wsfb: add support for optional "devcmap" property

A hardware driver can supply a pointer to a 16x 32-bit array to override
the default rasops device colour map in the "devcmap" property.

genfb: Handle uninitialized softc in genfb_enable/disable_polling.

This can happen due to janky MD kludgerosity like x86
x86_genfb_ddb_trap_callback, which should really be cleaned up, but
at least this might help with the recursive traps we've been seeing
in syzbot.

genfb(4): Make internal parts of struct genfb_softc private.

This way the ABI has no ifdefs, so it has a chance to be usable in
modules. This also makes genfb much easier to maintain without
worrying about ABI breakage.

genfb(4): KASSERT(a && b) -> KASSERT(a); KASSERT(b)

Better diagnostics. No other functional change.

Fix GENFB_GLYPHCACHE with shadow fb.

Add helper to detach genfb console.

wsfb: Prefer wide fonts when EDID data is available.

To give us a better chance of picking a readable font, prefer fonts that
will render at least 3mm wide instead of picking the font that will be
closest to that size.

Add 10-bit pixel format support.

Merge thorpej-cfargs2.

branches: 1.82.8;
Merge thorpej-cfargs branch:

Simplify and make extensible the config_search() / config_found() /
config_attach() interfaces: rather than having different variants for
which arguments you want pass along, just have a single call that
takes a variadic list of tag-value arguments.

Adjust all call sites:
- Simplify wherever possible; don't pass along arguments that aren't
actually needed.
- Don't be explicit about what interface attribute is attaching if
the device only has one. (More simplification.)
- Add a config_probe() function to be used in indirect configuiration
situations, making is visibly easier to see when indirect config is
in play, and allowing for future change in semantics. (As of now,
this is just a wrapper around config_match(), but that is an
implementation detail.)

Remove unnecessary or redundant interface attributes where they're not
needed.

There are currently 5 "cfargs" defined:
- CFARG_SUBMATCH (submatch function for direct config)
- CFARG_SEARCH (search function for indirect config)
- CFARG_IATTR (interface attribte)
- CFARG_LOCATORS (locators array)
- CFARG_DEVHANDLE (devhandle_t - wraps OFW, ACPI, etc. handles)

...and a sentinel value CFARG_EOL.

Add some extra sanity checking to ensure that interface attributes
aren't ambiguous.

Use CFARG_DEVHANDLE in MI FDT, OFW, and ACPI code, and macppc and shark
ports to associate those device handles with device_t instance. This
will trickle trough to more places over time (need back-end for pre-OFW
Sun OBP; any others?).

branches: 1.81.2;
add optional glyph cache for genfb
enable with options GENFB_GLYPHCACHE=n
with n being the desired size of the cache in MB. Should be enough to cache
at least 900 glyphs in whatever video mode used in order to be effective
in 32bit per pixel that's about 1MB

remove outdated comment

Prefer printing the framebuffer's PA instead of VA. It's much more useful..

branches: 1.78.2;
Add "is_swapped" property which indicates 32-bpp framebuffer is
byte-swapped.

Revert rev 1.75; do not set WSFB_VRAM_IS_RAM flag bit. It indicates
memory type obtained by mmap. I just misunderstood its intention.

Thanks to jmcneill for pointing it out!

Remove stray TABs. No binary changes.

For WSDISPLAYIO_GET_FBINFO ioctl, set WSFB_VRAM_IS_RAM to fbi_flags
when shadow FB is used.

Update for proplib(3) API changes.

reduce stack usage in genfb_calc_hsize()

Get genfb's address offset correctly when the vaule >= 4G. OK's by jmcneill.

Tested on Intel BXNUC10I3FNK (Comet Lake U).

branches: 1.71.4;
Do not replay the msgbuf if we are booting silent or quiet.

branches: 1.70.2;
Fix bug introduced in rev. 1.69:
http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/dev/wsfb/genfb.c#rev1.69

is_bgr should be initialized to false. Otherwise, color becomes strange
for depths 24 and 32 unless backend explicitly set "is_bgr" property.

We support anti-aliasing for depth 2.
Use switch appropriately.

Notify size of shadow framebuffer if enabled.

branches: 1.67.2;
Enable anti-aliasing for depth 24.
Style.

sys/dev/rasops routines support anti-aliasing for depths 15 and 16.

If an EDID is available, attempt to use the screen size to influence font
selection by passing a "wantcols" value to rasops_init that is satisfied
by a font that is at least 3mm wide.

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: 1.63.2;
Fix cmap bounds checking.

branches: 1.62.2;
CID-1427771: Remove incorrect pre-condition (index cannot be 256!)

branches: 1.61.2;
remove checks for failure after memory allocation calls that cannot fail:

kmem_alloc() with KM_SLEEP
kmem_zalloc() with KM_SLEEP
percpu_alloc()
pserialize_create()
psref_class_create()

all of these paths include an assertion that the allocation has not failed,
so callers should not assert that again.

enable font loading and screen resizing

genfb(4): Enabling/Disabling shadowfb can be controlled via prop.

branches: 1.58.2; 1.58.4;
Fix splashscreen on resolutions where text is centered.

This commit was approved by macallan@

support 24bpp framebuffers

branches: 1.56.2;
handle WSDISPLAYIO_GET_EDID and WSDISPLAYIO_GET_FBINFO only if the bus
frontend doesn't ( as in, returns EPASSTHROUGH or doesn't register a
handler )
Now flags passed from things like bcm2835_genfb.c are actually seen by
xf86-video-wsfb again and automatically disabling shadowfb works again,
which results in a major speedup on this kind of hardware.

branches: 1.55.2;
Add enable/disable polling hooks to genfb_ops.

Do not printf(foobar()). Bad bad bad bad bad!

branches: 1.53.2;
Merge riastradh-drm2 to HEAD.

