Add generic packet buffering API
cover Add generic packet buffering API
Hi,
This patch series works towards two[1] goals:
1) Merging the code in qemu-kvm which drains the tap file descriptor when it is readable and buffers a packet if it fails to add it to the NIC receive queue
2) Using the new TAP_SETSNDBUF ioctl() to put a limit on the number of in-flight packets allowed on a tap device; in this case, if a NIC pops a packet from its transmit queue, we need to be able to buffer said packet if the tap queue is full
Rather than having each host backend and each NIC implement
packet buffering itself, the idea is to add extend the VLAN send queue to do buffering.
The API is fairly simple:
- A qemu_send_packet() variant which takes a completion callback is added; if this returns zero, the sender should stop it's send queue until the completion callback is invoked
- The VLANClientState receive() handler can now return zero, indicating that no space is available to receive the packet; when space becomes available later, the receiver must call qemu_flush_queued_packets() to re-queue any buffered packets which, in turn, will also cause the sender to re-start
Note, these patches depend on two patches sent to the list by
Jan Kiszka:
[PATCH 01/11] net: Don't deliver to disabled interfaces in qemu_sendv_packet [PATCH 02/11] net: Fix and improved ordered packet delivery
Cheers,
Mark.
[1] - Eagle-eyed readers will notice that I've neglected to actually meet goal (2) with the patch set. Patience, patience :-)
PATCH 1/13 net: factor tap_read_packet() out of tap_send()
Move portability clutter out into its own function.
--- net.c | 28 +++++++++++++++++++--------- 1 files changed, 19 insertions(+), 9 deletions(-)
net.c +19 -9
@@ -806,21 +806,31 @@ static void tap_receive(void *opaque, const uint8_t *buf, int size) } } -static void tap_send (void *opaque) -{ -TAPState *s = opaque; -uint8_t buf[4096 ]; -int size; -#ifdef __sun__ +static ssize_t tap_read_packet (int tapfd,uint8_t *buf,int maxlen) +{struct strbuf sbuf;int f =0 ; - sbuf.maxlen =sizeof (buf); + + sbuf.maxlen = maxlen; sbuf.buf = (char *)buf; - size =getmsg (s->fd,NULL , &sbuf, &f) >=0 ? sbuf.len :-1 ; + +return getmsg (tapfd,NULL , &sbuf, &f) >=0 ? sbuf.len :-1 ; +}#else - size =read (s->fd, buf,sizeof (buf)); +static ssize_t tap_read_packet (int tapfd,uint8_t *buf,int maxlen) +{ +return read (tapfd, buf, maxlen); +}#endif + +static void tap_send (void *opaque) +{ +TAPState *s = opaque; +uint8_t buf[4096 ]; +int size; + + size =tap_read_packet (s->fd, buf,sizeof (buf));if (size >0 ) {qemu_send_packet (s->vc, buf, size); }
Signature · 1 line
1.6.0.6
PATCH 2/13 net: move the tap buffer into TAPState
KVM uses a 64k buffer for reading from tapfd (for GSO support) and allocates the buffer with TAPState rather than on the stack.
Not allocating it on the stack probably makes sense for qemu anyway, so merge it in advance of GSO support.
--- net.c | 6 +++--- 1 files changed, 3 insertions(+), 3 deletions(-)
net.c +3 -3
@@ -776,6 +776,7 @@ typedef struct TAPState {int fd;char down_script[1024 ];char down_script_arg[128 ]; +uint8_t buf[4096 ]; } TAPState;static int launch_script (const char *setup_script,const char *ifname,int fd);
@@ -827,12 +828,11 @@ static ssize_t tap_read_packet(int tapfd, uint8_t *buf, int maxlen)static void tap_send (void *opaque) {TAPState *s = opaque; -uint8_t buf[4096 ];int size; - size =tap_read_packet (s->fd, buf,sizeof (buf)); + size =tap_read_packet (s->fd, s->buf,sizeof (s->buf));if (size >0 ) { -qemu_send_packet (s->vc, buf, size); +qemu_send_packet (s->vc, s->buf, size); } }
Signature · 1 line
1.6.0.6
PATCH 3/13 net: vlan clients with no fd_can_read() can always receive
If a vlan client has no fd_can_read(), that means it can always receive packets. The current code assumes it can *never* receive packets.
--- net.c | 16 ++++++++++------ 1 files changed, 10 insertions(+), 6 deletions(-)
net.c +10 -6
@@ -389,15 +389,19 @@ VLANClientState *qemu_find_vlan_client(VLANState *vlan, void *opaque)return NULL ; } -int qemu_can_send_packet (VLANClientState *vc1) +int qemu_can_send_packet (VLANClientState *sender) { -VLANState *vlan = vc1->vlan; +VLANState *vlan = sender->vlan;VLANClientState *vc; -for (vc = vlan->first_client; vc !=NULL ; vc = vc->next) { -if (vc != vc1) { -if (vc->fd_can_read && vc->fd_can_read (vc->opaque)) -return 1 ; +for (vc = vlan->first_client; vc !=NULL ; vc = vc->next) { +if (vc == sender) { +continue ; + } + +/* no fd_can_read() handler, they can always receive */ +if (!vc->fd_can_read || vc->fd_can_read (vc->opaque)) { +return 1 ; } }return 0 ;
Signature · 1 line
1.6.0.6
PATCH 4/13 net: only read from tapfd when we can send
Reduce the number of packets dropped under heavy network traffic by only reading a packet from the tapfd when a client can actually handle it.
--- net.c | 9 ++++++++- 1 files changed, 8 insertions(+), 1 deletions(-)
net.c +8 -1
@@ -811,6 +811,13 @@ static void tap_receive(void *opaque, const uint8_t *buf, int size) } } +static int tap_can_send (void *opaque) +{ +TAPState *s = opaque; + +return qemu_can_send_packet (s->vc); +} +#ifdef __sun__static ssize_t tap_read_packet (int tapfd,uint8_t *buf,int maxlen) {
@@ -866,7 +873,7 @@ static TAPState *net_tap_fd_init(VLANState *vlan, s->vc =qemu_new_vlan_client (vlan, model, name, tap_receive,NULL , tap_cleanup, s); s->vc->fd_readv = tap_receive_iov; -qemu_set_fd_handler (s->fd, tap_send,NULL , s); +qemu_set_fd_handler2 (s->fd, tap_can_send, tap_send,NULL , s);snprintf (s->vc->info_str,sizeof (s->vc->info_str),"fd=%d" , fd);return s; }
Signature · 1 line
1.6.0.6
PATCH 5/13 net: add fd_readv() handler to qemu_new_vlan_client() args
This, apparently, is the style we prefer - all VLANClientState should be an argument to qemu_new_vlan_client().
--- hw/dp8393x.c | 4 ++-- hw/e1000.c | 4 ++-- hw/eepro100.c | 2 +- hw/etraxfs_eth.c | 2 +- hw/mcf_fec.c | 2 +- hw/mipsnet.c | 2 +- hw/musicpal.c | 2 +- hw/ne2000.c | 4 ++-- hw/pcnet.c | 2 +- hw/qdev.c | 7 ++++--- hw/rtl8139.c | 2 +- hw/smc91c111.c | 2 +- hw/stellaris_enet.c | 2 +- hw/usb-net.c | 3 ++- hw/virtio-net.c | 2 +- hw/xen_nic.c | 4 ++-- net.c | 21 +++++++++++---------- net.h | 6 ++++-- tap-win32.c | 2 +- 19 files changed, 40 insertions(+), 35 deletions(-)
hw/dp8393x.c +2 -2
@@ -888,8 +888,8 @@ void dp83932_init(NICInfo *nd, target_phys_addr_t base, int it_shift, s->watchdog =qemu_new_timer (vm_clock, dp8393x_watchdog, s); s->regs[SONIC_SR ] =0x0004 ;/* only revision recognized by Linux */ - s->vc =qemu_new_vlan_client (nd->vlan, nd->model, nd->name, - nic_receive, nic_can_receive, nic_cleanup, s); + s->vc =qemu_new_vlan_client (nd->vlan, nd->model, nd->name, nic_can_receive, + nic_receive,NULL , nic_cleanup, s);qemu_format_nic_info_str (s->vc, nd->macaddr);qemu_register_reset (nic_reset, s);
hw/e1000.c +2 -2
@@ -1099,8 +1099,8 @@ static void pci_e1000_init(PCIDevice *pci_dev)memset (&d->tx,0 ,sizeof d->tx); d->vc =qdev_get_vlan_client (&d->dev.qdev, - e1000_receive, e1000_can_receive, - e1000_cleanup, d); + e1000_can_receive, e1000_receive, +NULL , e1000_cleanup, d); d->vc->link_status_changed = e1000_set_link_status;qemu_format_nic_info_str (d->vc, macaddr);
hw/eepro100.c +1 -1
@@ -1766,7 +1766,7 @@ static void nic_init(PCIDevice *pci_dev, uint32_t device)nic_reset (s); s->vc =qdev_get_vlan_client (&d->dev.qdev, - nic_receive, nic_can_receive, + nic_can_receive, nic_receive,NULL , nic_cleanup, s);qemu_format_nic_info_str (s->vc, s->macaddr);
hw/etraxfs_eth.c +1 -1
@@ -593,7 +593,7 @@ void *etraxfs_eth_init(NICInfo *nd, CPUState *env,cpu_register_physical_memory (base,0x5c , eth->ethregs); eth->vc =qemu_new_vlan_client (nd->vlan, nd->model, nd->name, - eth_receive, eth_can_receive, + eth_can_receive, eth_receive,NULL , eth_cleanup, eth); eth->vc->opaque = eth; eth->vc->link_status_changed = eth_set_link;
hw/mcf_fec.c +1 -1
@@ -462,7 +462,7 @@ void mcf_fec_init(NICInfo *nd, target_phys_addr_t base, qemu_irq *irq)cpu_register_physical_memory (base,0x400 , s->mmio_index); s->vc =qemu_new_vlan_client (nd->vlan, nd->model, nd->name, - mcf_fec_receive, mcf_fec_can_receive, + mcf_fec_can_receive, mcf_fec_receive,NULL , mcf_fec_cleanup, s);memcpy (s->macaddr, nd->macaddr,6 );qemu_format_nic_info_str (s->vc, s->macaddr);
hw/mipsnet.c +1 -1
@@ -262,7 +262,7 @@ void mipsnet_init (int base, qemu_irq irq, NICInfo *nd) s->irq = irq;if (nd && nd->vlan) { s->vc =qemu_new_vlan_client (nd->vlan, nd->model, nd->name, - mipsnet_receive, mipsnet_can_receive, + mipsnet_can_receive, mipsnet_receive,NULL , mipsnet_cleanup, s); }else { s->vc =NULL ;
hw/musicpal.c +1 -1
@@ -753,7 +753,7 @@ static void mv88w8618_eth_init(SysBusDevice *dev)sysbus_init_irq (dev, &s->irq); s->vc =qdev_get_vlan_client (&dev->qdev, - eth_receive, eth_can_receive, + eth_can_receive, eth_receive,NULL , eth_cleanup, s); s->mmio_index =cpu_register_io_memory (0 , mv88w8618_eth_readfn, mv88w8618_eth_writefn, s);
hw/ne2000.c +2 -2
@@ -757,7 +757,7 @@ void isa_ne2000_init(int base, qemu_irq irq, NICInfo *nd)ne2000_reset (s); s->vc =qemu_new_vlan_client (nd->vlan, nd->model, nd->name, - ne2000_receive, ne2000_can_receive, + ne2000_can_receive, ne2000_receive,NULL , isa_ne2000_cleanup, s);qemu_format_nic_info_str (s->vc, s->macaddr);
@@ -821,7 +821,7 @@ static void pci_ne2000_init(PCIDevice *pci_dev)qdev_get_macaddr (&d->dev.qdev, s->macaddr);ne2000_reset (s); s->vc =qdev_get_vlan_client (&d->dev.qdev, - ne2000_receive, ne2000_can_receive, + ne2000_can_receive, ne2000_receive,NULL , ne2000_cleanup, s);qemu_format_nic_info_str (s->vc, s->macaddr);
