kernel.git/tools/testing/selftests/kvm, branch linux-5.18.y

selftests: kvm: set rax before vmcall

2022-08-17T12:41:55Z

[ Upstream commit 281106f938d3daaea6f8b6723a8217a2a1ef6936 ] kvm_hypercall has to place the hypercall number in rax. Trace events show that kvm_pv_test doesn't work properly: kvm_pv_test-53132: kvm_hypercall: nr 0x0 a0 0x0 a1 0x0 a2 0x0 a3 0x0 kvm_pv_test-53132: kvm_hypercall: nr 0x0 a0 0x0 a1 0x0 a2 0x0 a3 0x0 kvm_pv_test-53132: kvm_hypercall: nr 0x0 a0 0x0 a1 0x0 a2 0x0 a3 0x0 With this change, it starts working as expected: kvm_pv_test-54285: kvm_hypercall: nr 0x5 a0 0x0 a1 0x0 a2 0x0 a3 0x0 kvm_pv_test-54285: kvm_hypercall: nr 0xa a0 0x0 a1 0x0 a2 0x0 a3 0x0 kvm_pv_test-54285: kvm_hypercall: nr 0xb a0 0x0 a1 0x0 a2 0x0 a3 0x0 Signed-off-by: Andrei Vagin Message-Id: <20220722230241.1944655-5-avagin@google.com> Fixes: ac4a4d6de22e ("selftests: kvm: test enforcement of paravirtual cpuid features") Signed-off-by: Paolo Bonzini Signed-off-by: Sasha Levin

selftests: KVM: Handle compiler optimizations in ucall

2022-08-11T11:20:42Z

[ Upstream commit 9e2f6498efbbc880d7caa7935839e682b64fe5a6 ] The selftests, when built with newer versions of clang, is found to have over optimized guests' ucall() function, and eliminating the stores for uc.cmd (perhaps due to no immediate readers). This resulted in the userspace side always reading a value of '0', and causing multiple test failures. As a result, prevent the compiler from optimizing the stores in ucall() with WRITE_ONCE(). Suggested-by: Ricardo Koller Suggested-by: Reiji Watanabe Signed-off-by: Raghavendra Rao Ananta Message-Id: <20220615185706.1099208-1-rananta@google.com> Reviewed-by: Andrew Jones Signed-off-by: Paolo Bonzini Signed-off-by: Sasha Levin

KVM: selftests: Restrict test region to 48-bit physical addresses when using nested

2022-08-11T11:20:41Z

[ Upstream commit e0f3f46e42064a51573914766897b4ab95d943e3 ] The selftests nested code only supports 4-level paging at the moment. This means it cannot map nested guest physical addresses with more than 48 bits. Allow perf_test_util nested mode to work on hosts with more than 48 physical addresses by restricting the guest test region to 48-bits. While here, opportunistically fix an off-by-one error when dealing with vm_get_max_gfn(). perf_test_util.c was treating this as the maximum number of GFNs, rather than the maximum allowed GFN. This didn't result in any correctness issues, but it did end up shifting the test region down slightly when using huge pages. Suggested-by: Sean Christopherson Signed-off-by: David Matlack Message-Id: <20220520233249.3776001-12-dmatlack@google.com> Signed-off-by: Paolo Bonzini Signed-off-by: Sasha Levin

KVM: selftests: Make hyperv_clock selftest more stable

2022-08-11T11:20:41Z

[ Upstream commit eae260be3a0111a28fe95923e117a55dddec0384 ] hyperv_clock doesn't always give a stable test result, especially with AMD CPUs. The test compares Hyper-V MSR clocksource (acquired either with rdmsr() from within the guest or KVM_GET_MSRS from the host) against rdtsc(). To increase the accuracy, increase the measured delay (done with nop loop) by two orders of magnitude and take the mean rdtsc() value before and after rdmsr()/KVM_GET_MSRS. Reported-by: Maxim Levitsky Signed-off-by: Vitaly Kuznetsov Reviewed-by: Maxim Levitsky Tested-by: Maxim Levitsky Message-Id: <20220601144322.1968742-1-vkuznets@redhat.com> Signed-off-by: Paolo Bonzini Signed-off-by: Sasha Levin

KVM: selftests: Fix target thread to be migrated in rseq_test

2022-07-29T15:28:13Z

commit e923b0537d28e15c9d31ce8b38f810b325816903 upstream. In rseq_test, there are two threads, which are vCPU thread and migration worker separately. Unfortunately, the test has the wrong PID passed to sched_setaffinity() in the migration worker. It forces migration on the migration worker because zeroed PID represents the calling thread, which is the migration worker itself. It means the vCPU thread is never enforced to migration and it can migrate at any time, which eventually leads to failure as the following logs show. host# uname -r 5.19.0-rc6-gavin+ host# # cat /proc/cpuinfo | grep processor | tail -n 1 processor : 223 host# pwd /home/gavin/sandbox/linux.main/tools/testing/selftests/kvm host# for i in `seq 1 100`; do \ echo "--------> $i"; ./rseq_test; done --------> 1 --------> 2 --------> 3 --------> 4 --------> 5 --------> 6 ==== Test Assertion Failure ==== rseq_test.c:265: rseq_cpu == cpu pid=3925 tid=3925 errno=4 - Interrupted system call 1 0x0000000000401963: main at rseq_test.c:265 (discriminator 2) 2 0x0000ffffb044affb: ?? ??:0 3 0x0000ffffb044b0c7: ?? ??:0 4 0x0000000000401a6f: _start at ??:? rseq CPU = 4, sched CPU = 27 Fix the issue by passing correct parameter, TID of the vCPU thread, to sched_setaffinity() in the migration worker. Fixes: 61e52f1630f5 ("KVM: selftests: Add a test for KVM_RUN+rseq to detect task migration bugs") Suggested-by: Sean Christopherson Signed-off-by: Gavin Shan Reviewed-by: Oliver Upton Message-Id: <20220719020830.3479482-1-gshan@redhat.com> Reviewed-by: Andrew Jones Signed-off-by: Paolo Bonzini Signed-off-by: Greg Kroah-Hartman

