Skip to contentSkip to conversation

Fix --show-gateway for IPv6 on NetBSD/i386.

PATCH In review Gert Doering 2 participants 3 messages Sep 13, 2020 – Sep 14, 2020

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Fix --show-gateway for IPv6 on NetBSD/i386.

Our ROUNDUP() macro to achieve the required system-specific alignment for data structures sent to the routing socket was wrong for NetBSD - unlike OpenBSD/FreeBSD, NetBSD is not using "long" (32/64 bit depending on OS architecture), and not "uint32_t" either (32/32) like MacOS, but uint64_t.

So our use of "long" always worked on NetBSD/amd64 and stopped working on NetBSD/i386 when this was changed on the OS side...

NetBSD conveniently exports a RT_ROUNDUP() macro from <net/route.h> - use that, and avoid trying to second-guess OS requirements.

While at it, add M_ERRNO to ominous "GDG6: problem writing to routing socket" error message to differenciate between "EINVAL" and other errors.

Trac: #734

---
 src/openvpn/route.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)
src/openvpn/route.c +12 -2
index f127a90a..87febf49 100644
@@ -49,6 +49,10 @@
 #include <linux/rtnetlink.h>            /* RTM_GETROUTE etc. */
 #endif
 
+#if defined(TARGET_NETBSD)
+#include <net/route.h>			/* RT_ROUNDUP(), RT_ADVANCE() */
+#endif
+
 #ifdef _WIN32
 #include "openvpn-msg.h"
 
@@ -3408,11 +3412,15 @@ struct rtmsg {
 
 /* the route socket code is identical for all 4 supported BSDs and for
  * MacOS X (Darwin), with one crucial difference: when going from
- * 32 bit to 64 bit, the BSDs increased the structure size but kept
+ * 32 bit to 64 bit, FreeBSD/OpenBSD increased the structure size but kept
  * source code compatibility by keeping the use of "long", while
  * MacOS X decided to keep binary compatibility by *changing* the API
  * to use "uint32_t", thus 32 bit on all OS X variants
  *
+ * NetBSD does the MacOS way of "fixed number of bits, no matter if
+ * 32 or 64 bit OS", but chose uint64_t.  For maximum portability, we
+ * just use the OS RT_ROUNDUP() macro, which is guaranteed to be correct.
+ *
  * We used to have a large amount of duplicate code here which really
  * differed only in this (long) vs. (uint32_t) - IMHO, worse than
  * having a combined block for all BSDs with this single #ifdef inside
@@ -3421,6 +3429,8 @@ struct rtmsg {
 #if defined(TARGET_DARWIN)
 #define ROUNDUP(a) \
     ((a) > 0 ? (1 + (((a) - 1) | (sizeof(uint32_t) - 1))) : sizeof(uint32_t))
+#elif defined(TARGET_NETBSD)
+#define ROUNDUP(a) RT_ROUNDUP(a)
 #else
 #define ROUNDUP(a) \
     ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
@@ -3729,7 +3739,7 @@ get_default_gateway_ipv6(struct route_ipv6_gateway_info *rgi6,
     }
     if (write(sockfd, (char *)&m_rtmsg, l) < 0)
     {
-        msg(M_WARN, "GDG6: problem writing to routing socket");
+        msg(M_WARN|M_ERRNO, "GDG6: problem writing to routing socket");
         goto done;
     }
 
Signature · 1 line
2.22.0
Acked-by on v1 1/1
#1 lore raw

Am 13.09.20 um 16:56 schrieb Gert Doering:

Quoted 6 lines

Our ROUNDUP() macro to achieve the required system-specific alignment for data structures sent to the routing socket was wrong for NetBSD - unlike OpenBSD/FreeBSD, NetBSD is not using "long" (32/64 bit depending on OS architecture), and not "uint32_t" either (32/32) like MacOS, but uint64_t.

Quoted 8 lines

So our use of "long" always worked on NetBSD/amd64 and stopped working on NetBSD/i386 when this was changed on the OS side...

NetBSD conveniently exports a RT_ROUNDUP() macro from <net/route.h> - use that, and avoid trying to second-guess OS requirements.

While at it, add M_ERRNO to ominous "GDG6: problem writing to routing socket" error message to differenciate between "EINVAL" and other errors.

Have not tested it but looks sane and using OS code should be more sane indeed.

Arne

#2 lore raw
In reply to Gert Doering

Patch has been applied to many branches... (bugfix).

commit 37aab49b083a9e385970e3ab2dd727ea1a95ff35 (master) commit 9a695902402faafa343783c38761c4c0ccbb2b83 (HEAD -> release/2.5) commit 9e9fd488d7d939ac9992f8bd4c7394a39b73d7e4 (HEAD -> release/2.4)

Author: Gert Doering
Date:   Sun Sep 13 16:56:21 2020 +0200
     Fix --show-gateway for IPv6 on NetBSD/i386.

     Signed-off-by: Gert Doering <[email protected]...>
     Acked-by: Arne Schwabe <[email protected]...>
     Message-Id: <[email protected]...>
     URL: https://www.mail-archive.com/[email protected]/msg20983.html
     Signed-off-by: Gert Doering <[email protected]...>

--
kind regards,

Gert Doering