hw/pcnet.c +1 -1
@@ -1952,7 +1952,7 @@ static void pcnet_common_init(DeviceState *dev, PCNetState *s,qdev_get_macaddr (dev, s->macaddr); s->vc =qdev_get_vlan_client (dev, - pcnet_receive, pcnet_can_receive, + pcnet_can_receive, pcnet_receive,NULL , cleanup, s);pcnet_h_reset (s);register_savevm ("pcnet" ,-1 ,2 , pcnet_save, pcnet_load, s);
hw/qdev.c +4 -3
@@ -233,15 +233,16 @@ void qdev_connect_gpio_out(DeviceState * dev, int n, qemu_irq pin) }VLANClientState *qdev_get_vlan_client (DeviceState *dev, -IOReadHandler *fd_read,IOCanRWHandler *fd_can_read, +IOReadHandler *fd_read, +IOReadvHandler *fd_readv,NetCleanup *cleanup,void *opaque) {NICInfo *nd = dev->nd;assert (nd); -return qemu_new_vlan_client (nd->vlan, nd->model, nd->name, - fd_read, fd_can_read, cleanup, opaque); +return qemu_new_vlan_client (nd->vlan, nd->model, nd->name, fd_can_read, + fd_read, fd_readv, cleanup, opaque); }
hw/rtl8139.c +1 -1
@@ -3475,7 +3475,7 @@ static void pci_rtl8139_init(PCIDevice *dev)qdev_get_macaddr (&dev->qdev, s->macaddr);rtl8139_reset (s); s->vc =qdev_get_vlan_client (&dev->qdev, - rtl8139_receive, rtl8139_can_receive, + rtl8139_can_receive, rtl8139_receive,NULL , rtl8139_cleanup, s);qemu_format_nic_info_str (s->vc, s->macaddr);
hw/smc91c111.c +1 -1
@@ -711,7 +711,7 @@ static void smc91c111_init1(SysBusDevice *dev)smc91c111_reset (s); s->vc =qdev_get_vlan_client (&dev->qdev, - smc91c111_receive, smc91c111_can_receive, + smc91c111_can_receive, smc91c111_receive,NULL , smc91c111_cleanup, s);qemu_format_nic_info_str (s->vc, s->macaddr);/* ??? Save/restore. */
hw/stellaris_enet.c +1 -1
@@ -405,8 +405,8 @@ static void stellaris_enet_init(SysBusDevice *dev)qdev_get_macaddr (&dev->qdev, s->macaddr); s->vc =qdev_get_vlan_client (&dev->qdev, - stellaris_enet_receive, stellaris_enet_can_receive, + stellaris_enet_receive,NULL , stellaris_enet_cleanup, s);qemu_format_nic_info_str (s->vc, s->macaddr);
hw/usb-net.c +2 -1
@@ -1458,8 +1458,9 @@ USBDevice *usb_net_init(NICInfo *nd)pstrcpy (s->dev.devname,sizeof (s->dev.devname),"QEMU USB Network Interface" ); s->vc =qemu_new_vlan_client (nd->vlan, nd->model, nd->name, - usbnet_receive, usbnet_can_receive, + usbnet_receive, +NULL , usbnet_cleanup, s);qemu_format_nic_info_str (s->vc, s->mac);
hw/virtio-net.c +1 -1
@@ -606,8 +606,8 @@ VirtIODevice *virtio_net_init(DeviceState *dev)qdev_get_macaddr (dev, n->mac); n->status =VIRTIO_NET_S_LINK_UP ; n->vc =qdev_get_vlan_client (dev, - virtio_net_receive, virtio_net_can_receive, + virtio_net_receive,NULL , virtio_net_cleanup, n); n->vc->link_status_changed = virtio_net_set_link_status;
hw/xen_nic.c +2 -2
@@ -303,8 +303,8 @@ static int net_init(struct XenDevice *xendev) vlan =qemu_find_vlan (netdev->xendev.dev); netdev->vs =qemu_new_vlan_client (vlan,"xen" ,NULL , - net_rx_packet, net_rx_ok,NULL , - netdev); + net_rx_ok, net_rx_packet,NULL , +NULL , netdev);snprintf (netdev->vs->info_str,sizeof (netdev->vs->info_str),"nic: xenbus vif macaddr=%s" , netdev->mac);
net.c +11 -10
@@ -332,8 +332,9 @@ static char *assign_name(VLANClientState *vc1, const char *model)VLANClientState *qemu_new_vlan_client (VLANState *vlan,const char *model,const char *name, -IOReadHandler *fd_read,IOCanRWHandler *fd_can_read, +IOReadHandler *fd_read, +IOReadvHandler *fd_readv,NetCleanup *cleanup,void *opaque) {
@@ -344,8 +345,9 @@ VLANClientState *qemu_new_vlan_client(VLANState *vlan, vc->name =strdup (name); else vc->name = assign_name(vc, model); - vc->fd_read = fd_read; vc->fd_can_read = fd_can_read; + vc->fd_read = fd_read; + vc->fd_readv = fd_readv; vc->cleanup = cleanup; vc->opaque = opaque; vc->vlan = vlan;
@@ -592,7 +594,7 @@ static int net_slirp_init(VLANState *vlan, const char *model, const char *name) slirp_inited =1 ;slirp_init (slirp_restrict, slirp_ip); } - slirp_vc =qemu_new_vlan_client (vlan, model, name, + slirp_vc =qemu_new_vlan_client (vlan, model, name,NULL , slirp_receive,NULL , net_slirp_cleanup,NULL ); slirp_vc->info_str[0 ] ='\0' ; slirp_in_use =1 ;
@@ -870,9 +872,8 @@ static TAPState *net_tap_fd_init(VLANState *vlan, s =qemu_mallocz (sizeof (TAPState)); s->fd = fd; - s->vc =qemu_new_vlan_client (vlan, model, name, tap_receive, -NULL , tap_cleanup, s); - s->vc->fd_readv = tap_receive_iov; + s->vc =qemu_new_vlan_client (vlan, model, name,NULL , tap_receive, + tap_receive_iov, tap_cleanup, s);qemu_set_fd_handler2 (s->fd, tap_can_send, tap_send,NULL , s);snprintf (s->vc->info_str,sizeof (s->vc->info_str),"fd=%d" , fd);return s;
@@ -1198,7 +1199,7 @@ static int net_vde_init(VLANState *vlan, const char *model,free (s);return -1 ; } - s->vc =qemu_new_vlan_client (vlan, model, name, vde_from_qemu, + s->vc =qemu_new_vlan_client (vlan, model, name,NULL , vde_from_qemu,NULL , vde_cleanup, s);qemu_set_fd_handler (vde_datafd (s->vde), vde_to_qemu,NULL , s); snprintf(s->vc->info_str, sizeof(s->vc->info_str), "sock=%s,fd=%d",
@@ -1436,7 +1437,7 @@ static NetSocketState *net_socket_fd_init_dgram(VLANState *vlan, s =qemu_mallocz (sizeof (NetSocketState)); s->fd = fd; - s->vc =qemu_new_vlan_client (vlan, model, name, net_socket_receive_dgram, + s->vc =qemu_new_vlan_client (vlan, model, name,NULL , net_socket_receive_dgram,NULL , net_socket_cleanup, s);qemu_set_fd_handler (s->fd, net_socket_send_dgram,NULL , s);
@@ -1464,7 +1465,7 @@ static NetSocketState *net_socket_fd_init_stream(VLANState *vlan,NetSocketState *s; s =qemu_mallocz (sizeof (NetSocketState)); s->fd = fd; - s->vc =qemu_new_vlan_client (vlan, model, name, net_socket_receive, + s->vc =qemu_new_vlan_client (vlan, model, name,NULL , net_socket_receive,NULL , net_socket_cleanup, s);snprintf (s->vc->info_str,sizeof (s->vc->info_str),"socket: fd=%d" , fd);
@@ -1743,7 +1744,7 @@ static int net_dump_init(VLANState *vlan, const char *device,return -1 ; } - s->pcap_vc =qemu_new_vlan_client (vlan, device, name, dump_receive,NULL , + s->pcap_vc =qemu_new_vlan_client (vlan, device, name,NULL , dump_receive,NULL , net_dump_cleanup, s);snprintf (s->pcap_vc->info_str,sizeof (s->pcap_vc->info_str),"dump to %s (len=%d)" , filename, len);
net.h +4 -2
@@ -51,8 +51,9 @@ VLANState *qemu_find_vlan(int id);VLANClientState *qemu_new_vlan_client (VLANState *vlan,const char *model,const char *name, -IOReadHandler *fd_read,IOCanRWHandler *fd_can_read, +IOReadHandler *fd_read, +IOReadvHandler *fd_readv,NetCleanup *cleanup,void *opaque);void qemu_del_vlan_client (VLANClientState *vc);
@@ -129,8 +130,9 @@ void net_host_device_remove(Monitor *mon, int vlan_id, const char *device);void qdev_get_macaddr (DeviceState *dev,uint8_t *macaddr);VLANClientState *qdev_get_vlan_client (DeviceState *dev, -IOReadHandler *fd_read,IOCanRWHandler *fd_can_read, +IOReadHandler *fd_read, +IOReadvHandler *fd_readv,NetCleanup *cleanup,void *opaque);
tap-win32.c +1 -1
@@ -684,7 +684,7 @@ int tap_win32_init(VLANState *vlan, const char *model,return -1 ; } - s->vc =qemu_new_vlan_client (vlan, model, name, tap_receive, + s->vc =qemu_new_vlan_client (vlan, model, name,NULL , tap_receive,NULL , tap_cleanup, s); snprintf(s->vc->info_str, sizeof(s->vc->info_str),
Signature · 1 line
1.6.0.6
PATCH 6/13 net: re-name vc->fd_read() to vc->receive()
VLANClientState's fd_read() handler doesn't read from file descriptors, it adds a buffer to the client's receive queue.
Re-name the handlers to make things a little less confusing.
--- hw/dp8393x.c | 4 ++-- hw/e1000.c | 6 +++--- hw/eepro100.c | 2 +- hw/etraxfs_eth.c | 2 +- hw/mcf_fec.c | 2 +- hw/mipsnet.c | 2 +- hw/musicpal.c | 2 +- hw/ne2000.c | 2 +- hw/pcnet.c | 2 +- hw/qdev.c | 10 +++++----- hw/rtl8139.c | 2 +- hw/smc91c111.c | 2 +- hw/stellaris_enet.c | 2 +- hw/usb-net.c | 2 +- hw/virtio-net.c | 2 +- hw/xen_nic.c | 6 +++--- net.c | 36 ++++++++++++++++++------------------ net.h | 23 ++++++++++++----------- savevm.c | 2 +- tap-win32.c | 2 +- 20 files changed, 57 insertions(+), 56 deletions(-)
hw/dp8393x.c +2 -2
@@ -409,7 +409,7 @@ static void do_transmit_packets(dp8393xState *s) s->regs[SONIC_TCR ] |=SONIC_TCR_CRSL ;if (s->vc->fd_can_read (s)) { s->loopback_packet =1 ; - s->vc->fd_read (s, s->tx_buffer, tx_len); + s->vc->receive (s, s->tx_buffer, tx_len); } }else {/* Transmit packet */
@@ -725,7 +725,7 @@ static int receive_filter(dp8393xState *s, const uint8_t * buf, int size)return -1 ; } -static void nic_receive (void *opaque,const uint8_t * buf,int size) +static void nic_receive (void *opaque,const uint8_t * buf,size_t size) {uint16_t data[10 ];dp8393xState *s = opaque;
hw/e1000.c +3 -3
@@ -593,7 +593,7 @@ e1000_can_receive(void *opaque) }static void -e1000_receive (void *opaque,const uint8_t *buf,int size) +e1000_receive (void *opaque,const uint8_t *buf,size_t size) {E1000State *s = opaque;struct e1000_rx_desc desc;
@@ -607,8 +607,8 @@ e1000_receive(void *opaque, const uint8_t *buf, int size)return ;if (size > s->rxbuf_size) { -DBGOUT (RX ,"packet too large for buffers (%d > %d)\n" , size, - s->rxbuf_size); +DBGOUT (RX ,"packet too large for buffers (%lu > %d)\n" , + (unsigned long )size, s->rxbuf_size);return ; }
hw/eepro100.c +1 -1
@@ -1441,7 +1441,7 @@ static int nic_can_receive(void *opaque)//~ return !eepro100_buffer_full(s); } -static void nic_receive (void *opaque,const uint8_t * buf,int size) +static void nic_receive (void *opaque,const uint8_t * buf,size_t size) {/* TODO: * - Magic packets should set bit 30 in power management driver register.
