summary refs log tree commit diff stats
path: root/gitlab/issues/target_missing/host_missing/accel_missing/2860.toml
diff options
context:
space:
mode:
Diffstat (limited to 'gitlab/issues/target_missing/host_missing/accel_missing/2860.toml')
-rw-r--r--gitlab/issues/target_missing/host_missing/accel_missing/2860.toml41
1 files changed, 41 insertions, 0 deletions
diff --git a/gitlab/issues/target_missing/host_missing/accel_missing/2860.toml b/gitlab/issues/target_missing/host_missing/accel_missing/2860.toml
new file mode 100644
index 00000000..0e39f3bf
--- /dev/null
+++ b/gitlab/issues/target_missing/host_missing/accel_missing/2860.toml
@@ -0,0 +1,41 @@
+id = 2860
+title = "ps2 keyboard not work after boot and use libspice to connect it"
+state = "opened"
+created_at = "2025-03-13T11:30:45.533Z"
+closed_at = "n/a"
+labels = ["spice"]
+url = "https://gitlab.com/qemu-project/qemu/-/issues/2860"
+host-os = "centos8"
+host-arch = "x86"
+qemu-version = "n/a"
+guest-os = "windows"
+guest-arch = "x86"
+description = """When I start almost 10 qemu virtual machines, there will always be one or two that have the ps2 keyboard not work well after booted.But I use mstsc to connect to the desktop, the keyboard works fine. But when reboot or migrate it well recovery."""
+reproduce = """1.Asynchronously start 40 qemu virtual machines, each with 4 cores and 4 threads
+
+2.there will always be one or two that have the ps2 keyboard not work well.
+
+4.And when i gdb debug it, i found i hang at the func "prepare_mmio_access"
+
+5.reboot or migrate it well recovery"""
+additional = """the gdb debug as fllow:
+
+gdb attach $pid 
+
+gdb>b kbd_push_key      //spice input
+
+gdb>b kbd_read_data
+
+gdb>b ps2_keyboard_event
+
+gdb>c
+
+After continue, the code run on ps2_keyboard_event,but no work to "kbd_read_data".This Proves that the keyboard input has been added to the queue, but has not been read from the queue.
+
+gdb> thread 4   //switch to thread "CPU 0/KVM"
+
+gdb> bt
+
+![image](/uploads/82527d2b382cacd2d4e40793d78d3e59/image.png)
+
+I guess there is no event to notify the device to read after writing to the queue, or is it deadlocked? I'm not sure"""