gfwleak/zhangyang-libzt
Archived
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Added explanation and #defines for new guarded receive buffer

Browse Source
This commit is contained in:
Joseph Henry
2018-02-07 14:23:07 -08:00
parent 7e0da0cfd3
commit bdfdbea2dc
2 changed files with 18 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
include/libztDefs.h
View File

@@ -50,8 +50,18 @@
#define STACK_PICO 0 #define STACK_PICO 0
#define NO_STACK 0 // for layer-2 only (this will omit all userspace network stack code) #define NO_STACK 0 // for layer-2 only (this will omit all userspace network stack code)
/* The following three quantities are related and govern how incoming frames are fed into the
network stack's core:
Every LWIP_GUARDED_BUF_CHECK_INTERVAL milliseconds, a callback will be called from the core and
will input a maximum of LWIP_FRAMES_HANDLED_PER_CORE_CALL frames before returning control back
to the core. Meanwhile, incoming frames from the ZeroTier wire will be allocated and their
pointers will be cached in the receive frame buffer of the size LWIP_MAX_GUARDED_RX_BUF_SZ to
await the next callback from the core */
#define LWIP_GUARDED_BUF_CHECK_INTERVAL 50 // in ms #define LWIP_GUARDED_BUF_CHECK_INTERVAL 50 // in ms
#define LWIP_MAX_GUARDED_RX_BUF_SZ 1024 // number of frame pointers that can be cached waiting for receipt into core #define LWIP_MAX_GUARDED_RX_BUF_SZ 1024 // number of frame pointers that can be cached waiting for receipt into core
#define LWIP_FRAMES_HANDLED_PER_CORE_CALL 16
/* sanity checks for userspace network stack and socket API layer choices /* sanity checks for userspace network stack and socket API layer choices

27
src/lwIP.cpp
View File

@@ -78,7 +78,7 @@ struct netif lwipInterfaces[10];
int lwipInterfacesCount = 0; int lwipInterfacesCount = 0;
ZeroTier::Mutex _rx_input_lock_m; ZeroTier::Mutex _rx_input_lock_m;
struct pbuf* lwip_frame_rxbuf[1024]; struct pbuf* lwip_frame_rxbuf[MAX_GUARDED_RX_BUF_SZ];
int lwip_frame_rxbuf_tot = 0; int lwip_frame_rxbuf_tot = 0;
@@ -119,20 +119,15 @@ static void tcpip_init_done(void *arg)
void my_tcpip_callback(void *arg) void my_tcpip_callback(void *arg)
{ {
Mutex::Lock _l(_rx_input_lock_m); Mutex::Lock _l(_rx_input_lock_m);
int loop_score = 16; // max num of packets to read per polling call int loop_score = LWIP_FRAMES_HANDLED_PER_CORE_CALL; // max num of packets to read per polling call
// TODO: Optimize (use Ringbuffer) // TODO: Optimize (use Ringbuffer)
int pkt_num = 0; int pkt_num = 0;
int count_initial = lwip_frame_rxbuf_tot; int count_initial = lwip_frame_rxbuf_tot;
int count_final = 0;
while (lwip_frame_rxbuf_tot > 0 && loop_score > 0) { while (lwip_frame_rxbuf_tot > 0 && loop_score > 0) {
struct pbuf *p = lwip_frame_rxbuf[pkt_num]; struct pbuf *p = lwip_frame_rxbuf[pkt_num];
pkt_num += 1; pkt_num++;
// DEBUG_INFO("copying received packet FROM guarded buffer (addr=%p) total frames in buffer = %d", p, lwip_frame_rxbuf_tot);
// Packet routing logic. Inputs packet into correct lwip netif interface depending on protocol type // Packet routing logic. Inputs packet into correct lwip netif interface depending on protocol type
struct ip_hdr *iphdr; struct ip_hdr *iphdr;
//ip_addr_t iphdr_dest;
switch (((struct eth_hdr *)p->payload)->type) switch (((struct eth_hdr *)p->payload)->type)
{ {
case PP_HTONS(ETHTYPE_IPV6): { case PP_HTONS(ETHTYPE_IPV6): {
@@ -175,16 +170,14 @@ void my_tcpip_callback(void *arg)
default: default:
break; break;
} }
lwip_frame_rxbuf_tot-=1; lwip_frame_rxbuf_tot--;;
loop_score--; loop_score--;
} }
count_final = count_initial - lwip_frame_rxbuf_tot; int count_final = count_initial - lwip_frame_rxbuf_tot;
//DEBUG_INFO("adjusting buffer by (%d) elements, total frames in buffer = %d", count_final, lwip_frame_rxbuf_tot); // move pbuf frame pointer address buffer by the number of frames successfully fed into the stack core
// move pbuf address buffer by the number of packets successfully fed into the stack core
memmove(lwip_frame_rxbuf, lwip_frame_rxbuf + count_final, sizeof(lwip_frame_rxbuf) - count_final); memmove(lwip_frame_rxbuf, lwip_frame_rxbuf + count_final, sizeof(lwip_frame_rxbuf) - count_final);
} }
// main thread which starts the initialization process // main thread which starts the initialization process
static void main_thread(void *arg) static void main_thread(void *arg)
{ {
@@ -198,8 +191,7 @@ static void main_thread(void *arg)
sys_sem_wait(&sem); sys_sem_wait(&sem);
DEBUG_EXTRA("stack thread init complete"); DEBUG_EXTRA("stack thread init complete");
while(1) while(1) {
{
usleep(LWIP_GUARDED_BUF_CHECK_INTERVAL*1000); usleep(LWIP_GUARDED_BUF_CHECK_INTERVAL*1000);
// Handle incoming packets from the core's thread context. // Handle incoming packets from the core's thread context.
// If you feed frames into the core directly you will violate the core's thread model // If you feed frames into the core directly you will violate the core's thread model
@@ -223,8 +215,6 @@ void lwip_driver_init()
NULL, DEFAULT_THREAD_STACKSIZE, DEFAULT_THREAD_PRIO); NULL, DEFAULT_THREAD_STACKSIZE, DEFAULT_THREAD_PRIO);
} }
err_t lwip_eth_tx(struct netif *netif, struct pbuf *p) err_t lwip_eth_tx(struct netif *netif, struct pbuf *p)
{ {
struct pbuf *q; struct pbuf *q;
@@ -316,11 +306,10 @@ void lwip_eth_rx(VirtualTap *tap, const ZeroTier::MAC &from, const ZeroTier::MAC
} }
Mutex::Lock _l(_rx_input_lock_m); Mutex::Lock _l(_rx_input_lock_m);
if (lwip_frame_rxbuf_tot == LWIP_MAX_GUARDED_RX_BUF_SZ) { if (lwip_frame_rxbuf_tot == LWIP_MAX_GUARDED_RX_BUF_SZ) {
//DEBUG_ERROR("guarded receive buffer full, adjust MAX_GUARDED_RX_BUF_SZ or "); DEBUG_ERROR("guarded receive buffer full, adjust MAX_GUARDED_RX_BUF_SZ or LWIP_GUARDED_BUF_CHECK_INTERVAL");
} }
lwip_frame_rxbuf[lwip_frame_rxbuf_tot] = p; lwip_frame_rxbuf[lwip_frame_rxbuf_tot] = p;
lwip_frame_rxbuf_tot += 1; lwip_frame_rxbuf_tot += 1;
//DEBUG_INFO("copying received packet TO guarded buffer (addr=%p) total frames in buffer = %d", p, lwip_frame_rxbuf_tot);
} }
/* /*