| Message ID | 1494553288-30764-2-git-send-email-zhangchen.fnst@cn.fujitsu.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Hi Zhang On 05/11/2017 10:41 PM, Zhang Chen wrote: > Add vnet_hdr_len and using_vnet_hdr arguments in NetClientState > that make othermodule get real vnet_hdr_len easily. > > Signed-off-by: Zhang Chen <zhangchen.fnst@cn.fujitsu.com> > --- > include/net/net.h | 2 ++ > net/net.c | 2 ++ > 2 files changed, 4 insertions(+) > > diff --git a/include/net/net.h b/include/net/net.h > index 99b28d5..70edfc0 100644 > --- a/include/net/net.h > +++ b/include/net/net.h > @@ -100,6 +100,8 @@ struct NetClientState { > unsigned int queue_index; > unsigned rxfilter_notify_enabled:1; > int vring_enable; > + bool using_vnet_hdr; > + int vnet_hdr_len; > QTAILQ_HEAD(NetFilterHead, NetFilterState) filters; > }; > > diff --git a/net/net.c b/net/net.c > index 0ac3b9e..a00a0c9 100644 > --- a/net/net.c > +++ b/net/net.c > @@ -472,6 +472,7 @@ void qemu_using_vnet_hdr(NetClientState *nc, bool enable) > return; > } > > + nc->using_vnet_hdr = enable; > nc->info->using_vnet_hdr(nc, enable); > } > > @@ -491,6 +492,7 @@ void qemu_set_vnet_hdr_len(NetClientState *nc, int len) > return; > } > > + nc->vnet_hdr_len = len; > nc->info->set_vnet_hdr_len(nc, len); > } > For what it's worth, now having those fields in NetClientState it is possible to remove a deref to NetClientInfo in qemu_has_vnet_hdr() and qemu_has_vnet_hdr_len(). Anyway, Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
On 2017年05月14日 05:24, Philippe Mathieu-Daudé wrote: > Hi Zhang > > On 05/11/2017 10:41 PM, Zhang Chen wrote: >> Add vnet_hdr_len and using_vnet_hdr arguments in NetClientState >> that make othermodule get real vnet_hdr_len easily. >> >> Signed-off-by: Zhang Chen <zhangchen.fnst@cn.fujitsu.com> >> --- >> include/net/net.h | 2 ++ >> net/net.c | 2 ++ >> 2 files changed, 4 insertions(+) >> >> diff --git a/include/net/net.h b/include/net/net.h >> index 99b28d5..70edfc0 100644 >> --- a/include/net/net.h >> +++ b/include/net/net.h >> @@ -100,6 +100,8 @@ struct NetClientState { >> unsigned int queue_index; >> unsigned rxfilter_notify_enabled:1; >> int vring_enable; >> + bool using_vnet_hdr; >> + int vnet_hdr_len; >> QTAILQ_HEAD(NetFilterHead, NetFilterState) filters; >> }; >> >> diff --git a/net/net.c b/net/net.c >> index 0ac3b9e..a00a0c9 100644 >> --- a/net/net.c >> +++ b/net/net.c >> @@ -472,6 +472,7 @@ void qemu_using_vnet_hdr(NetClientState *nc, bool >> enable) >> return; >> } >> >> + nc->using_vnet_hdr = enable; >> nc->info->using_vnet_hdr(nc, enable); >> } >> >> @@ -491,6 +492,7 @@ void qemu_set_vnet_hdr_len(NetClientState *nc, >> int len) >> return; >> } >> >> + nc->vnet_hdr_len = len; >> nc->info->set_vnet_hdr_len(nc, len); >> } >> > > For what it's worth, now having those fields in NetClientState it is > possible to remove a deref to NetClientInfo in qemu_has_vnet_hdr() and > qemu_has_vnet_hdr_len(). > > Anyway, > Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> > > Yes and this could be done on top with removing private e.g vnet_hdr_len. Thanks
On 05/15/2017 11:24 AM, Jason Wang wrote: > > > On 2017年05月14日 05:24, Philippe Mathieu-Daudé wrote: >> Hi Zhang >> >> On 05/11/2017 10:41 PM, Zhang Chen wrote: >>> Add vnet_hdr_len and using_vnet_hdr arguments in NetClientState >>> that make othermodule get real vnet_hdr_len easily. >>> >>> Signed-off-by: Zhang Chen <zhangchen.fnst@cn.fujitsu.com> >>> --- >>> include/net/net.h | 2 ++ >>> net/net.c | 2 ++ >>> 2 files changed, 4 insertions(+) >>> >>> diff --git a/include/net/net.h b/include/net/net.h >>> index 99b28d5..70edfc0 100644 >>> --- a/include/net/net.h >>> +++ b/include/net/net.h >>> @@ -100,6 +100,8 @@ struct NetClientState { >>> unsigned int queue_index; >>> unsigned rxfilter_notify_enabled:1; >>> int vring_enable; >>> + bool using_vnet_hdr; >>> + int vnet_hdr_len; >>> QTAILQ_HEAD(NetFilterHead, NetFilterState) filters; >>> }; >>> >>> diff --git a/net/net.c b/net/net.c >>> index 0ac3b9e..a00a0c9 100644 >>> --- a/net/net.c >>> +++ b/net/net.c >>> @@ -472,6 +472,7 @@ void qemu_using_vnet_hdr(NetClientState *nc, >>> bool enable) >>> return; >>> } >>> >>> + nc->using_vnet_hdr = enable; >>> nc->info->using_vnet_hdr(nc, enable); >>> } >>> >>> @@ -491,6 +492,7 @@ void qemu_set_vnet_hdr_len(NetClientState *nc, >>> int len) >>> return; >>> } >>> >>> + nc->vnet_hdr_len = len; >>> nc->info->set_vnet_hdr_len(nc, len); >>> } >>> >> >> For what it's worth, now having those fields in NetClientState it is >> possible to remove a deref to NetClientInfo in qemu_has_vnet_hdr() >> and qemu_has_vnet_hdr_len(). >> >> Anyway, >> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> >> >> > > Yes and this could be done on top with removing private e.g vnet_hdr_len. Do you means we will remove the qemu_has_vnet_hdr() and qemu_has_vnet_hdr_len() in the future? Thanks Zhang Chen > > Thanks > > > . >
On 2017年05月15日 14:56, Zhang Chen wrote: > > > On 05/15/2017 11:24 AM, Jason Wang wrote: >> >> >> On 2017年05月14日 05:24, Philippe Mathieu-Daudé wrote: >>> Hi Zhang >>> >>> On 05/11/2017 10:41 PM, Zhang Chen wrote: >>>> Add vnet_hdr_len and using_vnet_hdr arguments in NetClientState >>>> that make othermodule get real vnet_hdr_len easily. >>>> >>>> Signed-off-by: Zhang Chen <zhangchen.fnst@cn.fujitsu.com> >>>> --- >>>> include/net/net.h | 2 ++ >>>> net/net.c | 2 ++ >>>> 2 files changed, 4 insertions(+) >>>> >>>> diff --git a/include/net/net.h b/include/net/net.h >>>> index 99b28d5..70edfc0 100644 >>>> --- a/include/net/net.h >>>> +++ b/include/net/net.h >>>> @@ -100,6 +100,8 @@ struct NetClientState { >>>> unsigned int queue_index; >>>> unsigned rxfilter_notify_enabled:1; >>>> int vring_enable; >>>> + bool using_vnet_hdr; >>>> + int vnet_hdr_len; >>>> QTAILQ_HEAD(NetFilterHead, NetFilterState) filters; >>>> }; >>>> >>>> diff --git a/net/net.c b/net/net.c >>>> index 0ac3b9e..a00a0c9 100644 >>>> --- a/net/net.c >>>> +++ b/net/net.c >>>> @@ -472,6 +472,7 @@ void qemu_using_vnet_hdr(NetClientState *nc, >>>> bool enable) >>>> return; >>>> } >>>> >>>> + nc->using_vnet_hdr = enable; >>>> nc->info->using_vnet_hdr(nc, enable); >>>> } >>>> >>>> @@ -491,6 +492,7 @@ void qemu_set_vnet_hdr_len(NetClientState *nc, >>>> int len) >>>> return; >>>> } >>>> >>>> + nc->vnet_hdr_len = len; >>>> nc->info->set_vnet_hdr_len(nc, len); >>>> } >>>> >>> >>> For what it's worth, now having those fields in NetClientState it is >>> possible to remove a deref to NetClientInfo in qemu_has_vnet_hdr() >>> and qemu_has_vnet_hdr_len(). >>> >>> Anyway, >>> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> >>> >>> >> >> Yes and this could be done on top with removing private e.g >> vnet_hdr_len. > > Do you means we will remove the qemu_has_vnet_hdr() and > qemu_has_vnet_hdr_len() in the future? > > Thanks > Zhang Chen Yes and e.g both tap and netmap have its private vnet header field. We can remove them too. Thanks > >> >> Thanks >> >> >> . >> >
