summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/amd/amdgpu
AgeCommit message (Collapse)Author
2025-11-14PCI: Fix restoring BARs on BAR resize rollback pathIlpo Järvinen
BAR resize operation is implemented in the pci_resize_resource() and pbus_reassign_bridge_resources() functions. pci_resize_resource() can be called either from __resource_resize_store() from sysfs or directly by the driver for the Endpoint Device. The pci_resize_resource() requires that caller has released the device resources that share the bridge window with the BAR to be resized as otherwise the bridge window is pinned in place and cannot be changed. pbus_reassign_bridge_resources() rolls back resources if the resize operation fails, but rollback is performed only for the bridge windows. Because releasing the device resources are done by the caller of the BAR resize interface, these functions performing the BAR resize do not have access to the device resources as they were before the resize. pbus_reassign_bridge_resources() could try __pci_bridge_assign_resources() after rolling back the bridge windows as they were, however, it will not guarantee the resource are assigned due to differences in how FW and the kernel assign the resources (alignment of the start address and tail). To perform rollback robustly, the BAR resize interface has to be altered to also release the device resources that share the bridge window with the BAR to be resized. Also, remove restoring from the entries failed list as saved list should now contain both the bridge windows and device resources so the extra restore is duplicated work. Some drivers (currently only amdgpu) want to prevent releasing some resources. Add exclude_bars param to pci_resize_resource() and make amdgpu pass its register BAR (BAR 2 or 5), which should never be released during resize operation. Normally 64-bit prefetchable resources do not share a bridge window with the 32-bit only register BAR, but there are various fallbacks in the resource assignment logic which may make the resources share the bridge window in rare cases. This change (together with the driver side changes) is to counter the resource releases that had to be done to prevent resource tree corruption in the ("PCI: Release assigned resource before restoring them") change. As such, it likely restores functionality in cases where device resources were released to avoid resource tree conflicts which appeared to be "working" when such conflicts were not correctly detected by the kernel. Reported-by: Simon Richter <Simon.Richter@hogyros.de> Link: https://lore.kernel.org/linux-pci/f9a8c975-f5d3-4dd2-988e-4371a1433a60@hogyros.de/ Reported-by: Alex Bennée <alex.bennee@linaro.org> Link: https://lore.kernel.org/linux-pci/874irqop6b.fsf@draig.linaro.org/ Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> [bhelgaas: squash amdgpu BAR selection from https://lore.kernel.org/r/20251114103053.13778-1-ilpo.jarvinen@linux.intel.com] Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Tested-by: Alex Bennée <alex.bennee@linaro.org> # AVA, AMD GPU Reviewed-by: Christian König <christian.koenig@amd.com> Link: https://patch.msgid.link/20251113162628.5946-7-ilpo.jarvinen@linux.intel.com
2025-11-14drm/amdgpu: Use amdgpu by default on SI dedicated GPUs (v2)Timur Kristóf
Now that the DC analog connector support and VCE1 support landed, amdgpu is at feature parity with the old radeon driver on SI dGPUs. Enabling the amdgpu driver by default for SI dGPUs has the following benefits: - More stable OpenGL support through RadeonSI - Vulkan support through RADV - Improved performance - Better display features through DC Users who want to keep using the old driver can do so using: amdgpu.si_support=0 radeon.si_support=1 v2: - Update documentation in Kconfig file Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Timur Kristóf <timur.kristof@gmail.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-14drm/amdgpu: Use amdgpu by default on CIK dedicated GPUsTimur Kristóf