selftests: kvm/x86: Verify the pmu event filter matches the correct event

2022-05-20T11:06:55Z

Add a test to demonstrate that when the guest programs an event select it is matched correctly in the pmu event filter and not inadvertently filtered. This could happen on AMD if the high nybble[1] in the event select gets truncated away only leaving the bottom byte[2] left for matching. This is a contrived example used for the convenience of demonstrating this issue, however, this can be applied to event selects 0x28A (OC Mode Switch) and 0x08A (L1 BTB Correction), where 0x08A could end up being denied when the event select was only set up to deny 0x28A. [1] bits 35:32 in the event select register and bits 11:8 in the event select. [2] bits 7:0 in the event select register and bits 7:0 in the event select. Signed-off-by: Aaron Lewis Message-Id: <20220517051238.2566934-3-aaronlewis@google.com> Signed-off-by: Paolo Bonzini

selftests: kvm/x86: Add the helper function create_pmu_event_filter

2022-05-20T11:06:55Z

Add a helper function that creates a pmu event filter given an event list. Currently, a pmu event filter can only be created with the same hard coded event list. Add a way to create one given a different event list. Also, rename make_pmu_event_filter to alloc_pmu_event_filter to clarify it's purpose given the introduction of create_pmu_event_filter. No functional changes intended. Signed-off-by: Aaron Lewis Message-Id: <20220517051238.2566934-2-aaronlewis@google.com> Signed-off-by: Paolo Bonzini

kvm: selftests: introduce and use more page size-related constants

2022-04-21T19:41:01Z

Clean up code that was hardcoding masks for various fields, now that the masks are included in processor.h. For more cleanup, define PAGE_SIZE and PAGE_MASK just like in Linux. PAGE_SIZE in particular was defined by several tests. Suggested-by: Sean Christopherson Reviewed-by: Peter Xu Signed-off-by: Paolo Bonzini

kvm: selftests: do not use bitfields larger than 32-bits for PTEs

2022-04-21T19:41:01Z

Red Hat's QE team reported test failure on access_tracking_perf_test: Testing guest mode: PA-bits:ANY, VA-bits:48, 4K pages guest physical test memory offset: 0x3fffbffff000 Populating memory : 0.684014577s Writing to populated memory : 0.006230175s Reading from populated memory : 0.004557805s ==== Test Assertion Failure ==== lib/kvm_util.c:1411: false pid=125806 tid=125809 errno=4 - Interrupted system call 1 0x0000000000402f7c: addr_gpa2hva at kvm_util.c:1411 2 (inlined by) addr_gpa2hva at kvm_util.c:1405 3 0x0000000000401f52: lookup_pfn at access_tracking_perf_test.c:98 4 (inlined by) mark_vcpu_memory_idle at access_tracking_perf_test.c:152 5 (inlined by) vcpu_thread_main at access_tracking_perf_test.c:232 6 0x00007fefe9ff81ce: ?? ??:0 7 0x00007fefe9c64d82: ?? ??:0 No vm physical memory at 0xffbffff000 I can easily reproduce it with a Intel(R) Xeon(R) CPU E5-2630 with 46 bits PA. It turns out that the address translation for clearing idle page tracking returned a wrong result; addr_gva2gpa()'s last step, which is based on "pte[index[0]].pfn", did the calculation with 40 bits length and the high 12 bits got truncated. In above case the GPA address to be returned should be 0x3fffbffff000 for GVA 0xc0000000, but it got truncated into 0xffbffff000 and the subsequent gpa2hva lookup failed. The width of operations on bit fields greater than 32-bit is implementation defined, and differs between GCC (which uses the bitfield precision) and clang (which uses 64-bit arithmetic), so this is a potential minefield. Remove the bit fields and using manual masking instead. Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=2075036 Reported-by: Nana Liu Reviewed-by: Peter Xu Tested-by: Peter Xu Signed-off-by: Paolo Bonzini

KVM: selftests: Silence compiler warning in the kvm_page_table_test

2022-04-21T17:16:14Z

When compiling kvm_page_table_test.c, I get this compiler warning with gcc 11.2: kvm_page_table_test.c: In function 'pre_init_before_test': ../../../../tools/include/linux/kernel.h:44:24: warning: comparison of distinct pointer types lacks a cast 44 | (void) (&_max1 == &_max2); \ | ^~ kvm_page_table_test.c:281:21: note: in expansion of macro 'max' 281 | alignment = max(0x100000, alignment); | ^~~ Fix it by adjusting the type of the absolute value. Signed-off-by: Thomas Huth Reviewed-by: Claudio Imbrenda Message-Id: <20220414103031.565037-1-thuth@redhat.com> Signed-off-by: Paolo Bonzini