hw/etraxfs_eth.c +1 -1
@@ -501,7 +501,7 @@ static int eth_can_receive(void *opaque)return 1 ; } -static void eth_receive (void *opaque,const uint8_t *buf,int size) +static void eth_receive (void *opaque,const uint8_t *buf,size_t size) {unsigned char sa_bcast[6 ] = {0xff ,0xff ,0xff ,0xff ,0xff ,0xff };struct fs_eth *eth = opaque;
hw/mcf_fec.c +1 -1
@@ -353,7 +353,7 @@ static int mcf_fec_can_receive(void *opaque)return s->rx_enabled; } -static void mcf_fec_receive (void *opaque,const uint8_t *buf,int size) +static void mcf_fec_receive (void *opaque,const uint8_t *buf,size_t size) {mcf_fec_state *s = (mcf_fec_state *)opaque;mcf_fec_bd bd;
hw/mipsnet.c +1 -1
@@ -75,7 +75,7 @@ static int mipsnet_can_receive(void *opaque)return !mipsnet_buffer_full (s); } -static void mipsnet_receive (void *opaque,const uint8_t *buf,int size) +static void mipsnet_receive (void *opaque,const uint8_t *buf,size_t size) {MIPSnetState *s = opaque;
hw/musicpal.c +1 -1
@@ -562,7 +562,7 @@ static int eth_can_receive(void *opaque)return 1 ; } -static void eth_receive (void *opaque,const uint8_t *buf,int size) +static void eth_receive (void *opaque,const uint8_t *buf,size_t size) {mv88w8618_eth_state *s = opaque;uint32_t desc_addr;
hw/ne2000.c +1 -1
@@ -224,7 +224,7 @@ static int ne2000_can_receive(void *opaque)#define MIN_BUF_SIZE 60 -static void ne2000_receive (void *opaque,const uint8_t *buf,int size) +static void ne2000_receive (void *opaque,const uint8_t *buf,size_t size) {NE2000State *s = opaque;uint8_t *p;
hw/pcnet.c +1 -1
@@ -1076,7 +1076,7 @@ static int pcnet_can_receive(void *opaque)#define MIN_BUF_SIZE 60 -static void pcnet_receive (void *opaque,const uint8_t *buf,int size) +static void pcnet_receive (void *opaque,const uint8_t *buf,size_t size) {PCNetState *s = opaque;int is_padr =0 , is_bcast =0 , is_ladr =0 ;
hw/qdev.c +5 -5
@@ -233,16 +233,16 @@ void qdev_connect_gpio_out(DeviceState * dev, int n, qemu_irq pin) }VLANClientState *qdev_get_vlan_client (DeviceState *dev, -IOCanRWHandler *fd_can_read, -IOReadHandler *fd_read, -IOReadvHandler *fd_readv, +NetCanReceive *can_receive, +NetReceive *receive, +NetReceiveIOV *receive_iov,NetCleanup *cleanup,void *opaque) {NICInfo *nd = dev->nd;assert (nd); -return qemu_new_vlan_client (nd->vlan, nd->model, nd->name, fd_can_read, - fd_read, fd_readv, cleanup, opaque); +return qemu_new_vlan_client (nd->vlan, nd->model, nd->name, can_receive, + receive, receive_iov, cleanup, opaque); }
hw/rtl8139.c +1 -1
@@ -1158,7 +1158,7 @@ static void rtl8139_do_receive(void *opaque, const uint8_t *buf, int size, int d } } -static void rtl8139_receive (void *opaque,const uint8_t *buf,int size) +static void rtl8139_receive (void *opaque,const uint8_t *buf,size_t size) {rtl8139_do_receive (opaque, buf, size,1 ); }
hw/smc91c111.c +1 -1
@@ -602,7 +602,7 @@ static int smc91c111_can_receive(void *opaque)return 1 ; } -static void smc91c111_receive (void *opaque,const uint8_t *buf,int size) +static void smc91c111_receive (void *opaque,const uint8_t *buf,size_t size) {smc91c111_state *s = (smc91c111_state *)opaque;int status;
hw/stellaris_enet.c +1 -1
@@ -78,7 +78,7 @@ static void stellaris_enet_update(stellaris_enet_state *s) }/* TODO: Implement MAC address filtering. */ -static void stellaris_enet_receive (void *opaque,const uint8_t *buf,int size) +static void stellaris_enet_receive (void *opaque,const uint8_t *buf,size_t size) {stellaris_enet_state *s = (stellaris_enet_state *)opaque;int n;
hw/usb-net.c +1 -1
@@ -1369,7 +1369,7 @@ static int usb_net_handle_data(USBDevice *dev, USBPacket *p)return ret; } -static void usbnet_receive (void *opaque,const uint8_t *buf,int size) +static void usbnet_receive (void *opaque,const uint8_t *buf,size_t size) {USBNetState *s = opaque;struct rndis_packet_msg_type *msg;
hw/virtio-net.c +1 -1
@@ -361,7 +361,7 @@ static int receive_filter(VirtIONet *n, const uint8_t *buf, int size)return 0 ; } -static void virtio_net_receive (void *opaque,const uint8_t *buf,int size) +static void virtio_net_receive (void *opaque,const uint8_t *buf,size_t size) {VirtIONet *n = opaque;struct virtio_net_hdr_mrg_rxbuf *mhdr =NULL ;
hw/xen_nic.c +3 -3
@@ -245,7 +245,7 @@ static int net_rx_ok(void *opaque)return 1 ; } -static void net_rx_packet (void *opaque,const uint8_t *buf,int size) +static void net_rx_packet (void *opaque,const uint8_t *buf,size_t size) {struct XenNetDev *netdev = opaque;netif_rx_request_t rxreq;
@@ -264,8 +264,8 @@ static void net_rx_packet(void *opaque, const uint8_t *buf, int size)return ; }if (size >XC_PAGE_SIZE -NET_IP_ALIGN ) { -xen_be_printf (&netdev->xendev,0 ,"packet too big (%d > %ld)" , - size,XC_PAGE_SIZE -NET_IP_ALIGN ); +xen_be_printf (&netdev->xendev,0 ,"packet too big (%lu > %ld)" , + (unsigned long )size,XC_PAGE_SIZE -NET_IP_ALIGN );return ; }
net.c +18 -18
@@ -332,9 +332,9 @@ static char *assign_name(VLANClientState *vc1, const char *model)VLANClientState *qemu_new_vlan_client (VLANState *vlan,const char *model,const char *name, -IOCanRWHandler *fd_can_read, -IOReadHandler *fd_read, -IOReadvHandler *fd_readv, +NetCanReceive *can_receive, +NetReceive *receive, +NetReceiveIOV *receive_iov,NetCleanup *cleanup,void *opaque) {
@@ -345,9 +345,9 @@ VLANClientState *qemu_new_vlan_client(VLANState *vlan, vc->name =strdup (name); else vc->name = assign_name(vc, model); - vc->fd_can_read = fd_can_read; - vc->fd_read = fd_read; - vc->fd_readv = fd_readv; + vc->can_receive = can_receive; + vc->receive = receive; + vc->receive_iov = receive_iov; vc->cleanup = cleanup; vc->opaque = opaque; vc->vlan = vlan;
@@ -401,8 +401,8 @@ int qemu_can_send_packet(VLANClientState *sender)continue ; } -/* no fd_can_read() handler, they can always receive */ -if (!vc->fd_can_read || vc->fd_can_read (vc->opaque)) { +/* no can_receive() handler, they can always receive */ +if (!vc->can_receive || vc->can_receive (vc->opaque)) {return 1 ; } }
@@ -416,7 +416,7 @@ qemu_deliver_packet(VLANClientState *sender, const uint8_t *buf, int size)for (vc = sender->vlan->first_client; vc !=NULL ; vc = vc->next) {if (vc != sender && !vc->link_down) { - vc->fd_read (vc->opaque, buf, size); + vc->receive (vc->opaque, buf, size); } } }
@@ -467,7 +467,7 @@ static ssize_t vc_sendv_compat(VLANClientState *vc, const struct iovec *iov, offset += len; } - vc->fd_read (vc->opaque, buffer, offset); + vc->receive (vc->opaque, buffer, offset);return offset; }
@@ -519,9 +519,9 @@ ssize_t qemu_sendv_packet(VLANClientState *sender, const struct iovec *iov, }if (vc->link_down) { len =calc_iov_length (iov, iovcnt); - }else if (vc->fd_readv) { - len = vc->fd_readv (vc->opaque, iov, iovcnt); - }else if (vc->fd_read) { + }else if (vc->receive_iov) { + len = vc->receive_iov (vc->opaque, iov, iovcnt); + }else if (vc->receive) { len =vc_sendv_compat (vc, iov, iovcnt); } max_len =MAX (max_len, len);
@@ -568,7 +568,7 @@ int slirp_is_inited(void)return slirp_inited; } -static void slirp_receive (void *opaque,const uint8_t *buf,int size) +static void slirp_receive (void *opaque,const uint8_t *buf,size_t size) {#ifdef DEBUG_SLIRP printf ("slirp input:\n" );
@@ -800,7 +800,7 @@ static ssize_t tap_receive_iov(void *opaque, const struct iovec *iov,return len; } -static void tap_receive (void *opaque,const uint8_t *buf,int size) +static void tap_receive (void *opaque,const uint8_t *buf,size_t size) {TAPState *s = opaque;int ret;
@@ -1227,7 +1227,7 @@ typedef struct NetSocketListenState { } NetSocketListenState;/* XXX: we consider we can send the whole packet without blocking */ -static void net_socket_receive (void *opaque,const uint8_t *buf,int size) +static void net_socket_receive (void *opaque,const uint8_t *buf,size_t size) {NetSocketState *s = opaque;uint32_t len;
@@ -1237,7 +1237,7 @@ static void net_socket_receive(void *opaque, const uint8_t *buf, int size)send_all (s->fd, buf, size); } -static void net_socket_receive_dgram (void *opaque,const uint8_t *buf,int size) +static void net_socket_receive_dgram (void *opaque,const uint8_t *buf,size_t size) {NetSocketState *s = opaque; sendto(s->fd, buf, size, 0,
@@ -1678,7 +1678,7 @@ struct pcap_sf_pkthdr {uint32_t len; }; -static void dump_receive (void *opaque,const uint8_t *buf,int size) +static void dump_receive (void *opaque,const uint8_t *buf,size_t size) {DumpState *s = opaque;struct pcap_sf_pkthdr hdr;
net.h +12 -11
@@ -5,19 +5,20 @@/* VLANs support */ -typedef ssize_t (IOReadvHandler )(void *,const struct iovec *,int ); -typedef struct VLANClientState VLANClientState ; +typedef int (NetCanReceive )(void *); +typedef void (NetReceive )(void *,const uint8_t *,size_t ); +typedef ssize_t (NetReceiveIOV )(void *,const struct iovec *,int );typedef void (NetCleanup ) (VLANClientState *);typedef void (LinkStatusChanged )(VLANClientState *);struct VLANClientState { -IOReadHandler *fd_read; -IOReadvHandler *fd_readv; +NetReceive *receive; +NetReceiveIOV *receive_iov;/* Packets may still be sent if this returns zero. It's used to rate-limit the slirp code. */ -IOCanRWHandler *fd_can_read; +NetCanReceive *can_receive;NetCleanup *cleanup;LinkStatusChanged *link_status_changed;int link_down;
@@ -51,9 +52,9 @@ VLANState *qemu_find_vlan(int id);VLANClientState *qemu_new_vlan_client (VLANState *vlan,const char *model,const char *name, -IOCanRWHandler *fd_can_read, -IOReadHandler *fd_read, -IOReadvHandler *fd_readv, +NetCanReceive *can_receive, +NetReceive *receive, +NetReceiveIOV *receive_iov,NetCleanup *cleanup,void *opaque);void qemu_del_vlan_client (VLANClientState *vc);
@@ -130,9 +131,9 @@ void net_host_device_remove(Monitor *mon, int vlan_id, const char *device);void qdev_get_macaddr (DeviceState *dev,uint8_t *macaddr);VLANClientState *qdev_get_vlan_client (DeviceState *dev, -IOCanRWHandler *fd_can_read, -IOReadHandler *fd_read, -IOReadvHandler *fd_readv, +NetCanReceive *can_receive, +NetReceive *receive, +NetReceiveIOV *receive_iov,NetCleanup *cleanup,void *opaque);
savevm.c +1 -1
@@ -129,7 +129,7 @@ void qemu_announce_self(void) vlan = nd_table[i].vlan;for (vc = vlan->first_client; vc !=NULL ; vc = vc->next) {for (j=0 ; j <SELF_ANNOUNCE_ROUNDS ; j++) - vc->fd_read (vc->opaque, buf, len); + vc->receive (vc->opaque, buf, len); } } }
tap-win32.c +1 -1
@@ -650,7 +650,7 @@ static void tap_cleanup(VLANClientState *vc)qemu_free (s); } -static void tap_receive (void *opaque,const uint8_t *buf,int size) +static void tap_receive (void *opaque,const uint8_t *buf,size_t size) {TAPState *s = opaque;
Signature · 1 line
1.6.0.6
PATCH 7/13 net: pass VLANClientState* as first arg to receive handlers
Give static type checking a chance to catch errors.
