summary refs log tree commit diff stats
path: root/results/classifier/118/graphic/1324727
diff options
context:
space:
mode:
Diffstat (limited to 'results/classifier/118/graphic/1324727')
-rw-r--r--results/classifier/118/graphic/132472763
1 files changed, 63 insertions, 0 deletions
diff --git a/results/classifier/118/graphic/1324727 b/results/classifier/118/graphic/1324727
new file mode 100644
index 00000000..b48bb750
--- /dev/null
+++ b/results/classifier/118/graphic/1324727
@@ -0,0 +1,63 @@
+graphic: 0.966
+arm: 0.908
+kernel: 0.902
+debug: 0.895
+architecture: 0.888
+device: 0.888
+semantic: 0.869
+performance: 0.859
+boot: 0.858
+PID: 0.824
+user-level: 0.813
+ppc: 0.801
+register: 0.796
+permissions: 0.779
+peripherals: 0.753
+files: 0.722
+socket: 0.706
+mistranslation: 0.656
+vnc: 0.654
+assembly: 0.587
+virtual: 0.578
+network: 0.493
+VMM: 0.489
+hypervisor: 0.469
+risc-v: 0.441
+TCG: 0.426
+i386: 0.333
+KVM: 0.166
+x86: 0.068
+
+qemu-system-arm segfaults without KVM on ARM
+
+I'm running on Odroid-XU, Debian Jessie armhf
+qemu built from today's head d7d3d6092cb7edc75dc49fb90c86dd5425ab4805
+
+sudo  qemu-system-arm -M vexpress-a15 -drive if=none,file=arm.img,cache=writeback,id=foo -device virtio-blk-device,drive=foo -netdev user,id=user.0 -device virtio-net-device,netdev=user.0 -nographic -append 'root=/dev/vda rw console=ttyAMA0 rootwait' -kernel /usr/src/build/arm/linux-guest/arch/arm/boot/zImage -dtb a15x2.dtb
+audio: Could not init `oss' audio driver
+Uncompressing Linux... done, booting the kernel.
+Segmentation fault
+
+If I run under GDB, the linux guest instance panics or hangs -- the behaviour is variable run to run.
+
+If I do:
+sudo  qemu-system-arm --enable-kvm -M vexpress-a15 -drive if=none,file=arm.img,cache=writeback,id=foo -device virtio-blk-device,drive=foo -netdev user,id=user.0 -device virtio-net-device,netdev=user.0 -nographic -append 'root=/dev/vda rw console=ttyAMA0 rootwait' -kernel /usr/src/build/arm/linux-guest/arch/arm/boot/zImage -dtb a15x2.dtb
+
+then the guest boots as expected.
+
+I tried to get a backtrace by allowinghte SEGV to dump core, and using gdb to inspect it:
+Core was generated by `qemu-system-arm -M vexpress-a15 -drive if=none,file=arm.img,cache=writeback,id='.
+Program terminated with signal 11, Segmentation fault.
+#0  0xb53399c0 in ?? ()
+(gdb) bt
+#0  0xb53399c0 in ?? ()
+Cannot access memory at address 0x28
+#1  0x0016d87e in cpu_tb_exec (
+    tb_ptr=0xc786fe90 <Address 0xc786fe90 out of bounds>, cpu=0x24450d8)
+    at /mnt/qemu/cpu-exec.c:67
+#2  cpu_arm_exec (env=<optimized out>) at /mnt/qemu/cpu-exec.c:642
+#3  0x00000000 in ?? ()
+
+This is a two year old bug which doesn't have an attached repro case and I haven't seen QEMU segfault like this, so I'm going to assume we've fixed this bug. Please reopen if you still have a problem with a newer QEMU, and provide a link to the guest binary that demonstrates the crash.
+
+