summaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2019-05-24kvm: svm/avic: fix off-by-one in checking host APIC IDSuthikulpanit, Suravee
Current logic does not allow VCPU to be loaded onto CPU with APIC ID 255. This should be allowed since the host physical APIC ID field in the AVIC Physical APIC table entry is an 8-bit value, and APIC ID 255 is valid in system with x2APIC enabled. Instead, do not allow VCPU load if the host APIC ID cannot be represented by an 8-bit value. Also, use the more appropriate AVIC_PHYSICAL_ID_ENTRY_HOST_PHYSICAL_ID_MASK instead of AVIC_MAX_PHYSICAL_ID_COUNT. Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com> Cc: stable@vger.kernel.org Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-05-24KVM: selftests: do not blindly clobber registers in guest asmPaolo Bonzini
The guest_code of sync_regs_test is assuming that the compiler will not touch %r11 outside the asm that increments it, which is a bit brittle. Instead, we can increment a variable and use a dummy asm to ensure the increment is not optimized away. However, we also need to use a callee-save register or the compiler will insert a save/restore around the vmexit, breaking the whole idea behind the test. (Yes, "if it ain't broken...", but I would like the test to be clean before it is copied into the upcoming s390 selftests). Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-05-24KVM: selftests: Remove duplicated TEST_ASSERT in hyperv_cpuid.cThomas Huth
The check for entry->index == 0 is done twice. One time should be sufficient. Suggested-by: Vitaly Kuznetsov <vkuznets@redhat.com> Signed-off-by: Thomas Huth <thuth@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-05-24KVM: LAPIC: Expose per-vCPU timer_advance_ns to userspaceWanpeng Li
Expose per-vCPU timer_advance_ns to userspace, so it is able to query the auto-adjusted value. Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Radim Krčmář <rkrcmar@redhat.com> Cc: Sean Christopherson <sean.j.christopherson@intel.com> Cc: Liran Alon <liran.alon@oracle.com> Signed-off-by: Wanpeng Li <wanpengli@tencent.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-05-24KVM: LAPIC: Fix lapic_timer_advance_ns parameter overflowWanpeng Li
After commit c3941d9e0 (KVM: lapic: Allow user to disable adaptive tuning of timer advancement), '-1' enables adaptive tuning starting from default advancment of 1000ns. However, we should expose an int instead of an overflow uint module parameter. Before patch: /sys/module/kvm/parameters/lapic_timer_advance_ns:4294967295 After patch: /sys/module/kvm/parameters/lapic_timer_advance_ns:-1 Fixes: c3941d9e0 (KVM: lapic: Allow user to disable adaptive tuning of timer advancement) Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Radim Krčmář <rkrcmar@redhat.com> Cc: Sean Christopherson <sean.j.christopherson@intel.com> Cc: Liran Alon <liran.alon@oracle.com> Reviewed-by: Sean Christopherson <sean.j.christopherson@intel.com> Signed-off-by: Wanpeng Li <wanpengli@tencent.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-05-24kvm: vmx: Fix -Wmissing-prototypes warningsYi Wang
We get a warning when build kernel W=1: arch/x86/kvm/vmx/vmx.c:6365:6: warning: no previous prototype for ‘vmx_update_host_rsp’ [-Wmissing-prototypes] void vmx_update_host_rsp(struct vcpu_vmx *vmx, unsigned long host_rsp) Add the missing declaration to fix this. Signed-off-by: Yi Wang <wang.yi59@zte.com.cn> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-05-24KVM: nVMX: Fix using __this_cpu_read() in preemptible contextWanpeng Li