--- hw/dp8393x.c | 12 ++++++------ hw/e1000.c | 8 ++++---- hw/eepro100.c | 8 ++++---- hw/etraxfs_eth.c | 6 +++--- hw/mcf_fec.c | 8 ++++---- hw/mipsnet.c | 10 +++++----- hw/musicpal.c | 6 +++--- hw/ne2000.c | 8 ++++---- hw/pcnet.c | 10 +++++----- hw/rtl8139.c | 14 +++++++------- hw/smc91c111.c | 8 ++++---- hw/stellaris_enet.c | 12 ++++++------ hw/usb-net.c | 8 ++++---- hw/virtio-net.c | 8 ++++---- hw/xen_nic.c | 8 ++++---- net.c | 36 ++++++++++++++++++------------------ net.h | 6 +++--- savevm.c | 2 +- tap-win32.c | 2 +- 19 files changed, 90 insertions(+), 90 deletions(-)
hw/dp8393x.c +6 -6
@@ -407,9 +407,9 @@ static void do_transmit_packets(dp8393xState *s)if (s->regs[SONIC_RCR ] & (SONIC_RCR_LB1 |SONIC_RCR_LB0 )) {/* Loopback */ s->regs[SONIC_TCR ] |=SONIC_TCR_CRSL ; -if (s->vc->fd_can_read (s)) { +if (s->vc->can_receive (s->vc)) { s->loopback_packet =1 ; - s->vc->receive (s, s->tx_buffer, tx_len); + s->vc->receive (s->vc, s->tx_buffer, tx_len); } }else {/* Transmit packet */
@@ -676,9 +676,9 @@ static CPUWriteMemoryFunc *dp8393x_write[3] = { dp8393x_writel, }; -static int nic_can_receive (void *opaque) +static int nic_can_receive (VLANClientState *vc) { -dp8393xState *s = opaque; +dp8393xState *s = vc->opaque;if (!(s->regs[SONIC_CR ] &SONIC_CR_RXEN ))return 0 ;
@@ -725,10 +725,10 @@ static int receive_filter(dp8393xState *s, const uint8_t * buf, int size)return -1 ; } -static void nic_receive (void *opaque,const uint8_t * buf,size_t size) +static void nic_receive (VLANClientState *vc,const uint8_t * buf,size_t size) {uint16_t data[10 ]; -dp8393xState *s = opaque; +dp8393xState *s = vc->opaque;int packet_type;uint32_t available, address;int width, rx_len = size;
hw/e1000.c +4 -4
@@ -585,17 +585,17 @@ e1000_set_link_status(VLANClientState *vc) }static int -e1000_can_receive (void *opaque) +e1000_can_receive (VLANClientState *vc) { -E1000State *s = opaque; +E1000State *s = vc->opaque;return (s->mac_reg[RCTL ] &E1000_RCTL_EN ); }static void -e1000_receive (void *opaque,const uint8_t *buf,size_t size) +e1000_receive (VLANClientState *vc,const uint8_t *buf,size_t size) { -E1000State *s = opaque; +E1000State *s = vc->opaque;struct e1000_rx_desc desc;target_phys_addr_t base;unsigned int n, rdt;
hw/eepro100.c +4 -4
@@ -1433,21 +1433,21 @@ static void pci_mmio_map(PCIDevice * pci_dev, int region_num, } } -static int nic_can_receive (void *opaque) +static int nic_can_receive (VLANClientState *vc) { -EEPRO100State *s = opaque; +EEPRO100State *s = vc->opaque;logout ("%p\n" , s);return get_ru_state (s) == ru_ready;//~ return !eepro100_buffer_full(s); } -static void nic_receive (void *opaque,const uint8_t * buf,size_t size) +static void nic_receive (VLANClientState *vc,const uint8_t * buf,size_t size) {/* TODO: * - Magic packets should set bit 30 in power management driver register. * - Interesting packets should set bit 29 in power management driver register. */ -EEPRO100State *s = opaque; +EEPRO100State *s = vc->opaque;uint16_t rfd_status =0xa000 ;static const uint8_t broadcast_macaddr[6 ] = {0xff ,0xff ,0xff ,0xff ,0xff ,0xff };
hw/etraxfs_eth.c +3 -3
@@ -496,15 +496,15 @@ static int eth_match_groupaddr(struct fs_eth *eth, const unsigned char *sa)return match; } -static int eth_can_receive (void *opaque) +static int eth_can_receive (VLANClientState *vc) {return 1 ; } -static void eth_receive (void *opaque,const uint8_t *buf,size_t size) +static void eth_receive (VLANClientState *vc,const uint8_t *buf,size_t size) {unsigned char sa_bcast[6 ] = {0xff ,0xff ,0xff ,0xff ,0xff ,0xff }; -struct fs_eth *eth = opaque; +struct fs_eth *eth = vc->opaque;int use_ma0 = eth->regs[RW_REC_CTRL ] &1 ;int use_ma1 = eth->regs[RW_REC_CTRL ] &2 ;int r_bcast = eth->regs[RW_REC_CTRL ] &8 ;
hw/mcf_fec.c +4 -4
@@ -347,15 +347,15 @@ static void mcf_fec_write(void *opaque, target_phys_addr_t addr, uint32_t value)mcf_fec_update (s); } -static int mcf_fec_can_receive (void *opaque) +static int mcf_fec_can_receive (VLANClientState *vc) { -mcf_fec_state *s = (mcf_fec_state *)opaque; +mcf_fec_state *s = vc->opaque;return s->rx_enabled; } -static void mcf_fec_receive (void *opaque,const uint8_t *buf,size_t size) +static void mcf_fec_receive (VLANClientState *vc,const uint8_t *buf,size_t size) { -mcf_fec_state *s = (mcf_fec_state *)opaque; +mcf_fec_state *s = vc->opaque;mcf_fec_bd bd;uint32_t flags =0 ;uint32_t addr;
hw/mipsnet.c +5 -5
@@ -66,23 +66,23 @@ static int mipsnet_buffer_full(MIPSnetState *s)return 0 ; } -static int mipsnet_can_receive (void *opaque) +static int mipsnet_can_receive (VLANClientState *vc) { -MIPSnetState *s = opaque; +MIPSnetState *s = vc->opaque;if (s->busy)return 0 ;return !mipsnet_buffer_full (s); } -static void mipsnet_receive (void *opaque,const uint8_t *buf,size_t size) +static void mipsnet_receive (VLANClientState *vc,const uint8_t *buf,size_t size) { -MIPSnetState *s = opaque; +MIPSnetState *s = vc->opaque;#ifdef DEBUG_MIPSNET_RECEIVE printf ("mipsnet: receiving len=%d\n" , size);#endif -if (!mipsnet_can_receive (opaque)) +if (!mipsnet_can_receive (vc))return ; s->busy =1 ;
hw/musicpal.c +3 -3
@@ -557,14 +557,14 @@ static void eth_rx_desc_get(uint32_t addr, mv88w8618_rx_desc *desc)le32_to_cpus (&desc->next); } -static int eth_can_receive (void *opaque) +static int eth_can_receive (VLANClientState *vc) {return 1 ; } -static void eth_receive (void *opaque,const uint8_t *buf,size_t size) +static void eth_receive (VLANClientState *vc,const uint8_t *buf,size_t size) { -mv88w8618_eth_state *s = opaque; +mv88w8618_eth_state *s = vc->opaque;uint32_t desc_addr;mv88w8618_rx_desc desc;int i;
hw/ne2000.c +4 -4
@@ -213,9 +213,9 @@ static int ne2000_buffer_full(NE2000State *s)return 0 ; } -static int ne2000_can_receive (void *opaque) +static int ne2000_can_receive (VLANClientState *vc) { -NE2000State *s = opaque; +NE2000State *s = vc->opaque;if (s->cmd &E8390_STOP )return 1 ;
@@ -224,9 +224,9 @@ static int ne2000_can_receive(void *opaque)#define MIN_BUF_SIZE 60 -static void ne2000_receive (void *opaque,const uint8_t *buf,size_t size) +static void ne2000_receive (VLANClientState *vc,const uint8_t *buf,size_t size) { -NE2000State *s = opaque; +NE2000State *s = vc->opaque;uint8_t *p;unsigned int total_len, next, avail, len, index, mcast_idx;uint8_t buf1[60 ];
hw/pcnet.c +5 -5
@@ -1062,9 +1062,9 @@ static int pcnet_tdte_poll(PCNetState *s)return !!(CSR_CXST (s) &0x8000 ); } -static int pcnet_can_receive (void *opaque) +static int pcnet_can_receive (VLANClientState *vc) { -PCNetState *s = opaque; +PCNetState *s = vc->opaque;if (CSR_STOP (s) ||CSR_SPND (s))return 0 ;
@@ -1076,9 +1076,9 @@ static int pcnet_can_receive(void *opaque)#define MIN_BUF_SIZE 60 -static void pcnet_receive (void *opaque,const uint8_t *buf,size_t size) +static void pcnet_receive (VLANClientState *vc,const uint8_t *buf,size_t size) { -PCNetState *s = opaque; +PCNetState *s = vc->opaque;int is_padr =0 , is_bcast =0 , is_ladr =0 ;uint8_t buf1[60 ];int remaining;
@@ -1302,7 +1302,7 @@ static void pcnet_transmit(PCNetState *s)if (BCR_SWSTYLE (s) ==1 ) add_crc = !GET_FIELD (tmd.status,TMDS ,NOFCS ); s->looptest = add_crc ?PCNET_LOOPTEST_CRC :PCNET_LOOPTEST_NOCRC ; -pcnet_receive (s, s->buffer, s->xmit_pos); +pcnet_receive (s->vc, s->buffer, s->xmit_pos); s->looptest =0 ; }else if (s->vc)
hw/rtl8139.c +7 -7
@@ -790,9 +790,9 @@ static inline target_phys_addr_t rtl8139_addr64(uint32_t low, uint32_t high)#endif } -static int rtl8139_can_receive (void *opaque) +static int rtl8139_can_receive (VLANClientState *vc) { -RTL8139State *s = opaque; +RTL8139State *s = vc->opaque;int avail;/* Receive (drop) packets if card is disabled. */
@@ -812,9 +812,9 @@ static int rtl8139_can_receive(void *opaque) } } -static void rtl8139_do_receive (void *opaque,const uint8_t *buf,int size,int do_interrupt) +static void rtl8139_do_receive (VLANClientState *vc,const uint8_t *buf,int size,int do_interrupt) { -RTL8139State *s = opaque; +RTL8139State *s = vc->opaque;uint32_t packet_header =0 ;
@@ -1158,9 +1158,9 @@ static void rtl8139_do_receive(void *opaque, const uint8_t *buf, int size, int d } } -static void rtl8139_receive (void *opaque,const uint8_t *buf,size_t size) +static void rtl8139_receive (VLANClientState *vc,const uint8_t *buf,size_t size) { -rtl8139_do_receive (opaque, buf, size,1 ); +rtl8139_do_receive (vc, buf, size,1 ); }static void rtl8139_reset_rxring (RTL8139State *s,uint32_t bufferSize)
@@ -1757,7 +1757,7 @@ static void rtl8139_transfer_frame(RTL8139State *s, const uint8_t *buf, int sizeif (TxLoopBack == (s->TxConfig & TxLoopBack)) {DEBUG_PRINT (("RTL8139: +++ transmit loopback mode\n" )); -rtl8139_do_receive (s, buf, size, do_interrupt); +rtl8139_do_receive (s->vc, buf, size, do_interrupt); }else {
hw/smc91c111.c +4 -4
@@ -591,9 +591,9 @@ static uint32_t smc91c111_readl(void *opaque, target_phys_addr_t offset)return val; } -static int smc91c111_can_receive (void *opaque) +static int smc91c111_can_receive (VLANClientState *vc) { -smc91c111_state *s = (smc91c111_state *)opaque; +smc91c111_state *s = vc->opaque;if ((s->rcr &RCR_RXEN ) ==0 || (s->rcr &RCR_SOFT_RST ))return 1 ;
@@ -602,9 +602,9 @@ static int smc91c111_can_receive(void *opaque)return 1 ; } -static void smc91c111_receive (void *opaque,const uint8_t *buf,size_t size) +static void smc91c111_receive (VLANClientState *vc,const uint8_t *buf,size_t size) { -smc91c111_state *s = (smc91c111_state *)opaque; +smc91c111_state *s = vc->opaque;int status;int packetsize;uint32_t crc;
hw/stellaris_enet.c +6 -6
@@ -78,9 +78,9 @@ static void stellaris_enet_update(stellaris_enet_state *s) }/* TODO: Implement MAC address filtering. */ -static void stellaris_enet_receive (void *opaque,const uint8_t *buf,size_t size) +static void stellaris_enet_receive (VLANClientState *vc,const uint8_t *buf,size_t size) { -stellaris_enet_state *s = (stellaris_enet_state *)opaque; +stellaris_enet_state *s = vc->opaque;int n;uint8_t *p;uint32_t crc;
@@ -118,9 +118,9 @@ static void stellaris_enet_receive(void *opaque, const uint8_t *buf, size_t sizestellaris_enet_update (s); } -static int stellaris_enet_can_receive (void *opaque) +static int stellaris_enet_can_receive (VLANClientState *vc) { -stellaris_enet_state *s = (stellaris_enet_state *)opaque; +stellaris_enet_state *s = vc->opaque;if ((s->rctl &SE_RCTL_RXEN ) ==0 )return 1 ;
@@ -128,9 +128,9 @@ static int stellaris_enet_can_receive(void *opaque)return (s->np <31 ); } -static uint32_t stellaris_enet_read (void *opaque,target_phys_addr_t offset) +static uint32_t stellaris_enet_read (VLANClientState *vc,target_phys_addr_t offset) { -stellaris_enet_state *s = (stellaris_enet_state *)opaque; +stellaris_enet_state *s = vc->opaque;uint32_t val;switch (offset) {
hw/usb-net.c +4 -4
