Goes a long way to correcting NVS295 memory reclocking issues.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Martin Peres <martin.peres@labri.fr>
There's some "extended" GDDR3 chipsets out there with EMRS2 settings that
change the layout of MRS/EMRS1 bitmaps.. Sigh.. Still need to track down
how exactly we're supposed to handle this.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Martin Peres <martin.peres@labri.fr>
Idea from Martin Peres, different implementation by me.
v2: Martin Peres:
- fix mast calculation
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Martin Peres <martin.peres@labri.fr>
This will probably result in more lines of code, however, we're going to
have at least 3 slightly different implementations of this very soon and
I'd rather keep the ram reclocking logic separate from the hw specifics.
DDR2/DDR3/GDDR3 implemented thus far, others will be added as necessary.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Martin Peres <martin.peres@labri.fr>
Statically generating the PFB register and MR values for each timing set
turns out to be insufficient. There's at least one (so far) known piece
of information which effects MR values which is stored in the perflvl
entry on some chipsets (and in another table on later ones), which is
disconnected from the timing table entries.
After this change we will generate a timing set based on an input clock
frequency instead, and have this data stored in the performance level
data.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Martin Peres <martin.peres@labri.fr>
We might want/need the boot data to generate the other perflevels.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Martin Peres <martin.peres@labri.fr>
Roy Spliet:
- Implement according to specs
- Simplify
- Make array for mc latency registers
Martin Peres:
- squash and split all the commits from Roy
- rework following Ben Skeggs comments
- add a form of timings validation
- store the initial timings for later use
Ben Skeggs
- merge slightly modified tidy-up patch with this one
- remove perflvl-dropping logic for the moment
Signed-off-by: Roy Spliet <r.spliet@student.tudelft.nl>
Signed-off-by: Martin Peres <martin.peres@labri.fr>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
It turns out we need access to some additional information in various VBIOS
tables to handle PFB memory timings correctly.
Rather than hack in parsing of the new stuff in some kludgy way, I've
restructured the VBIOS parsing to be more primitive, so we can use them in
more flexible ways in the future.
The perflvl->timing association code is disabled for the moment until it can
be reworked. We don't use this stuff yet anyway, so no harm done.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Martin Peres <martin.peres@labri.fr>
Module parameter descriptions don't take a trailing \n, otherwise it
breaks formatting of modinfo's output. Also remove trailing space.
Signed-off-by: Jean Delvare <jdelvare@suse.de>
Cc: David Airlie <airlied@linux.ie>
Cc: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
- Rename several VBIOS entries to closer match the real world
- Add the missing 0x100238 and 0x100240 register values
- Parse bit 14 of the VBIOS timing table
- "Magic value" -> tCWL, fixing some minor bugs in the process
- Also name a few more by their name rather than their number.
- Some values seem to be dependent on the memory type. Fix
Edits by Martin Peres <martin.peres@labri.fr>:
- this is a squash commit
- reworked for fixing some style issues
Signed-off-by: Roy Spliet <r.spliet@student.tudelft.nl>
Signed-off-by: Martin Peres <martin.peres@labri.fr>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Uses only the VBIOS tables, from what I can tell this is what NVIDIA do
too, I was able to change the detected memory type by modifying this table
on a NVC1 chipset.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
M version 2 appears to have a table with some form of memory type info
available.
NVIDIA appear to ignore the table information except for this DDR2/DDR3
case (which has the same value in 0x100714). My guess is this is due to
some of the supported memory types not being represented in the table.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
DDR1/DDR[23] confirmed on NVA8 (see note about DDR3 in source) by changing
the value and watching the binary driver's behaviour.
GDDR3/4 values confirmed on a NV96 via the same method above. That GDDR4
is present is interesting, as far as I can see no boards using it were ever
released.
GDDR5 value is based on VBIOS images of known GDDR5 boards.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
NV20/NV30 is partially educated guesswork at this point, based on any
information around about available memory types and a horribly unspeakable
amount of vbios image scouring. I'm not entirely certain the GDDR3 define
is correct, I have not spotted a single vbios with that value yet (though
it is mentioned in some 1218-using nv4x vbios), but there are reports that
some nv3x did use it..