BUG: using __this_cpu_read() in preemptible [00000000] code: qemu-system-x86/4590 caller is nested_vmx_enter_non_root_mode+0xebd/0x1790 [kvm_intel] CPU: 4 PID: 4590 Comm: qemu-system-x86 Tainted: G OE 5.1.0-rc4+ #1 Call Trace: dump_stack+0x67/0x95 __this_cpu_preempt_check+0xd2/0xe0 nested_vmx_enter_non_root_mode+0xebd/0x1790 [kvm_intel] nested_vmx_run+0xda/0x2b0 [kvm_intel] handle_vmlaunch+0x13/0x20 [kvm_intel] vmx_handle_exit+0xbd/0x660 [kvm_intel] kvm_arch_vcpu_ioctl_run+0xa2c/0x1e50 [kvm] kvm_vcpu_ioctl+0x3ad/0x6d0 [kvm] do_vfs_ioctl+0xa5/0x6e0 ksys_ioctl+0x6d/0x80 __x64_sys_ioctl+0x1a/0x20 do_syscall_64+0x6f/0x6c0 entry_SYSCALL_64_after_hwframe+0x49/0xbe Accessing per-cpu variable should disable preemption, this patch extends the preemption disable region for __this_cpu_read(). Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Radim Krčmář <rkrcmar@redhat.com> Signed-off-by: Wanpeng Li <wanpengli@tencent.com> Fixes: 52017608da33 ("KVM: nVMX: add option to perform early consistency checks via H/W") Cc: stable@vger.kernel.org Reviewed-by: Sean Christopherson <sean.j.christopherson@intel.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-05-24kvm: fix compilation on s390Paolo Bonzini
s390 does not have memremap, even though in this particular case it would be useful. Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-05-24kvm: x86: Include CPUID leaf 0x8000001e in kvm's supported CPUIDJim Mattson
Kvm now supports extended CPUID functions through 0x8000001f. CPUID leaf 0x8000001e is AMD's Processor Topology Information leaf. This contains similar information to CPUID leaf 0xb (Intel's Extended Topology Enumeration leaf), and should be included in the output of KVM_GET_SUPPORTED_CPUID, even though userspace is likely to override some of this information based upon the configuration of the particular VM. Cc: Brijesh Singh <brijesh.singh@amd.com> Cc: Borislav Petkov <bp@suse.de> Fixes: 8765d75329a38 ("KVM: X86: Extend CPUID range to include new leaf") Signed-off-by: Jim Mattson <jmattson@google.com> Reviewed-by: Marc Orr <marcorr@google.com> Reviewed-by: Borislav Petkov <bp@suse.de> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-05-24kvm: x86: Include multiple indices with CPUID leaf 0x8000001dJim Mattson
Per the APM, "CPUID Fn8000_001D_E[D,C,B,A]X reports cache topology information for the cache enumerated by the value passed to the instruction in ECX, referred to as Cache n in the following description. To gather information for all cache levels, software must repeatedly execute CPUID with 8000_001Dh in EAX and ECX set to increasing values beginning with 0 until a value of 00h is returned in the field CacheType (EAX[4:0]) indicating no more cache descriptions are available for this processor." The termination condition is the same as leaf 4, so we can reuse that code block for leaf 0x8000001d. Fixes: 8765d75329a38 ("KVM: X86: Extend CPUID range to include new leaf") Cc: Brijesh Singh <brijesh.singh@amd.com> Cc: Borislav Petkov <bp@suse.de> Signed-off-by: Jim Mattson <jmattson@google.com> Reviewed-by: Marc Orr <marcorr@google.com> Reviewed-by: Borislav Petkov <bp@suse.de> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-05-24KVM: selftests: Compile code with warnings enabledThomas Huth
So far the KVM selftests are compiled without any compiler warnings enabled. That's quite bad, since we miss a lot of possible bugs this way. Let's enable at least "-Wall" and some other useful warning flags now, and fix at least the trivial problems in the code (like unused variables). Signed-off-by: Thomas Huth <thuth@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-05-24kvm: selftests: avoid type punningPaolo Bonzini
Avoid warnings from -Wstrict-aliasing by using memcpy. Reviewed-by: Thomas Huth <thuth@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-05-24KVM: selftests: Fix a condition in test_hv_cpuid()Dan Carpenter
The code is trying to check that all the padding is zeroed out and it does this: entry->padding[0] == entry->padding[1] == entry->padding[2] == 0 Assume everything is zeroed correctly, then the first comparison is true, the next comparison is false and false is equal to zero so the overall condition is true. This bug doesn't affect run time very badly, but the code should instead just check that all three paddings are zero individually. Also the error message was copy and pasted from an earlier error and it wasn't correct. Fixes: 7edcb7343327 ("KVM: selftests: Add hyperv_cpuid test") Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com> Reviewed-by: Thomas Huth <thuth@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-05-24KVM: Fix spinlock taken warning during host resumeWanpeng Li