@@ -1369,9 +1369,9 @@ static int usb_net_handle_data(USBDevice *dev, USBPacket *p)return ret; } -static void usbnet_receive (void *opaque,const uint8_t *buf,size_t size) +static void usbnet_receive (VLANClientState *vc,const uint8_t *buf,size_t size) { -USBNetState *s = opaque; +USBNetState *s = vc->opaque;struct rndis_packet_msg_type *msg;if (s->rndis) {
@@ -1405,9 +1405,9 @@ static void usbnet_receive(void *opaque, const uint8_t *buf, size_t size) s->in_ptr =0 ; } -static int usbnet_can_receive (void *opaque) +static int usbnet_can_receive (VLANClientState *vc) { -USBNetState *s = opaque; +USBNetState *s = vc->opaque;if (s->rndis && !s->rndis_state ==RNDIS_DATA_INITIALIZED )return 1 ;
hw/virtio-net.c +4 -4
@@ -288,9 +288,9 @@ static int do_virtio_net_can_receive(VirtIONet *n, int bufsize)return 1 ; } -static int virtio_net_can_receive (void *opaque) +static int virtio_net_can_receive (VLANClientState *vc) { -VirtIONet *n = opaque; +VirtIONet *n = vc->opaque;return do_virtio_net_can_receive (n,VIRTIO_NET_MAX_BUFSIZE ); }
@@ -361,9 +361,9 @@ static int receive_filter(VirtIONet *n, const uint8_t *buf, int size)return 0 ; } -static void virtio_net_receive (void *opaque,const uint8_t *buf,size_t size) +static void virtio_net_receive (VLANClientState *vc,const uint8_t *buf,size_t size) { -VirtIONet *n = opaque; +VirtIONet *n = vc->opaque;struct virtio_net_hdr_mrg_rxbuf *mhdr =NULL ;size_t hdr_len, offset, i;
hw/xen_nic.c +4 -4
@@ -225,9 +225,9 @@ static void net_rx_response(struct XenNetDev *netdev,#define NET_IP_ALIGN 2 -static int net_rx_ok (void *opaque) +static int net_rx_ok (VLANClientState *vc) { -struct XenNetDev *netdev = opaque; +struct XenNetDev *netdev = vc->opaque;RING_IDX rc, rp;if (netdev->xendev.be_state != XenbusStateConnected)
@@ -245,9 +245,9 @@ static int net_rx_ok(void *opaque)return 1 ; } -static void net_rx_packet (void *opaque,const uint8_t *buf,size_t size) +static void net_rx_packet (VLANClientState *vc,const uint8_t *buf,size_t size) { -struct XenNetDev *netdev = opaque; +struct XenNetDev *netdev = vc->opaque;netif_rx_request_t rxreq;RING_IDX rc, rp;void *page;
net.c +18 -18
@@ -402,7 +402,7 @@ int qemu_can_send_packet(VLANClientState *sender) }/* no can_receive() handler, they can always receive */ -if (!vc->can_receive || vc->can_receive (vc->opaque)) { +if (!vc->can_receive || vc->can_receive (vc)) {return 1 ; } }
@@ -416,7 +416,7 @@ qemu_deliver_packet(VLANClientState *sender, const uint8_t *buf, int size)for (vc = sender->vlan->first_client; vc !=NULL ; vc = vc->next) {if (vc != sender && !vc->link_down) { - vc->receive (vc->opaque, buf, size); + vc->receive (vc, buf, size); } } }
@@ -467,7 +467,7 @@ static ssize_t vc_sendv_compat(VLANClientState *vc, const struct iovec *iov, offset += len; } - vc->receive (vc->opaque, buffer, offset); + vc->receive (vc, buffer, offset);return offset; }
@@ -520,7 +520,7 @@ ssize_t qemu_sendv_packet(VLANClientState *sender, const struct iovec *iov,if (vc->link_down) { len =calc_iov_length (iov, iovcnt); }else if (vc->receive_iov) { - len = vc->receive_iov (vc->opaque, iov, iovcnt); + len = vc->receive_iov (vc, iov, iovcnt); }else if (vc->receive) { len =vc_sendv_compat (vc, iov, iovcnt); }
@@ -568,7 +568,7 @@ int slirp_is_inited(void)return slirp_inited; } -static void slirp_receive (void *opaque,const uint8_t *buf,size_t size) +static void slirp_receive (VLANClientState *vc,const uint8_t *buf,size_t size) {#ifdef DEBUG_SLIRP printf ("slirp input:\n" );
@@ -787,10 +787,10 @@ typedef struct TAPState {static int launch_script (const char *setup_script,const char *ifname,int fd); -static ssize_t tap_receive_iov (void *opaque,const struct iovec *iov, +static ssize_t tap_receive_iov (VLANClientState *vc,const struct iovec *iov,int iovcnt) { -TAPState *s = opaque; +TAPState *s = vc->opaque;ssize_t len; do {
@@ -800,9 +800,9 @@ static ssize_t tap_receive_iov(void *opaque, const struct iovec *iov,return len; } -static void tap_receive (void *opaque,const uint8_t *buf,size_t size) +static void tap_receive (VLANClientState *vc,const uint8_t *buf,size_t size) { -TAPState *s = opaque; +TAPState *s = vc->opaque;int ret;for (;;) { ret =write (s->fd, buf, size);
@@ -1158,9 +1158,9 @@ static void vde_to_qemu(void *opaque) } } -static void vde_from_qemu (void *opaque,const uint8_t *buf,int size) +static void vde_receive (VLANClientState *vc,const uint8_t *buf,int size) { -VDEState *s = opaque; +VDEState *s = vc->opaque;int ret;for (;;) { ret =vde_send (s->vde, (const char *)buf, size,0 );
@@ -1199,7 +1199,7 @@ static int net_vde_init(VLANState *vlan, const char *model,free (s);return -1 ; } - s->vc =qemu_new_vlan_client (vlan, model, name,NULL , vde_from_qemu, + s->vc =qemu_new_vlan_client (vlan, model, name,NULL , vde_receive,NULL , vde_cleanup, s);qemu_set_fd_handler (vde_datafd (s->vde), vde_to_qemu,NULL , s); snprintf(s->vc->info_str, sizeof(s->vc->info_str), "sock=%s,fd=%d",
@@ -1227,9 +1227,9 @@ typedef struct NetSocketListenState { } NetSocketListenState;/* XXX: we consider we can send the whole packet without blocking */ -static void net_socket_receive (void *opaque,const uint8_t *buf,size_t size) +static void net_socket_receive (VLANClientState *vc,const uint8_t *buf,size_t size) { -NetSocketState *s = opaque; +NetSocketState *s = vc->opaque;uint32_t len; len =htonl (size);
@@ -1237,9 +1237,9 @@ static void net_socket_receive(void *opaque, const uint8_t *buf, size_t size)send_all (s->fd, buf, size); } -static void net_socket_receive_dgram (void *opaque,const uint8_t *buf,size_t size) +static void net_socket_receive_dgram (VLANClientState *vc,const uint8_t *buf,size_t size) { -NetSocketState *s = opaque; +NetSocketState *s = vc->opaque;sendto (s->fd, buf, size,0 , (struct sockaddr *)&s->dgram_dst,sizeof (s->dgram_dst)); }
@@ -1678,9 +1678,9 @@ struct pcap_sf_pkthdr {uint32_t len; }; -static void dump_receive (void *opaque,const uint8_t *buf,size_t size) +static void dump_receive (VLANClientState *vc,const uint8_t *buf,size_t size) { -DumpState *s = opaque; +DumpState *s = vc->opaque;struct pcap_sf_pkthdr hdr;int64_t ts;int caplen;
net.h +3 -3
@@ -7,9 +7,9 @@typedef struct VLANClientState VLANClientState ; -typedef int (NetCanReceive )(void *); -typedef void (NetReceive )(void *,const uint8_t *,size_t ); -typedef ssize_t (NetReceiveIOV )(void *,const struct iovec *,int ); +typedef int (NetCanReceive )(VLANClientState *); +typedef void (NetReceive )(VLANClientState *,const uint8_t *,size_t ); +typedef ssize_t (NetReceiveIOV )(VLANClientState *,const struct iovec *,int );typedef void (NetCleanup ) (VLANClientState *);typedef void (LinkStatusChanged )(VLANClientState *);
savevm.c +1 -1
@@ -129,7 +129,7 @@ void qemu_announce_self(void) vlan = nd_table[i].vlan;for (vc = vlan->first_client; vc !=NULL ; vc = vc->next) {for (j=0 ; j <SELF_ANNOUNCE_ROUNDS ; j++) - vc->receive (vc->opaque, buf, len); + vc->receive (vc, buf, len); } } }
tap-win32.c +1 -1
@@ -650,7 +650,7 @@ static void tap_cleanup(VLANClientState *vc)qemu_free (s); } -static void tap_receive (void *opaque,const uint8_t *buf,size_t size) +static void tap_receive (VLANClientState *vc,const uint8_t *buf,size_t size) {TAPState *s = opaque;
Signature · 1 line
1.6.0.6
PATCH 8/13 net: add return value to packet receive handler
This allows us to handle queue full conditions rather than dropping the packet on the floor.
--- hw/dp8393x.c | 8 ++++-- hw/e1000.c | 14 +++++++----- hw/eepro100.c | 15 +++++++------ hw/etraxfs_eth.c | 8 ++++-- hw/mcf_fec.c | 3 +- hw/mipsnet.c | 6 +++- hw/musicpal.c | 5 ++- hw/ne2000.c | 15 ++++++++----- hw/pcnet.c | 7 ++++- hw/rtl8139.c | 29 +++++++++++++++------------ hw/smc91c111.c | 10 +++++--- hw/stellaris_enet.c | 8 ++++-- hw/usb-net.c | 9 ++++--- hw/virtio-net.c | 10 +++++--- hw/xen_nic.c | 12 ++++++---- net.c | 54 ++++++++++++++++++++++++++------------------------ net.h | 2 +- tap-win32.c | 6 ++-- 18 files changed, 126 insertions(+), 95 deletions(-)
hw/dp8393x.c +5 -3
@@ -725,7 +725,7 @@ static int receive_filter(dp8393xState *s, const uint8_t * buf, int size)return -1 ; } -static void nic_receive (VLANClientState *vc,const uint8_t * buf,size_t size) +static ssize_t nic_receive (VLANClientState *vc,const uint8_t * buf,size_t size) {uint16_t data[10 ];dp8393xState *s = vc->opaque;
@@ -742,7 +742,7 @@ static void nic_receive(VLANClientState *vc, const uint8_t * buf, size_t size) packet_type =receive_filter (s, buf, size);if (packet_type <0 ) {DPRINTF ("packet not for netcard\n" ); -return ; +return -1 ; }/* XXX: Check byte ordering */
@@ -755,7 +755,7 @@ static void nic_receive(VLANClientState *vc, const uint8_t * buf, size_t size) s->memory_rw (s->mem_opaque, address, (uint8_t *)data, size,0 );if (data[0 * width] &0x1 ) {/* Still EOL ; stop reception */ -return ; +return -1 ; }else { s->regs[SONIC_CRDA ] = s->regs[SONIC_LLFA ]; }
@@ -833,6 +833,8 @@ static void nic_receive(VLANClientState *vc, const uint8_t * buf, size_t size)/* Done */ dp8393x_update_irq (s); + +return size; }static void nic_reset (void *opaque)
hw/e1000.c +8 -6
@@ -592,7 +592,7 @@ e1000_can_receive(VLANClientState *vc)return (s->mac_reg[RCTL ] &E1000_RCTL_EN ); } -static void +static ssize_t e1000_receive (VLANClientState *vc,const uint8_t *buf,size_t size) {E1000State *s = vc->opaque;
@@ -604,16 +604,16 @@ e1000_receive(VLANClientState *vc, const uint8_t *buf, size_t size)uint8_t vlan_status =0 , vlan_offset =0 ;if (!(s->mac_reg[RCTL ] &E1000_RCTL_EN )) -return ; +return -1 ;if (size > s->rxbuf_size) {DBGOUT (RX ,"packet too large for buffers (%lu > %d)\n" , (unsigned long )size, s->rxbuf_size); -return ; +return -1 ; }if (!receive_filter (s, buf, size)) -return ; +return size;if (vlan_enabled (s) &&is_vlan_packet (s, buf)) { vlan_special =cpu_to_le16 (be16_to_cpup ((uint16_t *)(buf +14 )));
@@ -628,7 +628,7 @@ e1000_receive(VLANClientState *vc, const uint8_t *buf, size_t size) do { if (s->mac_reg[RDH] == s->mac_reg[RDT] && s->check_rxov) { set_ics(s, 0, E1000_ICS_RXO); - return; + return -1; } base = ((uint64_t)s->mac_reg[RDBAH] << 32) + s->mac_reg[RDBAL] + sizeof(desc) * s->mac_reg[RDH];
@@ -652,7 +652,7 @@ e1000_receive(VLANClientState *vc, const uint8_t *buf, size_t size)DBGOUT (RXERR ,"RDH wraparound @%x, RDT %x, RDLEN %x\n" , rdh_start, s->mac_reg[RDT ], s->mac_reg[RDLEN ]);set_ics (s,0 ,E1000_ICS_RXO ); -return ; +return -1 ; } }while (desc.buffer_addr ==0 );
@@ -670,6 +670,8 @@ e1000_receive(VLANClientState *vc, const uint8_t *buf, size_t size) n |=E1000_ICS_RXDMT0 ;set_ics (s,0 , n); + +return size; } static uint32_t
hw/eepro100.c +8 -7
@@ -1441,7 +1441,7 @@ static int nic_can_receive(VLANClientState *vc)//~ return !eepro100_buffer_full(s); } -static void nic_receive (VLANClientState *vc,const uint8_t * buf,size_t size) +static ssize_t nic_receive (VLANClientState *vc,const uint8_t * buf,size_t size) {/* TODO: * - Magic packets should set bit 30 in power management driver register.
