Some fences might be alive even after we have stopped the scheduler leading
to warnings about leaked objects from the SLUB allocator.

Fix this by allocating/freeing the SLUB allocator from the module
init/fini functions just like we do it for hw fences.

v2: make variable static, add link to bug

Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=97500Reported-by: Grazvydas Ignotas <notasas@gmail.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com> (v1)
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

Otherwise callers end up using uninitialized data.
Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

This should only happen on boards TV connectors which do not
have a ddc bus for those connectors.  None of the asics supported
by amdgpu support tv, so we shouldn't hit this, but check
to be on the safe side (e.g., bios bug for example).
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Since commit a481daa8 ("drm/radeon: always apply pci shutdown
callbacks"), a Dell Latitude D600 laptop has crashed on shutdown. The
PCI Identification of the graphics adapter is "VGA compatible controller
[0300]: Advanced Micro Devices, Inc. [AMD/ATI] RV250/M9 GL [Mobility
FireGL 9000/Radeon 9000] [1002:4c66] (rev 01)".

Prior to commit b0c80bd5 ("drm/radeon: fix up dp aux tear down (v2)"),
I have no idea where the panic happened as the screen was blanked before
the crash.  Since that more recent change, the panic has been in routine
radeon_connector_unregister(), and has been shown to be due to a NULL
value in the ddc_bus member of struct drm_connector.

Bug: https://bugzilla.kernel.org/show_bug.cgi?id=178421
Fixes: a481daa8 ("drm/radeon: always apply pci shutdown callbacks")
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

If the platform does not support hybrid graphics or ATPX dGPU
power control.
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

If the platform does not support hybrid graphics or ATPX dGPU
power control.

bug: https://bugzilla.kernel.org/show_bug.cgi?id=51381Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

Newer iceland parts require new smc firmware.
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

Newer tonga parts require new smc firmware.
Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Eric Huang <JinHuiEric.Huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Consolidate existing quirks. Fixes stability issues
on some kickers.
Acked-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Consolidate existing quirks. Fixes stability issues
on some kickers.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

the value of last_mclk_dpm_enable_mask will be changed if
other clients(vce,dal) trigger set power state between enable
and disable uvd dpm.
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

The read is taking a considerable amount of time (about 50us on this
machine). The register does not ever hold anything other than the ring
ID that is updated in this exact function, so there is no need for
the read modify write cycle.

This chops off a big chunk of the time spent in hardirq disabled
context, as this function is called multiple times in the interrupt
handler. With this change applied radeon won't show up in the list
of the worst IRQ latency offenders anymore, where it was a regular
before.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

Older firmware versions don't support 3 rings.

fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=98016

v2: use define for fw version
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

To free fences, call_rcu() is used, which calls amdgpu_fence_free()
after a grace period. During teardown, there is no guarantee all
callbacks have finished, so amdgpu_fence_slab may be destroyed before
all fences have been freed. If we are lucky, this results in some slab
warnings, if not, we get a crash in one of rcu threads because callback
is called after amdgpu has already been unloaded.

Fix it with a rcu_barrier().

Fixes: b4413535 ("drm/amdgpu: RCU protected amdgpu_fence_release")
Acked-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

The names were wrong.
Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Looks like .last_flush reference is left at teardown.
Leak reported by CONFIG_SLUB_DEBUG.

Fixes: 41d9eb2c ("drm/amdgpu: add a fence after the VM flush")
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

To free fences, call_rcu() is used, which calls amd_sched_fence_free()
after a grace period. During teardown, there is no guarantee all
callbacks have finished, so sched_fence_slab may be destroyed before
all fences have been freed. If we are lucky, this results in some slab
warnings, if not, we get a crash in one of rcu threads because callback
is called after amdgpu has already been unloaded.

Fix it with a rcu_barrier().

Fixes: 189e0fb7 ("drm/amdgpu: RCU protected amd_sched_fence_release")
Acked-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

This reverts commit 1a738347.

It caused at least some Kaveri laptops to incorrectly report DisplayPort
connectors as connected.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=97857
Cc: stable@vger.kernel.org
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Flush any outstanding thermal work before tearing down
the dpm driver.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Cancel any pending reset work when we tear down the driver.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Currently the driver crashes if smu7_enable_dpm_tasks() returns early,
which it does if DPM is already active. It seems to be better just to
continue anyway, at least I haven't noticed any ill effects. It's also
unclear at what state the hardware was left by the previous driver, so
IMO it's better to always fully initialize.