The amdgpu driver has been working well on CIK dGPUs for years. Now that the DC analog connector support landed, amdgpu is at feature parity with the old radeon driver on CIK dGPUs. Enabling the amdgpu driver by default for CIK dGPUs has the following benefits: - More stable OpenGL support through RadeonSI - Vulkan support through RADV - Improved performance - Better display features through DC Users who want to keep using the old driver can do so using: amdgpu.cik_support=0 radeon.cik_support=1 v2: - Update documentation in Kconfig file v3: - Rebase documentation updates (Alex) Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Timur Kristóf <timur.kristof@gmail.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-14drm/amdgpu: Fix the issue of missing ras message on sriov hostYiPeng Chai
This code only applies to amdgpu processing poison consumption after uniras is enabled, but not to sriov. Signed-off-by: YiPeng Chai <YiPeng.Chai@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-14drm/amdgpu: Add lock to serialize sriov command executionYiPeng Chai
Add lock to serialize sriov command execution. Signed-off-by: YiPeng Chai <YiPeng.Chai@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-14drm/amdgpu: Synchronize sriov host to add block_mmsch bit fieldYiPeng Chai
Synchronize sriov host to add block_mmsch bit field. Signed-off-by: YiPeng Chai <YiPeng.Chai@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-14drm/amdgpu: use GFP_ATOMIC instead of NOWAIT in the critical pathChristian König
Otherwise job submissions can fail with ENOMEM. We probably need to re-design the per VMID tracking at some point. Signed-off-by: Christian König <christian.koenig@amd.com> Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4258 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-14drm/amdgpu: avoid memory allocation in the critical code path v3Christian König
When we run out of VMIDs we need to wait for some to become available. Previously we were using a dma_fence_array for that, but this means that we have to allocate memory. Instead just wait for the first not signaled fence from the least recently used VMID to signal. That is not as efficient since we end up in this function multiple times again, but allocating memory can easily fail or deadlock if we have to wait for memory to become available. v2: remove now unused VM manager fields v3: fix dma_fence reference Signed-off-by: Christian König <christian.koenig@amd.com> Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4258 Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-14drm/amdgpu: Enable xgmi extended peer links for sriov guestWill Aitken
The amd-smi tool relies on extended peer link information to report xgmi link metrics. The necessary xgmi ta command, GET_EXTEND_PEER_LINKS, has been enabled in the host driver and this change is necessary for the guest to make use of it. To handle the case where the host driver does not have the latest xgmi ta, the guest driver checks for guest support through a pf2vf feature flag before invoking psp. Signed-off-by: Will Aitken <wiaitken@amd.com> Reviewed-by: Lijo Lazar <lijo.lazar@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-14drm/amdgpu: Update headers for sriov xgmi ext peer link support feature flagWill Aitken
Adds new sriov msg flag to match host, feature flag in the amdgim enum, and a wrapper macro to check it. Signed-off-by: Will Aitken <wiaitken@amd.com> Reviewed-by: Lijo Lazar <lijo.lazar@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-14drm/amdgpu: Refactor sriov xgmi topology filling to common codeWill Aitken
amdgpu_xgmi_fill_topology_info and psp_xgmi_reflect_topology_info perform the same logic of copying topology info of one node to every other node in the hive. Instead of having two functions that purport to do the same thing, this refactoring moves the logic of the fill function to the reflect function and adds reflecting port number info as well for complete functionality. Signed-off-by: Will Aitken <wiaitken@amd.com> Reviewed-by: Lijo Lazar <lijo.lazar@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-14drm/amdgpu: Use amdgpu by default on CIK dedicated GPUsTimur Kristóf
