The code was trying to keep a strict limit on the amount of mob
memory that was used in the guest by making it match the host
settings. There's technically no reason to do that (guests can
certainly use more than the host can have resident in renderers
at the same time).
In particular this is problematic because our userspace is not
great at handling OOM conditions and running out of MOB space
results in GL apps crashing, e.g. gnome-shell likes to allocate
huge surfaces (~61MB for the desktop on 2560x1600 with two workspaces)
and running out of memory there means that the gnome-shell crashes
on startup taking us back to the login and resulting in a system
where one can not login in graphically anymore.
Instead of letting the userspace crash we can extend available
MOB space, we just don't want to use all of the RAM for graphics,
so we're going to limit it to half of RAM.
With the addition of some extra logging this should make the
"guest has been configured with not enough graphics memory"
errors a lot easier to diagnose in cases where the automatic
expansion of MOB space fails.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210723165153.113198-3-zackr@vmware.com
The code was using the old DRM logging functions, which made it
hard to figure out what was coming from vmwgfx. The newer logging
helpers include the driver name in the logs and make it explicit
which driver they're coming from. This allows us to standardize
our logging a bit and clean it up in the process.
vmwgfx is a little special because technically the hardware it's
running on can be anything from the last 12 years or so which is
why we need to include capabilities in the logs in the first
place or otherwise we'd have no way of knowing what were
the capabilities of the platform the guest was running in.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210723165153.113198-2-zackr@vmware.com
The macro has been accounting for DRM_COMMAND_BASE for a long time
now so there's no reason to still be duplicating it. Plus we were
leaving the name undefined which meant that all the DRM ioctl
warnings/errors were always listing "null" ioctl at the culprit.
This fixes the undefined ioctl name and removes duplicated code.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210723165153.113198-1-zackr@vmware.com
Drop the DRM IRQ midlayer in favor of Linux IRQ interfaces. DRM's
IRQ helpers are mostly useful for UMS drivers. Modern KMS drivers
don't benefit from using it.
Vmwgfx already uses Linux IRQ functions. All that's left to replace
is the reference to struct drm_device.irq. Use irq value of struct
pci_dev instead.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Zack Rusin <zackr@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210706072253.6844-1-tzimmermann@suse.de
Change
2ef4fb9236 ("drm/vmwgfx: Make sure bo's are unpinned before putting them back")
caused a conflict in one of the drm trees and the merge commit
68a32ba141 ("Merge tag 'drm-next-2021-04-28' of git://anongit.freedesktop.org/drm/drm")
accidently re-added code that the original change was removing.
Fixed by removing the incorrect buffer unpin - it has already been unpinned
two lines above.
Fixes: 68a32ba141 ("Merge tag 'drm-next-2021-04-28' of git://anongit.freedesktop.org/drm/drm")
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210615182336.995192-4-zackr@vmware.com
Historically our device headers have been forked versions of the
internal device headers, this has made maintaining them a bit
of a burden. To fix the situation, going forward, the device headers
will be verbatim copies of the internal headers.
To do that the driver code has to be adapted to use pristine
device headers. This will make future update to the device
headers trivial and automatic.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210615182336.995192-2-zackr@vmware.com
Original vmw_mksstat_remove_ioctl expected pid to match the corresponding vmw_mksstat_add_ioctl.
That made impossible en-masse removals by one pid, which is a valid use case, so pid match was
discarded. Current change enforces a broader pgid match as a form of protection from arbitrary
processes interrupting an ongoing mks-guest-stats.