WARNING: CPU: 0 PID: 13554 at kvm/arch/x86/kvm//../../../virt/kvm/kvm_main.c:4183 kvm_resume+0x3c/0x40 [kvm] CPU: 0 PID: 13554 Comm: step_after_susp Tainted: G OE 5.1.0-rc4+ #1 RIP: 0010:kvm_resume+0x3c/0x40 [kvm] Call Trace: syscore_resume+0x63/0x2d0 suspend_devices_and_enter+0x9d1/0xa40 pm_suspend+0x33a/0x3b0 state_store+0x82/0xf0 kobj_attr_store+0x12/0x20 sysfs_kf_write+0x4b/0x60 kernfs_fop_write+0x120/0x1a0 __vfs_write+0x1b/0x40 vfs_write+0xcd/0x1d0 ksys_write+0x5f/0xe0 __x64_sys_write+0x1a/0x20 do_syscall_64+0x6f/0x6c0 entry_SYSCALL_64_after_hwframe+0x49/0xbe Commit ca84d1a24 (KVM: x86: Add clock sync request to hardware enable) mentioned that "we always hold kvm_lock when hardware_enable is called. The one place that doesn't need to worry about it is resume, as resuming a frozen CPU, the spinlock won't be taken." However, commit 6706dae9 (virt/kvm: Replace spin_is_locked() with lockdep) introduces a bug, it asserts when the lock is not held which is contrary to the original goal. This patch fixes it by WARN_ON when the lock is held. Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Radim Krčmář <rkrcmar@redhat.com> Cc: Paul E. McKenney <paulmck@linux.ibm.com> Signed-off-by: Wanpeng Li <wanpengli@tencent.com> Fixes: 6706dae9 ("virt/kvm: Replace spin_is_locked() with lockdep") [Wrap with #ifdef CONFIG_LOCKDEP - Paolo] Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-05-24KVM: nVMX: Clear nested_run_pending if setting nested state failsSean Christopherson
VMX's nested_run_pending flag is subtly consumed when stuffing state to enter guest mode, i.e. needs to be set according before KVM knows if setting guest state is successful. If setting guest state fails, clear the flag as a nested run is obviously not pending. Reported-by: Aaron Lewis <aaronlewis@google.com> Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-05-24KVM: nVMX: really fix the size checks on KVM_SET_NESTED_STATEPaolo Bonzini
The offset for reading the shadow VMCS is sizeof(*kvm_state)+VMCS12_SIZE, so the correct size must be that plus sizeof(*vmcs12). This could lead to KVM reading garbage data from userspace and not reporting an error, but is otherwise not sensitive. Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-05-24Merge tag 'kvmarm-fixes-for-5.2' of ↵Paolo Bonzini
git://git.kernel.org/pub/scm/linux/kernel/git/kvmarm/kvmarm into HEAD KVM/arm updates for 5.2-rc2 - Correctly annotate HYP-callable code to be non-traceable - Remove Christoffer from the MAINTAINERS file as his request
2019-05-24Merge tag 'arm64-fixes' of ↵Linus Torvalds
git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux Pull more arm64 fixes from Will Deacon: - Fix incorrect LDADD instruction encoding in our disassembly macros - Disable the broken ARM64_PSEUDO_NMI support for now - Add workaround for Cortex-A76 CPU erratum #1463225 - Handle Cortex-A76/Neoverse-N1 erratum #1418040 w/ existing workaround - Fix IORT build failure if IOMMU_SUPPORT=n - Fix place-relative module relocation range checking and its interaction with KASLR * tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux: arm64: insn: Add BUILD_BUG_ON() for invalid masks arm64: insn: Fix ldadd instruction encoding arm64: Kconfig: Make ARM64_PSEUDO_NMI depend on BROKEN for now arm64: Handle erratum 1418040 as a superset of erratum 1188873 arm64/module: deal with ambiguity in PRELxx relocation ranges ACPI/IORT: Fix build error when IOMMU_SUPPORT is disabled arm64/kernel: kaslr: reduce module randomization range to 2 GB arm64: errata: Add workaround for Cortex-A76 erratum #1463225 arm64: Remove useless message during oops
2019-05-24drm/amd/doc: Add RAS documentation to guideTom St Denis
Acked-by: Slava Abramov <slava.abramov@amd.com> Signed-off-by: Tom St Denis <tom.stdenis@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amd/doc: Add XGMI sysfs documentationTom St Denis
Acked-by: Slava Abramov <slava.abramov@amd.com> Signed-off-by: Tom St Denis <tom.stdenis@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdgpu: sort probed modes before adding common modesYogesh Mohan Marimuthu
