viafb: simplify lcd size "detection"

Remove all occurences of get_lcd_size_method as only the values
GET_LCD_SIZE_BY_VGA_BIOS and GET_LCD_SIZE_BY_USER_SETTING were used
which had the identical code so there is no need to make things look
more complicated than they actually are.
Just a bit of of cleanup, really no regressions expected.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Joseph Chan <JosephChan@via.com.tw>

These are the files which should be available to subdevices compiled
outside of drivers/video/via.

Cc: ScottFang@viatech.com.cn
Cc: JosephChan@via.com.tw
Cc: Harald Welte <laforge@gnumonks.org>
Acked-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: NJonathan Corbet <corbet@lwn.net>

viafb: improve misc register handling

This patch improves the misc register handling by adding a modify
function for this to via_io.h and moving expanded definitions of the
relevant ports there. The code was changed to use those to improve
readability.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>

viafb: replace inb/outb

This patch replaces occurences of inb/outb with via_write_reg and
via_write_reg_mask where this is possible to improve code
readability.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>

viafb: move some modesetting functions to a seperate file

This patch moves the modesetting functions which are already cleaned up
to a seperate file.
Just the beginning to bring some structure in this mess.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>

viafb: unify modesetting functions

This patch unifies some cleaned up modesetting functions to prepare for
moving them to an extra file. This includes make them use via_io and
changing there names to reflect that they do not depend on anything
framebuffer specific.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>

This patch puts redesigned versions of the basic io functions that
are used overall the driver in an extra header. It is prefixed with
via_ as no framebuffer dependend stuff is in there. They were inlined
as they are really simple which reduced the module size about 2.5%.
The parameter order of read and write was fixed as it really doesn't
make sense to change the order as they are parts of the same address
and not source and destination.
Wrapper which use the new functions were added to hw.h to replicate
the old interface and avoid changing all old code.

[jc: added one comment]
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>

This patch moves data of interest into a new viafb_dev structure which
describes the device as a whole; the idea here is to create a separation
between what all devices may need and what the framebuffer device in
particular needs.

I've also made some small steps toward thinning out the global.h mess.

Cc: ScottFang@viatech.com.cn
Cc: JosephChan@via.com.tw
Cc: Harald Welte <laforge@gnumonks.org>
Acked-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: NJonathan Corbet <corbet@lwn.net>

[jc: extensive merge conflict fixes]
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: ScottFang@viatech.com.cn
Cc: JosephChan@via.com.tw
Signed-off-by: NChris Ball <cjb@laptop.org>

This will help us for the upcoming support for 2D acceleration using
the M1 engine.

[jc: fixed merge conflicts]
Signed-off-by: NHarald Welte <HaraldWelte@viatech.com>

This patch alters viafb to use the proper Linux in-kernel API to access
PCI configuration space, rather than poking at I/O ports by itself.

Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: ScottFang@viatech.com.cn
Cc: JosephChan@via.com.tw
Signed-off-by: NHarald Welte <HaraldWelte@viatech.com>

Add support for the color depth 15 on IGA1 and 30 on IGA1 and IGA2.  To
allow the usage of those the driver now refuses color depth that are
totally off and otherwise the selection in viafb_check_var is used.
Therefore the first call to this for the first framebuffer was delayed a
bit.  It only enables the new formats if they are requested exactly
(viafb_bpp=15|30).

As this is a new feature, no regressions are expected.  The color depth 15
was successfully tested.  Didn't get anything usable for 30 but that might
be the programs fault.  I would like to get some feedback whether it works
as expected or not if somebody knows a program/configuration where it
should.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This is a rewritten version of viafb_setcolreg.  The hardware register
writes were split up and moved to hw.c where they belong as this is really
low level stuff.  It was made dual fb aware.

