net/socket.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-)
Since commit 0f8c289ad "net: fix -netdev socket,fd= for UDP sockets"
we allow more than one parameter for -netdev socket. But now
we run into an assert when no parameter at all is specified
> qemu-system-x86_64 -net socket
socket.c:729: net_init_socket: Assertion `sock->has_udp' failed.
Change the check and error message to reflect that at least
one of the options has to be specified. An example for when we need
more than on is when we hand in a UDP socket. Then we have fd= and udp=
as parameters.
Cc: Jason Wang <jasowang@redhat.com>
Fixes: 0f8c289ad539feb5135c545bea947b310a893f4b
Reported-by: Mao Zhongyi <maozy.fnst@cn.fujitsu.com>
Signed-off-by: Jens Freimann <jfreimann@redhat.com>
---
v1->v2:
- add check to prevent listen= and connect= being used at the same time
net/socket.c | 11 ++++++++---
1 file changed, 8 insertions(+), 3 deletions(-)
diff --git a/net/socket.c b/net/socket.c
index e6b471c63d..74a2eff2e0 100644
--- a/net/socket.c
+++ b/net/socket.c
@@ -695,13 +695,18 @@ int net_init_socket(const Netdev *netdev, const char *name,
assert(netdev->type == NET_CLIENT_DRIVER_SOCKET);
sock = &netdev->u.socket;
- if (sock->has_listen + sock->has_connect + sock->has_mcast +
- sock->has_udp > 1) {
- error_setg(errp, "exactly one of listen=, connect=, mcast= or udp="
+ if (sock->has_fd + sock->has_listen + sock->has_connect + sock->has_mcast +
+ sock->has_udp == 0) {
+ error_setg(errp, "at least one of fd=, listen=, connect=, mcast= or udp="
" is required");
return -1;
}
+ if (sock->has_listen && sock->has_connect) {
+ error_setg(errp, "listen= and connect= can't be used together");
+ return -1;
+ }
+
if (sock->has_localaddr && !sock->has_mcast && !sock->has_udp) {
error_setg(errp, "localaddr= is only valid with mcast= or udp=");
return -1;
--
2.13.5
Hi, Jens On 09/26/2017 08:59 PM, Jens Freimann wrote: > Since commit 0f8c289ad "net: fix -netdev socket,fd= for UDP sockets" > we allow more than one parameter for -netdev socket. But now > we run into an assert when no parameter at all is specified > >> qemu-system-x86_64 -net socket > socket.c:729: net_init_socket: Assertion `sock->has_udp' failed. > > Change the check and error message to reflect that at least > one of the options has to be specified. An example for when we need > more than on is when we hand in a UDP socket. Then we have fd= and udp= s/on/one > as parameters. > > Cc: Jason Wang <jasowang@redhat.com> > Fixes: 0f8c289ad539feb5135c545bea947b310a893f4b > Reported-by: Mao Zhongyi <maozy.fnst@cn.fujitsu.com> > Signed-off-by: Jens Freimann <jfreimann@redhat.com> > --- > v1->v2: > - add check to prevent listen= and connect= being used at the same time > > net/socket.c | 11 ++++++++--- > 1 file changed, 8 insertions(+), 3 deletions(-) > > diff --git a/net/socket.c b/net/socket.c > index e6b471c63d..74a2eff2e0 100644 > --- a/net/socket.c > +++ b/net/socket.c > @@ -695,13 +695,18 @@ int net_init_socket(const Netdev *netdev, const char *name, > assert(netdev->type == NET_CLIENT_DRIVER_SOCKET); > sock = &netdev->u.socket; > > - if (sock->has_listen + sock->has_connect + sock->has_mcast + > - sock->has_udp > 1) { > - error_setg(errp, "exactly one of listen=, connect=, mcast= or udp=" > + if (sock->has_fd + sock->has_listen + sock->has_connect + sock->has_mcast + > + sock->has_udp == 0) { > + error_setg(errp, "at least one of fd=, listen=, connect=, mcast= or udp=" > " is required"); > return -1; > } Looks like this allow "listen" and "mcast" can be used at the same time? I'm not sure. the original: if (sock->has_fd + sock->has_listen + sock->has_connect + sock->has_mcast + sock->has_udp != 1) It means that only one of these elements can be used to a QEmu instance at the same time, i.e these elements are mutually exclusive. After this patch, are some cases missing? It's just my personal understanding, please judge. :) Thanks, Mao > > + if (sock->has_listen && sock->has_connect) { > + error_setg(errp, "listen= and connect= can't be used together"); > + return -1; > + } > + > if (sock->has_localaddr && !sock->has_mcast && !sock->has_udp) { > error_setg(errp, "localaddr= is only valid with mcast= or udp="); > return -1; >
On Tue, Sep 26, 2017 at 02:38:33PM +0000, Mao Zhongyi wrote: > >Hi, Jens > > >On 09/26/2017 08:59 PM, Jens Freimann wrote: >>Since commit 0f8c289ad "net: fix -netdev socket,fd= for UDP sockets" >>we allow more than one parameter for -netdev socket. But now >>we run into an assert when no parameter at all is specified >> >>>qemu-system-x86_64 -net socket >>socket.c:729: net_init_socket: Assertion `sock->has_udp' failed. >> >>Change the check and error message to reflect that at least >>one of the options has to be specified. An example for when we need >>more than on is when we hand in a UDP socket. Then we have fd= and udp= > >s/on/one > >>as parameters. >> >>Cc: Jason Wang <jasowang@redhat.com> >>Fixes: 0f8c289ad539feb5135c545bea947b310a893f4b >>Reported-by: Mao Zhongyi <maozy.fnst@cn.fujitsu.com> >>Signed-off-by: Jens Freimann <jfreimann@redhat.com> >>--- >>v1->v2: >>- add check to prevent listen= and connect= being used at the same time >> >> net/socket.c | 11 ++++++++--- >> 1 file changed, 8 insertions(+), 3 deletions(-) >> >>diff --git a/net/socket.c b/net/socket.c >>index e6b471c63d..74a2eff2e0 100644 >>--- a/net/socket.c >>+++ b/net/socket.c >>@@ -695,13 +695,18 @@ int net_init_socket(const Netdev *netdev, const char *name, >> assert(netdev->type == NET_CLIENT_DRIVER_SOCKET); >> sock = &netdev->u.socket; >> >>- if (sock->has_listen + sock->has_connect + sock->has_mcast + >>- sock->has_udp > 1) { >>- error_setg(errp, "exactly one of listen=, connect=, mcast= or udp=" >>+ if (sock->has_fd + sock->has_listen + sock->has_connect + sock->has_mcast + >>+ sock->has_udp == 0) { >>+ error_setg(errp, "at least one of fd=, listen=, connect=, mcast= or udp=" >> " is required"); >> return -1; >> } > >Looks like this allow "listen" and "mcast" can be used at the same time? >I'm not sure. > >the original: >if (sock->has_fd + sock->has_listen + sock->has_connect + sock->has_mcast + > sock->has_udp != 1) > >It means that only one of these elements can be used to a QEmu instance at >the same time, i.e these elements are mutually exclusive. After this patch, >are some cases missing? > >It's just my personal understanding, please judge. :) Looking at this again and re-testing with the (not yet applied) pxe test for vhost-user bridge, I think we can just revert this check back to != 1. More than one parameters is not needed. I'll send a patch with only this change. Thanks for the review! regards, Jens
© 2016 - 2024 Red Hat, Inc.