[Why] There are monitors which can have more than one preferred mode set. There are chances in these monitors that if common modes are added in function amdgpu_dm_connector_add_common_modes(), these common modes can be calculated with different preferred mode than the one used in function decide_crtc_timing_for_drm_display_mode(). The preferred mode can be different because after common modes are added, the mode list is sorted and this changes the order of preferred modes in the list. The first mode in the list with preferred flag set is selected as preferred mode. Due to this the preferred mode selected varies. If same preferred mode is not selected in common mode calculation and crtc timing, then during mode set instead of setting preferred timing, common mode timing will be applied which can cause "out of range" message in the monitor with monitor blanking out. [How] Sort the modes before adding common modes. The same sorting function is called during common mode addition and deciding crtc timing. Signed-off-by: Yogesh Mohan Marimuthu <yogesh.mohanmarimuthu@amd.com> Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/i915: Update DRIVER_DATE to 20190524Jani Nikula
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
2019-05-24drm/amd/display: Use new connector state when getting color depthNicholas Kazlauskas
[Why] The current state on the connector is queried when getting the max bpc rather than the new state. This means that a new max bpc value can only currently take effect on the commit *after* it changes. The new state should be passed in instead. [How] Pass down the dm_state as drm state to where we do color depth lookup. The passed in state can still be NULL when called from amdgpu_dm_connector_mode_valid, so make sure that we have reasonable defaults in place. That should probably be addressed at some point. This change now (correctly) causes a modeset to occur when changing the max bpc for a connector. v2: Drop extra TODO. Cc: Leo Li <sunpeng.li@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amd/display: Switch the custom "max bpc" property to the DRM propNicholas Kazlauskas
[Why] The custom "max bpc" property was added to limit color depth while the DRM one was still being merged. It's been a few kernel versions since then and this TODO was still sticking around. [How] Attach the DRM max bpc property to the connector and drop all of our custom property management. Set the max bpc to 8 by default since DRM defaults to the max in the range which would be 16 in this case. No behavioral changes are intended with this patch, it should just be a refactor. v2: Don't force 8bpc when no state is given Cc: Leo Li <sunpeng.li@amd.com> Cc: Harry Wentland <harry.wentland@amd.com> Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Harry Wentland <harry.wentland@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amd/powerplay: Fix code error for translating int type to bool type ↵Chengming Gui
correctly Fix code error to support value < 0 or > 1. Signed-off-by: Chengming Gui <Jack.Gui@amd.com> Reviewed-by: Hawking Zhang <hawking.zhang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdgpu: Add Unique Identifier sysfs file unique_id v2Kent Russell
Add a file that provides a Unique ID for the GPU. This will persist across machines and is guaranteed to be unique. This is only available for GFX9 and newer, so older ASICs will not have this file in the sysfs pool v2: Store it in adev for ASICs that don't have a hwmgr Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Kent Russell <kent.russell@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: Add domain number into gpu_idAmber Lin
A multi-socket server can have multiple PCIe segments so BFD is not enough to distingush each GPU. Also add domain number into account when generating gpu_id. Signed-off-by: Amber Lin <Amber.Lin@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: Add VegaM supportKent Russell