The amdgpu driver has been working well on CIK dGPUs for years. Now that the DC analog connector support landed, these GPUs are at feature parity with the old radeon driver. Additionally, amdgpu yields extra performance, supports Vulkan and provides more display features through DC as well as more robust power management. Signed-off-by: Timur Kristóf <timur.kristof@gmail.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-14drm/amdgpu: Refactor how SI and CIK support is determinedTimur Kristóf
Move the determination into a separate function. Change amdgpu.si_support and amdgpu.cik_support so that their default value is -1 (default). This prepares the code for changing the default driver based on the chip. Also adjust the module param documentation. Signed-off-by: Timur Kristóf <timur.kristof@gmail.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-14drm/amdgpu: Avoid xgmi register accessLijo Lazar
On single GPU systems, avoid accesses to XGMI link registers. Signed-off-by: Lijo Lazar <lijo.lazar@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu/jpeg: Add parse_cs for JPEG5_0_1Sathishkumar S
enable parse_cs callback for JPEG5_0_1. Signed-off-by: Sathishkumar S <sathishkumar.sundararaju@amd.com> Reviewed-by: Leo Liu <leo.liu@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> (cherry picked from commit 547985579932c1de13f57f8bcf62cd9361b9d3d3) Cc: stable@vger.kernel.org
2025-11-11drm/amd/amdgpu: Ensure isp_kernel_buffer_alloc() creates a new BOSultan Alsawaf
When the BO pointer provided to amdgpu_bo_create_kernel() points to non-NULL, amdgpu_bo_create_kernel() takes it as a hint to pin that address rather than allocate a new BO. This functionality is never desired for allocating ISP buffers. A new BO should always be created when isp_kernel_buffer_alloc() is called, per the description for isp_kernel_buffer_alloc(). Ensure this by zeroing *bo right before the amdgpu_bo_create_kernel() call. Fixes: 55d42f616976 ("drm/amd/amdgpu: Add helper functions for isp buffers") Reviewed-by: Mario Limonciello (AMD) <superm1@kernel.org> Reviewed-by: Pratap Nirujogi <pratap.nirujogi@amd.com> Signed-off-by: Sultan Alsawaf <sultan@kerneltoast.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> (cherry picked from commit 73c8c29baac7f0c7e703d92eba009008cbb5228e)
2025-11-11drm/amdgpu: fix lock warning in amdgpu_userq_fence_driver_processJesse.Zhang
Fix a potential deadlock caused by inconsistent spinlock usage between interrupt and process contexts in the userq fence driver. The issue occurs when amdgpu_userq_fence_driver_process() is called from both: - Interrupt context: gfx_v11_0_eop_irq() -> amdgpu_userq_fence_driver_process() - Process context: amdgpu_eviction_fence_suspend_worker() -> amdgpu_userq_fence_driver_force_completion() -> amdgpu_userq_fence_driver_process() In interrupt context, the spinlock was acquired without disabling interrupts, leaving it in {IN-HARDIRQ-W} state. When the same lock is acquired in process context, the kernel detects inconsistent locking since the process context acquisition would enable interrupts while holding a lock previously acquired in interrupt context. Kernel log shows: [ 4039.310790] inconsistent {IN-HARDIRQ-W} -> {HARDIRQ-ON-W} usage. [ 4039.310804] kworker/7:2/409 [HC0[0]:SC0[0]:HE1:SE1] takes: [ 4039.310818] ffff9284e1bed000 (&fence_drv->fence_list_lock){?...}-{3:3}, [ 4039.310993] {IN-HARDIRQ-W} state was registered at: [ 4039.311004] lock_acquire+0xc6/0x300 [ 4039.311018] _raw_spin_lock+0x39/0x80 [ 4039.311031] amdgpu_userq_fence_driver_process.part.0+0x30/0x180 [amdgpu] [ 4039.311146] amdgpu_userq_fence_driver_process+0x17/0x30 [amdgpu] [ 4039.311257] gfx_v11_0_eop_irq+0x132/0x170 [amdgpu] Fix by using spin_lock_irqsave()/spin_unlock_irqrestore() to properly manage interrupt state regardless of calling context. Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Jesse Zhang <Jesse.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> (cherry picked from commit ded3ad780cf97a04927773c4600823b84f7f3cc2) Cc: stable@vger.kernel.org
2025-11-11drm/amdgpu: jump to the correct label on failurePierre-Eric Pelloux-Prayer
