Re: linux-next: Tree for Apr 15 (vdpa)

[Michael S. Tsirkin]

On Thu, Apr 16, 2020 at 12:07:06PM +0800, Jason Wang wrote:
> 
> On 2020/4/16 äå12:16, Randy Dunlap wrote:
> > On 4/14/20 10:22 PM, Stephen Rothwell wrote:
> > > Hi all,
> > > 
> > > Changes since 20200414:
> > > 
> > on x86_64:
> > 
> > ERROR: modpost: "vringh_set_iotlb" [drivers/vdpa/vdpa_sim/vdpa_sim.ko] undefined!
> > ERROR: modpost: "vringh_init_iotlb" [drivers/vdpa/vdpa_sim/vdpa_sim.ko] undefined!
> > ERROR: modpost: "vringh_iov_push_iotlb" [drivers/vdpa/vdpa_sim/vdpa_sim.ko] undefined!
> > ERROR: modpost: "vringh_iov_pull_iotlb" [drivers/vdpa/vdpa_sim/vdpa_sim.ko] undefined!
> > ERROR: modpost: "vringh_complete_iotlb" [drivers/vdpa/vdpa_sim/vdpa_sim.ko] undefined!
> > ERROR: modpost: "vringh_getdesc_iotlb" [drivers/vdpa/vdpa_sim/vdpa_sim.ko] undefined!
> > 
> > 
> > Full randconfig file is attached.
> > 
> 
> The config has
> 
> CONFIG_VHOST_IOTLB=m
> CONFIG_VHOST_RING=y
> 
> But we don't select VHOST_IOTLB in VHOST_RING after commit
> e6faeaa128417("vhost: drop vring dependency on iotlb"). Which seems wrong.
> 
> Thanks

Well selecting IOTLB from ring breaks configs which don't need IOTLB.

Legal configurations are:

CONFIG_VHOST_IOTLB=y
CONFIG_VHOST_RING=n

CONFIG_VHOST_IOTLB=m
CONFIG_VHOST_RING=n

CONFIG_VHOST_IOTLB=n
CONFIG_VHOST_RING=n

CONFIG_VHOST_IOTLB=y
CONFIG_VHOST_RING=y

CONFIG_VHOST_IOTLB=n
CONFIG_VHOST_RING=y

CONFIG_VHOST_IOTLB=n
CONFIG_VHOST_RING=m

CONFIG_VHOST_IOTLB=y
CONFIG_VHOST_RING=m

CONFIG_VHOST_IOTLB=m
CONFIG_VHOST_RING=m


So VHOST_RING=y and VHOST_IOTLB=m is the only illegal one.


-- 
MST