Provide a wsdisplay_accessops.pollc

This makes sure that the screen gets painted properly during early boot
input (e.g. boot -a) on the Raspberry PI.

support WSDISPLAYIO_GET_FBINFO

branches: 1.50.2; 1.50.8;
for WSDISPLAYIO_SMODE, if the bus ioctl handler returns EPASSTHROUGH, dont treat it as an error

aprint_verbose -> aprint_debug

branches: 1.48.2;
in 32 bit colour assume RGB ordering, allow override using a device property

branches: 1.47.2;
On 8-bit screens only enable font anti-aliasing when the cmap_callback
property is defined. It is needed for a suitable R3G3B2 color map.
Ok by releng@

use rasops_init(0, 0)

support anti-aliased fonts in 8 bit, generate an appropriate colour map

if the cursor row passed in a device property doesn't fit on the screen just
ignore it and clear the screen

support anti-aliased fonts in 32bit colour

branches: 1.42.2; 1.42.6;
use callbacks that match the ioctl() interface to control backlight

some steps to make multiple instances work:
- don't bail if we're not the console. If we get the right paramters there is no
reason not to work
- move wsdisplay_accessops into the softc
- call wsdisplay_cnattach only if we're the console

branches: 1.40.2;
fix build for LP32
thanks to uebayasi@

look for a 'virtual_address' property and use it as fb address

genfb_enable/disable_polling only matters if genfb is the console screen,
so make it a noop if it's not

call vcons_hard_switch to switch to the console screen when entering polling
mode, makes ddb-from-X work with genfb on x86

Fixed NULL-pointer dereferencing when a mode-callback is not defined.

add vcons_{enable,disable}_polling and genfb_{enable,disable}_polling
functions, to switch between intr and polling modes when VCONS_DRAW_INTR
is defined

add a 'setmode' callback to genfb and use it to setup write-combining
MTRRs on x86 whenever switching to WSDISPLAYIO_MODE_EMUL

- add support for using compressed images as splash images
- retire SPLASHSCREEN_PROGRESS and SPLASHSCREEN_IMAGE options

branches: 1.32.2; 1.32.4;
finish device_t-ification

add support for backlight control via machine dependent callbacks

make use of a shadow framebuffer optional, use VCONS_DONT_READ by default

branches: 1.29.2;
Restore splashscreen support with genfb.
genfb patch from Pierre Pronchery, other small changes to make it
compile by me.

Closes PR kern/42605.

XXX doesnt work in 8bit, probably a cmap issue

branches: 1.28.2;
Expose genfb_restore_palette and allow MD code to specify PMF callbacks.

call vcons_replay_msgbuf()

s/long/intptr_t/

sparc complains "cast to pointer from integer of different size"; cast the
int to (void *)(long).

If the bus glue provides an ioctl handler, don't return before we have a
chance to perform generic handling in the WSDISPLAYIO_SMODE case.

ok macallan

Use kmem_alloc instead of malloc for shadowfb

Print the physical address if available instead of a virtual address.

Add clear-screen and cursor-row properties to allow genfb to persist
screen state from the MD console driver.

Provide callbacks so MD code can explicitly prevent genfb from matching;
works around vga/genfb deathmatches.

Introduce genfb_ops genfb_borrow callback to allow the bus frontend to
lend mappings to drm.

Provide a stub genfb_cnattach for md code to call to give a hint to the
bus driver's match function to return a higher confidence (so it can
beat out vga(4), for example). genfb needs-flag so md code can determine
if the genfb driver is present.

branches: 1.17.8; 1.17.14;
Convert to new 2 clause license

branches: 1.16.2; 1.16.4;
use aprint_*_dev and device_xname

missing newline in message

Do not attach genfb when the frame buffer is not flagged as being console
(e.g. when a serial console is used).
Approved by macallan.

branches: 1.13.2; 1.13.6;
deal with a bug in the Sun Raptor's firmware which reports
linebytes == width even in 32bit colour

branches: 1.12.2; 1.12.6;
sprinkle \n

branches: 1.11.2;
don't panic if we're not finding all the properties we need. Just complain
quietly and return.

branches: 1.10.2; 1.10.6;
Clear the screen during initialization so that the first messages are not
printed over "unknown garbage" (e.g. the Mac OS desktop). Agreed with
macallan@.

Print some details of the framebuffer configuration during attach.
OK'ed by macallan@.

Properly initialize the palette registers by calling the colormap callback
on all of them. A typo made it only initialize half of them. OK'ed by
macallan@.

branches: 1.7.2;
use bool with prop_dictionary_get_bool()

branches: 1.6.2;
fix typo - fbsize is /height/ * stride. Doh.

add shadow fb support
now genfb is actually usable even on slow PCI framebuffers. Yay Jared :)

branches: 1.4.2; 1.4.4; 1.4.8;
add a callback to change the colour map
so now X with wsfb on top of genfb should look right

add WSDISPLAYIO_LINEBYTES support so we can run XFree86 with wsfb

branches: 1.2.2;
- move all memory mapping into the bus frontend
- get rid of macppc-isms like *_mmap() returning physical addresses

add genfb, a generic framebuffer console driver that's configured via
device properties - all it needs is width and height in pixels, stride in
bytes, depth in bits and the framebuffer address.
Features:
- unaccelerated fb console with colours and virtual consoles
- split into console handling and bus frontend
- so far there is only a PCI bus frontend which allows to mmap BARs belonging
to the graphics controller, others should be easy to add.

TODO:
- find an opaque way to change the palette ( probably pass a function
pointer as property ) since this would need to call some firmware interface
which is pretty much always MD
- find a clean way to mmap the framebuffer, just returning the physical
address works on macppc but not on other architectures. Problem is, that
we cannot rely on the address being a bus address - on macppc it is,
but for instance on sparc64 it's not.

genfb: Handle uninitialized softc in genfb_enable/disable_polling.

This can happen due to janky MD kludgerosity like x86
x86_genfb_ddb_trap_callback, which should really be cleaned up, but
at least this might help with the recursive traps we've been seeing
in syzbot.

genfb(4): Make internal parts of struct genfb_softc private.

This way the ABI has no ifdefs, so it has a chance to be usable in
modules. This also makes genfb much easier to maintain without
worrying about ABI breakage.

genfb(4): KASSERT(a && b) -> KASSERT(a); KASSERT(b)

Better diagnostics. No other functional change.

Fix GENFB_GLYPHCACHE with shadow fb.

Add helper to detach genfb console.

wsfb: Prefer wide fonts when EDID data is available.

To give us a better chance of picking a readable font, prefer fonts that
will render at least 3mm wide instead of picking the font that will be
closest to that size.

Add 10-bit pixel format support.

Merge thorpej-cfargs2.

branches: 1.82.8;
Merge thorpej-cfargs branch:

Simplify and make extensible the config_search() / config_found() /
config_attach() interfaces: rather than having different variants for
which arguments you want pass along, just have a single call that
takes a variadic list of tag-value arguments.

Adjust all call sites:
- Simplify wherever possible; don't pass along arguments that aren't
actually needed.
- Don't be explicit about what interface attribute is attaching if
the device only has one. (More simplification.)
- Add a config_probe() function to be used in indirect configuiration
situations, making is visibly easier to see when indirect config is
in play, and allowing for future change in semantics. (As of now,
this is just a wrapper around config_match(), but that is an
implementation detail.)

