Re: [Qemu-devel] [PATCH v6 2/4] cryptodev: add vhost support

[Zhoujian (jay)]

> -----Original Message-----
> From: Michael S. Tsirkin [mailto:address@hidden
> Sent: Wednesday, February 14, 2018 12:44 AM
> To: Zhoujian (jay) <address@hidden>
> Cc: address@hidden; address@hidden; Huangweidong (C)
> <address@hidden>; address@hidden; address@hidden;
> longpeng <address@hidden>; address@hidden; address@hidden;
> Gonglei (Arei) <address@hidden>; wangxin (U)
> <address@hidden>
> Subject: Re: [PATCH v6 2/4] cryptodev: add vhost support
> 
> On Sun, Jan 21, 2018 at 08:54:48PM +0800, Jay Zhou wrote:
> > diff --git a/hw/virtio/Makefile.objs b/hw/virtio/Makefile.objs index
> > 765d363..c65dca2 100644
> > --- a/hw/virtio/Makefile.objs
> > +++ b/hw/virtio/Makefile.objs
> > @@ -7,7 +7,7 @@ common-obj-y += virtio-mmio.o  obj-y += virtio.o
> > virtio-balloon.o
> >  obj-$(CONFIG_LINUX) += vhost.o vhost-backend.o vhost-user.o
> >  obj-$(CONFIG_VHOST_VSOCK) += vhost-vsock.o -obj-y += virtio-crypto.o
> > +obj-$(CONFIG_LINUX) += virtio-crypto.o
> >  obj-$(CONFIG_VIRTIO_PCI) += virtio-crypto-pci.o  endif
> >
> 
> This disables virtio crypto completely on non-Linux, which is not nice. We
> should not break working configs.

So If I understand correctly, the virtio crypto device should be compiled in
unconditionally, which is like this:

    obj-y += virtio-crypto.o

> 
> In particular this causes test failures on non-Linux hosts. Peter Maydell was
> kind enough to debug this and reported this backtrace:
> 
> gdb --args ./aarch64-softmmu/qemu-system-aarch64 -device virtio-crypto-pci -
> machine virt [...]
> #0  0x00007f7ff450e6fa in _lwp_kill () from /usr/lib/libc.so.12
> #1  0x00007f7ff450e385 in abort () from /usr/lib/libc.so.12
> #2  0x00007f7ff5c65da2 in g_assertion_message () from /usr/pkg/lib/libglib-
> 2.0.so.0
> #3  0x00007f7ff5c65e11 in g_assertion_message_expr () from
> /usr/pkg/lib/libglib-2.0.so.0
> #4  0x000000000074dc16 in object_initialize_with_type
> (address@hidden, size=<optimized out>, type=0x0)
>     at /root/qemu/qom/object.c:372
> #5  0x000000000074de33 in object_initialize (address@hidden,
> size=<optimized out>, typename=<optimized out>)
>     at /root/qemu/qom/object.c:392
> #6  0x00000000004d2293 in virtio_instance_init_common
> (proxy_obj=0x7f7ff339a000, data=0x7f7ff33a2170, vdev_size=<optimized out>,
>     vdev_name=<optimized out>) at /root/qemu/hw/virtio/virtio.c:2232
> #7  0x000000000074db0d in object_initialize_with_type
> (address@hidden, size=33664, address@hidden)
>     at /root/qemu/qom/object.c:384
> #8  0x000000000074dc66 in object_new_with_type (type=0x7f7ff7b79a80) at
> /root/qemu/qom/object.c:492
> #9  0x000000000074deb9 in object_new (address@hidden
> "virtio-crypto-pci") at /root/qemu/qom/object.c:502
> #10 0x00000000005924d6 in qdev_device_add (opts=0x7f7ff7b4c070,
> address@hidden) at /root/qemu/qdev-monitor.c:615
> #11 0x0000000000594d31 in device_init_func (opaque=<optimized out>,
> opts=<optimized out>, errp=<optimized out>) at /root/qemu/vl.c:2373
> #12 0x0000000000826e56 in qemu_opts_foreach (list=<optimized out>,
> address@hidden <device_init_func>, address@hidden,
>     address@hidden) at /root/qemu/util/qemu-option.c:1073
> #13 0x00000000008b723d in main (argc=<optimized out>, argv=<optimized out>,
> envp=<optimized out>) at /root/qemu/vl.c:4642
> 
> 
> He explained:
> 
> 
> <pm215> ... this is almost certainly the classic "device A depends on device
> B, device B is conditionally compiled but device A isn't"
> <pm215> the type that is missing is virtio-crypto-device <pm215> virtio-
> crypto.o is built only if CONFIG_LINUX, but virtio-crypto-pci is in virtio-
> crypto-pci.c which is built if CONFIG_VIRTIO_PCI

Okay, I see. Thanks for Peter's help.

Regards,
Jay

> 
> 
> --
> MST