1 files changed, 112 insertions, 0 deletions
diff --git a/results/classifier/118/all/1739378 b/results/classifier/118/all/1739378
new file mode 100644
index 00000000..16496de5
--- /dev/null
+++ b/results/classifier/118/all/1739378
@@ -0,0 +1,112 @@
+debug: 0.985
+virtual: 0.980
+semantic: 0.980
+assembly: 0.977
+graphic: 0.976
+performance: 0.975
+device: 0.975
+register: 0.974
+PID: 0.974
+architecture: 0.973
+arm: 0.969
+kernel: 0.966
+boot: 0.965
+user-level: 0.961
+socket: 0.955
+files: 0.955
+permissions: 0.954
+vnc: 0.950
+network: 0.949
+peripherals: 0.946
+risc-v: 0.937
+TCG: 0.930
+mistranslation: 0.928
+VMM: 0.924
+KVM: 0.918
+ppc: 0.903
+hypervisor: 0.887
+x86: 0.842
+i386: 0.802
+
+migration state save/load of sdcard device is broken
+
+I'm having different issues trying to have QEMU snapshots working using qemu-system-arm with vexpress-a15 board.
+
+In this opportunity, I'm trying the git master head version:
+# git rev-parse HEAD
+af352675efb7e92a1f5f6461a042a12015ab3d12
+
+$ /usr/local/bin/qemu-system-arm -kernel kernel/vmlinuz-4.10.0-42-generic -initrd kernel/initrd.img-4.10.0-42-generic -M vexpress-a15 -m 2048 -append 'root=/dev/mmcblk0 rootwait console=tty0' -sd vexpress-4G.qcow2 -dtb device-tree/vexpress-v2p-ca15-tc1.dtb  
+audio: Could not init `oss' audio driver
+
+Later on, when the machine finishes booting I savevm ss and quit. However, when I try to restore it, I have that Missing section footer error:
+
+$ /usr/local/bin/qemu-system-arm -kernel kernel/vmlinuz-4.10.0-42-generic -initrd kernel/initrd.img-4.10.0-42-generic -M vexpress-a15 -m 2048 -append 'root=/dev/mmcblk0 rootwait console=tty0' -sd vexpress-4G.qcow2 -dtb device-tree/vexpress-v2p-ca15-tc1.dtb  -loadvm ss
+audio: Could not init `oss' audio driver
+qemu-system-arm: Missing section footer for sd-card
+qemu-system-arm: Error -22 while loading VM state
+
+
+OS: Ubuntu 16.04.3 LTS (xenial)
+
+$ gcc --version
+gcc (Ubuntu 5.4.0-6ubuntu1~16.04.5) 5.4.0 20160609
+
+I've also tried a different ./configure line, explicitly enabling some of the features, i.e. smartcard, with the same results:
+
+./configure '--disable-user' '--enable-system' '--enable-linux-user' '--enable-modules' '--enable-linux-aio' '--audio-drv-list=pa' '--enable-attr' '--enable-brlapi' '--enable-virtfs' '--enable-cap-ng' '--enable-curl' '--enable-fdt' '--enable-gnutls' '--disable-gtk' '--disable-vte' '--enable-libiscsi' '--enable-curses' '--enable-smartcard' '--enable-rbd' '--enable-vnc-sasl' '--enable-seccomp' '--enable-spice' '--enable-libusb' '--enable-usb-redir' '--enable-xfsctl' '--enable-vnc' '--enable-vnc-jpeg' '--enable-vnc-png' '--enable-kvm' '--enable-vhost-net'
+
+How have I built it?
+# git clone git://git.qemu.org/qemu.git
+# cd qemu
+# git submodule update --init --checkout
+# make clean && ./configure --target-list=arm-softmmu && make -j8
+# sudo make install
+
+As a reference, and just in case these may be in some way related, I've just submitted another ticket for a different issue with snapshots using Ubuntu Qemu version (https://bugs.launchpad.net/qemu/+bug/1739371)
+
+Cheers,
+Gus
+
+
+
+From a quick look my guess would be the wpgrps_size in sd_vmstate is different on the source and destination for some reason - it's the only thing that seems to be variable.
+(It could also be whether the subsection is transmitted or not, but that's supposed to sort itself out automatically).
+
+So, it should be only related to SD? Which virtual device/storage backend do you recommend for snapshots?
+
+
+On 20 December 2017 at 18:44, Dr. David Alan Gilbert
+<email address hidden> wrote:
+> From a quick look my guess would be the wpgrps_size in sd_vmstate is
+> different on the source and destination for some reason
+
+Good guess: I added a pre_save hook to sdcard.c's vmstate,
+which reports wpgrps_size as 201, but in the pre_load hook
+it is reported as 0.
+
+This seems to be because nothing ever calls the sd_reset()
+function. I suspect this is down to the "legacy" handling
+of sd card setup in sd_init(), or possibly the generic handling
+of reset for devices that connect to SD buses.
+
+(sd_reset() eventually gets called when the guest OS
+probes the SD controller and issues a card reset command,
+which is why it's correctly set by the time we do a vmsave.)
+
+thanks
+-- PMM
+
+
+I think the sd card migration issue should be fixed by http://patchwork.ozlabs.org/patch/857554/
+
+NB that you'll also need to use "-machine secure=off" on the command line, as there's a second bug where we don't successfully migrate with trustzone emulation enabled (which is the default now).
+
+
+Finally I could test it. Yeah man, awesome, that did the trick. It is working like a charm.
+You can close the ticket.
+Thanks
+
+The patch has been merged here:
+https://git.qemu.org/?p=qemu.git;a=commitdiff;h=0cb57cc701839e7358918
+