On 05/15/2017 04:06 PM, Jason Wang wrote: > > > On 2017年05月15日 14:56, Zhang Chen wrote: >> >> >> On 05/15/2017 11:24 AM, Jason Wang wrote: >>> >>> >>> On 2017年05月14日 05:24, Philippe Mathieu-Daudé wrote: >>>> Hi Zhang >>>> >>>> On 05/11/2017 10:41 PM, Zhang Chen wrote: >>>>> Add vnet_hdr_len and using_vnet_hdr arguments in NetClientState >>>>> that make othermodule get real vnet_hdr_len easily. >>>>> >>>>> Signed-off-by: Zhang Chen <zhangchen.fnst@cn.fujitsu.com> >>>>> --- >>>>> include/net/net.h | 2 ++ >>>>> net/net.c | 2 ++ >>>>> 2 files changed, 4 insertions(+) >>>>> >>>>> diff --git a/include/net/net.h b/include/net/net.h >>>>> index 99b28d5..70edfc0 100644 >>>>> --- a/include/net/net.h >>>>> +++ b/include/net/net.h >>>>> @@ -100,6 +100,8 @@ struct NetClientState { >>>>> unsigned int queue_index; >>>>> unsigned rxfilter_notify_enabled:1; >>>>> int vring_enable; >>>>> + bool using_vnet_hdr; >>>>> + int vnet_hdr_len; >>>>> QTAILQ_HEAD(NetFilterHead, NetFilterState) filters; >>>>> }; >>>>> >>>>> diff --git a/net/net.c b/net/net.c >>>>> index 0ac3b9e..a00a0c9 100644 >>>>> --- a/net/net.c >>>>> +++ b/net/net.c >>>>> @@ -472,6 +472,7 @@ void qemu_using_vnet_hdr(NetClientState *nc, >>>>> bool enable) >>>>> return; >>>>> } >>>>> >>>>> + nc->using_vnet_hdr = enable; >>>>> nc->info->using_vnet_hdr(nc, enable); >>>>> } >>>>> >>>>> @@ -491,6 +492,7 @@ void qemu_set_vnet_hdr_len(NetClientState *nc, >>>>> int len) >>>>> return; >>>>> } >>>>> >>>>> + nc->vnet_hdr_len = len; >>>>> nc->info->set_vnet_hdr_len(nc, len); >>>>> } >>>>> >>>> >>>> For what it's worth, now having those fields in NetClientState it >>>> is possible to remove a deref to NetClientInfo in >>>> qemu_has_vnet_hdr() and qemu_has_vnet_hdr_len(). >>>> >>>> Anyway, >>>> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> >>>> >>>> >>> >>> Yes and this could be done on top with removing private e.g >>> vnet_hdr_len. >> >> Do you means we will remove the qemu_has_vnet_hdr() and >> qemu_has_vnet_hdr_len() in the future? >> >> Thanks >> Zhang Chen > > Yes and e.g both tap and netmap have its private vnet header field. We > can remove them too. Maybe I should do this job in this series next version or a independent series? Thanks Zhang Chen > > Thanks > >> >>> >>> Thanks >>> >>> >>> . >>> >> > > > > . >
On 2017年05月15日 16:14, Zhang Chen wrote: >> Yes and e.g both tap and netmap have its private vnet header field. >> We can remove them too. > > Maybe I should do this job in this series next version or a > independent series? > > Thanks > Zhang Chen Your call. Thanks
On 05/16/2017 02:49 PM, Jason Wang wrote: > > > On 2017年05月15日 16:14, Zhang Chen wrote: >>> Yes and e.g both tap and netmap have its private vnet header field. >>> We can remove them too. >> >> Maybe I should do this job in this series next version or a >> independent series? >> >> Thanks >> Zhang Chen > > Your call. I will send a independent series after this series have been merged. Thanks Zhang Chen > > Thanks > > >
diff --git a/include/net/net.h b/include/net/net.h index 99b28d5..70edfc0 100644 --- a/include/net/net.h +++ b/include/net/net.h @@ -100,6 +100,8 @@ struct NetClientState { unsigned int queue_index; unsigned rxfilter_notify_enabled:1; int vring_enable; + bool using_vnet_hdr; + int vnet_hdr_len; QTAILQ_HEAD(NetFilterHead, NetFilterState) filters; }; diff --git a/net/net.c b/net/net.c index 0ac3b9e..a00a0c9 100644 --- a/net/net.c +++ b/net/net.c @@ -472,6 +472,7 @@ void qemu_using_vnet_hdr(NetClientState *nc, bool enable) return; } + nc->using_vnet_hdr = enable; nc->info->using_vnet_hdr(nc, enable); } @@ -491,6 +492,7 @@ void qemu_set_vnet_hdr_len(NetClientState *nc, int len) return; } + nc->vnet_hdr_len = len; nc->info->set_vnet_hdr_len(nc, len); }
Add vnet_hdr_len and using_vnet_hdr arguments in NetClientState that make othermodule get real vnet_hdr_len easily. Signed-off-by: Zhang Chen <zhangchen.fnst@cn.fujitsu.com> --- include/net/net.h | 2 ++ net/net.c | 2 ++ 2 files changed, 4 insertions(+)