Apr 3 15:26:26 CDT 2011, I could not compile these kernels
with or without my patches in place:

### evbmips-el GDIUM

nbmake: nbmake: don't know how to make /home/dyoung/pristine-nbsd/src/sys/arch/mips/mips/softintr.c. Stop

### evbarm-el MPCSA_GENERIC
src/sys/arch/evbarm/conf/MPCSA_GENERIC:318: ds1672rtc*: unknown device `ds1672rtc'

### ia64 GENERIC

/tmp/genassym.28085/assym.c: In function 'f111':
/tmp/genassym.28085/assym.c:67: error: invalid application of 'sizeof' to incomplete type 'struct pcb'
/tmp/genassym.28085/assym.c:76: error: dereferencing pointer to incomplete type

### sgimips GENERIC32_IP3x

crmfb.o: In function `crmfb_attach':
crmfb.c:(.text+0x2304): undefined reference to `ddc_read_edid'
crmfb.c:(.text+0x2304): relocation truncated to fit: R_MIPS_26 against `ddc_read_edid'
crmfb.c:(.text+0x234c): undefined reference to `edid_parse'
crmfb.c:(.text+0x234c): relocation truncated to fit: R_MIPS_26 against `edid_parse'
crmfb.c:(.text+0x2354): undefined reference to `edid_print'
crmfb.c:(.text+0x2354): relocation truncated to fit: R_MIPS_26 against `edid_print'

remove unnecessary iommuvar.h and psycho{reg,var}.h includes.

don't try to turn on parity or set the latency timer, or dump pci config
space for the pci bridge in sparc64_pci_enumerate_bus(). it doesn't work
on PCIe adapters. #if 0'ed for now, should move this into psycho/schizo.

branches: 1.68.2;
allow sparc64 spc_find_ino() to be NULL; schizo doesn't need it.

branches: 1.67.4;
Cope with up to four "interrupts" properties for a node.
Just use the first one for now. This should be improved.
OK: mrg@.

move the guts of pci_intr_map() into pci_machdep.c, and move the schizo
ign addition into a hook. also, don't double shift the ign in set_intr.

tested on U60 and SB2500.

convert pci_intr_map() into a chipset tag function pointer, and implement
the schizo version slightly differently.

pull out the schizo's IGN from the upaid, not the apparently broken device
ID register. from openbsd.

with this i appear to have valid working interrupts on the SB2500.

tested on U60 and SB2500.


XXX: we can probably kill (*spc_find_ino)() now that pci_intr_map() itself
XXX: is no longer a first class function.

give more debug output about mapping interrupts.

clean up some debug messages a little bit.

pci_make_tag:
- add assertion for checking spc_busnode initialization.

ofpci_make_tag():
don't write PCI_COMMAND_STATUS_REG with
PCI_COMMAND_(MEM|MASTER|IO)_ENABLE, or it in.
sparc_pci_childspace():
move psycho.c:get_childspace() into pci_machdep.c but rename it
psycho_alloc_bus_tag():
use M_ZERO

a few clean up/rearrange changes, somewhat inspired from openbsd's
schizo.c:

- rename pp_busnode{} -> spc_busnode{}, and move it to
sparc_pci_chipset{}

- move public pci functions pci_conf_read(), pci_conf_write() and
pci_intr_establish() into psycho.c, renaming them in the process,
and add pointers through sparc_pci_chipset{} to access them.
as part of this, add PDB_INTMAP and PDB_CONF debug options to
psycho.c

- move and rename sparc64 private pci_find_ino() into
sparc_pci_chipset{}, like the above

- update psycho_alloc_chipset() to set the new 4 function members


no functional change intended. XXX: not tested on cardbus.

branches: 1.59.4; 1.59.6;
pci_intr_setattr(), allows PCI interrupts to be marked MPSAFE on x86, and
other platforms if the code is added.

pci_intr_map(...)
pci_intr_setattr(pc, ih, PCI_INTR_MPSAFE, 1);
pci_intr_establish(...)

remove clause #3 from my license where there are no other
copyright holders involved.

branches: 1.57.2; 1.57.4; 1.57.6;
Change previous to use aprint_error_dev.

use device_xname()

branches: 1.55.6;
Do not panic when trying to disestablish a pci interrupt.

branches: 1.54.14; 1.54.20;
pci_find_ino(): Use interrupt pin from pci_attach_args when mapping interrupts.
Use *ihp as a pure output. Fix case where interrupt pin #D was incorrectly
mapped. Some cosmetic changes while here. Ok'd by martin.

branches: 1.53.4; 1.53.8; 1.53.10;
remove "struct cacheline" and the "cacheline" variable. (it was only set
in cpu_attach(), overwritten by multiple cpus, and only one member of the
12 or more were ever used again.) add an ecache_min_line_size to replace
this functionality.

clean up some non-SMP friendly code from cpu_attach():
- only reset the fpu state on the boot cpu
- don't bother printing "FPU version 0" always
- don't call getver() and ignore the result
- print the UPA id on the first cpu0: line

only print mp_tramp address #ifdef DEBUG.

reset the fpu state in cpu_hatch(), like cpu_attach() does for the boot cpu.


cpu_attach() is now free of SMP-unfriendly code.

Missing initializers

branches: 1.51.14; 1.51.16;
ANSIfication: u_intN_t -> uintN_t, use ANSI function declarations/definitions
instead of K&R ones.

branches: 1.50.2; 1.50.4; 1.50.6;
merge ktrace-lwp.

branches: 1.49.12;
make PCI devices attachable/detachable (as far as the particular drivers
allow), and allow to rescan a bus selectively (ie only the device/
function I'm looking at)

give the special PCI bus enumerator a sparc64_ prefix and make the
PCI_MACHDEP_ENUMERATE_BUS macro point to it

pci_find_ino: support pci-pci bridges.

Revert a part of rev. 1.44.

