| Message ID | 20191205151610.19299-1-thuth@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [kvm-unit-tests] gitlab-ci.yml: Remove ioapic from the x86 tests | expand |
On 05/12/19 16:16, Thomas Huth wrote: > The test recently started to fail (likely do to a recent change to > "x86/ioapic.c). According to Nitesh, it's not required to keep this > test running with TCG, and we already check it with KVM on Travis, > so let's simply disable it here now. It works for me though: $ /usr/bin/qemu-system-x86_64 -nodefaults -device pc-testdev -device isa-debug-exit,iobase=0xf4,iosize=0x4 -vnc none -serial stdio -device pci-testdev -machine accel=tcg -kernel x86/ioapic.flat enabling apic paging enabled cr0 = 80010011 cr3 = 61d000 cr4 = 20 x2apic not detected PASS: version register read only test PASS: id register only bits [24:27] writable PASS: arbitration register set by id PASS: arbtration register read only PASS: edge triggered intr PASS: level triggered intr PASS: ioapic simultaneous edge interrupts PASS: coalesce simultaneous level interrupts PASS: sequential level interrupts PASS: retriggered level interrupts without masking PASS: masked level interrupt PASS: unmasked level interrupt PASS: masked level interrupt PASS: unmasked level interrupt PASS: retriggered level interrupts with mask PASS: TMR for ioapic edge interrupts (expected false) PASS: TMR for ioapic level interrupts (expected false) PASS: TMR for ioapic level interrupts (expected true) PASS: TMR for ioapic edge interrupts (expected true) SUMMARY: 19 tests
On 09/12/2019 18.12, Paolo Bonzini wrote: > On 05/12/19 16:16, Thomas Huth wrote: >> The test recently started to fail (likely do to a recent change to >> "x86/ioapic.c). According to Nitesh, it's not required to keep this >> test running with TCG, and we already check it with KVM on Travis, >> so let's simply disable it here now. > > It works for me though: > > $ /usr/bin/qemu-system-x86_64 -nodefaults -device pc-testdev -device > isa-debug-exit,iobase=0xf4,iosize=0x4 -vnc none -serial stdio -device > pci-testdev -machine accel=tcg -kernel x86/ioapic.flat You have to run the test with "-smp 4" (like in x86/unittests.cfg), then it hangs. Thomas
On 09/12/19 18:33, Thomas Huth wrote: >> It works for me though: >> >> $ /usr/bin/qemu-system-x86_64 -nodefaults -device pc-testdev -device >> isa-debug-exit,iobase=0xf4,iosize=0x4 -vnc none -serial stdio -device >> pci-testdev -machine accel=tcg -kernel x86/ioapic.flat > You have to run the test with "-smp 4" (like in x86/unittests.cfg), then > it hangs. The problem was actually that I hadn't recompiled x86/ioapic.flat. Now I can see the hang even with "-machine accel=kvm,kernel_irqchip=off|split", I'll take a look if Nitesh doesn't beat me. Paolo
On 12/9/19 12:45 PM, Paolo Bonzini wrote: > On 09/12/19 18:33, Thomas Huth wrote: >>> It works for me though: >>> >>> $ /usr/bin/qemu-system-x86_64 -nodefaults -device pc-testdev -device >>> isa-debug-exit,iobase=0xf4,iosize=0x4 -vnc none -serial stdio -device >>> pci-testdev -machine accel=tcg -kernel x86/ioapic.flat >> You have to run the test with "-smp 4" (like in x86/unittests.cfg), then >> it hangs. > The problem was actually that I hadn't recompiled x86/ioapic.flat. Now > I can see the hang even with "-machine > accel=kvm,kernel_irqchip=off|split", I'll take a look if Nitesh doesn't > beat me. Sorry for the delay in response, I am currently on PTO. I haven't specifically tried with 'kernel_irqchip=off|split', previously I have only tried with -machine accel=kvm. I can give it a try once I am back. > Paolo >
On 12/9/19 12:45 PM, Paolo Bonzini wrote: > On 09/12/19 18:33, Thomas Huth wrote: >>> It works for me though: >>> >>> $ /usr/bin/qemu-system-x86_64 -nodefaults -device pc-testdev -device >>> isa-debug-exit,iobase=0xf4,iosize=0x4 -vnc none -serial stdio -device >>> pci-testdev -machine accel=tcg -kernel x86/ioapic.flat >> You have to run the test with "-smp 4" (like in x86/unittests.cfg), then >> it hangs. > The problem was actually that I hadn't recompiled x86/ioapic.flat. Now > I can see the hang even with "-machine > accel=kvm,kernel_irqchip=off|split", I'll take a look if Nitesh doesn't > beat me. > > Paolo I just started looking into the issue and the following is my understanding so far. - The issue has something to do with test_ioapic_self_reconfigure() failure when ran with kernel_irqchip=split. If I remove this particular test or move it below my set of tests then both physical and logical destination mode test pass normally without any hang. My understanding is that the failure of test_ioapic_self_reconfigure() leads to an uncleaned interrupt state which prevents the processing of the next request (physical and logical destination mode test). I still don't have a justification for why test_ioapic_self_reconfigure is failing with kernel_irqchip=split. I will have to investigate that.
diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml index fbf3328..001c272 100644 --- a/.gitlab-ci.yml +++ b/.gitlab-ci.yml @@ -64,7 +64,7 @@ build-x86_64: - ./configure --arch=x86_64 - make -j2 - ACCEL=tcg ./run_tests.sh - ioapic-split ioapic smptest smptest3 vmexit_cpuid vmexit_mov_from_cr8 + ioapic-split smptest smptest3 vmexit_cpuid vmexit_mov_from_cr8 vmexit_mov_to_cr8 vmexit_inl_pmtimer vmexit_ipi vmexit_ipi_halt vmexit_ple_round_robin vmexit_tscdeadline vmexit_tscdeadline_immed eventinj msr port80 syscall tsc rmap_chain umip intel_iommu
The test recently started to fail (likely do to a recent change to "x86/ioapic.c). According to Nitesh, it's not required to keep this test running with TCG, and we already check it with KVM on Travis, so let's simply disable it here now. Signed-off-by: Thomas Huth <thuth@redhat.com> --- .gitlab-ci.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)