drm_sched_entity_init wasn't called yet, so the only thing to do is to release allocated memory. This doesn't fix any bug since entity is zero allocated and drm_sched_entity_fini does nothing in this case. Signed-off-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com> Acked-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> (cherry picked from commit ec49374ccb8da86b465beaf09c367f3dfd648d8f)
2025-11-11drm/amdgpu: disable peer-to-peer access for DCC-enabled GC12 VRAM surfacesVitaly Prosyak
Certain multi-GPU configurations (especially GFX12) may hit data corruption when a DCC-compressed VRAM surface is shared across GPUs using peer-to-peer (P2P) DMA transfers. Such surfaces rely on device-local metadata and cannot be safely accessed through a remote GPU’s page tables. Attempting to import a DCC-enabled surface through P2P leads to incorrect rendering or GPU faults. This change disables P2P for DCC-enabled VRAM buffers that are contiguous and allocated on GFX12+ hardware. In these cases, the importer falls back to the standard system-memory path, avoiding invalid access to compressed surfaces. Future work could consider optional migration (VRAM→System→VRAM) if a performance regression is observed when `attach->peer2peer = false`. Tested on: - Dual RX 9700 XT (Navi4x) setup - GNOME and Wayland compositor scenarios - Confirmed no corruption after disabling P2P under these conditions v2: Remove check TTM_PL_VRAM & TTM_PL_FLAG_CONTIGUOUS. v3: simplify for upsteam and fix ip version check (Alex) Suggested-by: Christian König <christian.koenig@amd.com> Signed-off-by: Vitaly Prosyak <vitaly.prosyak@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> (cherry picked from commit 9dff2bb709e6fbd97e263fd12bf12802d2b5a0cf) Cc: stable@vger.kernel.org
2025-11-11drm/amdgpu/vce1: Workaround PLL timeout on FirePro W9000Timur Kristóf
Sometimes the VCE PLL times out waiting for CTLACK/CTLACK2. When it happens, the VCE still works, but much slower. Observed on a Tahiti GPU, but not all: - FirePro W9000 has the issue - Radeon R9 280X not affected - Radeon HD 7990 not affected As a workaround, on the affected chip just don't put the VCE PLL in sleep mode. Leaving the VCE PLL in bypass mode or reset mode both work. Using bypass mode is simpler. Signed-off-by: Timur Kristóf <timur.kristof@gmail.com> Acked-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu/vce1: Enable VCE1 on Tahiti, Pitcairn, Cape Verde GPUsTimur Kristóf
Add the VCE1 IP block to the SI GPUs that have it. Advertise the encoder capabilities corresponding to VCE1, so the userspace applications can detect and use it. Signed-off-by: Timur Kristóf <timur.kristof@gmail.com> Co-developed-by: Alexandre Demers <alexandre.f.demers@gmail.com> Signed-off-by: Alexandre Demers <alexandre.f.demers@gmail.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu/vce1: Ensure VCPU BO is in lower 32-bit address space (v3)Timur Kristóf
Based on research and ideas by Alexandre and Christian. VCE1 actually executes its code from the VCPU BO. Due to various hardware limitations, the VCE1 requires the VCPU BO to be in the low 32 bit address range. However, VRAM is typically mapped at the high address range, which means the VCPU can't access VRAM through the FB aperture. To solve this, we write a few page table entries to map the VCPU BO in the GART address range. And we make sure that the GART is located at the low address range. That way the VCE1 can access the VCPU BO. v2: - Adjust to v2 of the GART helper commit. - Add empty line to multi-line comment. v3: - Instead of relying on gmc_v6 to set the GART space before GTT, add a new function amdgpu_vce_required_gart_pages() which is called from amdgpu_gtt_mgr_init() directly. Signed-off-by: Timur Kristóf <timur.kristof@gmail.com> Co-developed-by: Alexandre Demers <alexandre.f.demers@gmail.com> Signed-off-by: Alexandre Demers <alexandre.f.demers@gmail.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu: Check if AID is active before accessLijo Lazar
Access XGMI registers only if AID is active. Signed-off-by: Lijo Lazar <lijo.lazar@amd.com> Reviewed-by: Asad Kamal <asad.kamal@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu/vce1: Implement VCE1 IP block (v2)Timur Kristóf