OF_getprop can get only the first "reg" property with a statically
allocated buffer, but prom_getprop cannot.
This should fix a "kernel trap 30: data access exception" when you
invoke a pcictl(8).

branches: 1.45.2;
Set prom_getprop args correctly.

Some more promlib interface conversions.

Two commas went missing in last commit.

Partial switch to the promlib interface.

For pci devices without ofw mapping find ino by pci bus/slot/device
information, from OpenBSD.

Nuke bcopy/bzero.

Cosmetic: move DEBUG defines.

__KERNEL_RCSID()

branches: 1.37.2;
When overwriting the cache line size configuration, do not use the magic
value 0x10, but instead calculate a line size that is properly aligned
to the IOMMU strbuf cache and our external cache line size.
From FreeBSD.

Move the bus speed query code into it's own function and make it default
to 33 MHz. Cosmetics, no functional change.

PCI-CardBus bridge support for sparc64:
- handle devices which has no OBP node.
- move PCI latency-timer initialization from pci_intr_map to
pci_enumerate_bus.
- make PCI bus free space extents for cardbus devices.
- fix PCI config space map size.
- some code integrations.

Remove the `flags' argument from bus_intr_establish().

Set the PCI latency timer for bus master devices.

Slightly different to the patch suggested by Takeshi Nakayama in PR 18451
(idea from FreeBSD).

Additional input from Jason Thorpe: do not hard code the bus frequency,
instead get it from OpenFirmware.

Turn on parity checking for the PCI bus.

branches: 1.31.2;
Make the PCI code actually work.

branches: 1.30.2;
* Add "pcitag_t *pba_bridgetag" to pci_attach_args. This is set to
NULL for root PCI busses. For busses behind a bridge, it points to
a persistent copy of the bridge's pcitag_t. This can be very useful
for machine-dependent PCI bus enumeration code.
* Implement a machine-dependent pci_enumerate_bus() for sparc64 which
uses OFW device nodes to enumerate the bus. When a PCI bus that is
behind a bridge is attached, pci_attach_hook() allocates a new PCI
chipset tag for the new bus and sets it's "curnode" to the OFW node
of the bridge. This is used as a starting point when enumerating
that bus. Root busses get the OFW node of the host bridge (psycho).
* Garbage-collect "ofpci" and "ofppb" from the sparc64 port.

Garbage-collect __PCI_DEV_FUNCORDER.

Implement pci_decompose_tag().

Add ofpci_make_tag().

Overhaul bus space.

bus_space_handle_t now holds an address and two ASIs, one for normal accesses
and one for streaming accesses. This allows to map individual handles
different ways, so some can use MMU bypass accesses and others use virtual
addresses. bus_space_map() will now create handles that use bypass accesses
unles BUS_SPACE_MAP_LINEAR is passed in. So only pass in BUS_SPACE_MAP_LINEAR
if you absolutely *need* to use bus_space_vaddr(). This removes at least one
extra level of indirection and should reduce TLB misses.

32-bit kernels have problems accessing 64-bit addresses, so they always use
virtual addresses.

branches: 1.25.4;
Fix the semantics of __PCI_DEV_FUNCORDER so that they're actually useful:
If __PCI_DEV_FUNCORDER is defined, don't do the song-and-dance to check if
a device is multi-function; machdep code is going to tell us exactly which
functions to probe.

Note this required changing how pci_func_devorder() works in the
sparc64 PCI machdep code; now the "curnode" is assumed to point
to the bus, rather than some function (typically 0) on the device,
just as pci_bus_devorder() makes that assumption.

All this should allow the PCI code to actually locate the second
HME device on a Sun Netra t1, which is at 3,1 -- previously, the
PCI code would have missed it because there is no device at 3,0.

(Sun deserves a brick to the head for this one -- this seems clearly
out of line with the PCI spec.)

in pci_conf_read() and pci_conf_write(), convert all printf()'s to only
occur if SPDB_CONF (and hence DEBUG). also convert the panic in
pci_conf_write() to a SPDB_CONF warning and a return.

this cleans up pcictl(8) support.

branches: 1.23.2;
Fix backwards compare. (From Jason Wright)

PCI overhaul.

branches: 1.21.2;
adapt to systems with out interrupt-map & interrupt-map-mask properties.

rework pci_attach_hook() to just deal with the INO, not the full
interrupt number. properly find interrupts for the E250. modify
pci_intr_map() accordingly. retire psycho_intr_map(). deal with
INO values upto 0x3f, not upto 0x32. restructure sabre_init() and
psycho_init() to be more similar, and display each psycho's IGN.
psycho_intr_establish() deals with INO upto 0x3f, values from 0x32
and higher get 0 for IPL.

tested on E250 & U5.

branches: 1.19.2;
use pci_bus_devorder() and pci_dev_funcorder()

Enable IO space mapping by default.
Fixes PR 11940.

Fix format warnings to make it compile again.

Change pci_intr_map to get interrupt source information from a "struct
pci_attach_args *" instead of from four separate parameters which in
all cases were extracted from the same "struct pci_attach_args".

This both simplifies the driver api, and allows for alternate PCI
interrupt mapping schemes, such as one using the tables described in
the Intel Multiprocessor Spec which describe interrupt wirings for
devices behind pci-pci bridges based on the device's location rather
the bridge's location.

Tested on alpha and i386; welcome to 1.5Q

On-board devices on psycho machines seem to have the `interrupt' property
in the parent bus format (i.e. an INO) rather than being represented as
an PCI interrupt line. Provide a hack to work around this in pci_attach_hook().

Use probeget() in pci_config_read() for the psycho, to avoid bus
faults when probing PCI space.

Add a `device class' interrupt level argument (from machine/intr.h)
to bus_interrupt_establish().