Remove unnecessary or redundant interface attributes where they're not
needed.

There are currently 5 "cfargs" defined:
- CFARG_SUBMATCH (submatch function for direct config)
- CFARG_SEARCH (search function for indirect config)
- CFARG_IATTR (interface attribte)
- CFARG_LOCATORS (locators array)
- CFARG_DEVHANDLE (devhandle_t - wraps OFW, ACPI, etc. handles)

...and a sentinel value CFARG_EOL.

Add some extra sanity checking to ensure that interface attributes
aren't ambiguous.

Use CFARG_DEVHANDLE in MI FDT, OFW, and ACPI code, and macppc and shark
ports to associate those device handles with device_t instance. This
will trickle trough to more places over time (need back-end for pre-OFW
Sun OBP; any others?).

branches: 1.81.2;
add optional glyph cache for genfb
enable with options GENFB_GLYPHCACHE=n
with n being the desired size of the cache in MB. Should be enough to cache
at least 900 glyphs in whatever video mode used in order to be effective
in 32bit per pixel that's about 1MB

remove outdated comment

Prefer printing the framebuffer's PA instead of VA. It's much more useful..

branches: 1.78.2;
Add "is_swapped" property which indicates 32-bpp framebuffer is
byte-swapped.

Revert rev 1.75; do not set WSFB_VRAM_IS_RAM flag bit. It indicates
memory type obtained by mmap. I just misunderstood its intention.

Thanks to jmcneill for pointing it out!

Remove stray TABs. No binary changes.

For WSDISPLAYIO_GET_FBINFO ioctl, set WSFB_VRAM_IS_RAM to fbi_flags
when shadow FB is used.

Update for proplib(3) API changes.

reduce stack usage in genfb_calc_hsize()

Get genfb's address offset correctly when the vaule >= 4G. OK's by jmcneill.

Tested on Intel BXNUC10I3FNK (Comet Lake U).

branches: 1.71.4;
Do not replay the msgbuf if we are booting silent or quiet.

branches: 1.70.2;
Fix bug introduced in rev. 1.69:
http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/dev/wsfb/genfb.c#rev1.69

is_bgr should be initialized to false. Otherwise, color becomes strange
for depths 24 and 32 unless backend explicitly set "is_bgr" property.

We support anti-aliasing for depth 2.
Use switch appropriately.

Notify size of shadow framebuffer if enabled.

branches: 1.67.2;
Enable anti-aliasing for depth 24.
Style.

sys/dev/rasops routines support anti-aliasing for depths 15 and 16.

If an EDID is available, attempt to use the screen size to influence font
selection by passing a "wantcols" value to rasops_init that is satisfied
by a font that is at least 3mm wide.

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: 1.63.2;
Fix cmap bounds checking.

branches: 1.62.2;
CID-1427771: Remove incorrect pre-condition (index cannot be 256!)

branches: 1.61.2;
remove checks for failure after memory allocation calls that cannot fail:

kmem_alloc() with KM_SLEEP
kmem_zalloc() with KM_SLEEP
percpu_alloc()
pserialize_create()
psref_class_create()

all of these paths include an assertion that the allocation has not failed,
so callers should not assert that again.

enable font loading and screen resizing

genfb(4): Enabling/Disabling shadowfb can be controlled via prop.

branches: 1.58.2; 1.58.4;
Fix splashscreen on resolutions where text is centered.

This commit was approved by macallan@

support 24bpp framebuffers

branches: 1.56.2;
handle WSDISPLAYIO_GET_EDID and WSDISPLAYIO_GET_FBINFO only if the bus
frontend doesn't ( as in, returns EPASSTHROUGH or doesn't register a
handler )
Now flags passed from things like bcm2835_genfb.c are actually seen by
xf86-video-wsfb again and automatically disabling shadowfb works again,
which results in a major speedup on this kind of hardware.

branches: 1.55.2;
Add enable/disable polling hooks to genfb_ops.

Do not printf(foobar()). Bad bad bad bad bad!

branches: 1.53.2;
Merge riastradh-drm2 to HEAD.

Provide a wsdisplay_accessops.pollc

This makes sure that the screen gets painted properly during early boot
input (e.g. boot -a) on the Raspberry PI.

support WSDISPLAYIO_GET_FBINFO

branches: 1.50.2; 1.50.8;
for WSDISPLAYIO_SMODE, if the bus ioctl handler returns EPASSTHROUGH, dont treat it as an error

aprint_verbose -> aprint_debug

branches: 1.48.2;
in 32 bit colour assume RGB ordering, allow override using a device property

branches: 1.47.2;
On 8-bit screens only enable font anti-aliasing when the cmap_callback
property is defined. It is needed for a suitable R3G3B2 color map.
Ok by releng@

use rasops_init(0, 0)

support anti-aliased fonts in 8 bit, generate an appropriate colour map

if the cursor row passed in a device property doesn't fit on the screen just
ignore it and clear the screen

support anti-aliased fonts in 32bit colour

branches: 1.42.2; 1.42.6;
use callbacks that match the ioctl() interface to control backlight

some steps to make multiple instances work:
- don't bail if we're not the console. If we get the right paramters there is no
reason not to work
- move wsdisplay_accessops into the softc
- call wsdisplay_cnattach only if we're the console

branches: 1.40.2;
fix build for LP32
thanks to uebayasi@

look for a 'virtual_address' property and use it as fb address

genfb_enable/disable_polling only matters if genfb is the console screen,
so make it a noop if it's not

call vcons_hard_switch to switch to the console screen when entering polling
mode, makes ddb-from-X work with genfb on x86

Fixed NULL-pointer dereferencing when a mode-callback is not defined.

add vcons_{enable,disable}_polling and genfb_{enable,disable}_polling
functions, to switch between intr and polling modes when VCONS_DRAW_INTR
is defined

add a 'setmode' callback to genfb and use it to setup write-combining
MTRRs on x86 whenever switching to WSDISPLAYIO_MODE_EMUL

- add support for using compressed images as splash images
- retire SPLASHSCREEN_PROGRESS and SPLASHSCREEN_IMAGE options

branches: 1.32.2; 1.32.4;
finish device_t-ification

add support for backlight control via machine dependent callbacks

make use of a shadow framebuffer optional, use VCONS_DONT_READ by default

branches: 1.29.2;
Restore splashscreen support with genfb.
genfb patch from Pierre Pronchery, other small changes to make it
compile by me.

