Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

There's really no good reason for us to be in here anymore, we have to
maintain this ABI anyway to avoid angering people.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

The time has come to get a proper version number that we can change to
indicate new features etc, rather than the lock-step 0.0.XX that we
previously had.

libdrm has recognised this version as compatible with 0.0.16 since 2.4.22,
so hopefully any breakage people see should be very minimal.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

This happens somehow during init on a machine I have, and leads to a
divide-by-zero.

Lets avoid that...

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

All available subchannels are now available for userspace to do with as it
pleases on NVC0+.

On all earlier chipsets, the kernel still uses a software object on subc 0
to implement the page flip completion method.  I hope to find some decent
way of addressing this too, but it's a tad tricker prior to fermi.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Removes need for M2MF subchannel usage on NVC0+.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

I want to be able to use REF_CNT from other places in the kernel without
pushing a fence object onto the list of emitted fences.

The current code makes an assumption that every time the acked sequence is
bumped that there's at least one fence on the list that'll be signalled.

This will no longer be true in the near future.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

These are FIFO methods, it doesn't matter what subchannel is being used.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

There was once good reasons for wanting the drm to be able to use M2MF etc
on user channels, but they're not relevant anymore.  For the general
buffer move case, we've already lost by transferring between vram/sysmem
already so the context switching overhead is minimal in comparison.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

NVIDIA appear to do these around the same place they do the MODE_CTRL
methods, and for DP at least we need to bash some extra bits in "syncs"
to keep EVO happy.

It's a bit of a guess as to the 6/8bpc, but i have no better idea yet.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

GF119 will use this too.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

The shift from hwsq_data = 0x1400 to 0x080000 actually happened in nv94, not nv92
This fixes some reclocking issues on my newly acquired nv92

Signed-off-by: Martin Peres <martin.peres@labri.fr>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Off-chip encoders (which we don't support yet anyway), and newer chipsets
(such as NVD9...), will need their own code for this.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

More code to do the same thing, but will make it easier to handle various
changes that could possibly happen the the VBIOS tables.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Refactored to allow shadowing of VBIOS images longer than 64KiB, which
allows us to pass the VBIOS checksum test on certain boards.

There's also a workaround for reading the PROM VBIOS on some chipsets.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

There's cards out there with completely messed up PCIROM images that have
a perfectly valid signature.. Sigh!

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Theoretically handles CRTC2/CRTC3, should any GF119 out there actually
have them enabled.  The room is there for the regs etc, so why not :)

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

This patch fixes two small issues in timing generation as spotted on
several NVCx cards.

In addition, the header of the file is updated to also contain (some of)
the current developers of this code.

Signed-off-by: Roy Spliet <r.spliet@student.tudelft.nl>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Fixes reclocking failure on some chips where we attempted to set PDAEMON
to PLL mode.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

The comparison (lpre == DP_TRAIN_PRE_EMPHASIS_9_5) is always false:
lpre is initialized as (lane & 0x0c) >> 2, which is at most 3, while
DP_TRAIN_PRE_EMPHASIS_9_5 is defined as (3 << 3).

Signed-off-by: Xi Wang <xi.wang@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

There's a HP laptop out there where the MXM version in the VBIOS doesn't
match what the ACPI implementation is expecting.  These tables will accept
0x00 to MXMS to return latest version, but *only* if MXMI has been called
first..

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

This patch fixes an oops cause by pm_trigger accessing the (uninitialised)
crtc list.

Reported-by: Roy Spliet <r.spliet@student.tudelft.nl>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Signed-off-by: Christoph Bumiller <e0425955@student.tuwien.ac.at>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

v2 (Emil Velikov <emil.l.velikov@gmail.com>):
- Fixed a regression on certain nv50 IGP due to not passing the correct
  target type to nv50_vm_addr()

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Tested-by: Johannes Obermayr <johannesobermayr@gmx.de>

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Martin Peres <martin.peres@labri.fr>

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Martin Peres <martin.peres@labri.fr>