Remove unnecessary sockaddr memory allocation #7219
Conversation
twose
force-pushed
the
sockaddr_malloc
branch
2 times, most recently
from
cf1a979 to
f923b27
Compare
The size of sockaddr_in and sockaddr_in6 is known on compile-time and it is not that big for the stack.
| switch (sa->sa_family) { | ||
| #if HAVE_GETADDRINFO && HAVE_IPV6 | ||
| case AF_INET6: | ||
| ((struct sockaddr_in6 *)sa)->sin6_family = sa->sa_family; |
There was a problem hiding this comment.
That doesn't seem to make sense, it is equal to a = a, so I removed all of such code, correct me if I am wrong.
| ((struct sockaddr_in *)sa)->sin_port = htons(port); | ||
| socklen = sizeof(struct sockaddr_in); | ||
| break; | ||
| default: | ||
| /* Unknown family */ | ||
| socklen = 0; | ||
| sa = NULL; |
There was a problem hiding this comment.
if sa is NULL, it will close the socket and create a new socket in the next loop. We can check it first, then we create the socket.
main/network.c
Outdated
| } | ||
| } | ||
| #endif | ||
|
|
||
| if (!local_address || bind(sock, local_address, local_address_len)) { | ||
| if (local_address.len == 0) { |
There was a problem hiding this comment.
Maybe this makes len more meaningful... And it also killed duplicated code of error report.
| #ifdef HAVE_INET_PTON | ||
| if (!inet_pton(AF_INET, bindto, &in4->sin_addr)) { |
There was a problem hiding this comment.
The check ways of the two inet_pton calls are not consistent, I think it might be better to judge whether it returns 1
|
Another optimization point, the socket can re-call connect after connect failed, so we need not re-create the socket if the address family did not change. But the manual does not stipulate such behavior, I only know that the windows documentation makes this clear.
|
The size of sockaddr_in and sockaddr_in6 is known on compile-time and it is not that big for the stack.