Closes PR kern/42605.

XXX doesnt work in 8bit, probably a cmap issue

branches: 1.28.2;
Expose genfb_restore_palette and allow MD code to specify PMF callbacks.

call vcons_replay_msgbuf()

s/long/intptr_t/

sparc complains "cast to pointer from integer of different size"; cast the
int to (void *)(long).

If the bus glue provides an ioctl handler, don't return before we have a
chance to perform generic handling in the WSDISPLAYIO_SMODE case.

ok macallan

Use kmem_alloc instead of malloc for shadowfb

Print the physical address if available instead of a virtual address.

Add clear-screen and cursor-row properties to allow genfb to persist
screen state from the MD console driver.

Provide callbacks so MD code can explicitly prevent genfb from matching;
works around vga/genfb deathmatches.

Introduce genfb_ops genfb_borrow callback to allow the bus frontend to
lend mappings to drm.

Provide a stub genfb_cnattach for md code to call to give a hint to the
bus driver's match function to return a higher confidence (so it can
beat out vga(4), for example). genfb needs-flag so md code can determine
if the genfb driver is present.

branches: 1.17.8; 1.17.14;
Convert to new 2 clause license

branches: 1.16.2; 1.16.4;
use aprint_*_dev and device_xname

missing newline in message

Do not attach genfb when the frame buffer is not flagged as being console
(e.g. when a serial console is used).
Approved by macallan.

branches: 1.13.2; 1.13.6;
deal with a bug in the Sun Raptor's firmware which reports
linebytes == width even in 32bit colour

branches: 1.12.2; 1.12.6;
sprinkle \n

branches: 1.11.2;
don't panic if we're not finding all the properties we need. Just complain
quietly and return.

branches: 1.10.2; 1.10.6;
Clear the screen during initialization so that the first messages are not
printed over "unknown garbage" (e.g. the Mac OS desktop). Agreed with
macallan@.

Print some details of the framebuffer configuration during attach.
OK'ed by macallan@.

Properly initialize the palette registers by calling the colormap callback
on all of them. A typo made it only initialize half of them. OK'ed by
macallan@.

branches: 1.7.2;
use bool with prop_dictionary_get_bool()

branches: 1.6.2;
fix typo - fbsize is /height/ * stride. Doh.

add shadow fb support
now genfb is actually usable even on slow PCI framebuffers. Yay Jared :)

branches: 1.4.2; 1.4.4; 1.4.8;
add a callback to change the colour map
so now X with wsfb on top of genfb should look right

add WSDISPLAYIO_LINEBYTES support so we can run XFree86 with wsfb

branches: 1.2.2;
- move all memory mapping into the bus frontend
- get rid of macppc-isms like *_mmap() returning physical addresses

add genfb, a generic framebuffer console driver that's configured via
device properties - all it needs is width and height in pixels, stride in
bytes, depth in bits and the framebuffer address.
Features:
- unaccelerated fb console with colours and virtual consoles
- split into console handling and bus frontend
- so far there is only a PCI bus frontend which allows to mmap BARs belonging
to the graphics controller, others should be easy to add.

TODO:
- find an opaque way to change the palette ( probably pass a function
pointer as property ) since this would need to call some firmware interface
which is pretty much always MD
- find a clean way to mmap the framebuffer, just returning the physical
address works on macppc but not on other architectures. Problem is, that
we cannot rely on the address being a bus address - on macppc it is,
but for instance on sparc64 it's not.

genfb(4): Make internal parts of struct genfb_softc private.

This way the ABI has no ifdefs, so it has a chance to be usable in
modules. This also makes genfb much easier to maintain without
worrying about ABI breakage.

genfb(4): KASSERT(a && b) -> KASSERT(a); KASSERT(b)

Better diagnostics. No other functional change.

Fix GENFB_GLYPHCACHE with shadow fb.

Add helper to detach genfb console.

wsfb: Prefer wide fonts when EDID data is available.

To give us a better chance of picking a readable font, prefer fonts that
will render at least 3mm wide instead of picking the font that will be
closest to that size.

Add 10-bit pixel format support.

Merge thorpej-cfargs2.

branches: 1.82.8;
Merge thorpej-cfargs branch:

Simplify and make extensible the config_search() / config_found() /
config_attach() interfaces: rather than having different variants for
which arguments you want pass along, just have a single call that
takes a variadic list of tag-value arguments.

Adjust all call sites:
- Simplify wherever possible; don't pass along arguments that aren't
actually needed.
- Don't be explicit about what interface attribute is attaching if
the device only has one. (More simplification.)
- Add a config_probe() function to be used in indirect configuiration
situations, making is visibly easier to see when indirect config is
in play, and allowing for future change in semantics. (As of now,
this is just a wrapper around config_match(), but that is an
implementation detail.)

Remove unnecessary or redundant interface attributes where they're not
needed.

There are currently 5 "cfargs" defined:
- CFARG_SUBMATCH (submatch function for direct config)
- CFARG_SEARCH (search function for indirect config)
- CFARG_IATTR (interface attribte)
- CFARG_LOCATORS (locators array)
- CFARG_DEVHANDLE (devhandle_t - wraps OFW, ACPI, etc. handles)

...and a sentinel value CFARG_EOL.

Add some extra sanity checking to ensure that interface attributes
aren't ambiguous.

Use CFARG_DEVHANDLE in MI FDT, OFW, and ACPI code, and macppc and shark
ports to associate those device handles with device_t instance. This
will trickle trough to more places over time (need back-end for pre-OFW
Sun OBP; any others?).

branches: 1.81.2;
add optional glyph cache for genfb
enable with options GENFB_GLYPHCACHE=n
with n being the desired size of the cache in MB. Should be enough to cache
at least 900 glyphs in whatever video mode used in order to be effective
in 32bit per pixel that's about 1MB

remove outdated comment

Prefer printing the framebuffer's PA instead of VA. It's much more useful..

branches: 1.78.2;
Add "is_swapped" property which indicates 32-bpp framebuffer is
byte-swapped.

Revert rev 1.75; do not set WSFB_VRAM_IS_RAM flag bit. It indicates
memory type obtained by mmap. I just misunderstood its intention.

Thanks to jmcneill for pointing it out!

Remove stray TABs. No binary changes.

For WSDISPLAYIO_GET_FBINFO ioctl, set WSFB_VRAM_IS_RAM to fbi_flags
when shadow FB is used.