Add the VegaM information to KFD Signed-off-by: Kent Russell <kent.russell@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: Fix a circular lock dependencyFelix Kuehling
Fix a circular lock dependency exposed under userptr memory pressure. The DQM lock is the only one taken inside the MMU notifier. We need to make sure that no reclaim is done under this lock, and that no other locks are taken under which reclaim is possible. Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Reviewed-by: Philip Yang <Philip.Yang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: Delete alloc_format field from map_queue structOak Zeng
Alloc format was never really supported by MEC FW. FW always does one per pipe allocation. Signed-off-by: Oak Zeng <Oak.Zeng@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: Expose sdma engine numbers to topologyOak Zeng
Expose available numbers of both SDMA queue types in the topology. Signed-off-by: Oak Zeng <Oak.Zeng@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: Introduce XGMI SDMA queue typeOak Zeng
Existing QUEUE_TYPE_SDMA means PCIe optimized SDMA queues. Introduce a new QUEUE_TYPE_SDMA_XGMI, which is optimized for non-PCIe transfer such as XGMI. Signed-off-by: Oak Zeng <Oak.Zeng@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: Fix sdma queue map issueOak Zeng
Previous codes assumes there are two sdma engines. This is not true e.g., Raven only has 1 SDMA engine. Fix the issue by using sdma engine number info in device_info. Signed-off-by: Oak Zeng <Oak.Zeng@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: Move sdma_queue_id calculation into allocate_sdma_queue()Yong Zhao
This avoids duplicated code. Signed-off-by: Yong Zhao <Yong.Zhao@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: Allocate hiq and sdma mqd from mqd trunkOak Zeng
Instead of allocat hiq and sdma mqd from sub-allocator, allocate them from a mqd trunk pool. This is done for all asics Signed-off-by: Oak Zeng <ozeng@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: Move non-sdma mqd allocation out of init_mqdOak Zeng
This is preparation work to introduce more mqd allocation scheme Signed-off-by: Oak Zeng <ozeng@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: Fix a potential memory leakOak Zeng
Free mqd_mem_obj it GTT buffer allocation for MQD+control stack fails. Signed-off-by: Oak Zeng <ozeng@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: Allocate MQD trunk for HIQ and SDMAOak Zeng
MEC FW for some new asic requires all SDMA MQDs to be in a continuous trunk of memory right after HIQ MQD. Add a field in device queue manager to hold the HIQ/SDMA MQD memory object and allocate MQD trunk on device queue manager initialization. Signed-off-by: Oak Zeng <ozeng@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: Add mqd size in mqd manager structOak Zeng
Also initialize mqd size on mqd manager initialization Signed-off-by: Oak Zeng <ozeng@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: Init mqd managers in device queue manager initOak Zeng
Previously mqd managers was initialized on demand. As there are only a few type of mqd managers, the on demand initialization doesn't save too much memory. Initialize them on device queue initialization instead and delete the get_mqd_manager interface. This makes codes more organized for future changes. Signed-off-by: Oak Zeng <ozeng@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: Introduce DIQ type mqd managerOak Zeng
With introduction of new mqd allocation scheme for HIQ, DIQ and HIQ use different mqd allocation scheme, DIQ can't reuse HIQ mqd manager Signed-off-by: Oak Zeng <ozeng@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: Introduce asic-specific mqd_manager_init functionOak Zeng
Global function mqd_manager_init just calls asic-specific functions and it is not necessary. Delete it and introduce a mqd_manager_init interface in dqm for asic-specific mqd manager init. Call mqd_manager_init interface directly to initialize mqd manager Signed-off-by: Oak Zeng <ozeng@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdgpu: Improve error handling for HMMFelix Kuehling