Way to reproduce:
with GPU passthrough.
forced power off the VM or forced reset the VM
without shutting down the Os.
Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Reviewed-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Reviewed-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Signed-off-by: JimQu <Jim.Qu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

No need to retain previous settings as this is the first time
we set pg_flags.  Probably a copy/paste typo from the CZ code.
Avoids confusion.

No change in behavior as adev is kzallocated.
Reviewed-by: Tom St Denis <tom.stdenis@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

It's technically an MC register so make sure we initialize it
in the MC module rather than the gfx module.  Since other bits
in the same register are used to enable ATC CG features make
sure we apply the golden setting first.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

We need this for more than just DCE.  Move it out of the DCE modules
and into the device code.  This way we can be sure the scratch registers
are initialized properly before we run asic_init which happens before
DCE IPs are restored.

Fixes atombios hangs in asic_init.
Reviewed-by: JimQu <Jim.Qu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

This is already handled by the dce IP modules in their
suspend and resume code.  No need to do it again.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Limit clocks on a specific HD86xx part to avoid
crashes (while awaiting an appropriate PP fix).
Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

Casting of voltage values to a larger size results in
overwriting adjacent memory in the structure.
Reviewed-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

2 more patches to stabilize the new MMUv2 support.

* 'drm-etnaviv-fixes' of git://git.pengutronix.de/lst/linux:
  drm/etnaviv: block 64K of address space behind each cmdstream
  drm/etnaviv: ensure write caches are flushed at end of user cmdstream

vmwgfx cleanups and fixes.

* 'drm-vmwgfx-fixes' of ssh://people.freedesktop.org/~syeh/repos_linux:
  drm/vmwgfx: Adjust checks for null pointers in 13 functions
  drm/vmwgfx: Use memdup_user() rather than duplicating its implementation
  drm/vmwgfx: Use kmalloc_array() in vmw_surface_define_ioctl()
  drm/vmwgfx: Avoid validating views on view destruction
  drm/vmwgfx: Limit the user-space command buffer size
  drm/vmwgfx: Remove a leftover debug printout
  drm/vmwgfx: Allow resource relocations on byte boundaries
  drm/vmwgfx: Enable SVGA_3D_CMD_DX_TRANSFER_FROM_BUFFER command
  drm/vmwgfx: Remove call to reservation_object_test_signaled_rcu before wait
  drm/vmwgfx: Replace numeric parameter like 0444 with macro

One small fix for Armada, where the clock prepare/enable counts were
going awry.

* 'drm-armada-fixes' of git://git.armlinux.org.uk/~rmk/linux-arm:
  drm/armada: fix clock counts

This are some fixes which I hoped to still get into v4.9. I used to
test them here since about 2 weeks and Meng came around to test it
on the second platform making use of this IP too, so they are well
tested now.

* 'fixes-for-v4.9-rc2' of http://git.agner.ch/git/linux-drm-fsl-dcu:
  drm/fsl-dcu: enable pixel clock when enabling CRTC
  drm/fsl-dcu: do not transfer registers in mode_set_nofb
  drm/fsl-dcu: do not transfer registers on plane init
  drm/fsl-dcu: enable TCON bypass mode by default

Pull power management fix from Rafael Wysocki:
 "This fixes the pointer arithmetics mess-up in the cpufreq core
  introduced by one of recent commits and leading to all kinds of
  breakage from kernel crashes to incorrect governor decisions (Sergey
  Senozhatsky)"

* tag 'pm-4.9-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  cpufreq: fix overflow in cpufreq_table_find_index_dl()

* pm-cpufreq:
  cpufreq: fix overflow in cpufreq_table_find_index_dl()

Pull arm64 fixes from Will Deacon:
 "Most of these are CC'd for stable, but there are a few fixing issues
  introduced during the recent merge window too.

  There's also a fix for the xgene PMU driver, but it seemed daft to
  send as a separate pull request, so I've included it here with the
  rest of the fixes.