Furthermore viafb_setcmap was removed as the problem with 8bpp originated
from a bug in writing multiple color registers at once.  The removal of
viafb_setcmap might introduce a small performance regression but its
certainly better to receive the correct result a bit slower than a garbled
picture fast.  It should give us a working 8bpp mode and is more
extensible than the old hardcoded code.  No other regressions are expected
but as the hardware might be a bit picky it might cause some regressions
in 8bpp mode on some hardware although I doubt that.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Make color checking a bit more tolerant in what values it allows and more
fine grained to later support 15 and 30 bits formats.  It splits the
filling of the color information in var to a seperate function and sets
some color related values in var that where previously untouched.

This could be a bug fix but at least I don't know any applications that
was fooled by not correctly setting the fields in var.  At least no
regressions are expected.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Some code depended on IGA1_IGA2 which was never set (at least with the
symbolic name).  Remove this dead code although it might one day be useful
to get a hint on how some things might work.  However as this is dead it
is likely full of bugs and would prevent a clean structure (as it has some
very strange things).

Dead code -> no regressions, at least if VIA doesn't do anything very ugly.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Cc: Joseph Chan <JosephChan@via.com.tw>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This patch splits color mode setting up in seperate functions.  Some
hardware initialization that was previously mixed with it is moved to
viafb_setmode.  As are the calls to the newly created function.  This is
yet another little step towards controlling each IGA on its own.

As this patch really aims too mimic the old behaviour no regressions are
expected.  However I noticed that 8bpp (or 6bpp?) seems actually a bit
broken before and after the patch.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Cc: Joseph Chan <JosephChan@via.com.tw>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Set always the correct video address.  Especially do panning right on
multiple IGAs.

This should have no effect on single monitor mode (no SAMM, no dual fb).
For SAMM without dual fb this might break something as I really cannot
image what we are supposed to do for different resolutions with a single
framebuffer as we can't get data out of nowhere (no, they are not set up
in something one would call "expanded").  Previously I got for that funny
colored pictures as the second IGA pointed to video memory that was never
written to.  After the patch it'll work as cloning if the first and second
mode are identical (this was working already without SAMM).  Finally for
dual fb this should push us a step in the right direction.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Cc: Joseph Chan <JosephChan@via.com.tw>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Remove some unused variables, move some dvi code around and store the
detected maximum resolution for later use.  The vertical resolution is
handled as the old code did it but I hope it can be read from the hardware
some day.

No runtime change expected.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This is the first step to remove an artificial global index that was used
in two ways:

1. As a pseudo index in the mode table.  Pseudo as you had to search
   through the table to find the referenced entry.  This was replaced by
   using a pointer to the entry.

2. As a shortcut to compare a combination of horizontal and vertical
   resolution at the same time.

   This was replaced by a "(hres<<16) | vres" which is good enough for
   now and the near future.  If vres or hres become greater than 2^16 this
   might indeed cause problems but this solution allows to split this
   indexing mess up without the requirement to do even more code changes.

This is a big change that will allow more clean ups.  It should be a bit
faster but that is probably not relevant for normal operation.  No
regressions expected but as this is a relatively big step heavy testing is
appreciated.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Signed-off-by: NHarald Welte <HaraldWelte@viatech.com>
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Add support for a new VIA integrated graphics chipset, the VX855.
Signed-off-by: NHaraldWelte <HaraldWelte@viatech.com>
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Split the pitch handling up and replaces the calculation from virtual xres
and bpp with fix.line_length which already contains the pitch and does not
add any constrains for the virtual resolution.

Also add a bit to the second pitch which the documentation mentions but
which was ignored by the driver.

Although it is a bit unclear what the right pitch for some LCD modes is
this patch should have no negative runtime impact.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Harald Welte <laforge@gnumonks.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This removes the completly useless io variable as well as the temporary
used variables mmio_base and mmio_len in favor to use directly the fb_info
variables.

This is a code cleanup only, no runtime change expected.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Harald Welte <laforge@gnumonks.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Move individual start address setting to viafb_set_primary_address and
viafb_set_secondary_address and make it more flexible to reuse it for
panning.  Using central functions makes it easier to follow HW
manipulations.