Update for proplib(3) API changes.

reduce stack usage in genfb_calc_hsize()

Get genfb's address offset correctly when the vaule >= 4G. OK's by jmcneill.

Tested on Intel BXNUC10I3FNK (Comet Lake U).

branches: 1.71.4;
Do not replay the msgbuf if we are booting silent or quiet.

branches: 1.70.2;
Fix bug introduced in rev. 1.69:
http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/dev/wsfb/genfb.c#rev1.69

is_bgr should be initialized to false. Otherwise, color becomes strange
for depths 24 and 32 unless backend explicitly set "is_bgr" property.

We support anti-aliasing for depth 2.
Use switch appropriately.

Notify size of shadow framebuffer if enabled.

branches: 1.67.2;
Enable anti-aliasing for depth 24.
Style.

sys/dev/rasops routines support anti-aliasing for depths 15 and 16.

If an EDID is available, attempt to use the screen size to influence font
selection by passing a "wantcols" value to rasops_init that is satisfied
by a font that is at least 3mm wide.

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: 1.63.2;
Fix cmap bounds checking.

branches: 1.62.2;
CID-1427771: Remove incorrect pre-condition (index cannot be 256!)

branches: 1.61.2;
remove checks for failure after memory allocation calls that cannot fail:

kmem_alloc() with KM_SLEEP
kmem_zalloc() with KM_SLEEP
percpu_alloc()
pserialize_create()
psref_class_create()

all of these paths include an assertion that the allocation has not failed,
so callers should not assert that again.

enable font loading and screen resizing

genfb(4): Enabling/Disabling shadowfb can be controlled via prop.

branches: 1.58.2; 1.58.4;
Fix splashscreen on resolutions where text is centered.

This commit was approved by macallan@

support 24bpp framebuffers

branches: 1.56.2;
handle WSDISPLAYIO_GET_EDID and WSDISPLAYIO_GET_FBINFO only if the bus
frontend doesn't ( as in, returns EPASSTHROUGH or doesn't register a
handler )
Now flags passed from things like bcm2835_genfb.c are actually seen by
xf86-video-wsfb again and automatically disabling shadowfb works again,
which results in a major speedup on this kind of hardware.

branches: 1.55.2;
Add enable/disable polling hooks to genfb_ops.

Do not printf(foobar()). Bad bad bad bad bad!

branches: 1.53.2;
Merge riastradh-drm2 to HEAD.

Provide a wsdisplay_accessops.pollc

This makes sure that the screen gets painted properly during early boot
input (e.g. boot -a) on the Raspberry PI.

support WSDISPLAYIO_GET_FBINFO

branches: 1.50.2; 1.50.8;
for WSDISPLAYIO_SMODE, if the bus ioctl handler returns EPASSTHROUGH, dont treat it as an error

aprint_verbose -> aprint_debug

branches: 1.48.2;
in 32 bit colour assume RGB ordering, allow override using a device property

branches: 1.47.2;
On 8-bit screens only enable font anti-aliasing when the cmap_callback
property is defined. It is needed for a suitable R3G3B2 color map.
Ok by releng@

use rasops_init(0, 0)

support anti-aliased fonts in 8 bit, generate an appropriate colour map

if the cursor row passed in a device property doesn't fit on the screen just
ignore it and clear the screen

support anti-aliased fonts in 32bit colour

branches: 1.42.2; 1.42.6;
use callbacks that match the ioctl() interface to control backlight

some steps to make multiple instances work:
- don't bail if we're not the console. If we get the right paramters there is no
reason not to work
- move wsdisplay_accessops into the softc
- call wsdisplay_cnattach only if we're the console

branches: 1.40.2;
fix build for LP32
thanks to uebayasi@

look for a 'virtual_address' property and use it as fb address

genfb_enable/disable_polling only matters if genfb is the console screen,
so make it a noop if it's not

call vcons_hard_switch to switch to the console screen when entering polling
mode, makes ddb-from-X work with genfb on x86

Fixed NULL-pointer dereferencing when a mode-callback is not defined.

add vcons_{enable,disable}_polling and genfb_{enable,disable}_polling
functions, to switch between intr and polling modes when VCONS_DRAW_INTR
is defined

add a 'setmode' callback to genfb and use it to setup write-combining
MTRRs on x86 whenever switching to WSDISPLAYIO_MODE_EMUL

- add support for using compressed images as splash images
- retire SPLASHSCREEN_PROGRESS and SPLASHSCREEN_IMAGE options

branches: 1.32.2; 1.32.4;
finish device_t-ification

add support for backlight control via machine dependent callbacks

make use of a shadow framebuffer optional, use VCONS_DONT_READ by default

branches: 1.29.2;
Restore splashscreen support with genfb.
genfb patch from Pierre Pronchery, other small changes to make it
compile by me.

Closes PR kern/42605.

XXX doesnt work in 8bit, probably a cmap issue

branches: 1.28.2;
Expose genfb_restore_palette and allow MD code to specify PMF callbacks.

call vcons_replay_msgbuf()

s/long/intptr_t/

sparc complains "cast to pointer from integer of different size"; cast the
int to (void *)(long).

If the bus glue provides an ioctl handler, don't return before we have a
chance to perform generic handling in the WSDISPLAYIO_SMODE case.

ok macallan

Use kmem_alloc instead of malloc for shadowfb

Print the physical address if available instead of a virtual address.

Add clear-screen and cursor-row properties to allow genfb to persist
screen state from the MD console driver.

Provide callbacks so MD code can explicitly prevent genfb from matching;
works around vga/genfb deathmatches.

Introduce genfb_ops genfb_borrow callback to allow the bus frontend to
lend mappings to drm.

Provide a stub genfb_cnattach for md code to call to give a hint to the
bus driver's match function to return a higher confidence (so it can
beat out vga(4), for example). genfb needs-flag so md code can determine
if the genfb driver is present.

branches: 1.17.8; 1.17.14;
Convert to new 2 clause license

branches: 1.16.2; 1.16.4;
use aprint_*_dev and device_xname

missing newline in message

Do not attach genfb when the frame buffer is not flagged as being console
(e.g. when a serial console is used).
Approved by macallan.

branches: 1.13.2; 1.13.6;
deal with a bug in the Sun Raptor's firmware which reports
linebytes == width even in 32bit colour

branches: 1.12.2; 1.12.6;
sprinkle \n

branches: 1.11.2;
don't panic if we're not finding all the properties we need. Just complain
quietly and return.

branches: 1.10.2; 1.10.6;
Clear the screen during initialization so that the first messages are not
printed over "unknown garbage" (e.g. the Mac OS desktop). Agreed with
macallan@.