Use unsigned long for number of pages. Check that pfns are valid after hmm_vma_fault. If they are not, return an error instead of continuing with invalid page pointers and PTEs. Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com> Reviewed-by: Philip Yang <Philip.Yang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdgpu: more descriptive message if HMM not enabledPhilip Yang
If using old kernel config file, CONFIG_ZONE_DEVICE is not selected, so CONFIG_HMM and CONFIG_HMM_MIRROR is not enabled, the current driver error message "Failed to register MMU notifier" is not clear. Inform user with more descriptive message on how to fix the missing kernel config option. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=109808 Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdgpu: support userptr cross VMAs case with HMMPhilip Yang
userptr may cross two VMAs if the forked child process (not call exec after fork) malloc buffer, then free it, and then malloc larger size buf, kerenl will create new VMA adjacent to old VMA which was cloned from parent process, some pages of userptr are in the first VMA, the rest pages are in the second VMA. HMM expects range only have one VMA, loop over all VMAs in the address range, create multiple ranges to handle this case. See is_mergeable_anon_vma in mm/mmap.c for details. Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: support concurrent userptr update for HMMPhilip Yang
Userptr restore may have concurrent userptr invalidation after hmm_vma_fault adds the range to the hmm->ranges list, needs call hmm_vma_range_done to remove the range from hmm->ranges list first, then reschedule the restore worker. Otherwise hmm_vma_fault will add same range to the list, this will cause loop in the list because range->next point to range itself. Add function untrack_invalid_user_pages to reduce code duplication. Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdgpu: fix HMM config dependency issuePhilip Yang
Only select HMM_MIRROR will get kernel config dependency warnings if CONFIG_HMM is missing in the config. Add depends on HMM will solve the issue. Add conditional compilation to fix compilation errors if HMM_MIRROR is not enabled as HMM config is not enabled. Remove unused function amdgpu_ttm_tt_mark_user_pages. Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdgpu: replace get_user_pages with HMM mirror helpersPhilip Yang
Use HMM helper function hmm_vma_fault() to get physical pages backing userptr and start CPU page table update track of those pages. Then use hmm_vma_range_done() to check if those pages are updated before amdgpu_cs_submit for gfx or before user queues are resumed for kfd. If userptr pages are updated, for gfx, amdgpu_cs_ioctl will restart from scratch, for kfd, restore worker is rescheduled to retry. HMM simplify the CPU page table concurrent update check, so remove guptasklock, mmu_invalidations, last_set_pages fields from amdgpu_ttm_tt struct. HMM does not pin the page (increase page ref count), so remove related operations like release_pages(), put_page(), mark_page_dirty(). Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdkfd: avoid HMM change cause circular lockPhilip Yang
There is circular lock between gfx and kfd path with HMM change: lock(dqm) -> bo::reserve -> amdgpu_mn_lock To avoid this, move init/unint_mqd() out of lock(dqm), to remove nested locking between mmap_sem and bo::reserve. The locking order is: bo::reserve -> amdgpu_mn_lock(p->mn) Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Acked-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2019-05-24drm/amdgpu: use HMM callback to replace mmu notifierPhilip Yang
Replace our MMU notifier with hmm_mirror_ops.sync_cpu_device_pagetables callback. Enable CONFIG_HMM and CONFIG_HMM_MIRROR as a dependency in DRM_AMDGPU_USERPTR Kconfig. It supports both KFD userptr and gfx userptr paths. Signed-off-by: Philip Yang <Philip.Yang@amd.com> Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
generated by cgit v1.2.3 (git 2.39.1) at 2025-07-29 07:26:52 +0000