It's currently only used in sparc64/dev/psycho.c to assign a CPU interrupt
level to devices in PCI slots.

remove include of <vm/vm.h>. <vm/vm.h> -> <uvm/uvm_extern.h>

remove/move more mach vm header files:

<vm/pglist.h> -> <uvm/uvm_pglist.h>
<vm/vm_inherit.h> -> <uvm/uvm_inherit.h>
<vm/vm_kern.h> -> into <uvm/uvm_extern.h>
<vm/vm_object.h> -> nothing
<vm/vm_pager.h> -> into <uvm/uvm_pager.h>

also includes a bunch of <vm/vm_page.h> include removals (due to redudancy
with <vm/vm.h>), and a scattering of other similar headers.

branches: 1.10.2;
kill dead code.

Turn on PCI MEM and DMA.

Implement the more flexiable `evcnt' interface as discussed (briefly) on
tech-kern and now documented in evcnt(9).

branches: 1.7.2;
Use the PCI function code to distinguish simba bus A from bus B instead
of the current hack.

add a note about US IIi and PCI_INTERRUPT_LINE register

- store the ebus's parent so we can find the IOMMU
- use generic iommu routines in the ebus code (uses the above)
- add some more comments.

clean this up some.

branches: 1.3.2; 1.3.4;
Make pbrobeget() and probeset() work for 64-bit values as well. To do this
the arguments are changed so the address is first and the ASI second so we
can have the address in %o0:%o1 and not worry about unused registers.

Also a bit of copyright cleanup.

clean up a bit, implement pci_conf_{read,write}() with probe[gs]et() (not yet used).

PCI driver for the UltraSPARC. this only works on the Ultra5/10 machines
(`SUNW,sabre') for now, and it doesn't really quite work there yet anyway.
the bus space/dma code is cloned from the sbus driver. the IOMMU code also
is cloned from the sbus code, but separated out into iommu.c so that we can
share it with the sbus driver. hopefully, much of the bus space/dma code
can also be re-shared with the sbus driver and the ebus driver but for now
these copies will do.

support for the real UltraSPARC PCI (`SUNW,psycho') is unwritten, though
most of this code is shared with it.

we can probe PCI config space and try to configue devices, but interrupts
don't work yet...

Don't try to make a tag based on node 0

Rename min/max -> uimin/uimax for better honesty.

These functions are defined on unsigned int. The generic name
min/max should not silently truncate to 32 bits on 64-bit systems.
This is purely a name change -- no functional change intended.

HOWEVER! Some subsystems have

#define min(a, b) ((a) < (b) ? (a) : (b))
#define max(a, b) ((a) > (b) ? (a) : (b))

even though our standard name for that is MIN/MAX. Although these
may invite multiple evaluation bugs, these do _not_ cause integer
truncation.

To avoid `fixing' these cases, I first changed the name in libkern,
and then compile-tested every file where min/max occurred in order to
confirm that it failed -- and thus confirm that nothing shadowed
min/max -- before changing it.

I have left a handful of bootloaders that are too annoying to
compile-test, and some dead code:

cobalt ews4800mips hp300 hppa ia64 luna68k vax
acorn32/if_ie.c (not included in any kernels)
macppc/if_gm.c (superseded by gem(4))

It should be easy to fix the fallout once identified -- this way of
doing things fails safe, and the goal here, after all, is to _avoid_
silent integer truncations, not introduce them.

Maybe one day we can reintroduce min/max as type-generic things that
never silently truncate. But we should avoid doing that for a while,
so that existing code has a chance to be detected by the compiler for
conversion to uimin/uimax without changing the semantics until we can
properly audit it all. (Who knows, maybe in some cases integer
truncation is actually intended!)

branches: 1.77.28; 1.77.30;
make pci_intr_string and eisa_intr_string take a buffer and a length
instead of relying in local static storage.

kill sprintf

branches: 1.75.2;
split device_t/softc for all remaining drivers.
replace "struct device *" with "device_t".
use device_xname(), device_unit(), etc.

branches: 1.74.2; 1.74.12;
convert some open-coded SPDB_PROBE printf calls to use DPRINTF() macro.

#include <sys/bus.h> instead of <machine/bus.h>.

when looking for a device's node we want to start with the child nodes of the
root bus, not its peers

Neither pci_dma64_available(), pci_probe_device(), pci_mapreg_map(9),
pci_find_rom(), pci_intr_map(9), pci_enumerate_bus(), nor the match
predicate passed to pciide_compat_intr_establish() should ever modify
their pci_attach_args argument, so make their pci_attach_args arguments
const and deal with the fallout throughout the kernel.

For the most part, these changes add a 'const' where there was no
'const' before, however, some drivers and MD code used to modify
pci_attach_args. Now those drivers either copy their pci_attach_args
and modify the copy, or refrain from modifying pci_attach_args:

Xen: according to Manuel Bouyer, writing to pci_attach_args in
pci_intr_map() was a leftover from Xen 2. Probably a bug. I
stopped writing it. I have not tested this change.

siside(4): sis_hostbr_match() needlessly wrote to pci_attach_args.
Probably a bug. I use a temporary variable. I have not tested this
change.

slide(4): sl82c105_chip_map() overwrote the caller's pci_attach_args.
Probably a bug. Use a local pci_attach_args. I have not tested
this change.

viaide(4): via_sata_chip_map() and via_sata_chip_map_new() overwrote the
caller's pci_attach_args. Probably a bug. Make a local copy of the
caller's pci_attach_args and modify the copy. I have not tested
this change.

While I'm here, make pci_mapreg_submap() static.

