summary refs log tree commit diff stats
path: root/results/classifier/118/graphic/2690
diff options
context:
space:
mode:
Diffstat (limited to 'results/classifier/118/graphic/2690')
-rw-r--r--results/classifier/118/graphic/269050
1 files changed, 50 insertions, 0 deletions
diff --git a/results/classifier/118/graphic/2690 b/results/classifier/118/graphic/2690
new file mode 100644
index 000000000..d1a7e9b2e
--- /dev/null
+++ b/results/classifier/118/graphic/2690
@@ -0,0 +1,50 @@
+graphic: 0.852
+semantic: 0.814
+device: 0.722
+performance: 0.685
+hypervisor: 0.658
+x86: 0.577
+socket: 0.553
+network: 0.550
+vnc: 0.512
+mistranslation: 0.495
+virtual: 0.478
+i386: 0.468
+architecture: 0.462
+debug: 0.438
+kernel: 0.415
+risc-v: 0.393
+ppc: 0.381
+user-level: 0.374
+VMM: 0.371
+arm: 0.365
+register: 0.358
+TCG: 0.341
+boot: 0.341
+PID: 0.333
+peripherals: 0.316
+permissions: 0.227
+KVM: 0.139
+assembly: 0.137
+files: 0.132
+
+"Guest says index 40947 is available"
+Description of problem:
+As discussed [here](https://github.com/danobi/vmtest/issues/96) I have been running several instances of QEMU in parallel at `SCHED_IDLE`, and I've been getting QGA setup failures.
+Steps to reproduce:
+1. Install [vmtest](https://github.com/danobi/vmtest)
+2. Run lots of copies of the command in the [github issues](https://github.com/danobi/vmtest/issues/96) via `chrt --idle 0`.
+3. Unclear if this is the cause, but then I use the computer in the meantime so probably starve the `SCHED_IDLE` QEMU threads running from 2.
+
+This leads to failures to connect to the guest agent and then at the end I see this:
+
+```
+Guest says index 40947 is available
+    qemu-system-x86_64: Guest says index 40947 is available
+    qemu-system-x86_64: Guest says index 40947 is available
+```
+
+
+The developer of vmtest seemed to think this may be of interest to QEMU developers based on the tone of the [comment they found](https://github.com/danobi/vmtest/issues/96#issuecomment-2483860554) in the QEMU code.
+
+I've now installed QEMU from Git master so I can report back whether the bug still appeared.