NV40(100914) confirmed by switching an NV49 to DDR1/DDR2 values and making
sure that the binary driver behaviour showed it had detected DDR1/DDR2
instead of GDDR3 before dying horribly.
NV40(100474) confirmed by doing much the same task as above on an NV44,
except this was *much* easier as changing the values didn't seem to have
any noticable effect on the memory controller aside from changing the
binary driver's behaviour.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
The current enabling of bus mastering in the drm midlayer allows a large
race condition under kexec. When a kexec'ed kernel re-enables bus mastering
for the GPU, previously setup dma blocks may cause writes to random pieces
of memory. On radeon the writeback mechanism can cause these sorts of issues.
This patch doesn't fix the problem, but it moves the bus master enable under
the individual drivers control so they can move enabling it until later in
their load cycle and close the race.
Fix for radeon kms driver will be in a follow-up patch.
Signed-off-by: Dave Airlie <airlied@redhat.com>
The drm drivers set the fb_info->pixmap fields without setting
fb_info->pixmap.addr. If this is not set the fb core will overwrite
these all fb_info->pixmap fields anyway, so there is not much point
in setting them in the first place.
[airlied: dropped nvidiafb piece - not mine]
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Creating a range property is a common pattern, so create
a convenience function for this and use it where appropriate.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Creating an enum property is a common pattern, so create
a convenience function for this and use it where appropriate.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Dave Airlie <airlied@redhat.com>
calc_mclk() returns zero on success and negative on failure but clk is
a u32.
v2: Martin Peres:
- clk should be an int, not a u32
Signed-off-by: Martin Peres <martin.peres@labri.fr>
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
There's at least one known case where our shadowing code is buggy, and we
fail init. Until we can be confident we're doing all this correctly, lets
succeed and risk crazy bios tables rather than failing for perfectly valid
configs too.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Both changes in dc97b3409a cause serious
regressions in the nouveau driver.
move_notify() was originally able to presume that bo->mem is the old node,
and new_mem is the new node. The above commit moves the call to
move_notify() to after move() has been done, which means that now, sometimes,
new_mem isn't the new node at all, bo->mem is, and new_mem points at a
stale, possibly-just-been-killed-by-move node.
This is clearly not a good situation. This patch reverts this change, and
replaces it with a cleanup in the move() failure path instead.
The second issue is that the call to move_notify() from cleanup_memtype_use()
causes the TTM ghost objects to get passed into the driver. This is clearly
bad as the driver knows nothing about these "fake" TTM BOs, and ends up
accessing uninitialised memory.
I worked around this in nouveau's move_notify() hook by ensuring the BO
destructor was nouveau's. I don't particularly like this solution, and
would rather TTM never pass the driver these objects. However, I don't
clearly understand the reason why we're calling move_notify() here anyway
and am happy to work around the problem in nouveau instead of breaking the
behaviour expected by other drivers.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Cc: Jerome Glisse <j.glisse@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Newer nVidia cards with Optimus do not support/use the DSM switching functions.
Instead, it require a DSM function to be called prior to bringing a device into
D3 state. No other _DSM calls are necessary before/after enabling/disabling a
device. Switching between discrete and integrated GPU is not supported by
this Optimus _DSM call, therefore return on the switching method.
Signed-off-by: Peter Lekensteyn <lekensteyn@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
According to the ACPI spec version 4, section 9.14.1, _DSM functions
must return a value with the first bit enabled if any DSM functions are
supported for the given UUID and revision ID. For a given function index n
to be marked supported, bit n must be enabled.
Signed-off-by: Peter Lekensteyn <lekensteyn@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
ttm tt rework modified the way we allocate and populate the
ttm_tt structure, the AGP side was missing some bit to properly
work. Fix those and fix radeon and nouveau AGP support.
Tested on radeon only so far.
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
"drm/ttm: callback move_notify any time bo placement change v4" failed to
avoid a NULL pointer dereference in nouveau caused by move_notify being
expected to handle that case now.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