With these changes in place, I have tested the compilation of these
kernels:

alpha GENERIC
amd64 GENERIC XEN3_DOM0
arc GENERIC
atari HADES MILAN-PCIIDE
bebox GENERIC
cats GENERIC
cobalt GENERIC
evbarm-eb NSLU2
evbarm-el ADI_BRH ARMADILLO9 CP3100 GEMINI GEMINI_MASTER GEMINI_SLAVE GUMSTIX
HDL_G IMX31LITE INTEGRATOR IQ31244 IQ80310 IQ80321 IXDP425 IXM1200
KUROBOX_PRO LUBBOCK MARVELL_NAS NAPPI SHEEVAPLUG SMDK2800 TEAMASA_NPWR
TEAMASA_NPWR_FC TS7200 TWINTAIL ZAO425
evbmips-el AP30 DBAU1500 DBAU1550 MALTA MERAKI MTX-1 OMSAL400 RB153 WGT624V3
evbmips64-el XLSATX
evbppc EV64260 MPC8536DS MPC8548CDS OPENBLOCKS200 OPENBLOCKS266
OPENBLOCKS266_OPT P2020RDB PMPPC RB800 WALNUT
hp700 GENERIC
i386 ALL XEN3_DOM0 XEN3_DOMU
ibmnws GENERIC
macppc GENERIC
mvmeppc GENERIC
netwinder GENERIC
ofppc GENERIC
prep GENERIC
sandpoint GENERIC
sgimips GENERIC32_IP2x
sparc GENERIC_SUN4U KRUPS
sparc64 GENERIC

As of Sun Apr 3 15:26:26 CDT 2011, I could not compile these kernels
with or without my patches in place:

### evbmips-el GDIUM

nbmake: nbmake: don't know how to make /home/dyoung/pristine-nbsd/src/sys/arch/mips/mips/softintr.c. Stop

### evbarm-el MPCSA_GENERIC
src/sys/arch/evbarm/conf/MPCSA_GENERIC:318: ds1672rtc*: unknown device `ds1672rtc'

### ia64 GENERIC

/tmp/genassym.28085/assym.c: In function 'f111':
/tmp/genassym.28085/assym.c:67: error: invalid application of 'sizeof' to incomplete type 'struct pcb'
/tmp/genassym.28085/assym.c:76: error: dereferencing pointer to incomplete type

### sgimips GENERIC32_IP3x

crmfb.o: In function `crmfb_attach':
crmfb.c:(.text+0x2304): undefined reference to `ddc_read_edid'
crmfb.c:(.text+0x2304): relocation truncated to fit: R_MIPS_26 against `ddc_read_edid'
crmfb.c:(.text+0x234c): undefined reference to `edid_parse'
crmfb.c:(.text+0x234c): relocation truncated to fit: R_MIPS_26 against `edid_parse'
crmfb.c:(.text+0x2354): undefined reference to `edid_print'
crmfb.c:(.text+0x2354): relocation truncated to fit: R_MIPS_26 against `edid_print'

remove unnecessary iommuvar.h and psycho{reg,var}.h includes.

don't try to turn on parity or set the latency timer, or dump pci config
space for the pci bridge in sparc64_pci_enumerate_bus(). it doesn't work
on PCIe adapters. #if 0'ed for now, should move this into psycho/schizo.

branches: 1.68.2;
allow sparc64 spc_find_ino() to be NULL; schizo doesn't need it.

branches: 1.67.4;
Cope with up to four "interrupts" properties for a node.
Just use the first one for now. This should be improved.
OK: mrg@.

move the guts of pci_intr_map() into pci_machdep.c, and move the schizo
ign addition into a hook. also, don't double shift the ign in set_intr.

tested on U60 and SB2500.

convert pci_intr_map() into a chipset tag function pointer, and implement
the schizo version slightly differently.

pull out the schizo's IGN from the upaid, not the apparently broken device
ID register. from openbsd.

with this i appear to have valid working interrupts on the SB2500.

tested on U60 and SB2500.


XXX: we can probably kill (*spc_find_ino)() now that pci_intr_map() itself
XXX: is no longer a first class function.

give more debug output about mapping interrupts.

clean up some debug messages a little bit.

pci_make_tag:
- add assertion for checking spc_busnode initialization.

ofpci_make_tag():
don't write PCI_COMMAND_STATUS_REG with
PCI_COMMAND_(MEM|MASTER|IO)_ENABLE, or it in.
sparc_pci_childspace():
move psycho.c:get_childspace() into pci_machdep.c but rename it
psycho_alloc_bus_tag():
use M_ZERO

a few clean up/rearrange changes, somewhat inspired from openbsd's
schizo.c:

- rename pp_busnode{} -> spc_busnode{}, and move it to
sparc_pci_chipset{}

- move public pci functions pci_conf_read(), pci_conf_write() and
pci_intr_establish() into psycho.c, renaming them in the process,
and add pointers through sparc_pci_chipset{} to access them.
as part of this, add PDB_INTMAP and PDB_CONF debug options to
psycho.c

- move and rename sparc64 private pci_find_ino() into
sparc_pci_chipset{}, like the above

- update psycho_alloc_chipset() to set the new 4 function members


no functional change intended. XXX: not tested on cardbus.

branches: 1.59.4; 1.59.6;
pci_intr_setattr(), allows PCI interrupts to be marked MPSAFE on x86, and
other platforms if the code is added.

pci_intr_map(...)
pci_intr_setattr(pc, ih, PCI_INTR_MPSAFE, 1);
pci_intr_establish(...)

remove clause #3 from my license where there are no other
copyright holders involved.

branches: 1.57.2; 1.57.4; 1.57.6;
Change previous to use aprint_error_dev.

use device_xname()

branches: 1.55.6;
Do not panic when trying to disestablish a pci interrupt.

branches: 1.54.14; 1.54.20;
pci_find_ino(): Use interrupt pin from pci_attach_args when mapping interrupts.
Use *ihp as a pure output. Fix case where interrupt pin #D was incorrectly
mapped. Some cosmetic changes while here. Ok'd by martin.

branches: 1.53.4; 1.53.8; 1.53.10;
remove "struct cacheline" and the "cacheline" variable. (it was only set
in cpu_attach(), overwritten by multiple cpus, and only one member of the
12 or more were ever used again.) add an ecache_min_line_size to replace
this functionality.

clean up some non-SMP friendly code from cpu_attach():
- only reset the fpu state on the boot cpu
- don't bother printing "FPU version 0" always
- don't call getver() and ignore the result
- print the UPA id on the first cpu0: line

only print mp_tramp address #ifdef DEBUG.

reset the fpu state in cpu_hatch(), like cpu_attach() does for the boot cpu.


cpu_attach() is now free of SMP-unfriendly code.