Implement the necessary functionality to support the VCE1. This implementation is based on: - VCE2 code from amdgpu - VCE1 code from radeon (the old driver) - Some trial and error A subsequent commit will ensure correct mapping for the VCPU BO, which will make this actually work. v2: - Use memset_io more. - Use memcpy_toio more. - Remove __func__ from warnings. - Don't reserve and map the VCPU BO anymore. - Add empty line to multi-line comments Signed-off-by: Timur Kristóf <timur.kristof@gmail.com> Co-developed-by: Alexandre Demers <alexandre.f.demers@gmail.com> Signed-off-by: Alexandre Demers <alexandre.f.demers@gmail.com> Acked-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu/vce1: Load VCE1 firmwareTimur Kristóf
Load VCE1 firmware using amdgpu_ucode_request, just like it is done for other VCE versions. All SI chips share the same VCE1 firmware file: vce_1_0_0.bin which will be sent to linux-firmware soon. Signed-off-by: Timur Kristóf <timur.kristof@gmail.com> Co-developed-by: Alexandre Demers <alexandre.f.demers@gmail.com> Signed-off-by: Alexandre Demers <alexandre.f.demers@gmail.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu/vce1: Clean up register definitionsTimur Kristóf
The sid.h header contained some VCE1 register definitions, but they were using byte offsets (probably copied from the old radeon driver). Move all of these to the proper VCE1 headers and ensure they are in dword offsets. Also add the register definitions that we need for the firmware validation mechanism in VCE1. Signed-off-by: Timur Kristóf <timur.kristof@gmail.com> Co-developed-by: Alexandre Demers <alexandre.f.demers@gmail.com> Signed-off-by: Alexandre Demers <alexandre.f.demers@gmail.com> Acked-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu/vce: Clear VCPU BO, don't unmap/unreserve (v4)Timur Kristóf
The VCPU BO doesn't only contain the VCE firmware but also other ranges that the VCE uses for its stack and data. Let's initialize this to zero to avoid having garbage in the VCPU BO. Additionally, don't unmap/unreserve the VCPU BO. The VCPU BO needs to stay at the same location before and after sleep/resume because the FW code is not relocatable once it's started. Signed-off-by: Timur Kristóf <timur.kristof@gmail.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu/vce: Move firmware load to amdgpu_vce_early_initTimur Kristóf
Try to load the VCE firmware at early_init. When the correct firmware is not found, return -ENOENT. This way, the driver initialization will complete even without VCE, and the GPU will be functional, albeit without video encoding capabilities. This is necessary because we are planning to add support for the VCE1, and AMD hasn't yet publised the correct firmware for this version. So we need to anticipate that users will try to boot amdgpu on SI GPUs without the correct VCE1 firmware present on their system. Signed-off-by: Timur Kristóf <timur.kristof@gmail.com> Reviewed-by: Christian König <christian.koenig@amd.com> Reviewed-by: Leo Liu <leo.liu@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu/ttm: Use GART helper to map VRAM pages (v2)Timur Kristóf
Use the GART helper function introduced in the previous commit to map the VRAM pages of the transfer window to GART. No functional changes, just code cleanup. Split this into a separate commit to make it easier to bisect, in case there are problems in the future. Signed-off-by: Timur Kristóf <timur.kristof@gmail.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu: Use DC by default on SI dGPUsTimur Kristóf
Now that DC supports analog connectors, it has reached feature parity with the legacy non-DC display driver on SI dGPUs. Use the DC display driver by default on SI dGPUs, unless it is explicitly disabled using the amdgpu.dc=0 module parameter. DC brings proper support for DP/HDMI audio, DP MST, 10-bit colors, some HDR features, atomic modesetting, etc. Also clarify the comment about what is missing to have full DC support for CIK APUs. Signed-off-by: Timur Kristóf <timur.kristof@gmail.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu/gart: Add helper to bind VRAM pages (v2)Timur Kristóf