@@ -1458,18 +1458,18 @@ static void nic_receive(VLANClientState *vc, const uint8_t * buf, size_t size)if (s->configuration[8 ] &0x80 ) {/* CSMA is disabled. */ logout ("%p received while CSMA is disabled\n" , s); -return ; +return -1 ; }else if (size <64 && (s->configuration[7 ] &1 )) {/* Short frame and configuration byte 7/0 (discard short receive) set: * Short frame is discarded */ logout ("%p received short frame (%d byte)\n" , s, size); s->statistics.rx_short_frame_errors++; -//~ return; +//~ return -1; }else if ((size >MAX_ETH_FRAME_SIZE +4 ) && !(s->configuration[18 ] &8 )) {/* Long frame and configuration byte 18/3 (long receive ok) not set: * Long frames are discarded. */ logout ("%p received long frame (%d byte), ignored\n" , s, size); -return ; +return -1 ; }else if (memcmp (buf, s->macaddr,6 ) ==0 ) {// !!! /* Frame matches individual address. */ /* TODO: check configuration byte 15/4 (ignore U/L). */
@@ -1485,7 +1485,7 @@ static void nic_receive(VLANClientState *vc, const uint8_t * buf, size_t size)assert (!(s->configuration[21 ] &BIT (3 )));int mcast_idx =compute_mcast_idx (buf);if (!(s->mult[mcast_idx >>3 ] & (1 << (mcast_idx &7 )))) { -return ; +return size; } rfd_status |=0x0002 ; }else if (s->configuration[15 ] &1 ) {
@@ -1495,7 +1495,7 @@ static void nic_receive(VLANClientState *vc, const uint8_t * buf, size_t size) }else {logout ("%p received frame, ignored, len=%d,%s\n" , s, size,nic_dump (buf, size)); -return ; +return size; }if (get_ru_state (s) != ru_ready) {
@@ -1503,7 +1503,7 @@ static void nic_receive(VLANClientState *vc, const uint8_t * buf, size_t size)logout ("no ressources, state=%u\n" ,get_ru_state (s)); s->statistics.rx_resource_errors++;//~ assert(!"no ressources"); -return ; +return -1 ; }//~ !!! //~ $3 = {status = 0x0, command = 0xc000, link = 0x2d220, rx_buf_addr = 0x207dc, count = 0x0, size = 0x5f8, packet = {0x0 <repeats 1518 times>}}
@@ -1540,6 +1540,7 @@ static void nic_receive(VLANClientState *vc, const uint8_t * buf, size_t size)/* S bit is set. */ set_ru_state (s, ru_suspended); } +return size; }static int nic_load (QEMUFile * f,void *opaque,int version_id)
hw/etraxfs_eth.c +5 -3
@@ -501,7 +501,7 @@ static int eth_can_receive(VLANClientState *vc)return 1 ; } -static void eth_receive (VLANClientState *vc,const uint8_t *buf,size_t size) +static ssize_t eth_receive (VLANClientState *vc,const uint8_t *buf,size_t size) {unsigned char sa_bcast[6 ] = {0xff ,0xff ,0xff ,0xff ,0xff ,0xff };struct fs_eth *eth = vc->opaque;
@@ -510,7 +510,7 @@ static void eth_receive(VLANClientState *vc, const uint8_t *buf, size_t size)int r_bcast = eth->regs[RW_REC_CTRL ] &8 ;if (size <12 ) -return ; +return -1 ; D(printf("%x.%x.%x.%x.%x.%x ma=%d %d bc=%d\n", buf[0], buf[1], buf[2], buf[3], buf[4], buf[5],
@@ -521,10 +521,12 @@ static void eth_receive(VLANClientState *vc, const uint8_t *buf, size_t size) && (!use_ma1 ||memcmp (buf, eth->macaddr[1 ],6 )) && (!r_bcast ||memcmp (buf, sa_bcast,6 )) && !eth_match_groupaddr(eth, buf)) - return; + return size;/* FIXME: Find another way to pass on the fake csum. */ etraxfs_dmac_input (eth->dma_in, (void *)buf, size +4 ,1 ); + +return size; }static int eth_tx_push (void *opaque,unsigned char *buf,int len)
hw/mcf_fec.c +2 -1
@@ -353,7 +353,7 @@ static int mcf_fec_can_receive(VLANClientState *vc)return s->rx_enabled; } -static void mcf_fec_receive (VLANClientState *vc,const uint8_t *buf,size_t size) +static ssize_t mcf_fec_receive (VLANClientState *vc,const uint8_t *buf,size_t size) {mcf_fec_state *s = vc->opaque;mcf_fec_bd bd;
@@ -426,6 +426,7 @@ static void mcf_fec_receive(VLANClientState *vc, const uint8_t *buf, size_t size s->rx_descriptor = addr;mcf_fec_enable_rx (s);mcf_fec_update (s); +return size; }static CPUReadMemoryFunc *mcf_fec_readfn[] = {
hw/mipsnet.c +4 -2
@@ -75,7 +75,7 @@ static int mipsnet_can_receive(VLANClientState *vc)return !mipsnet_buffer_full (s); } -static void mipsnet_receive (VLANClientState *vc,const uint8_t *buf,size_t size) +static ssize_t mipsnet_receive (VLANClientState *vc,const uint8_t *buf,size_t size) {MIPSnetState *s = vc->opaque;
@@ -83,7 +83,7 @@ static void mipsnet_receive(VLANClientState *vc, const uint8_t *buf, size_t sizeprintf ("mipsnet: receiving len=%d\n" , size);#endif if (!mipsnet_can_receive (vc)) -return ; +return -1 ; s->busy =1 ;
@@ -98,6 +98,8 @@ static void mipsnet_receive(VLANClientState *vc, const uint8_t *buf, size_t size/* Now we can signal we have received something. */ s->intctl |=MIPSNET_INTCTL_RXDONE ;mipsnet_update_irq (s); + +return size; }static uint32_t mipsnet_ioport_read (void *opaque,uint32_t addr)
hw/musicpal.c +3 -2
@@ -562,7 +562,7 @@ static int eth_can_receive(VLANClientState *vc)return 1 ; } -static void eth_receive (VLANClientState *vc,const uint8_t *buf,size_t size) +static ssize_t eth_receive (VLANClientState *vc,const uint8_t *buf,size_t size) {mv88w8618_eth_state *s = vc->opaque;uint32_t desc_addr;
@@ -586,11 +586,12 @@ static void eth_receive(VLANClientState *vc, const uint8_t *buf, size_t size)if (s->icr & s->imr)qemu_irq_raise (s->irq);eth_rx_desc_put (desc_addr, &desc); -return ; +return size; } desc_addr = desc.next; }while (desc_addr != s->rx_queue[i]); } +return size; }static void eth_tx_desc_put (uint32_t addr,mv88w8618_tx_desc *desc)
hw/ne2000.c +9 -6
@@ -224,9 +224,10 @@ static int ne2000_can_receive(VLANClientState *vc)#define MIN_BUF_SIZE 60 -static void ne2000_receive (VLANClientState *vc,const uint8_t *buf,size_t size) +static ssize_t ne2000_receive (VLANClientState *vc,const uint8_t *buf,size_t size_) {NE2000State *s = vc->opaque; +int size = size_;uint8_t *p;unsigned int total_len, next, avail, len, index, mcast_idx;uint8_t buf1[60 ];
@@ -238,7 +239,7 @@ static void ne2000_receive(VLANClientState *vc, const uint8_t *buf, size_t size)#endif if (s->cmd &E8390_STOP ||ne2000_buffer_full (s)) -return ; +return -1 ;/* XXX: check this */ if (s->rxcr &0x10 ) {
@@ -247,14 +248,14 @@ static void ne2000_receive(VLANClientState *vc, const uint8_t *buf, size_t size)if (!memcmp (buf, broadcast_macaddr,6 )) {/* broadcast address */ if (!(s->rxcr &0x04 )) -return ; +return size; }else if (buf[0 ] &0x01 ) {/* multicast */ if (!(s->rxcr &0x08 )) -return ; +return size; mcast_idx =compute_mcast_idx (buf);if (!(s->mult[mcast_idx >>3 ] & (1 << (mcast_idx &7 )))) -return ; +return size; } else if (s->mem[0] == buf[0] && s->mem[2] == buf[1] && s->mem[4] == buf[2] &&
@@ -263,7 +264,7 @@ static void ne2000_receive(VLANClientState *vc, const uint8_t *buf, size_t size) s->mem[10] == buf[5]) { /* match */ } else { -return ; +return size; } }
@@ -316,6 +317,8 @@ static void ne2000_receive(VLANClientState *vc, const uint8_t *buf, size_t size)/* now we can signal we have received something */ s->isr |=ENISR_RX ;ne2000_update_irq (s); + +return size_; }static void ne2000_ioport_write (void *opaque,uint32_t addr,uint32_t val)
hw/pcnet.c +5 -2
@@ -1076,16 +1076,17 @@ static int pcnet_can_receive(VLANClientState *vc)#define MIN_BUF_SIZE 60 -static void pcnet_receive (VLANClientState *vc,const uint8_t *buf,size_t size) +static ssize_t pcnet_receive (VLANClientState *vc,const uint8_t *buf,size_t size_) {PCNetState *s = vc->opaque;int is_padr =0 , is_bcast =0 , is_ladr =0 ;uint8_t buf1[60 ];int remaining;int crc_err =0 ; +int size = size_;if (CSR_DRX (s) ||CSR_STOP (s) ||CSR_SPND (s) || !size) -return ; +return -1 ;#ifdef PCNET_DEBUG printf ("pcnet_receive size=%d\n" , size);
@@ -1252,6 +1253,8 @@ static void pcnet_receive(VLANClientState *vc, const uint8_t *buf, size_t size)pcnet_poll (s);pcnet_update_irq (s); + +return size_; }static void pcnet_transmit (PCNetState *s)
hw/rtl8139.c +16 -13
@@ -812,9 +812,10 @@ static int rtl8139_can_receive(VLANClientState *vc) } } -static void rtl8139_do_receive (VLANClientState *vc,const uint8_t *buf,int size,int do_interrupt) +static ssize_t rtl8139_do_receive (VLANClientState *vc,const uint8_t *buf,size_t size_,int do_interrupt) {RTL8139State *s = vc->opaque; +int size = size_;uint32_t packet_header =0 ;
@@ -828,7 +829,7 @@ static void rtl8139_do_receive(VLANClientState *vc, const uint8_t *buf, int sizeif (!s->clock_enabled) {DEBUG_PRINT (("RTL8139: stopped ==========================\n" )); -return ; +return -1 ; }/* first check if receiver is enabled */
@@ -836,7 +837,7 @@ static void rtl8139_do_receive(VLANClientState *vc, const uint8_t *buf, int sizeif (!rtl8139_receiver_enabled (s)) {DEBUG_PRINT (("RTL8139: receiver disabled ================\n" )); -return ; +return -1 ; }/* XXX: check this */
@@ -854,7 +855,7 @@ static void rtl8139_do_receive(VLANClientState *vc, const uint8_t *buf, int size/* update tally counter */ ++s->tally_counters.RxERR; -return ; +return size; } packet_header |= RxBroadcast;
@@ -873,7 +874,7 @@ static void rtl8139_do_receive(VLANClientState *vc, const uint8_t *buf, int size/* update tally counter */ ++s->tally_counters.RxERR; -return ; +return size; }int mcast_idx =compute_mcast_idx (buf);
@@ -885,7 +886,7 @@ static void rtl8139_do_receive(VLANClientState *vc, const uint8_t *buf, int size/* update tally counter */ ++s->tally_counters.RxERR; -return ; +return size; } packet_header |= RxMulticast;
@@ -909,7 +910,7 @@ static void rtl8139_do_receive(VLANClientState *vc, const uint8_t *buf, int size/* update tally counter */ ++s->tally_counters.RxERR; -return ; +return size; } packet_header |= RxPhysical;
@@ -926,7 +927,7 @@ static void rtl8139_do_receive(VLANClientState *vc, const uint8_t *buf, int size/* update tally counter */ ++s->tally_counters.RxERR; -return ; +return size; } }
@@ -993,7 +994,7 @@ static void rtl8139_do_receive(VLANClientState *vc, const uint8_t *buf, int size ++s->tally_counters.MissPkt;rtl8139_update_irq (s); -return ; +return size_; }uint32_t rx_space = rxdw0 &CP_RX_BUFFER_SIZE_MASK ;
@@ -1013,7 +1014,7 @@ static void rtl8139_do_receive(VLANClientState *vc, const uint8_t *buf, int size ++s->tally_counters.MissPkt;rtl8139_update_irq (s); -return ; +return size_; }target_phys_addr_t rx_addr =rtl8139_addr64 (rxbufLO, rxbufHI);
@@ -1118,7 +1119,7 @@ static void rtl8139_do_receive(VLANClientState *vc, const uint8_t *buf, int size s->IntrStatus |= RxOverflow; ++s->RxMissed;rtl8139_update_irq (s); -return ; +return size_; } packet_header |= RxStatusOK;
@@ -1156,11 +1157,13 @@ static void rtl8139_do_receive(VLANClientState *vc, const uint8_t *buf, int size {rtl8139_update_irq (s); } + +return size_; } -static void rtl8139_receive (VLANClientState *vc,const uint8_t *buf,size_t size) +static ssize_t rtl8139_receive (VLANClientState *vc,const uint8_t *buf,size_t size) { -rtl8139_do_receive (vc, buf, size,1 ); +return rtl8139_do_receive (vc, buf, size,1 ); }static void rtl8139_reset_rxring (RTL8139State *s,uint32_t bufferSize)
hw/smc91c111.c +6 -4
@@ -602,7 +602,7 @@ static int smc91c111_can_receive(VLANClientState *vc)return 1 ; } -static void smc91c111_receive (VLANClientState *vc,const uint8_t *buf,size_t size) +static ssize_t smc91c111_receive (VLANClientState *vc,const uint8_t *buf,size_t size) {smc91c111_state *s = vc->opaque;int status;
@@ -612,7 +612,7 @@ static void smc91c111_receive(VLANClientState *vc, const uint8_t *buf, size_t siuint8_t *p;if ((s->rcr &RCR_RXEN ) ==0 || (s->rcr &RCR_SOFT_RST )) -return ; +return -1 ;/* Short packets are padded with zeros. Receiving a packet < 64 bytes long is considered an error condition. */ if (size <64 )
@@ -625,10 +625,10 @@ static void smc91c111_receive(VLANClientState *vc, const uint8_t *buf, size_t si packetsize +=4 ;/* TODO: Flag overrun and receive errors. */ if (packetsize >2048 ) -return ; +return -1 ; packetnum =smc91c111_allocate_packet (s);if (packetnum ==0x80 ) -return ; +return -1 ; s->rx_fifo[s->rx_fifo_len++] = packetnum; p = &s->data[packetnum][0 ];
@@ -676,6 +676,8 @@ static void smc91c111_receive(VLANClientState *vc, const uint8_t *buf, size_t si/* TODO: Raise early RX interrupt? */ s->int_level |=INT_RCV ;smc91c111_update (s); + +return size; }static CPUReadMemoryFunc *smc91c111_readfn[] = {
hw/stellaris_enet.c +5 -3
@@ -78,7 +78,7 @@ static void stellaris_enet_update(stellaris_enet_state *s) }/* TODO: Implement MAC address filtering. */ -static void stellaris_enet_receive (VLANClientState *vc,const uint8_t *buf,size_t size) +static ssize_t stellaris_enet_receive (VLANClientState *vc,const uint8_t *buf,size_t size) {stellaris_enet_state *s = vc->opaque;int n;
@@ -86,10 +86,10 @@ static void stellaris_enet_receive(VLANClientState *vc, const uint8_t *buf, sizeuint32_t crc;if ((s->rctl &SE_RCTL_RXEN ) ==0 ) -return ; +return -1 ;if (s->np >=31 ) {DPRINTF ("Packet dropped\n" ); -return ; +return -1 ; }DPRINTF ("Received packet len=%d\n" , size);
@@ -116,6 +116,8 @@ static void stellaris_enet_receive(VLANClientState *vc, const uint8_t *buf, size s->ris |=SE_INT_RX ;stellaris_enet_update (s); + +return size; }static int stellaris_enet_can_receive (VLANClientState *vc)
hw/usb-net.c +5 -4
@@ -1369,7 +1369,7 @@ static int usb_net_handle_data(USBDevice *dev, USBPacket *p)return ret; } -static void usbnet_receive (VLANClientState *vc,const uint8_t *buf,size_t size) +static ssize_t usbnet_receive (VLANClientState *vc,const uint8_t *buf,size_t size) {USBNetState *s = vc->opaque;struct rndis_packet_msg_type *msg;
@@ -1377,9 +1377,9 @@ static void usbnet_receive(VLANClientState *vc, const uint8_t *buf, size_t size)if (s->rndis) { msg = (struct rndis_packet_msg_type *) s->in_buf;if (!s->rndis_state ==RNDIS_DATA_INITIALIZED ) -return ; +return -1 ;if (size +sizeof (struct rndis_packet_msg_type ) >sizeof (s->in_buf)) -return ; +return -1 ;memset (msg,0 ,sizeof (struct rndis_packet_msg_type )); msg->MessageType =cpu_to_le32 (RNDIS_PACKET_MSG );
@@ -1398,11 +1398,12 @@ static void usbnet_receive(VLANClientState *vc, const uint8_t *buf, size_t size) s->in_len = size +sizeof (struct rndis_packet_msg_type ); }else {if (size >sizeof (s->in_buf)) -return ; +return -1 ;memcpy (s->in_buf, buf, size); s->in_len = size; } s->in_ptr =0 ; +return size; }static int usbnet_can_receive (VLANClientState *vc)
hw/virtio-net.c +6 -4
@@ -361,17 +361,17 @@ static int receive_filter(VirtIONet *n, const uint8_t *buf, int size)return 0 ; } -static void virtio_net_receive (VLANClientState *vc,const uint8_t *buf,size_t size) +static ssize_t virtio_net_receive (VLANClientState *vc,const uint8_t *buf,size_t size) {VirtIONet *n = vc->opaque;struct virtio_net_hdr_mrg_rxbuf *mhdr =NULL ;size_t hdr_len, offset, i;if (!do_virtio_net_can_receive (n, size)) -return ; +return -1 ;if (!receive_filter (n, buf, size)) -return ; +return size;/* hdr_len refers to the header we supply to the guest */ hdr_len = n->mergeable_rx_bufs ?