Missing initializers

branches: 1.51.14; 1.51.16;
ANSIfication: u_intN_t -> uintN_t, use ANSI function declarations/definitions
instead of K&R ones.

branches: 1.50.2; 1.50.4; 1.50.6;
merge ktrace-lwp.

branches: 1.49.12;
make PCI devices attachable/detachable (as far as the particular drivers
allow), and allow to rescan a bus selectively (ie only the device/
function I'm looking at)

give the special PCI bus enumerator a sparc64_ prefix and make the
PCI_MACHDEP_ENUMERATE_BUS macro point to it

pci_find_ino: support pci-pci bridges.

Revert a part of rev. 1.44.

OF_getprop can get only the first "reg" property with a statically
allocated buffer, but prom_getprop cannot.
This should fix a "kernel trap 30: data access exception" when you
invoke a pcictl(8).

branches: 1.45.2;
Set prom_getprop args correctly.

Some more promlib interface conversions.

Two commas went missing in last commit.

Partial switch to the promlib interface.

For pci devices without ofw mapping find ino by pci bus/slot/device
information, from OpenBSD.

Nuke bcopy/bzero.

Cosmetic: move DEBUG defines.

__KERNEL_RCSID()

branches: 1.37.2;
When overwriting the cache line size configuration, do not use the magic
value 0x10, but instead calculate a line size that is properly aligned
to the IOMMU strbuf cache and our external cache line size.
From FreeBSD.

Move the bus speed query code into it's own function and make it default
to 33 MHz. Cosmetics, no functional change.

PCI-CardBus bridge support for sparc64:
- handle devices which has no OBP node.
- move PCI latency-timer initialization from pci_intr_map to
pci_enumerate_bus.
- make PCI bus free space extents for cardbus devices.
- fix PCI config space map size.
- some code integrations.

Remove the `flags' argument from bus_intr_establish().

Set the PCI latency timer for bus master devices.

Slightly different to the patch suggested by Takeshi Nakayama in PR 18451
(idea from FreeBSD).

Additional input from Jason Thorpe: do not hard code the bus frequency,
instead get it from OpenFirmware.

Turn on parity checking for the PCI bus.

branches: 1.31.2;
Make the PCI code actually work.

branches: 1.30.2;
* Add "pcitag_t *pba_bridgetag" to pci_attach_args. This is set to
NULL for root PCI busses. For busses behind a bridge, it points to
a persistent copy of the bridge's pcitag_t. This can be very useful
for machine-dependent PCI bus enumeration code.
* Implement a machine-dependent pci_enumerate_bus() for sparc64 which
uses OFW device nodes to enumerate the bus. When a PCI bus that is
behind a bridge is attached, pci_attach_hook() allocates a new PCI
chipset tag for the new bus and sets it's "curnode" to the OFW node
of the bridge. This is used as a starting point when enumerating
that bus. Root busses get the OFW node of the host bridge (psycho).
* Garbage-collect "ofpci" and "ofppb" from the sparc64 port.

Garbage-collect __PCI_DEV_FUNCORDER.

Implement pci_decompose_tag().

Add ofpci_make_tag().

Overhaul bus space.

bus_space_handle_t now holds an address and two ASIs, one for normal accesses
and one for streaming accesses. This allows to map individual handles
different ways, so some can use MMU bypass accesses and others use virtual
addresses. bus_space_map() will now create handles that use bypass accesses
unles BUS_SPACE_MAP_LINEAR is passed in. So only pass in BUS_SPACE_MAP_LINEAR
if you absolutely *need* to use bus_space_vaddr(). This removes at least one
extra level of indirection and should reduce TLB misses.

32-bit kernels have problems accessing 64-bit addresses, so they always use
virtual addresses.

branches: 1.25.4;
Fix the semantics of __PCI_DEV_FUNCORDER so that they're actually useful:
If __PCI_DEV_FUNCORDER is defined, don't do the song-and-dance to check if
a device is multi-function; machdep code is going to tell us exactly which
functions to probe.

Note this required changing how pci_func_devorder() works in the
sparc64 PCI machdep code; now the "curnode" is assumed to point
to the bus, rather than some function (typically 0) on the device,
just as pci_bus_devorder() makes that assumption.

All this should allow the PCI code to actually locate the second
HME device on a Sun Netra t1, which is at 3,1 -- previously, the
PCI code would have missed it because there is no device at 3,0.

(Sun deserves a brick to the head for this one -- this seems clearly
out of line with the PCI spec.)

in pci_conf_read() and pci_conf_write(), convert all printf()'s to only
occur if SPDB_CONF (and hence DEBUG). also convert the panic in
pci_conf_write() to a SPDB_CONF warning and a return.

this cleans up pcictl(8) support.

branches: 1.23.2;
Fix backwards compare. (From Jason Wright)

PCI overhaul.

branches: 1.21.2;
adapt to systems with out interrupt-map & interrupt-map-mask properties.

rework pci_attach_hook() to just deal with the INO, not the full
interrupt number. properly find interrupts for the E250. modify
pci_intr_map() accordingly. retire psycho_intr_map(). deal with
INO values upto 0x3f, not upto 0x32. restructure sabre_init() and
psycho_init() to be more similar, and display each psycho's IGN.
psycho_intr_establish() deals with INO upto 0x3f, values from 0x32
and higher get 0 for IPL.

tested on E250 & U5.

branches: 1.19.2;
use pci_bus_devorder() and pci_dev_funcorder()

Enable IO space mapping by default.
Fixes PR 11940.