Binds pages that located in VRAM to the GART page table. Useful when a kernel BO is located in VRAM but needs to be accessed from the GART address space, for example to give a kernel BO a 32-bit address when GART is placed in LOW address space. v2: - Refactor function to be more reusable Signed-off-by: Timur Kristóf <timur.kristof@gmail.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu/gmc6: Place gart at low address rangeTimur Kristóf
Instead of using a best-fit algorithm to determine which part of the VMID 0 address space to use for GART, always use the low address range. A subsequent commit will use this to map the VCPU BO in GART for the VCE1 IP block. Split this into a separate patch to make it easier to bisect, in case there are any errors in the future. Signed-off-by: Timur Kristóf <timur.kristof@gmail.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu/userqueue: Remove duplicate amdgpu_reset.h headerJiapeng Chong
./drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c: amdgpu_reset.h is included more than once. Reported-by: Abaci Robot <abaci@linux.alibaba.com> Closes: https://bugzilla.openanolis.cn/show_bug.cgi?id=26930 Signed-off-by: Jiapeng Chong <jiapeng.chong@linux.alibaba.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu: resume MES scheduling after user queue hang detection and recoveryJesse.Zhang
This patch ensures the Micro-Engine Scheduler (MES) is properly resumed after detecting and recovering from a user queue hang condition. Key changes: 1. Track when a hung user queue is detected using found_hung_queue flag 2. Call amdgpu_mes_resume() to restart MES scheduling after completing the hang recovery process 3. This complements the existing recovery steps (fence force completion and device wedging) by ensuring the scheduler can process new work Without this resume call, the MES scheduler may remain in a paused state even after the hung queue has been handled, preventing newly submitted work from being processed and leading to system stalls. Acked-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Jesse Zhang <Jesse.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu/jpeg: Add parse_cs for JPEG5_0_1Sathishkumar S
enable parse_cs callback for JPEG5_0_1. Signed-off-by: Sathishkumar S <sathishkumar.sundararaju@amd.com> Reviewed-by: Leo Liu <leo.liu@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amd/amdgpu: Ensure isp_kernel_buffer_alloc() creates a new BOSultan Alsawaf
When the BO pointer provided to amdgpu_bo_create_kernel() points to non-NULL, amdgpu_bo_create_kernel() takes it as a hint to pin that address rather than allocate a new BO. This functionality is never desired for allocating ISP buffers. A new BO should always be created when isp_kernel_buffer_alloc() is called, per the description for isp_kernel_buffer_alloc(). Ensure this by zeroing *bo right before the amdgpu_bo_create_kernel() call. Fixes: 55d42f616976 ("drm/amd/amdgpu: Add helper functions for isp buffers") Reviewed-by: Mario Limonciello (AMD) <superm1@kernel.org> Reviewed-by: Pratap Nirujogi <pratap.nirujogi@amd.com> Signed-off-by: Sultan Alsawaf <sultan@kerneltoast.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu: optimize timeout implemention in ras_eeprom_update_record_numTao Zhou
The busy status returned by ras_eeprom_update_record_num may not be an error, increase timeout to exclude false busy status. Also add more comments to make the code readable. v2: define a macro for the timeout value. Signed-off-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu: add RAS bad page threshold handling for PMFW manages eepromTao Zhou
Check if bad page threshold is reached and take actions accordingly. v2: remove rma message sent to smu when pmfw manages eeprom. v3: add null pointer check for con. Signed-off-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu: fix lock warning in amdgpu_userq_fence_driver_processJesse.Zhang