Reviewed-by: Zack Rusin <zackr@vmware.com>
Signed-off-by: Martin Krastev <krastevm@vmware.com>
Signed-off-by: Zack Rusin <zackr@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210609172307.131929-8-zackr@vmware.com
This code has been unused for a while now. When the explicit checks
for whether the driver is running on top of non-coherent swiotlb
have been deprecated we lost the ability to fallback to physical
mappings. Instead of trying to readd a module parameter to force
usage of physical addresses it's better to just force coherent
TTM pages via the force_coherent module parameter making this
code pointless.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210609172307.131929-6-zackr@vmware.com
The has_dx variable was only set during the initialization which
meant that UPDATE_SUBRESOURCE was never used. We were emulating it
with UPDATE_GB_IMAGE but that's always been a stop-gap. Instead
of has_dx which has been deprecated a long time ago we need to check
for whether shader model 4.0 or newer is available to the device.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210609172307.131929-4-zackr@vmware.com
VMware mks-guest-stats mechanism allows the collection of performance stats from
guest userland GL contexts, as well as from vmwgfx kernelspace, via a set of sw-
defined performance counters. The userspace performance counters are (de)registerd
with vmware-vmx-stats hypervisor via new iocts. The vmwgfx kernelspace counters
are controlled at build-time via a new config DRM_VMWGFX_MKSSTATS.
* Add vmw_mksstat_{add|remove|reset}_ioctl controlling the tracking of
mks-guest-stats in guest winsys contexts
* Add DRM_VMWGFX_MKSSTATS config to drivers/gpu/drm/vmwgfx/Kconfig controlling
the instrumentation of vmwgfx for kernelspace mks-guest-stats counters
* Instrument vmwgfx vmw_execbuf_ioctl to collect mks-guest-stats according to
DRM_VMWGFX_MKSSTATS
Signed-off-by: Martin Krastev <krastevm@vmware.com>
Reviewed-by: Zack Rusin <zackr@vmware.com>
Signed-off-by: Zack Rusin <zackr@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210609172307.131929-3-zackr@vmware.com
Make devcaps code self-contained so that it's easier to cache
and operate on them.
As the number of devcaps got bigger the code dealing with them
got more and more tricky. Lets create a central place to deal
with all the complexity. This lets us remove the lock we used
to require to deal with register write races because we only
read the devcaps at initialization.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210609172307.131929-2-zackr@vmware.com
The allocation of fifo is lacking an allocation failure check, so
fix this by adding one.
In the case where fifo->static_buffer fails to be allocated the
error return path neglects to kfree the fifo object. Fix this by
adding in the missing kfree.
Kudos to Dan Carpenter for spotting the missing kzalloc failure
check.
Addresses-Coverity: ("Resource leak")
Fixes: 2cd80dbd35 ("drm/vmwgfx: Add basic support for SVGA3")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Zack Rusin <zackr@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210514144930.89813-1-colin.king@canonical.com
This change fixes all of the arm64 issues we've had in the driver.
ARM support is provided in svga version 3, for which support we've added
in previous changes. svga version 3 currently lacks many of the
advanced features (in particular 3D support is lacking) but
that will change in time.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210505035740.286923-7-zackr@vmware.com
SVGA3 is the next version of our PCI device. Some of the changes
include using MMIO for register accesses instead of ioports,
deprecating the FIFO MMIO and removing a lot of the old and
legacy functionality. SVGA3 doesn't support guest backed
objects right now so everything except 3D is working.
v2: Fixes all the static analyzer warnings
Signed-off-by: Zack Rusin <zackr@vmware.com>
Cc: Martin Krastev <krastevm@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210505191007.305872-1-zackr@vmware.com
Now since Christian reworked TTM to always keep objects on the LRU
list unless they are pinned we shouldn't need the reservation
semaphore. It makes the driver code a lot cleaner, especially
because it was a little hard to reason when and where the
reservation semaphore needed to be held.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210505035740.286923-5-zackr@vmware.com
In cases where the dirty linear memory range spans multiple sample sheets
in a surface, the dirty surface region is incorrectly computed.
To do this correctly and in an optimized fashion we would have to compute
the dirty region of each sample sheet and compute the union of those
regions.
But assuming that cpu writing to a multisample surface is rather a corner
case than a common case, just set the dirty region to the full surface.
This fixes OpenGL piglit errors with SVGA_FORCE_COHERENT=1
and the piglit test:
fbo-depthstencil blit default_fb -samples=2 -auto
Fixes: 9ca7d19ff8 ("drm/vmwgfx: Add surface dirty-tracking callbacks")
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Zack Rusin <zackr@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210505035740.286923-4-zackr@vmware.com
The SVGA3dCmdDXGenMips command uses a shader-resource view to access
the underlying surface. Normally accesses using that view-type are not
dirtying the underlying surface, but that particular command is an
exception.