Fix format warnings to make it compile again.

Change pci_intr_map to get interrupt source information from a "struct
pci_attach_args *" instead of from four separate parameters which in
all cases were extracted from the same "struct pci_attach_args".

This both simplifies the driver api, and allows for alternate PCI
interrupt mapping schemes, such as one using the tables described in
the Intel Multiprocessor Spec which describe interrupt wirings for
devices behind pci-pci bridges based on the device's location rather
the bridge's location.

Tested on alpha and i386; welcome to 1.5Q

On-board devices on psycho machines seem to have the `interrupt' property
in the parent bus format (i.e. an INO) rather than being represented as
an PCI interrupt line. Provide a hack to work around this in pci_attach_hook().

Use probeget() in pci_config_read() for the psycho, to avoid bus
faults when probing PCI space.

Add a `device class' interrupt level argument (from machine/intr.h)
to bus_interrupt_establish().

It's currently only used in sparc64/dev/psycho.c to assign a CPU interrupt
level to devices in PCI slots.

remove include of <vm/vm.h>. <vm/vm.h> -> <uvm/uvm_extern.h>

remove/move more mach vm header files:

<vm/pglist.h> -> <uvm/uvm_pglist.h>
<vm/vm_inherit.h> -> <uvm/uvm_inherit.h>
<vm/vm_kern.h> -> into <uvm/uvm_extern.h>
<vm/vm_object.h> -> nothing
<vm/vm_pager.h> -> into <uvm/uvm_pager.h>

also includes a bunch of <vm/vm_page.h> include removals (due to redudancy
with <vm/vm.h>), and a scattering of other similar headers.

branches: 1.10.2;
kill dead code.

Turn on PCI MEM and DMA.

Implement the more flexiable `evcnt' interface as discussed (briefly) on
tech-kern and now documented in evcnt(9).

branches: 1.7.2;
Use the PCI function code to distinguish simba bus A from bus B instead
of the current hack.

add a note about US IIi and PCI_INTERRUPT_LINE register

- store the ebus's parent so we can find the IOMMU
- use generic iommu routines in the ebus code (uses the above)
- add some more comments.

clean this up some.

branches: 1.3.2; 1.3.4;
Make pbrobeget() and probeset() work for 64-bit values as well. To do this
the arguments are changed so the address is first and the ASI second so we
can have the address in %o0:%o1 and not worry about unused registers.

Also a bit of copyright cleanup.

clean up a bit, implement pci_conf_{read,write}() with probe[gs]et() (not yet used).

PCI driver for the UltraSPARC. this only works on the Ultra5/10 machines
(`SUNW,sabre') for now, and it doesn't really quite work there yet anyway.
the bus space/dma code is cloned from the sbus driver. the IOMMU code also
is cloned from the sbus code, but separated out into iommu.c so that we can
share it with the sbus driver. hopefully, much of the bus space/dma code
can also be re-shared with the sbus driver and the ebus driver but for now
these copies will do.

support for the real UltraSPARC PCI (`SUNW,psycho') is unwritten, though
most of this code is shared with it.

we can probe PCI config space and try to configue devices, but interrupts
don't work yet...

Rename min/max -> uimin/uimax for better honesty.

These functions are defined on unsigned int. The generic name
min/max should not silently truncate to 32 bits on 64-bit systems.
This is purely a name change -- no functional change intended.

HOWEVER! Some subsystems have

#define min(a, b) ((a) < (b) ? (a) : (b))
#define max(a, b) ((a) > (b) ? (a) : (b))

even though our standard name for that is MIN/MAX. Although these
may invite multiple evaluation bugs, these do _not_ cause integer
truncation.

To avoid `fixing' these cases, I first changed the name in libkern,
and then compile-tested every file where min/max occurred in order to
confirm that it failed -- and thus confirm that nothing shadowed
min/max -- before changing it.

I have left a handful of bootloaders that are too annoying to
compile-test, and some dead code:

cobalt ews4800mips hp300 hppa ia64 luna68k vax
acorn32/if_ie.c (not included in any kernels)
macppc/if_gm.c (superseded by gem(4))

It should be easy to fix the fallout once identified -- this way of
doing things fails safe, and the goal here, after all, is to _avoid_
silent integer truncations, not introduce them.

Maybe one day we can reintroduce min/max as type-generic things that
never silently truncate. But we should avoid doing that for a while,
so that existing code has a chance to be detected by the compiler for
conversion to uimin/uimax without changing the semantics until we can
properly audit it all. (Who knows, maybe in some cases integer
truncation is actually intended!)

branches: 1.77.28;
make pci_intr_string and eisa_intr_string take a buffer and a length
instead of relying in local static storage.

kill sprintf

branches: 1.75.2;
split device_t/softc for all remaining drivers.
replace "struct device *" with "device_t".
use device_xname(), device_unit(), etc.

branches: 1.74.2; 1.74.12;
convert some open-coded SPDB_PROBE printf calls to use DPRINTF() macro.

#include <sys/bus.h> instead of <machine/bus.h>.

when looking for a device's node we want to start with the child nodes of the
root bus, not its peers

Neither pci_dma64_available(), pci_probe_device(), pci_mapreg_map(9),
pci_find_rom(), pci_intr_map(9), pci_enumerate_bus(), nor the match
predicate passed to pciide_compat_intr_establish() should ever modify
their pci_attach_args argument, so make their pci_attach_args arguments
const and deal with the fallout throughout the kernel.

