commit 2751c9882b947292fcfb084c4f604e01724af804 ("vhost: cross-endian support for legacy devices") introduced a minor regression: even with cross-endian disabled, and even on LE host, vhost_is_little_endian is checking is_le flag so there's always a branch. To fix, simply check virtio_legacy_is_little_endian first. Cc: Greg Kurz <gkurz at linux.vnet.ibm.com> Signed-off-by: Michael S. Tsirkin <mst at redhat.com> --- drivers/vhost/vhost.h | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/drivers/vhost/vhost.h b/drivers/vhost/vhost.h index 4772862..d3f7674 100644 --- a/drivers/vhost/vhost.h +++ b/drivers/vhost/vhost.h @@ -183,10 +183,17 @@ static inline bool vhost_has_feature(struct vhost_virtqueue *vq, int bit) return vq->acked_features & (1ULL << bit); } +#ifdef CONFIG_VHOST_CROSS_ENDIAN_LEGACY static inline bool vhost_is_little_endian(struct vhost_virtqueue *vq) { return vq->is_le; } +#else +static inline bool vhost_is_little_endian(struct vhost_virtqueue *vq) +{ + return virtio_legacy_is_little_endian() || vq->is_le; +} +#endif /* Memory accessors */ static inline u16 vhost16_to_cpu(struct vhost_virtqueue *vq, __virtio16 val) -- MST
On Tue, 27 Oct 2015 11:37:39 +0200 "Michael S. Tsirkin" <mst at redhat.com> wrote:> commit 2751c9882b947292fcfb084c4f604e01724af804 ("vhost: cross-endian > support for legacy devices") introduced a minor regression: even with > cross-endian disabled, and even on LE host, vhost_is_little_endian is > checking is_le flag so there's always a branch. > > To fix, simply check virtio_legacy_is_little_endian first. > > Cc: Greg Kurz <gkurz at linux.vnet.ibm.com> > Signed-off-by: Michael S. Tsirkin <mst at redhat.com>Oops my bad for this oversight... Reviewed-by: Greg Kurz <gkurz at linux.vnet.ibm.com>> --- > drivers/vhost/vhost.h | 7 +++++++ > 1 file changed, 7 insertions(+) > > diff --git a/drivers/vhost/vhost.h b/drivers/vhost/vhost.h > index 4772862..d3f7674 100644 > --- a/drivers/vhost/vhost.h > +++ b/drivers/vhost/vhost.h > @@ -183,10 +183,17 @@ static inline bool vhost_has_feature(struct vhost_virtqueue *vq, int bit) > return vq->acked_features & (1ULL << bit); > } > > +#ifdef CONFIG_VHOST_CROSS_ENDIAN_LEGACY > static inline bool vhost_is_little_endian(struct vhost_virtqueue *vq) > { > return vq->is_le; > } > +#else > +static inline bool vhost_is_little_endian(struct vhost_virtqueue *vq) > +{ > + return virtio_legacy_is_little_endian() || vq->is_le; > +} > +#endif > > /* Memory accessors */ > static inline u16 vhost16_to_cpu(struct vhost_virtqueue *vq, __virtio16 val)
From: "Michael S. Tsirkin" <mst at redhat.com> Date: Tue, 27 Oct 2015 11:37:39 +0200> commit 2751c9882b947292fcfb084c4f604e01724af804 ("vhost: cross-endian > support for legacy devices") introduced a minor regression: even with > cross-endian disabled, and even on LE host, vhost_is_little_endian is > checking is_le flag so there's always a branch. > > To fix, simply check virtio_legacy_is_little_endian first. > > Cc: Greg Kurz <gkurz at linux.vnet.ibm.com> > Signed-off-by: Michael S. Tsirkin <mst at redhat.com>Michael, do you want me to take this via the 'net' tree?
On Tue, Oct 27, 2015 at 06:32:55AM -0700, David Miller wrote:> From: "Michael S. Tsirkin" <mst at redhat.com> > Date: Tue, 27 Oct 2015 11:37:39 +0200 > > > commit 2751c9882b947292fcfb084c4f604e01724af804 ("vhost: cross-endian > > support for legacy devices") introduced a minor regression: even with > > cross-endian disabled, and even on LE host, vhost_is_little_endian is > > checking is_le flag so there's always a branch. > > > > To fix, simply check virtio_legacy_is_little_endian first. > > > > Cc: Greg Kurz <gkurz at linux.vnet.ibm.com> > > Signed-off-by: Michael S. Tsirkin <mst at redhat.com> > > Michael, do you want me to take this via the 'net' tree?Sure - especially since it looks like I have nothing else for this merge window (working on fixing shutdown races, but it's slow going). Thanks a lot! -- MST
From: "Michael S. Tsirkin" <mst at redhat.com> Date: Tue, 27 Oct 2015 11:37:39 +0200> commit 2751c9882b947292fcfb084c4f604e01724af804 ("vhost: cross-endian > support for legacy devices") introduced a minor regression: even with > cross-endian disabled, and even on LE host, vhost_is_little_endian is > checking is_le flag so there's always a branch. > > To fix, simply check virtio_legacy_is_little_endian first. > > Cc: Greg Kurz <gkurz at linux.vnet.ibm.com> > Signed-off-by: Michael S. Tsirkin <mst at redhat.com>Applied.
Reasonably Related Threads
- [PATCH] vhost: fix performance on LE hosts
- [PATCH] vhost: fix performance on LE hosts
- [PATCH v3 7/7] vhost: feature to set the vring endianness
- [PATCH v3 7/7] vhost: feature to set the vring endianness
- [PATCH v4 7/8] vhost: feature to set the vring endianness