@@ -389,7 +389,7 @@ static void virtio_net_receive(VLANClientState *vc, const uint8_t *buf, size_t sif ((i !=0 && !n->mergeable_rx_bufs) ||virtqueue_pop (n->rx_vq, &elem) ==0 ) {if (i ==0 ) -return ; +return -1 ;fprintf (stderr,"virtio-net truncating packet\n" );exit (1 ); }
@@ -431,6 +431,8 @@ static void virtio_net_receive(VLANClientState *vc, const uint8_t *buf, size_t svirtqueue_flush (n->rx_vq, i);virtio_notify (&n->vdev, n->rx_vq); + +return size; }/* TX */
hw/xen_nic.c +7 -5
@@ -245,7 +245,7 @@ static int net_rx_ok(VLANClientState *vc)return 1 ; } -static void net_rx_packet (VLANClientState *vc,const uint8_t *buf,size_t size) +static ssize_t net_rx_packet (VLANClientState *vc,const uint8_t *buf,size_t size) {struct XenNetDev *netdev = vc->opaque;netif_rx_request_t rxreq;
@@ -253,7 +253,7 @@ static void net_rx_packet(VLANClientState *vc, const uint8_t *buf, size_t size)void *page;if (netdev->xendev.be_state != XenbusStateConnected) -return ; +return -1 ; rc = netdev->rx_ring.req_cons; rp = netdev->rx_ring.sring->req_prod;
@@ -261,12 +261,12 @@ static void net_rx_packet(VLANClientState *vc, const uint8_t *buf, size_t size)if (rc == rp ||RING_REQUEST_CONS_OVERFLOW (&netdev->rx_ring, rc)) {xen_be_printf (&netdev->xendev,2 ,"no buffer, drop packet\n" ); -return ; +return -1 ; }if (size >XC_PAGE_SIZE -NET_IP_ALIGN ) {xen_be_printf (&netdev->xendev,0 ,"packet too big (%lu > %ld)" , (unsigned long )size,XC_PAGE_SIZE -NET_IP_ALIGN ); -return ; +return -1 ; }memcpy (&rxreq,RING_GET_REQUEST (&netdev->rx_ring, rc),sizeof (rxreq));
@@ -279,11 +279,13 @@ static void net_rx_packet(VLANClientState *vc, const uint8_t *buf, size_t size)xen_be_printf (&netdev->xendev,0 ,"error: rx gref dereference failed (%d)\n" , rxreq.gref);net_rx_response (netdev, &rxreq,NETIF_RSP_ERROR ,0 ,0 ,0 ); -return ; +return -1 ; }memcpy (page +NET_IP_ALIGN , buf, size);xc_gnttab_munmap (netdev->xendev.gnttabdev, page,1 );net_rx_response (netdev, &rxreq,NETIF_RSP_OKAY ,NET_IP_ALIGN , size,0 ); + +return size; }/* ------------------------------------------------------------- */
net.c +28 -26
116 lines not on this page · raw message
net.h +1 -1
@@ -8,7 +8,7 @@typedef struct VLANClientState VLANClientState ;typedef int (NetCanReceive )(VLANClientState *); -typedef void (NetReceive )(VLANClientState *,const uint8_t *,size_t ); +typedef ssize_t (NetReceive )(VLANClientState *,const uint8_t *,size_t );typedef ssize_t (NetReceiveIOV )(VLANClientState *,const struct iovec *,int );typedef void (NetCleanup ) (VLANClientState *);typedef void (LinkStatusChanged )(VLANClientState *);
tap-win32.c +3 -3
@@ -650,11 +650,11 @@ static void tap_cleanup(VLANClientState *vc)qemu_free (s); } -static void tap_receive (VLANClientState *vc,const uint8_t *buf,size_t size) +static ssize_t tap_receive (VLANClientState *vc,const uint8_t *buf,size_t size) { -TAPState *s = opaque; +TAPState *s = vc->opaque; -tap_win32_write (s->handle, buf, size); +return tap_win32_write (s->handle, buf, size); }static void tap_win32_send (void *opaque)
Signature · 1 line
1.6.0.6
PATCH 9/13 net: return status from qemu_deliver_packet()
Will allow qemu_send_packet() handle queue full condition.
--- net.c | 20 +++++++++++++++++--- 1 files changed, 17 insertions(+), 3 deletions(-)
net.c +17 -3
@@ -409,16 +409,30 @@ int qemu_can_send_packet(VLANClientState *sender)return 0 ; } -static void +static int qemu_deliver_packet (VLANClientState *sender,const uint8_t *buf,int size) {VLANClientState *vc; +int ret =-1 ;for (vc = sender->vlan->first_client; vc !=NULL ; vc = vc->next) { -if (vc != sender && !vc->link_down) { - vc->receive (vc, buf, size); +ssize_t len; + +if (vc == sender) { +continue ; + } + +if (vc->link_down) { + ret = size; +continue ; } + + len = vc->receive (vc, buf, size); + + ret = (ret >=0 ) ? ret : len; } + +return ret; }void qemu_send_packet (VLANClientState *vc,const uint8_t *buf,int size)
Signature · 1 line
1.6.0.6
PATCH 10/13 net: split out packet queueing and flushing into separate functions
We'll be doing more packet queueing in later commits.
--- net.c | 155 +++++++++++++++++++++++++++++++++++++++++------------------------ 1 files changed, 98 insertions(+), 57 deletions(-)
net.c +98 -57
@@ -415,6 +415,8 @@ qemu_deliver_packet(VLANClientState *sender, const uint8_t *buf, int size)VLANClientState *vc;int ret =-1 ; + sender->vlan->delivering =1 ; +for (vc = sender->vlan->first_client; vc !=NULL ; vc = vc->next) {ssize_t len;
@@ -432,13 +434,38 @@ qemu_deliver_packet(VLANClientState *sender, const uint8_t *buf, int size) ret = (ret >=0 ) ? ret : len; } + sender->vlan->delivering =0 ; +return ret; } +static void qemu_flush_queued_packets (VLANClientState *vc) +{ +VLANPacket *packet; + +while ((packet = vc->vlan->send_queue) !=NULL ) { + vc->vlan->send_queue = packet->next; +qemu_deliver_packet (packet->sender, packet->data, packet->size); +qemu_free (packet); + } +} + +static void +qemu_enqueue_packet (VLANClientState *sender,const uint8_t *buf,int size) +{ +VLANPacket *packet; + + packet =qemu_malloc (sizeof (VLANPacket) + size); + packet->next = sender->vlan->send_queue; + packet->sender = sender; + packet->size = size; +memcpy (packet->data, buf, size); + sender->vlan->send_queue = packet; +} +void qemu_send_packet (VLANClientState *vc,const uint8_t *buf,int size) {VLANState *vlan = vc->vlan; -VLANPacket *packet;if (vc->link_down)return ;
@@ -448,22 +475,12 @@ void qemu_send_packet(VLANClientState *vc, const uint8_t *buf, int size)hex_dump (stdout, buf, size);#endif if (vlan->delivering) { - packet =qemu_malloc (sizeof (VLANPacket) + size); - packet->next = vlan->send_queue; - packet->sender = vc; - packet->size = size; -memcpy (packet->data, buf, size); - vlan->send_queue = packet; - }else { - vlan->delivering =1 ; -qemu_deliver_packet (vc, buf, size); -while ((packet = vlan->send_queue) !=NULL ) { - vlan->send_queue = packet->next; -qemu_deliver_packet (packet->sender, packet->data, packet->size); -qemu_free (packet); - } - vlan->delivering =0 ; +qemu_enqueue_packet (vc, buf, size); +return ; } + +qemu_deliver_packet (vc, buf, size); +qemu_flush_queued_packets (vc); } static ssize_t vc_sendv_compat(VLANClientState *vc, const struct iovec *iov,
@@ -496,60 +513,84 @@ static ssize_t calc_iov_length(const struct iovec *iov, int iovcnt)return offset; } -ssize_t qemu_sendv_packet (VLANClientState *sender,const struct iovec *iov, -int iovcnt) +static int qemu_deliver_packet_iov (VLANClientState *sender, +const struct iovec *iov,int iovcnt) { -VLANState *vlan = sender->vlan;VLANClientState *vc; +int ret =-1 ; + + sender->vlan->delivering =1 ; + +for (vc = sender->vlan->first_client; vc !=NULL ; vc = vc->next) { +ssize_t len; + +if (vc == sender) { +continue ; + } + +if (vc->link_down) { + ret =calc_iov_length (iov, iovcnt); +continue ; + } + +if (vc->receive_iov) { + len = vc->receive_iov (vc, iov, iovcnt); + }else { + len =vc_sendv_compat (vc, iov, iovcnt); + } + + ret = (ret >=0 ) ? ret : len; + } + + sender->vlan->delivering =0 ; + +return ret; +} + +static ssize_t qemu_enqueue_packet_iov (VLANClientState *sender, +const struct iovec *iov,int iovcnt) +{VLANPacket *packet; -ssize_t max_len =0 ; +size_t max_len =0 ;int i; -if (sender->link_down) -return calc_iov_length (iov, iovcnt); + max_len =calc_iov_length (iov, iovcnt); -if (vlan->delivering) { - max_len =calc_iov_length (iov, iovcnt); + packet =qemu_malloc (sizeof (VLANPacket) + max_len); + packet->next = sender->vlan->send_queue; + packet->sender = sender; + packet->size =0 ; - packet =qemu_malloc (sizeof (VLANPacket) + max_len); - packet->next = vlan->send_queue; - packet->sender = sender; - packet->size =0 ; -for (i =0 ; i < iovcnt; i++) { -size_t len = iov[i].iov_len; +for (i =0 ; i < iovcnt; i++) { +size_t len = iov[i].iov_len; -memcpy (packet->data + packet->size, iov[i].iov_base, len); - packet->size += len; - } - vlan->send_queue = packet; - }else { - vlan->delivering =1 ; +memcpy (packet->data + packet->size, iov[i].iov_base, len); + packet->size += len; + } -for (vc = vlan->first_client; vc !=NULL ; vc = vc->next) { -ssize_t len =0 ; + sender->vlan->send_queue = packet; -if (vc == sender) { -continue ; - } -if (vc->link_down) { - len =calc_iov_length (iov, iovcnt); - }else if (vc->receive_iov) { - len = vc->receive_iov (vc, iov, iovcnt); - }else if (vc->receive) { - len =vc_sendv_compat (vc, iov, iovcnt); - } - max_len =MAX (max_len, len); - } +return packet->size; +} -while ((packet = vlan->send_queue) !=NULL ) { - vlan->send_queue = packet->next; -qemu_deliver_packet (packet->sender, packet->data, packet->size); -qemu_free (packet); - } - vlan->delivering =0 ; +ssize_t qemu_sendv_packet (VLANClientState *sender,const struct iovec *iov, +int iovcnt) +{ +int ret; + +if (sender->link_down) { +return calc_iov_length (iov, iovcnt); + } + +if (sender->vlan->delivering) { +return qemu_enqueue_packet_iov (sender, iov, iovcnt); } -return max_len; + ret =qemu_deliver_packet_iov (sender, iov, iovcnt); + +qemu_flush_queued_packets (sender); + +return ret; }#if defined(CONFIG_SLIRP )
Signature · 1 line
1.6.0.6
PATCH 11/13 net: add qemu_send_packet_async()
Add a qemu_send_packet() variant which will queue up the packet if it cannot be sent when all client queues are full. It later invokes the supplied callback when the packet has been sent.
