1 files changed, 46 insertions, 0 deletions
diff --git a/results/classifier/108/other/1639322 b/results/classifier/108/other/1639322
new file mode 100644
index 000000000..98124e7e6
--- /dev/null
+++ b/results/classifier/108/other/1639322
@@ -0,0 +1,46 @@
+graphic: 0.864
+other: 0.706
+performance: 0.667
+debug: 0.610
+semantic: 0.572
+device: 0.552
+permissions: 0.515
+PID: 0.487
+network: 0.456
+files: 0.425
+vnc: 0.398
+socket: 0.369
+boot: 0.211
+KVM: 0.102
+
+pasting into ppc64 serial console kills qemu
+
+- run qemu-system-ppc64
+- when X window appears press Ctrl+Alt+3
+- paste any text longer than 16 characters
+
+
+qemu-system-ppc64: /home/abuild/rpmbuild/BUILD/qemu-2.6.1/hw/char/spapr_vty.c:40: vty_receive: Assertion `(dev->in - dev->out) < 16' failed.
+Aborted (core dumped)
+
+Broken in SUSE Leap 42.2 and git 4eb28abd52d48657cff6ff45e8dbbbefe4dbb414
+
+What user interface are you using? VNC? SDL? GTK?
+
+This is gtk interface.
+
+However, the function on line 40 os spapr_vty.c looks really insane.
+
+It asserts that it is not given more data to input in a ring buffer than is size of the buffer and then stuffs all the data in regardless of the amount of data already present.
+
+It should probably loop or one of its callers but I did not find a decent comparable piece of code to cut and paste whatever callbacks are needed for the other side to consume the bytes.
+
+OK, seems like you need to compile QEMU with CONFIG_VTE enabled (i.e. with the vte-devel packages installed before running configure) to get copy-n-paste support in the GTK interface, that's why I was initially not able to reproduce this issue.
+Anyway, now I can trigger the assert(), too, and I've suggested a patch here:
+
+http://marc.info/?<email address hidden>
+
+FWIW, the crash should be fixed by this commit here:
+http://git.qemu.org/?p=qemu.git;a=commitdiff;h=7bacfd7f7289192c83330
+(but we still need to fix the gtk side, too, to only send as much characters at once as the receiving side can take)
+