Mark the surface gpu-dirty after a SVGA3dCmdDXGenMips command has been
submitted.
This fixes the piglit getteximage-formats test run with
SVGA_FORCE_COHERENT=1
Fixes: a9f58c456e ("drm/vmwgfx: Be more restrictive when dirtying resources")
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Zack Rusin <zackr@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210505035740.286923-3-zackr@vmware.com
SVGA_REG_ENABLE is a register name, and SVGA_REG_ENABLE_(ENABLE|
DISABLE|HIDE) are its valid values. We were incorrectly setting
the register value to itself. This happened to work because
the SVGA_REG_ENABLE is happens to to be the same value as
SVGA_REG_ENABLE_ENABLE, but is still semantically incorrect.
Lets use the correct enum when setting SVGA_REG_ENABLE.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210505035740.286923-2-zackr@vmware.com
Pull more drm updates from Dave Airlie:
"Looks like I missed a tegra feature request for next, but should still
be fine since it's pretty self contained.
Apart from that got a set of i915 and amdgpu fixes as per usual along
with a few misc fixes.
tegra:
- Tegra186 hardware cursor support
- better capability reporting for different SoC
- better framebuffer modifier support
- host1x fixes
ttm:
- fix unswappable BO handling
efifb:
- check for PCI before using it
amdgpu:
- Fixes for Aldebaran
- Display LTTPR fixes
- eDP fixes
- Fixes for Vangogh
- RAS fixes
- ASPM support
- Renoir SMU fixes
- Modifier fixes
- Misc code cleanups
- Freesync fixes
i915:
- Several fixes to GLK handling in recent display refactoring
- Rare watchdog timer race fix
- Cppcheck redundant condition fix
- Overlay error code propagation fix
- Documentation fix
- gvt: Remove one unused function warning
- gvt: Fix intel_gvt_init_device() return type
- gvt: Remove one duplicated register accessible check"
* tag 'drm-next-2021-04-30' of git://anongit.freedesktop.org/drm/drm: (111 commits)
efifb: Check efifb_pci_dev before using it
drm/i915: Fix docbook descriptions for i915_gem_shrinker
drm/i915: fix an error code in intel_overlay_do_put_image()
drm/i915/display/psr: Fix cppcheck warnings
drm/i915: Disable LTTPR detection on GLK once again
drm/i915: Restore lost glk ccs w/a
drm/i915: Restore lost glk FBC 16bpp w/a
drm/i915: Take request reference before arming the watchdog timer
drm/ttm: fix error handling if no BO can be swapped out v4
drm/i915/gvt: Remove duplicated register accessible check
drm/amdgpu/gmc9: remove dummy read workaround for newer chips
drm/amdgpu: Add mem sync flag for IB allocated by SA
drm/amdgpu: Fix SDMA RAS error reporting on Aldebaran
drm/amdgpu: Reset RAS error count and status regs
Revert "drm/amdgpu: workaround the TMR MC address issue (v2)"
drm/amd/display: 3.2.132
drm/amd/display: [FW Promotion] Release 0.0.62
drm/amd/display: add helper for enabling mst stream features
drm/amd/display: Report Proper Quantization Range in AVI Infoframe
drm/amd/display: Fix call to pass bpp in 16ths of a bit
...
idr_init() uses base 0 which is an invalid identifier for this driver.
The idr_alloc for this driver uses 1 as start value for ID range. The
new function idr_init_base allows IDR to set the ID lookup from base 1.
This avoids all lookups that otherwise starts from 0 since 0 is always
unused / available.
References: commit 6ce711f275 ("idr: Make 1-based IDRs more efficient")
Signed-off-by: Deepak R Varma <mh12gx2825@gmail.com>
Signed-off-by: Zack Rusin <zackr@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201105190718.GA89863@localhost