If qemu_send_packet_async() returns zero, the caller is expected to not send any more packets until the queued packet has been sent.
Packets are queued iff a receive() handler returns zero (indicating queue full) and the caller has provided a sent notification callback (indicating it will stop and start its own queue).
We need the packet sending API to support queueing because:
- a sending client should process all available packets in one go (e.g. virtio-net emptying its tx ring)
- a receiving client may not be able to handle the packet (e.g. -EAGAIN from write() to tapfd)
- the sending client could detect this condition in advance (e.g. by select() for writable on tapfd)
- that's too much overhead (e.g. a select() call per packet)
- therefore the sending client must handle the condition by dropping the packet or queueing it
- dropping packets is poor form; we should queue.
However, we don't want queueing to be completely transparent. We want the sending client to stop sending packets as soon as a packet is queued. This allows the sending client to be throttled by the receiver.
--- net.c | 83 ++++++++++++++++++++++++++++++++++++++++++++++++---------------- net.h | 8 ++++++ 2 files changed, 70 insertions(+), 21 deletions(-)
net.c +62 -21
@@ -439,19 +439,32 @@ qemu_deliver_packet(VLANClientState *sender, const uint8_t *buf, int size)return ret; } -static void qemu_flush_queued_packets (VLANClientState *vc) +void qemu_flush_queued_packets (VLANClientState *vc) {VLANPacket *packet;while ((packet = vc->vlan->send_queue) !=NULL ) { +int ret; + vc->vlan->send_queue = packet->next; -qemu_deliver_packet (packet->sender, packet->data, packet->size); + + ret =qemu_deliver_packet (packet->sender, packet->data, packet->size); +if (ret ==0 && packet->sent_cb !=NULL ) { + packet->next = vc->vlan->send_queue; + vc->vlan->send_queue = packet; +break ; + } + +if (packet->sent_cb) + packet->sent_cb (packet->sender); +qemu_free (packet); } } -static void -qemu_enqueue_packet (VLANClientState *sender,const uint8_t *buf,int size) +static void qemu_enqueue_packet (VLANClientState *sender, +const uint8_t *buf,int size, +NetPacketSent *sent_cb) {VLANPacket *packet;
@@ -459,28 +472,45 @@ qemu_enqueue_packet(VLANClientState *sender, const uint8_t *buf, int size) packet->next = sender->vlan->send_queue; packet->sender = sender; packet->size = size; + packet->sent_cb = sent_cb;memcpy (packet->data, buf, size); sender->vlan->send_queue = packet; } -void qemu_send_packet (VLANClientState *vc,const uint8_t *buf,int size) +ssize_t qemu_send_packet_async (VLANClientState *sender, +const uint8_t *buf,int size, +NetPacketSent *sent_cb) { -VLANState *vlan = vc->vlan; +int ret; -if (vc->link_down) -return ; +if (sender->link_down) { +return size; + }#ifdef DEBUG_NET -printf ("vlan %d send:\n" , vlan->id); +printf ("vlan %d send:\n" , sender->vlan->id);hex_dump (stdout, buf, size);#endif -if (vlan->delivering) { -qemu_enqueue_packet (vc, buf, size); -return ; + +if (sender->vlan->delivering) { +qemu_enqueue_packet (sender, buf, size,NULL ); +return size; } -qemu_deliver_packet (vc, buf, size); -qemu_flush_queued_packets (vc); + ret =qemu_deliver_packet (sender, buf, size); +if (ret ==0 && sent_cb !=NULL ) { +qemu_enqueue_packet (sender, buf, size, sent_cb); +return 0 ; + } + +qemu_flush_queued_packets (sender); + +return ret; +} + +void qemu_send_packet (VLANClientState *vc,const uint8_t *buf,int size) +{ +qemu_send_packet_async (vc, buf, size,NULL ); } static ssize_t vc_sendv_compat(VLANClientState *vc, const struct iovec *iov,
@@ -498,9 +528,7 @@ static ssize_t vc_sendv_compat(VLANClientState *vc, const struct iovec *iov, offset += len; } - vc->receive (vc, buffer, offset); - -return offset; +return vc->receive (vc, buffer, offset); }static ssize_t calc_iov_length (const struct iovec *iov,int iovcnt)
@@ -548,7 +576,8 @@ static int qemu_deliver_packet_iov(VLANClientState *sender, }static ssize_t qemu_enqueue_packet_iov (VLANClientState *sender, -const struct iovec *iov,int iovcnt) +const struct iovec *iov,int iovcnt, +NetPacketSent *sent_cb) {VLANPacket *packet;size_t max_len =0 ;
@@ -559,6 +588,7 @@ static ssize_t qemu_enqueue_packet_iov(VLANClientState *sender, packet =qemu_malloc (sizeof (VLANPacket) + max_len); packet->next = sender->vlan->send_queue; packet->sender = sender; + packet->sent_cb = sent_cb; packet->size =0 ;for (i =0 ; i < iovcnt; i++) {
@@ -573,8 +603,9 @@ static ssize_t qemu_enqueue_packet_iov(VLANClientState *sender,return packet->size; } -ssize_t qemu_sendv_packet (VLANClientState *sender,const struct iovec *iov, -int iovcnt) +ssize_t qemu_sendv_packet_async (VLANClientState *sender, +const struct iovec *iov,int iovcnt, +NetPacketSent *sent_cb) {int ret;
@@ -583,16 +614,26 @@ ssize_t qemu_sendv_packet(VLANClientState *sender, const struct iovec *iov, }if (sender->vlan->delivering) { -return qemu_enqueue_packet_iov (sender, iov, iovcnt); +return qemu_enqueue_packet_iov (sender, iov, iovcnt,NULL ); } ret =qemu_deliver_packet_iov (sender, iov, iovcnt); +if (ret ==0 && sent_cb !=NULL ) { +qemu_enqueue_packet_iov (sender, iov, iovcnt, sent_cb); +return 0 ; + }qemu_flush_queued_packets (sender);return ret; } +ssize_t +qemu_sendv_packet (VLANClientState *vc,const struct iovec *iov,int iovcnt) +{ +return qemu_sendv_packet_async (vc, iov, iovcnt,NULL ); +} +#if defined(CONFIG_SLIRP )/* slirp network adapter */
net.h +8 -0
@@ -32,10 +32,13 @@ struct VLANClientState {typedef struct VLANPacket VLANPacket ; +typedef void (NetPacketSent ) (VLANClientState *); +struct VLANPacket {struct VLANPacket *next;VLANClientState *sender;int size; +NetPacketSent *sent_cb;uint8_t data[0 ]; };
@@ -62,7 +65,12 @@ VLANClientState *qemu_find_vlan_client(VLANState *vlan, void *opaque);int qemu_can_send_packet (VLANClientState *vc);ssize_t qemu_sendv_packet (VLANClientState *vc,const struct iovec *iov,int iovcnt); +ssize_t qemu_sendv_packet_async (VLANClientState *vc,const struct iovec *iov, +int iovcnt,NetPacketSent *sent_cb);void qemu_send_packet (VLANClientState *vc,const uint8_t *buf,int size); +ssize_t qemu_send_packet_async (VLANClientState *vc,const uint8_t *buf, +int size,NetPacketSent *sent_cb); +void qemu_flush_queued_packets (VLANClientState *vc);void qemu_format_nic_info_str (VLANClientState *vc,uint8_t macaddr[6 ]);void qemu_check_nic_model (NICInfo *nd,const char *model); void qemu_check_nic_model_list(NICInfo *nd, const char * const *models,
Signature · 1 line
1.6.0.6
PATCH 12/13 net: make use of async packet sending API in tap client
If a packet is queued by qemu_send_packet(), remove I/O handler for the tap fd until we get notification that the packet has been sent.
A not insignificant side effect of this is we can now drain the tap send queue in one go without fear of packets being dropped.
--- net.c | 24 ++++++++++++++++++++---- 1 files changed, 20 insertions(+), 4 deletions(-)
net.c +20 -4
@@ -934,15 +934,31 @@ static ssize_t tap_read_packet(int tapfd, uint8_t *buf, int maxlen) }#endif +static void tap_send (void *opaque); + +static void tap_send_completed (VLANClientState *vc) +{ +TAPState *s = vc->opaque; + +qemu_set_fd_handler2 (s->fd, tap_can_send, tap_send,NULL , s); +} +static void tap_send (void *opaque) {TAPState *s = opaque;int size; - size =tap_read_packet (s->fd, s->buf,sizeof (s->buf)); -if (size >0 ) { -qemu_send_packet (s->vc, s->buf, size); - } +do { + size =tap_read_packet (s->fd, s->buf,sizeof (s->buf)); +if (size <=0 ) { +break ; + } + + size =qemu_send_packet_async (s->vc, s->buf, size, tap_send_completed); +if (size ==0 ) { +qemu_set_fd_handler2 (s->fd,NULL ,NULL ,NULL ,NULL ); + } + }while (size >0 ); }static void tap_cleanup (VLANClientState *vc)
Signature · 1 line
1.6.0.6
PATCH 13/13 virtio-net: implement rx packet queueing
If we don't have room to receive a packet, we return zero from virtio_net_receive() and call qemu_flush_queued_packets() as soon as space becomes available.
--- hw/virtio-net.c | 5 ++++- 1 files changed, 4 insertions(+), 1 deletions(-)
hw/virtio-net.c +4 -1
@@ -269,6 +269,9 @@ static void virtio_net_handle_ctrl(VirtIODevice *vdev, VirtQueue *vq)static void virtio_net_handle_rx (VirtIODevice *vdev,VirtQueue *vq) { +VirtIONet *n =to_virtio_net (vdev); + +qemu_flush_queued_packets (n->vc); }static int do_virtio_net_can_receive (VirtIONet *n,int bufsize)
@@ -368,7 +371,7 @@ static ssize_t virtio_net_receive(VLANClientState *vc, const uint8_t *buf, size_size_t hdr_len, offset, i;if (!do_virtio_net_can_receive (n, size)) -return -1 ; +return 0 ;if (!receive_filter (n, buf, size))return size;
Signature · 1 line
1.6.0.6
Quoted 12 lines · Mark McLoughlin
Hi,
This patch series works towards two[1] goals:1) Merging the code in qemu-kvm which drains the tap file descriptor when it is readable and buffers a packet if it fails to add it to the NIC receive queue
2) Using the new
TAP_SETSNDBUFioctl()to put a limit on the number of in-flight packets allowed on a tap device; in this case, if a NIC pops a packet from its transmit queue, we need to be able to buffer said packet if the tap queue is full
Isn't it better instead to unpop the buffer? The NIC tx ring already is a buffer, no need to add another one on top.
It will need adjustments to the device models; for example we'll need virtqueue_pop_commit() after we're certain the tap had enough room for our packet and virtqueue_pop_cancel() (to unmap the buffers) if we don't.
Signature · 1 line
error compiling committee.c: too many arguments to function
Quoted 14 lines · Avi Kivity
Quoted 12 lines · Mark McLoughlin
Hi,
This patch series works towards two[1] goals:1) Merging the code in qemu-kvm which drains the tap file descriptor when it is readable and buffers a packet if it fails to add it to the NIC receive queue
2) Using the new
TAP_SETSNDBUFioctl()to put a limit on the number of in-flight packets allowed on a tap device; in this case, if a NIC pops a packet from its transmit queue, we need to be able to buffer said packet if the tap queue is full
Quoted 6 lines
Isn't it better instead to unpop the buffer? The NIC tx ring already is a buffer, no need to add another one on top.
It will need adjustments to the device models; for example we'll need
virtqueue_pop_commit()after we're certain the tap had enough room for our packet andvirtqueue_pop_cancel()(to unmap the buffers) if we don't.
Yep, it's possible with virtio. However, you can't unpop the buffer from a tap file descriptor or socket.
The alternative in those cases is to implement buffering for each, or to always check the receiving side has buffers available before popping. I choose this option because checking in advance for each packet seems expensive - i.e. a syscall for tap/socket or a trawl through the ring for virtio.
Cheers,
Mark.
Quoted 12 lines · Mark McLoughlin
Quoted 2 lines
It will need adjustments to the device models; for example we'll need
virtqueue_pop_commit()after we're certain the tap had enough room for our packet andvirtqueue_pop_cancel()(to unmap the buffers) if we don't.Yep, it's possible with virtio. However, you can't unpop the buffer from a tap file descriptor or socket.
The alternative in those cases is to implement buffering for each, or to always check the receiving side has buffers available before popping. I choose this option because checking in advance for each packet seems expensive - i.e. a syscall for tap/socket or a trawl through the ring for virtio.
For virtio (or the other emulated devices) the check is pretty cheap; furthermore you only need to do it when something changes (notification or a pop).
So, with the disadvantage of an asymmetrical API, we have:
transmit():
conditional pop
tap->receive()
commit pop / cancel pop
receive():
tap->send()
if (no more room)
tap->stop()
receive_notify():
if (more room)
tap->start()
Signature · 1 line
error compiling committee.c: too many arguments to function
Quoted 4 lines · Mark McLoughlin
Hi,
This patch series works towards two[1] goals:
10-13 seems to have been lost in the ether....
Signature · 3 lines
Regards, Anthony Liguori
Quoted 7 lines · Anthony Liguori
Quoted 4 lines · Mark McLoughlin
Hi,
This patch series works towards two[1] goals:10-13 seems to have been lost in the ether....
Odd. I'll hunt them down.
Cheers,
Mark.