For the most part, these changes add a 'const' where there was no
'const' before, however, some drivers and MD code used to modify
pci_attach_args. Now those drivers either copy their pci_attach_args
and modify the copy, or refrain from modifying pci_attach_args:

Xen: according to Manuel Bouyer, writing to pci_attach_args in
pci_intr_map() was a leftover from Xen 2. Probably a bug. I
stopped writing it. I have not tested this change.

siside(4): sis_hostbr_match() needlessly wrote to pci_attach_args.
Probably a bug. I use a temporary variable. I have not tested this
change.

slide(4): sl82c105_chip_map() overwrote the caller's pci_attach_args.
Probably a bug. Use a local pci_attach_args. I have not tested
this change.

viaide(4): via_sata_chip_map() and via_sata_chip_map_new() overwrote the
caller's pci_attach_args. Probably a bug. Make a local copy of the
caller's pci_attach_args and modify the copy. I have not tested
this change.

While I'm here, make pci_mapreg_submap() static.

With these changes in place, I have tested the compilation of these
kernels:

alpha GENERIC
amd64 GENERIC XEN3_DOM0
arc GENERIC
atari HADES MILAN-PCIIDE
bebox GENERIC
cats GENERIC
cobalt GENERIC
evbarm-eb NSLU2
evbarm-el ADI_BRH ARMADILLO9 CP3100 GEMINI GEMINI_MASTER GEMINI_SLAVE GUMSTIX
HDL_G IMX31LITE INTEGRATOR IQ31244 IQ80310 IQ80321 IXDP425 IXM1200
KUROBOX_PRO LUBBOCK MARVELL_NAS NAPPI SHEEVAPLUG SMDK2800 TEAMASA_NPWR
TEAMASA_NPWR_FC TS7200 TWINTAIL ZAO425
evbmips-el AP30 DBAU1500 DBAU1550 MALTA MERAKI MTX-1 OMSAL400 RB153 WGT624V3
evbmips64-el XLSATX
evbppc EV64260 MPC8536DS MPC8548CDS OPENBLOCKS200 OPENBLOCKS266
OPENBLOCKS266_OPT P2020RDB PMPPC RB800 WALNUT
hp700 GENERIC
i386 ALL XEN3_DOM0 XEN3_DOMU
ibmnws GENERIC
macppc GENERIC
mvmeppc GENERIC
netwinder GENERIC
ofppc GENERIC
prep GENERIC
sandpoint GENERIC
sgimips GENERIC32_IP2x
sparc GENERIC_SUN4U KRUPS
sparc64 GENERIC

As of Sun Apr 3 15:26:26 CDT 2011, I could not compile these kernels
with or without my patches in place:

### evbmips-el GDIUM

nbmake: nbmake: don't know how to make /home/dyoung/pristine-nbsd/src/sys/arch/mips/mips/softintr.c. Stop

### evbarm-el MPCSA_GENERIC
src/sys/arch/evbarm/conf/MPCSA_GENERIC:318: ds1672rtc*: unknown device `ds1672rtc'

### ia64 GENERIC

/tmp/genassym.28085/assym.c: In function 'f111':
/tmp/genassym.28085/assym.c:67: error: invalid application of 'sizeof' to incomplete type 'struct pcb'
/tmp/genassym.28085/assym.c:76: error: dereferencing pointer to incomplete type

### sgimips GENERIC32_IP3x

crmfb.o: In function `crmfb_attach':
crmfb.c:(.text+0x2304): undefined reference to `ddc_read_edid'
crmfb.c:(.text+0x2304): relocation truncated to fit: R_MIPS_26 against `ddc_read_edid'
crmfb.c:(.text+0x234c): undefined reference to `edid_parse'
crmfb.c:(.text+0x234c): relocation truncated to fit: R_MIPS_26 against `edid_parse'
crmfb.c:(.text+0x2354): undefined reference to `edid_print'
crmfb.c:(.text+0x2354): relocation truncated to fit: R_MIPS_26 against `edid_print'

remove unnecessary iommuvar.h and psycho{reg,var}.h includes.

don't try to turn on parity or set the latency timer, or dump pci config
space for the pci bridge in sparc64_pci_enumerate_bus(). it doesn't work
on PCIe adapters. #if 0'ed for now, should move this into psycho/schizo.

branches: 1.68.2;
allow sparc64 spc_find_ino() to be NULL; schizo doesn't need it.

branches: 1.67.4;
Cope with up to four "interrupts" properties for a node.
Just use the first one for now. This should be improved.
OK: mrg@.

move the guts of pci_intr_map() into pci_machdep.c, and move the schizo
ign addition into a hook. also, don't double shift the ign in set_intr.

tested on U60 and SB2500.

convert pci_intr_map() into a chipset tag function pointer, and implement
the schizo version slightly differently.

pull out the schizo's IGN from the upaid, not the apparently broken device
ID register. from openbsd.

with this i appear to have valid working interrupts on the SB2500.

tested on U60 and SB2500.


XXX: we can probably kill (*spc_find_ino)() now that pci_intr_map() itself
XXX: is no longer a first class function.

give more debug output about mapping interrupts.

clean up some debug messages a little bit.

pci_make_tag:
- add assertion for checking spc_busnode initialization.

ofpci_make_tag():
don't write PCI_COMMAND_STATUS_REG with
PCI_COMMAND_(MEM|MASTER|IO)_ENABLE, or it in.
sparc_pci_childspace():
move psycho.c:get_childspace() into pci_machdep.c but rename it
psycho_alloc_bus_tag():
use M_ZERO

a few clean up/rearrange changes, somewhat inspired from openbsd's
schizo.c:

- rename pp_busnode{} -> spc_busnode{}, and move it to
sparc_pci_chipset{}

- move public pci functions pci_conf_read(), pci_conf_write() and
pci_intr_establish() into psycho.c, renaming them in the process,
and add pointers through sparc_pci_chipset{} to access them.
as part of this, add PDB_INTMAP and PDB_CONF debug options to
psycho.c

- move and rename sparc64 private pci_find_ino() into
sparc_pci_chipset{}, like the above

- update psycho_alloc_chipset() to set the new 4 function members


no functional change intended. XXX: not tested on cardbus.

branches: 1.59.4; 1.59.6;
pci_intr_setattr(), allows PCI interrupts to be marked MPSAFE on x86, and
other platforms if the code is added.

