Skip Menu |

This queue is for tickets about the Socket CPAN distribution.

Report information
The Basics
Id: 107058
Status: resolved
Priority: 0/
Queue: Socket

People
Owner: Nobody in particular
Requestors: BULKDD [...] cpan.org
Cc:
AdminCc:

Bug Information
Severity: (no value)
Broken in: (no value)
Fixed in: 2.021



Subject: [PATCH] Win32 inet_pton fallback misc fixes
This patch also fixes the issue I described in https://rt.cpan.org/Ticket/Display.html?id=91524#txn-1519279 so that ticket can be closed again.
Subject: 0001-Win32-inet_pton-fallback-misc-fixes.patch
From 6621966b3520e54fd62afe2ed7273324582a1115 Mon Sep 17 00:00:00 2001 From: Daniel Dragan <bulk88@hotmail.com> Date: Thu, 10 Sep 2015 03:36:36 -0400 Subject: [PATCH] Win32 inet_pton fallback misc fixes -VC complains in inet_pton "warning C4715: 'inet_pton' : not all control paths return a value" this isn't much a problem since Socket.xs doesn't allow anything but AF_INET and AF_INET6 on a XSUB level but fix the implementation anyway since the previous cargo culted off the internet version has flaws, so the Socket.xs version is usable elsewhere if necessery -remove copying the string to C auto array, unix inet_ptoa requires null terminated strings, so does WSAStringToAddress. WSAStringToAddress's docs dont mention a maximum length to the input string, so no reason to truncate and re-null terminate it -MSDN's docs for WSAStringToAddress mentions that filling in sin_family is required, even though lpAddress is an output arg, not input and a duplicate of arg AddressFamily, there is probably some legacy protocol driver out in the world that requires this -static the functions, these fallbacks dont need to be visible in any other .o, and with static they might be inlined/further optimized -provide fallbacks for Visual C 6 (circa 1998) with very old headers that were created before RFC 2553 was created --- cpan/Socket/Socket.xs | 43 ++++++++++++++++++++++++++++++++++--------- 1 files changed, 34 insertions(+), 9 deletions(-) diff --git a/cpan/Socket/Socket.xs b/cpan/Socket/Socket.xs index 5f60afa..52df483 100644 --- a/cpan/Socket/Socket.xs +++ b/cpan/Socket/Socket.xs @@ -52,18 +52,40 @@ #endif #ifdef WIN32 -int inet_pton(int af, const char *src, void *dst) + +/* VC 6 with its original headers doesn't know about sockaddr_storage, VC 2003 does*/ +#ifndef _SS_MAXSIZE + +# define _SS_MAXSIZE 128 +# define _SS_ALIGNSIZE (sizeof(__int64)) + +# define _SS_PAD1SIZE (_SS_ALIGNSIZE - sizeof (short)) +# define _SS_PAD2SIZE (_SS_MAXSIZE - (sizeof (short) + _SS_PAD1SIZE \ + + _SS_ALIGNSIZE)) + +struct sockaddr_storage { + short ss_family; + char __ss_pad1[_SS_PAD1SIZE]; + __int64 __ss_align; + char __ss_pad2[_SS_PAD2SIZE]; +}; + +typedef int socklen_t; + +#define in6_addr in_addr6 + +#define INET_ADDRSTRLEN 22 +#define INET6_ADDRSTRLEN 65 + +#endif + +static int inet_pton(int af, const char *src, void *dst) { struct sockaddr_storage ss; int size = sizeof(ss); - char src_copy[INET6_ADDRSTRLEN+1]; + ss.ss_family = af; /* per MSDN */ - ZeroMemory(&ss, sizeof(ss)); - /* stupid non-const API */ - strncpy(src_copy, src, INET6_ADDRSTRLEN+1); - src_copy[INET6_ADDRSTRLEN] = 0; - - if (WSAStringToAddress(src_copy, af, NULL, (struct sockaddr *)&ss, &size) != 0) + if (WSAStringToAddress((char*)src, af, NULL, (struct sockaddr *)&ss, &size) != 0) return 0; switch(af) { @@ -73,10 +95,13 @@ int inet_pton(int af, const char *src, void *dst) case AF_INET6: *(struct in6_addr *)dst = ((struct sockaddr_in6 *)&ss)->sin6_addr; return 1; + default: + WSASetLastError(WSAEAFNOSUPPORT); + return -1; } } -const char *inet_ntop(int af, const void *src, char *dst, socklen_t size) +static const char *inet_ntop(int af, const void *src, char *dst, socklen_t size) { struct sockaddr_storage ss; unsigned long s = size; -- 1.7.9.msysgit.0
On Sun Sep 13 00:12:35 2015, BULKDD wrote: Show quoted text
> This patch also fixes the issue I described in > https://rt.cpan.org/Ticket/Display.html?id=91524#txn-1519279 so that > ticket can be closed again.
Thanks, applied. -- Paul Evans
Fixed in 2.021 -- Paul Evans