Print some details of the framebuffer configuration during attach.
OK'ed by macallan@.

Properly initialize the palette registers by calling the colormap callback
on all of them. A typo made it only initialize half of them. OK'ed by
macallan@.

branches: 1.7.2;
use bool with prop_dictionary_get_bool()

branches: 1.6.2;
fix typo - fbsize is /height/ * stride. Doh.

add shadow fb support
now genfb is actually usable even on slow PCI framebuffers. Yay Jared :)

branches: 1.4.2; 1.4.4; 1.4.8;
add a callback to change the colour map
so now X with wsfb on top of genfb should look right

add WSDISPLAYIO_LINEBYTES support so we can run XFree86 with wsfb

branches: 1.2.2;
- move all memory mapping into the bus frontend
- get rid of macppc-isms like *_mmap() returning physical addresses

add genfb, a generic framebuffer console driver that's configured via
device properties - all it needs is width and height in pixels, stride in
bytes, depth in bits and the framebuffer address.
Features:
- unaccelerated fb console with colours and virtual consoles
- split into console handling and bus frontend
- so far there is only a PCI bus frontend which allows to mmap BARs belonging
to the graphics controller, others should be easy to add.

TODO:
- find an opaque way to change the palette ( probably pass a function
pointer as property ) since this would need to call some firmware interface
which is pretty much always MD
- find a clean way to mmap the framebuffer, just returning the physical
address works on macppc but not on other architectures. Problem is, that
we cannot rely on the address being a bus address - on macppc it is,
but for instance on sparc64 it's not.

Fix GENFB_GLYPHCACHE with shadow fb.

Add helper to detach genfb console.

wsfb: Prefer wide fonts when EDID data is available.

To give us a better chance of picking a readable font, prefer fonts that
will render at least 3mm wide instead of picking the font that will be
closest to that size.

Add 10-bit pixel format support.

Merge thorpej-cfargs2.

branches: 1.82.8;
Merge thorpej-cfargs branch:

Simplify and make extensible the config_search() / config_found() /
config_attach() interfaces: rather than having different variants for
which arguments you want pass along, just have a single call that
takes a variadic list of tag-value arguments.

Adjust all call sites:
- Simplify wherever possible; don't pass along arguments that aren't
actually needed.
- Don't be explicit about what interface attribute is attaching if
the device only has one. (More simplification.)
- Add a config_probe() function to be used in indirect configuiration
situations, making is visibly easier to see when indirect config is
in play, and allowing for future change in semantics. (As of now,
this is just a wrapper around config_match(), but that is an
implementation detail.)

Remove unnecessary or redundant interface attributes where they're not
needed.

There are currently 5 "cfargs" defined:
- CFARG_SUBMATCH (submatch function for direct config)
- CFARG_SEARCH (search function for indirect config)
- CFARG_IATTR (interface attribte)
- CFARG_LOCATORS (locators array)
- CFARG_DEVHANDLE (devhandle_t - wraps OFW, ACPI, etc. handles)

...and a sentinel value CFARG_EOL.

Add some extra sanity checking to ensure that interface attributes
aren't ambiguous.

Use CFARG_DEVHANDLE in MI FDT, OFW, and ACPI code, and macppc and shark
ports to associate those device handles with device_t instance. This
will trickle trough to more places over time (need back-end for pre-OFW
Sun OBP; any others?).

branches: 1.81.2;
add optional glyph cache for genfb
enable with options GENFB_GLYPHCACHE=n
with n being the desired size of the cache in MB. Should be enough to cache
at least 900 glyphs in whatever video mode used in order to be effective
in 32bit per pixel that's about 1MB

remove outdated comment

Prefer printing the framebuffer's PA instead of VA. It's much more useful..

branches: 1.78.2;
Add "is_swapped" property which indicates 32-bpp framebuffer is
byte-swapped.

Revert rev 1.75; do not set WSFB_VRAM_IS_RAM flag bit. It indicates
memory type obtained by mmap. I just misunderstood its intention.

Thanks to jmcneill for pointing it out!

Remove stray TABs. No binary changes.

For WSDISPLAYIO_GET_FBINFO ioctl, set WSFB_VRAM_IS_RAM to fbi_flags
when shadow FB is used.

Update for proplib(3) API changes.

reduce stack usage in genfb_calc_hsize()

Get genfb's address offset correctly when the vaule >= 4G. OK's by jmcneill.

Tested on Intel BXNUC10I3FNK (Comet Lake U).

branches: 1.71.4;
Do not replay the msgbuf if we are booting silent or quiet.

branches: 1.70.2;
Fix bug introduced in rev. 1.69:
http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/dev/wsfb/genfb.c#rev1.69

is_bgr should be initialized to false. Otherwise, color becomes strange
for depths 24 and 32 unless backend explicitly set "is_bgr" property.

We support anti-aliasing for depth 2.
Use switch appropriately.

Notify size of shadow framebuffer if enabled.

branches: 1.67.2;
Enable anti-aliasing for depth 24.
Style.

sys/dev/rasops routines support anti-aliasing for depths 15 and 16.

If an EDID is available, attempt to use the screen size to influence font
selection by passing a "wantcols" value to rasops_init that is satisfied
by a font that is at least 3mm wide.

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: 1.63.2;
Fix cmap bounds checking.

branches: 1.62.2;
CID-1427771: Remove incorrect pre-condition (index cannot be 256!)

branches: 1.61.2;
remove checks for failure after memory allocation calls that cannot fail:

kmem_alloc() with KM_SLEEP
kmem_zalloc() with KM_SLEEP
percpu_alloc()
pserialize_create()
psref_class_create()

all of these paths include an assertion that the allocation has not failed,
so callers should not assert that again.

enable font loading and screen resizing

genfb(4): Enabling/Disabling shadowfb can be controlled via prop.

branches: 1.58.2; 1.58.4;
Fix splashscreen on resolutions where text is centered.

This commit was approved by macallan@

support 24bpp framebuffers

branches: 1.56.2;
handle WSDISPLAYIO_GET_EDID and WSDISPLAYIO_GET_FBINFO only if the bus
frontend doesn't ( as in, returns EPASSTHROUGH or doesn't register a
handler )
Now flags passed from things like bcm2835_genfb.c are actually seen by
xf86-video-wsfb again and automatically disabling shadowfb works again,
which results in a major speedup on this kind of hardware.

branches: 1.55.2;
Add enable/disable polling hooks to genfb_ops.

