add 18th iteration of classifier

1 files changed, 55 insertions, 0 deletions

diff --git a/results/classifier/118/peripherals/2976 b/results/classifier/118/peripherals/2976
new file mode 100644
index 000000000..eafb9e390
--- /dev/null
+++ b/results/classifier/118/peripherals/2976
@@ -0,0 +1,55 @@
+peripherals: 0.935
+device: 0.799
+performance: 0.791
+architecture: 0.672
+network: 0.636
+files: 0.618
+permissions: 0.559
+ppc: 0.520
+socket: 0.520
+kernel: 0.492
+vnc: 0.459
+semantic: 0.453
+graphic: 0.414
+virtual: 0.402
+risc-v: 0.391
+VMM: 0.388
+register: 0.382
+PID: 0.381
+hypervisor: 0.372
+i386: 0.342
+user-level: 0.336
+debug: 0.320
+x86: 0.317
+assembly: 0.235
+boot: 0.218
+arm: 0.173
+mistranslation: 0.169
+TCG: 0.150
+KVM: 0.071
+
+attach-ns doesn't work correctly in SR-IOV, cannot attach NS to VF if it is attached to a PF
+Description of problem:
+We can't attach namespace to a VF (Secondary controller) unless it is not attached to a primary controller first
+
+Lately in the commit https://github.com/qemu/qemu/commit/6ccca4b6bb9f994cc04e71004e1767a3476d2b23 the file qemu/hw/nvme/ctrl.c got changed -\> in function "nvme_ns_attachment" -\> line 6819 (At the time I'm writing the bug) which is the condation in attach ns to check if the namespace is attached "`if (nvme_ns(n, nsid)) {`"
+
+This change will always result in checking the namespace attach to the PF even if we are trying to attach it to the VF.
+Steps to reproduce:
+1. Enable a VF:
+
+```
+echo 1 > /sys/bus/pci/devices/0000:01:00.0/reset
+sleep 1
+
+echo 2 > /sys/bus/pci/devices/0000:01:00.0/sriov_numvfs
+sleep 1
+
+nvme virt-mgmt /dev/nvme0 -c 1 -r 1 -a 8 -n 1
+nvme virt-mgmt /dev/nvme0 -c 1 -r 0 -a 8 -n 2
+nvme virt-mgmt /dev/nvme0 -c 1 -a 9
+sleep 1
+```
+
+2. attach namespace 1 to the PF (e.g. `vme attach-ns /dev/nvme0 -n1 -c0` )
+3. try to attach it using the nvme_cli command from the PF (e.g. `nvme attach-ns /dev/nvme0 -n1 -c1`)