Fix a potential deadlock caused by inconsistent spinlock usage between interrupt and process contexts in the userq fence driver. The issue occurs when amdgpu_userq_fence_driver_process() is called from both: - Interrupt context: gfx_v11_0_eop_irq() -> amdgpu_userq_fence_driver_process() - Process context: amdgpu_eviction_fence_suspend_worker() -> amdgpu_userq_fence_driver_force_completion() -> amdgpu_userq_fence_driver_process() In interrupt context, the spinlock was acquired without disabling interrupts, leaving it in {IN-HARDIRQ-W} state. When the same lock is acquired in process context, the kernel detects inconsistent locking since the process context acquisition would enable interrupts while holding a lock previously acquired in interrupt context. Kernel log shows: [ 4039.310790] inconsistent {IN-HARDIRQ-W} -> {HARDIRQ-ON-W} usage. [ 4039.310804] kworker/7:2/409 [HC0[0]:SC0[0]:HE1:SE1] takes: [ 4039.310818] ffff9284e1bed000 (&fence_drv->fence_list_lock){?...}-{3:3}, [ 4039.310993] {IN-HARDIRQ-W} state was registered at: [ 4039.311004] lock_acquire+0xc6/0x300 [ 4039.311018] _raw_spin_lock+0x39/0x80 [ 4039.311031] amdgpu_userq_fence_driver_process.part.0+0x30/0x180 [amdgpu] [ 4039.311146] amdgpu_userq_fence_driver_process+0x17/0x30 [amdgpu] [ 4039.311257] gfx_v11_0_eop_irq+0x132/0x170 [amdgpu] Fix by using spin_lock_irqsave()/spin_unlock_irqrestore() to properly manage interrupt state regardless of calling context. Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Jesse Zhang <Jesse.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu: try for more times if RAS bad page number is not updatedTao Zhou
RAS info update in PMFW is time cost, wait for it. Signed-off-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu: jump to the correct label on failurePierre-Eric Pelloux-Prayer
drm_sched_entity_init wasn't called yet, so the only thing to do is to release allocated memory. This doesn't fix any bug since entity is zero allocated and drm_sched_entity_fini does nothing in this case. Signed-off-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com> Acked-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu: get RAS bad page address from MCA addressTao Zhou
Instead of from physical address. v2: add comment to make the code more readable Signed-off-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amd: Clarify that amdgpu.audio only works for non-DCMario Limonciello (AMD)
The comment already explains it but the module parameter help text doesn't. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4684 Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org> Acked-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu: disable peer-to-peer access for DCC-enabled GC12 VRAM surfacesVitaly Prosyak
Certain multi-GPU configurations (especially GFX12) may hit data corruption when a DCC-compressed VRAM surface is shared across GPUs using peer-to-peer (P2P) DMA transfers. Such surfaces rely on device-local metadata and cannot be safely accessed through a remote GPU’s page tables. Attempting to import a DCC-enabled surface through P2P leads to incorrect rendering or GPU faults. This change disables P2P for DCC-enabled VRAM buffers that are contiguous and allocated on GFX12+ hardware. In these cases, the importer falls back to the standard system-memory path, avoiding invalid access to compressed surfaces. Future work could consider optional migration (VRAM→System→VRAM) if a performance regression is observed when `attach->peer2peer = false`. Tested on: - Dual RX 9700 XT (Navi4x) setup - GNOME and Wayland compositor scenarios - Confirmed no corruption after disabling P2P under these conditions v2: Remove check TTM_PL_VRAM & TTM_PL_FLAG_CONTIGUOUS. v3: simplify for upsteam and fix ip version check (Alex) Suggested-by: Christian König <christian.koenig@amd.com> Signed-off-by: Vitaly Prosyak <vitaly.prosyak@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11drm/amdgpu: load RAS bad page from PMFW in page retirementTao Zhou
In legacy way, bad page is queried from MCA registers, switch to getting it from PMFW when PMFW manages eeprom data. Signed-off-by: Tao Zhou <tao.zhou1@amd.com> Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2025-11-11Merge tag 'amd-drm-next-6.19-2025-11-07' of ↵Dave Airlie