Do not printf(foobar()). Bad bad bad bad bad!

branches: 1.53.2;
Merge riastradh-drm2 to HEAD.

Provide a wsdisplay_accessops.pollc

This makes sure that the screen gets painted properly during early boot
input (e.g. boot -a) on the Raspberry PI.

support WSDISPLAYIO_GET_FBINFO

branches: 1.50.2; 1.50.8;
for WSDISPLAYIO_SMODE, if the bus ioctl handler returns EPASSTHROUGH, dont treat it as an error

aprint_verbose -> aprint_debug

branches: 1.48.2;
in 32 bit colour assume RGB ordering, allow override using a device property

branches: 1.47.2;
On 8-bit screens only enable font anti-aliasing when the cmap_callback
property is defined. It is needed for a suitable R3G3B2 color map.
Ok by releng@

use rasops_init(0, 0)

support anti-aliased fonts in 8 bit, generate an appropriate colour map

if the cursor row passed in a device property doesn't fit on the screen just
ignore it and clear the screen

support anti-aliased fonts in 32bit colour

branches: 1.42.2; 1.42.6;
use callbacks that match the ioctl() interface to control backlight

some steps to make multiple instances work:
- don't bail if we're not the console. If we get the right paramters there is no
reason not to work
- move wsdisplay_accessops into the softc
- call wsdisplay_cnattach only if we're the console

branches: 1.40.2;
fix build for LP32
thanks to uebayasi@

look for a 'virtual_address' property and use it as fb address

genfb_enable/disable_polling only matters if genfb is the console screen,
so make it a noop if it's not

call vcons_hard_switch to switch to the console screen when entering polling
mode, makes ddb-from-X work with genfb on x86

Fixed NULL-pointer dereferencing when a mode-callback is not defined.

add vcons_{enable,disable}_polling and genfb_{enable,disable}_polling
functions, to switch between intr and polling modes when VCONS_DRAW_INTR
is defined

add a 'setmode' callback to genfb and use it to setup write-combining
MTRRs on x86 whenever switching to WSDISPLAYIO_MODE_EMUL

- add support for using compressed images as splash images
- retire SPLASHSCREEN_PROGRESS and SPLASHSCREEN_IMAGE options

branches: 1.32.2; 1.32.4;
finish device_t-ification

add support for backlight control via machine dependent callbacks

make use of a shadow framebuffer optional, use VCONS_DONT_READ by default

branches: 1.29.2;
Restore splashscreen support with genfb.
genfb patch from Pierre Pronchery, other small changes to make it
compile by me.

Closes PR kern/42605.

XXX doesnt work in 8bit, probably a cmap issue

branches: 1.28.2;
Expose genfb_restore_palette and allow MD code to specify PMF callbacks.

call vcons_replay_msgbuf()

s/long/intptr_t/

sparc complains "cast to pointer from integer of different size"; cast the
int to (void *)(long).

If the bus glue provides an ioctl handler, don't return before we have a
chance to perform generic handling in the WSDISPLAYIO_SMODE case.

ok macallan

Use kmem_alloc instead of malloc for shadowfb

Print the physical address if available instead of a virtual address.

Add clear-screen and cursor-row properties to allow genfb to persist
screen state from the MD console driver.

Provide callbacks so MD code can explicitly prevent genfb from matching;
works around vga/genfb deathmatches.

Introduce genfb_ops genfb_borrow callback to allow the bus frontend to
lend mappings to drm.

Provide a stub genfb_cnattach for md code to call to give a hint to the
bus driver's match function to return a higher confidence (so it can
beat out vga(4), for example). genfb needs-flag so md code can determine
if the genfb driver is present.

branches: 1.17.8; 1.17.14;
Convert to new 2 clause license

branches: 1.16.2; 1.16.4;
use aprint_*_dev and device_xname

missing newline in message

Do not attach genfb when the frame buffer is not flagged as being console
(e.g. when a serial console is used).
Approved by macallan.

branches: 1.13.2; 1.13.6;
deal with a bug in the Sun Raptor's firmware which reports
linebytes == width even in 32bit colour

branches: 1.12.2; 1.12.6;
sprinkle \n

branches: 1.11.2;
don't panic if we're not finding all the properties we need. Just complain
quietly and return.

branches: 1.10.2; 1.10.6;
Clear the screen during initialization so that the first messages are not
printed over "unknown garbage" (e.g. the Mac OS desktop). Agreed with
macallan@.

Print some details of the framebuffer configuration during attach.
OK'ed by macallan@.

Properly initialize the palette registers by calling the colormap callback
on all of them. A typo made it only initialize half of them. OK'ed by
macallan@.

branches: 1.7.2;
use bool with prop_dictionary_get_bool()

branches: 1.6.2;
fix typo - fbsize is /height/ * stride. Doh.

add shadow fb support
now genfb is actually usable even on slow PCI framebuffers. Yay Jared :)

branches: 1.4.2; 1.4.4; 1.4.8;
add a callback to change the colour map
so now X with wsfb on top of genfb should look right

add WSDISPLAYIO_LINEBYTES support so we can run XFree86 with wsfb

branches: 1.2.2;
- move all memory mapping into the bus frontend
- get rid of macppc-isms like *_mmap() returning physical addresses

add genfb, a generic framebuffer console driver that's configured via
device properties - all it needs is width and height in pixels, stride in
bytes, depth in bits and the framebuffer address.
Features:
- unaccelerated fb console with colours and virtual consoles
- split into console handling and bus frontend
- so far there is only a PCI bus frontend which allows to mmap BARs belonging
to the graphics controller, others should be easy to add.

TODO:
- find an opaque way to change the palette ( probably pass a function
pointer as property ) since this would need to call some firmware interface
which is pretty much always MD
- find a clean way to mmap the framebuffer, just returning the physical
address works on macppc but not on other architectures. Problem is, that
we cannot rely on the address being a bus address - on macppc it is,
but for instance on sparc64 it's not.

Add helper to detach genfb console.

wsfb: Prefer wide fonts when EDID data is available.

To give us a better chance of picking a readable font, prefer fonts that
will render at least 3mm wide instead of picking the font that will be
closest to that size.

Add 10-bit pixel format support.

Merge thorpej-cfargs2.

branches: 1.82.8;
Merge thorpej-cfargs branch:

Simplify and make extensible the config_search() / config_found() /
config_attach() interfaces: rather than having different variants for
which arguments you want pass along, just have a single call that
takes a variadic list of tag-value arguments.