Remove crt locking as it should be only needed for timing manipulation.
Move iga_path manipulation to via_pci_probe.

Remove memset for screen cleaning as it is currently done only for the
second screen.  This is not needed for normal operation but has a little
chance of causing unwanted display artifacts.  This can be fixed later
more consistent and more efficient (using viafb_fillrect) if needed.

This is a code clenup, no notable runtime changes expected.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Harald Welte <laforge@gnumonks.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The current code initializes the register for CX700 chips 2 times due to a
missing break as discovered by Harald Welte.

As CX700 and VX800 have exactly the same register initialization we can
use one for both to avoid duplicated code.

As this is a pure code cleanup no measurable runtime effects are expected.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Harald Welte <laforge@gnumonks.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This fixes a bug caused by changing pointers (viafb_mode, viafb_mode1)
assigned by module_param.  It reduces driver complexity by not needlessly
changing these vars as they are only read once and removing now
superfluous code.

On unpatched kernels loading viafb with viafb_mode or viafb_mode1 option
used and afterwards unloading it results in:

kernel BUG at mm/slub.c:2926!
invalid opcode: 0000 [#1] PREEMPT
last sysfs file: /sys/devices/virtual/block/loop0/removable
Modules linked in: snd_hda_codec_realtek snd_hda_intel snd_hda_codec
snd_hwdep snd_pcm rtl8187 snd_timer eeprom_93cx6 mmc_block snd soundcore
via_sdmmc fb snd_page_alloc i2c_algo_bit i2c_viapro ehci_hcd uhci_hcd
cfbcopyarea mmc_core cfbimgblt cfbfillrect video output [last unloaded:
viafb]

  Pid: 3355, comm: rmmod Not tainted (2.6.31-rc1 #0)
  EIP: 0060:[<c106a759>] EFLAGS: 00010246 CPU: 0
  EIP is at kfree+0x80/0xda
  EAX: c17c2da0 EBX: dc7edbdc ECX: 0000010f EDX: 00000000
  ESI: c102c700 EDI: dc7ed8fa EBP: d703ff2c ESP: d703ff20
   DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 0068
  Process rmmod (pid: 3355, ti=d703e000 task=db1412c0 task.ti=d703e000)
  Stack:
   dc7edbdc 00000014 00000016 d703ff40 c102c700 dc7f45d4 dc7f45d4 00000880
   d703ff4c c103e571 00000000 d703ffac c103e751 66616976 da140062 db89ba80
   00000328 d702edf8 db89ba80 d703ff9c c105d0f0 00000200 da14f898 00000014
  Call Trace:
   [<c102c700>] ? destroy_params+0x1e/0x2b
   [<c103e571>] ? free_module+0xa2/0xd7
   [<c103e751>] ? sys_delete_module+0x1ab/0x1da
   [<c105d0f0>] ? do_munmap+0x20a/0x225
   [<c10029b4>] ? sysenter_do_call+0x12/0x26
  Code: 10 76 7a 8d 87 00 00 00 40 c1 e8 0c c1 e0 05 03 05 1c 87 41 c1 66 83 38 00 79 03 8b 40 0c 8b 10 84 d2 78 12 66 f7 c2 00 c0 75 04 <0f> 0b eb fe e8 6f 5a fe ff eb 47 8b 55 04 8b 58 0c 9c 5e fa 3b
  EIP: [<c106a759>] kfree+0x80/0xda SS:ESP 0068:d703ff20

This is caused by the current code changing the pointers assigned by
module_param.  During unload it tries to free the memory the pointers
point at which is now part of an internal structure.

The patch simply avoids changing the pointers.  This is okay as they are
read only once during the initialization process.
Signed-off-by: NFlorian Tobias Schandinat <FlorianSchandinat@gmx.de>
Cc: Scott Fang <ScottFang@viatech.com.cn>
Cc: Joseph Chan <JosephChan@via.com.tw>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Display HW setting and other chips initialization.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: NJoseph Chan <josephchan@via.com.tw>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>