   - Fix ACPI boot due to recent broken NUMA changes
   - Fix remote enabling of CPU features requiring PSTATE bit manipulation
   - Add address range check when emulating user cache maintenance
   - Fix LL/SC loops that allow compiler to introduce memory accesses
   - Fix recently added write_sysreg_s macro
   - Ensure MDCR_EL2 is initialised on qemu targets without a PMU
   - Avoid kaslr breakage due to MODVERSIONs and DYNAMIC_FTRACE
   - Correctly drive recent ld when building relocatable Image
   - Remove junk IS_ERR check from xgene PMU driver added during merge window
   - pr_cont fixes after core changes in the merge window"

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  arm64: remove pr_cont abuse from mem_init
  arm64: fix show_regs fallout from KERN_CONT changes
  arm64: kernel: force ET_DYN ELF type for CONFIG_RELOCATABLE=y
  arm64: suspend: Reconfigure PSTATE after resume from idle
  arm64: mm: Set PSTATE.PAN from the cpu_enable_pan() call
  arm64: cpufeature: Schedule enable() calls instead of calling them via IPI
  arm64: Cortex-A53 errata workaround: check for kernel addresses
  arm64: percpu: rewrite ll/sc loops in assembly
  arm64: swp emulation: bound LL/SC retries before rescheduling
  arm64: sysreg: Fix use of XZR in write_sysreg_s
  arm64: kaslr: keep modules close to the kernel when DYNAMIC_FTRACE=y
  arm64: kernel: Init MDCR_EL2 even in the absence of a PMU
  perf: xgene: Remove bogus IS_ERR() check
  arm64: kernel: numa: fix ACPI boot cpu numa node mapping
  arm64: kaslr: fix breakage with CONFIG_MODVERSIONS=y

Pull Ceph fixes from Ilya Dryomov:
 "An rbd exclusive-lock edge case fix and several filesystem fixups.

  Nikolay's error path patch is tagged for stable, everything else but
  readdir vs frags race was introduced in this merge window"

* tag 'ceph-for-4.9-rc2' of git://github.com/ceph/ceph-client:
  ceph: fix non static symbol warning
  ceph: fix uninitialized dentry pointer in ceph_real_mount()
  ceph: fix readdir vs fragmentation race
  ceph: fix error handling in ceph_read_iter
  rbd: don't retry watch reregistration if header object is gone
  rbd: don't wait for the lock forever if blacklisted

Pull MMC fixes from Ulf Hansson:
 "Here are some mmc fixes intended for v4.9 rc2.

  This time I have also included a few changes for a memstick driver
  which has a corresponding mmc driver. They use the same USB device as
  parent, hence both needs to play nice with runtime PM, which they
  didn't.

  MMC core:
   - Update MAINTAINERS as the mmc tree moved to kernel.org
   - A few fixes for HS400es mode
   - A few other minor fixes

  MMC host:
   - sdhci: Fix an issue when dealing with stop commands
   - sdhci-pci: Fix a bus power failure issue
   - sdhci-esdhc-imx: Correct two register accesses
   - sdhci-of-arasan: Fix the 1.8V I/O signal switch behaviour
   - rtsx_usb_sdmmc: Fix runtime PM issues

  Other: (Because of no maintainer)
   - memstick: rtsx_usb_ms: Fix runtime PM issues"

* tag 'mmc-v4.9-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/mmc:
  MAINTAINERS: mmc: Move the mmc tree to kernel.org
  memstick: rtsx_usb_ms: Manage runtime PM when accessing the device
  memstick: rtsx_usb_ms: Runtime resume the device when polling for cards
  mmc: rtsx_usb_sdmmc: Handle runtime PM while changing the led
  mmc: rtsx_usb_sdmmc: Avoid keeping the device runtime resumed when unused
  mmc: sdhci: cast unsigned int to unsigned long long to avoid unexpeted error
  mmc: sdhci-esdhc-imx: Correct two register accesses
  mmc: sdhci-pci: Fix bus power failing to enable for some Intel controllers
  mmc: sdhci-pci: Let devices define their own sdhci_ops
  mmc: sdhci: Rename sdhci_set_power() to sdhci_set_power_noreg()
  mmc: sdhci: Fix SDHCI_QUIRK2_STOP_WITH_TC
  mmc: core: Annotate cmd_hdr as __le32
  mmc: sdhci-of-arasan: add sdhci_arasan_voltage_switch for arasan, 5.1
  mmc: core: changes frequency to hs_max_dtr when selecting hs400es
  mmc: core: switch to 1V8 or 1V2 for hs400es mode
  mmc: block: add missing header dependencies
  mmc: sdhci-of-arasan: Fix non static symbol warning