Adjust all call sites:
- Simplify wherever possible; don't pass along arguments that aren't
actually needed.
- Don't be explicit about what interface attribute is attaching if
the device only has one. (More simplification.)
- Add a config_probe() function to be used in indirect configuiration
situations, making is visibly easier to see when indirect config is
in play, and allowing for future change in semantics. (As of now,
this is just a wrapper around config_match(), but that is an
implementation detail.)

Remove unnecessary or redundant interface attributes where they're not
needed.

There are currently 5 "cfargs" defined:
- CFARG_SUBMATCH (submatch function for direct config)
- CFARG_SEARCH (search function for indirect config)
- CFARG_IATTR (interface attribte)
- CFARG_LOCATORS (locators array)
- CFARG_DEVHANDLE (devhandle_t - wraps OFW, ACPI, etc. handles)

...and a sentinel value CFARG_EOL.

Add some extra sanity checking to ensure that interface attributes
aren't ambiguous.

Use CFARG_DEVHANDLE in MI FDT, OFW, and ACPI code, and macppc and shark
ports to associate those device handles with device_t instance. This
will trickle trough to more places over time (need back-end for pre-OFW
Sun OBP; any others?).

branches: 1.81.2;
add optional glyph cache for genfb
enable with options GENFB_GLYPHCACHE=n
with n being the desired size of the cache in MB. Should be enough to cache
at least 900 glyphs in whatever video mode used in order to be effective
in 32bit per pixel that's about 1MB

remove outdated comment

Prefer printing the framebuffer's PA instead of VA. It's much more useful..

branches: 1.78.2;
Add "is_swapped" property which indicates 32-bpp framebuffer is
byte-swapped.

Revert rev 1.75; do not set WSFB_VRAM_IS_RAM flag bit. It indicates
memory type obtained by mmap. I just misunderstood its intention.

Thanks to jmcneill for pointing it out!

Remove stray TABs. No binary changes.

For WSDISPLAYIO_GET_FBINFO ioctl, set WSFB_VRAM_IS_RAM to fbi_flags
when shadow FB is used.

Update for proplib(3) API changes.

reduce stack usage in genfb_calc_hsize()

Get genfb's address offset correctly when the vaule >= 4G. OK's by jmcneill.

Tested on Intel BXNUC10I3FNK (Comet Lake U).

branches: 1.71.4;
Do not replay the msgbuf if we are booting silent or quiet.

branches: 1.70.2;
Fix bug introduced in rev. 1.69:
http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/dev/wsfb/genfb.c#rev1.69

is_bgr should be initialized to false. Otherwise, color becomes strange
for depths 24 and 32 unless backend explicitly set "is_bgr" property.

We support anti-aliasing for depth 2.
Use switch appropriately.

Notify size of shadow framebuffer if enabled.

branches: 1.67.2;
Enable anti-aliasing for depth 24.
Style.

sys/dev/rasops routines support anti-aliasing for depths 15 and 16.

If an EDID is available, attempt to use the screen size to influence font
selection by passing a "wantcols" value to rasops_init that is satisfied
by a font that is at least 3mm wide.

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: 1.63.2;
Fix cmap bounds checking.

branches: 1.62.2;
CID-1427771: Remove incorrect pre-condition (index cannot be 256!)

branches: 1.61.2;
remove checks for failure after memory allocation calls that cannot fail:

kmem_alloc() with KM_SLEEP
kmem_zalloc() with KM_SLEEP
percpu_alloc()
pserialize_create()
psref_class_create()

all of these paths include an assertion that the allocation has not failed,
so callers should not assert that again.

enable font loading and screen resizing

genfb(4): Enabling/Disabling shadowfb can be controlled via prop.

branches: 1.58.2; 1.58.4;
Fix splashscreen on resolutions where text is centered.

This commit was approved by macallan@

support 24bpp framebuffers

branches: 1.56.2;
handle WSDISPLAYIO_GET_EDID and WSDISPLAYIO_GET_FBINFO only if the bus
frontend doesn't ( as in, returns EPASSTHROUGH or doesn't register a
handler )
Now flags passed from things like bcm2835_genfb.c are actually seen by
xf86-video-wsfb again and automatically disabling shadowfb works again,
which results in a major speedup on this kind of hardware.

branches: 1.55.2;
Add enable/disable polling hooks to genfb_ops.

Do not printf(foobar()). Bad bad bad bad bad!

branches: 1.53.2;
Merge riastradh-drm2 to HEAD.

Provide a wsdisplay_accessops.pollc

This makes sure that the screen gets painted properly during early boot
input (e.g. boot -a) on the Raspberry PI.

support WSDISPLAYIO_GET_FBINFO

branches: 1.50.2; 1.50.8;
for WSDISPLAYIO_SMODE, if the bus ioctl handler returns EPASSTHROUGH, dont treat it as an error

aprint_verbose -> aprint_debug

branches: 1.48.2;
in 32 bit colour assume RGB ordering, allow override using a device property

branches: 1.47.2;
On 8-bit screens only enable font anti-aliasing when the cmap_callback
property is defined. It is needed for a suitable R3G3B2 color map.
Ok by releng@

use rasops_init(0, 0)

support anti-aliased fonts in 8 bit, generate an appropriate colour map

if the cursor row passed in a device property doesn't fit on the screen just
ignore it and clear the screen

support anti-aliased fonts in 32bit colour

branches: 1.42.2; 1.42.6;
use callbacks that match the ioctl() interface to control backlight

some steps to make multiple instances work:
- don't bail if we're not the console. If we get the right paramters there is no
reason not to work
- move wsdisplay_accessops into the softc
- call wsdisplay_cnattach only if we're the console

branches: 1.40.2;
fix build for LP32
thanks to uebayasi@

look for a 'virtual_address' property and use it as fb address

genfb_enable/disable_polling only matters if genfb is the console screen,
so make it a noop if it's not

call vcons_hard_switch to switch to the console screen when entering polling
mode, makes ddb-from-X work with genfb on x86

Fixed NULL-pointer dereferencing when a mode-callback is not defined.

add vcons_{enable,disable}_polling and genfb_{enable,disable}_polling
functions, to switch between intr and polling modes when VCONS_DRAW_INTR
is defined

add a 'setmode' callback to genfb and use it to setup write-combining
MTRRs on x86 whenever switching to WSDISPLAYIO_MODE_EMUL

- add support for using compressed images as splash images
- retire SPLASHSCREEN_PROGRESS and SPLASHSCREEN_IMAGE options