https://gitlab.freedesktop.org/agd5f/linux into drm-next amd-drm-next-6.19-2025-11-07: amdgpu: - Misc fixes - HMM cleanup - HDP flush rework - RAS updates - SMU 13.x updates - SI DPM cleanup - Suspend rework - UQ reset support - Replay/PSR fixes - HDCP updates - DC PMO fixes - DC pstate fixes - DCN4 fixes - GPUVM fixes - SMU 13 parition metrics - Fix possible fence leak in job cleanup - Hibernation fix - MST fix amdkfd: - HMM cleanup - Process cleanup fix Signed-off-by: Dave Airlie <airlied@redhat.com> From: Alex Deucher <alexander.deucher@amd.com> Link: https://patch.msgid.link/20251107145938.26669-1-alexander.deucher@amd.com
2025-11-07Merge tag 'drm-misc-next-2025-11-05-1' of ↵Dave Airlie
https://gitlab.freedesktop.org/drm/misc/kernel into drm-next drm-misc-next for v6.19-rc1: UAPI Changes: - Add userptr support to ivpu. - Add IOCTL's for resource and telemetry data in amdxdna. Core Changes: - Improve some atomic state checking handling. - drm/client updates. - Use forward declarations instead of including drm_print.h - RUse allocation flags in ttm_pool/device_init and allow specifying max useful pool size and propagate ENOSPC. - Updates and fixes to scheduler and bridge code. - Add support for quirking DisplayID checksum errors. Driver Changes: - Assorted cleanups and fixes in rcar-du, accel/ivpu, panel/nv3052cf, sti, imxm, accel/qaic, accel/amdxdna, imagination, tidss, sti, panthor, vkms. - Add Samsung S6E3FC2X01 DDIC/AMS641RW, Synaptics TDDI series DSI, TL121BVMS07-00 (IL79900A) panels. - Add mali MediaTek MT8196 SoC gpu support. - Add etnaviv GC8000 Nano Ultra VIP r6205 support. - Document powervr ge7800 support in the devicetree. Signed-off-by: Dave Airlie <airlied@redhat.com> From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: https://patch.msgid.link/5afae707-c9aa-4a47-b726-5e1f1aa7a106@linux.intel.com
2025-11-06drm/amdgpu: Fix wait after reset sequence in S3Lijo Lazar
For a mode-1 reset done at the end of S3 on PSPv11 dGPUs, only check if TOS is unloaded. Fixes: 32f73741d6ee ("drm/amdgpu: Wait for bootloader after PSPv11 reset") Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4649 Signed-off-by: Lijo Lazar <lijo.lazar@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> (cherry picked from commit 1ad25fd272753db14c5d1cc8c68e20ce01f3f888)
2025-11-06drm/amd: Fix suspend failure with secure display TAMario Limonciello
commit c760bcda83571 ("drm/amd: Check whether secure display TA loaded successfully") attempted to fix extra messages, but failed to port the cleanup that was in commit 5c6d52ff4b61e ("drm/amd: Don't try to enable secure display TA multiple times") to prevent multiple tries. Add that to the failure handling path even on a quick failure. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4679 Fixes: c760bcda8357 ("drm/amd: Check whether secure display TA loaded successfully") Signed-off-by: Mario Limonciello <mario.limonciello@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> (cherry picked from commit 4104c0a454f6a4d1e0d14895d03c0e7bdd0c8240)
2025-11-06drm/amdgpu: fix gpu page fault after hibernation on PF passthroughSamuel Zhang
On PF passthrough environment, after hibernate and then resume, coralgemm will cause gpu page fault. Mode1 reset happens during hibernate, but partition mode is not restored on resume, register mmCP_HYP_XCP_CTL and mmCP_PSP_XCP_CTL is not right after resume. When CP access the MQD BO, wrong stride size is used, this will cause out of bound access on the MQD BO, resulting page fault. The fix is to ensure gfx_v9_4_3_switch_compute_partition() is called when resume from a hibernation. KFD resume is called separately during a reset recovery or resume from suspend sequence. Hence it's not required to be called as part of partition switch. Signed-off-by: Samuel Zhang <guoqing.zhang@amd.com> Reviewed-by: Lijo Lazar <lijo.lazar@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> (cherry picked from commit 5d1b32cfe4a676fe552416cb5ae847b215463a1a)
generated by cgit v1.2.3 (git 2.46.0) at 2025-12-10 16:07:55 +0000