pci_intr_map(...)
pci_intr_setattr(pc, ih, PCI_INTR_MPSAFE, 1);
pci_intr_establish(...)

remove clause #3 from my license where there are no other
copyright holders involved.

branches: 1.57.2; 1.57.4; 1.57.6;
Change previous to use aprint_error_dev.

use device_xname()

branches: 1.55.6;
Do not panic when trying to disestablish a pci interrupt.

branches: 1.54.14; 1.54.20;
pci_find_ino(): Use interrupt pin from pci_attach_args when mapping interrupts.
Use *ihp as a pure output. Fix case where interrupt pin #D was incorrectly
mapped. Some cosmetic changes while here. Ok'd by martin.

branches: 1.53.4; 1.53.8; 1.53.10;
remove "struct cacheline" and the "cacheline" variable. (it was only set
in cpu_attach(), overwritten by multiple cpus, and only one member of the
12 or more were ever used again.) add an ecache_min_line_size to replace
this functionality.

clean up some non-SMP friendly code from cpu_attach():
- only reset the fpu state on the boot cpu
- don't bother printing "FPU version 0" always
- don't call getver() and ignore the result
- print the UPA id on the first cpu0: line

only print mp_tramp address #ifdef DEBUG.

reset the fpu state in cpu_hatch(), like cpu_attach() does for the boot cpu.


cpu_attach() is now free of SMP-unfriendly code.

Missing initializers

branches: 1.51.14; 1.51.16;
ANSIfication: u_intN_t -> uintN_t, use ANSI function declarations/definitions
instead of K&R ones.

branches: 1.50.2; 1.50.4; 1.50.6;
merge ktrace-lwp.

branches: 1.49.12;
make PCI devices attachable/detachable (as far as the particular drivers
allow), and allow to rescan a bus selectively (ie only the device/
function I'm looking at)

give the special PCI bus enumerator a sparc64_ prefix and make the
PCI_MACHDEP_ENUMERATE_BUS macro point to it

pci_find_ino: support pci-pci bridges.

Revert a part of rev. 1.44.

OF_getprop can get only the first "reg" property with a statically
allocated buffer, but prom_getprop cannot.
This should fix a "kernel trap 30: data access exception" when you
invoke a pcictl(8).

branches: 1.45.2;
Set prom_getprop args correctly.

Some more promlib interface conversions.

Two commas went missing in last commit.

Partial switch to the promlib interface.

For pci devices without ofw mapping find ino by pci bus/slot/device
information, from OpenBSD.

Nuke bcopy/bzero.

Cosmetic: move DEBUG defines.

__KERNEL_RCSID()

branches: 1.37.2;
When overwriting the cache line size configuration, do not use the magic
value 0x10, but instead calculate a line size that is properly aligned
to the IOMMU strbuf cache and our external cache line size.
From FreeBSD.

Move the bus speed query code into it's own function and make it default
to 33 MHz. Cosmetics, no functional change.

PCI-CardBus bridge support for sparc64:
- handle devices which has no OBP node.
- move PCI latency-timer initialization from pci_intr_map to
pci_enumerate_bus.
- make PCI bus free space extents for cardbus devices.
- fix PCI config space map size.
- some code integrations.

Remove the `flags' argument from bus_intr_establish().

Set the PCI latency timer for bus master devices.

Slightly different to the patch suggested by Takeshi Nakayama in PR 18451
(idea from FreeBSD).

Additional input from Jason Thorpe: do not hard code the bus frequency,
instead get it from OpenFirmware.

Turn on parity checking for the PCI bus.

branches: 1.31.2;
Make the PCI code actually work.

branches: 1.30.2;
* Add "pcitag_t *pba_bridgetag" to pci_attach_args. This is set to
NULL for root PCI busses. For busses behind a bridge, it points to
a persistent copy of the bridge's pcitag_t. This can be very useful
for machine-dependent PCI bus enumeration code.
* Implement a machine-dependent pci_enumerate_bus() for sparc64 which
uses OFW device nodes to enumerate the bus. When a PCI bus that is
behind a bridge is attached, pci_attach_hook() allocates a new PCI
chipset tag for the new bus and sets it's "curnode" to the OFW node
of the bridge. This is used as a starting point when enumerating
that bus. Root busses get the OFW node of the host bridge (psycho).
* Garbage-collect "ofpci" and "ofppb" from the sparc64 port.

Garbage-collect __PCI_DEV_FUNCORDER.

Implement pci_decompose_tag().

Add ofpci_make_tag().

Overhaul bus space.

bus_space_handle_t now holds an address and two ASIs, one for normal accesses
and one for streaming accesses. This allows to map individual handles
different ways, so some can use MMU bypass accesses and others use virtual
addresses. bus_space_map() will now create handles that use bypass accesses
unles BUS_SPACE_MAP_LINEAR is passed in. So only pass in BUS_SPACE_MAP_LINEAR
if you absolutely *need* to use bus_space_vaddr(). This removes at least one
extra level of indirection and should reduce TLB misses.

32-bit kernels have problems accessing 64-bit addresses, so they always use
virtual addresses.

branches: 1.25.4;
Fix the semantics of __PCI_DEV_FUNCORDER so that they're actually useful:
If __PCI_DEV_FUNCORDER is defined, don't do the song-and-dance to check if
a device is multi-function; machdep code is going to tell us exactly which
functions to probe.

Note this required changing how pci_func_devorder() works in the
sparc64 PCI machdep code; now the "curnode" is assumed to point
to the bus, rather than some function (typically 0) on the device,
just as pci_bus_devorder() makes that assumption.

All this should allow the PCI code to actually locate the second
HME device on a Sun Netra t1, which is at 3,1 -- previously, the
PCI code would have missed it because there is no device at 3,0.

(Sun deserves a brick to the head for this one -- this seems clearly
out of line with the PCI spec.)