CEPH filesystem development
 help / color / mirror / Atom feed
* [PATCH v2] net/ceph/messenger: support IPv6 Link-Local with scope identifier
@ 2026-07-06  7:44 Max Kellermann
  2026-07-06 16:18 ` [v2] " Alex Markuze
  0 siblings, 1 reply; 2+ messages in thread
From: Max Kellermann @ 2026-07-06  7:44 UTC (permalink / raw)
  To: idryomov, amarkuze, ceph-devel, linux-kernel; +Cc: Max Kellermann

Implement ceph_pton() using inet_pton_with_scope().  This minimal
patch preserves the ceph_pton() API and thus requires copying the
parameter to a null-terminated string.

Signed-off-by: Max Kellermann <max.kellermann@ionos.com>
---
v1->v2:
- check "ipend" before dereferencing it
---
 net/ceph/messenger.c | 37 ++++++++++++++++++++++++++++---------
 1 file changed, 28 insertions(+), 9 deletions(-)

diff --git a/net/ceph/messenger.c b/net/ceph/messenger.c
index 34b3097b4c7b..f7c7179589af 100644
--- a/net/ceph/messenger.c
+++ b/net/ceph/messenger.c
@@ -1221,19 +1221,38 @@ void ceph_addr_set_port(struct ceph_entity_addr *addr, int p)
 static int ceph_pton(const char *str, size_t len, struct ceph_entity_addr *addr,
 		char delim, const char **ipend)
 {
-	memset(&addr->in_addr, 0, sizeof(addr->in_addr));
+	const char *delim_p;
+	char *copy;
+	struct sockaddr_storage stor = { 0 };
+	int ret;
 
-	if (in4_pton(str, len, (u8 *)&((struct sockaddr_in *)&addr->in_addr)->sin_addr.s_addr, delim, ipend)) {
-		put_unaligned(AF_INET, &addr->in_addr.ss_family);
-		return 0;
-	}
+	delim_p = memchr(str, delim, len);
+	if (delim_p)
+		/* delimiter was found - stop parsing there */
+		len = delim_p - str;
 
-	if (in6_pton(str, len, (u8 *)&((struct sockaddr_in6 *)&addr->in_addr)->sin6_addr.s6_addr, delim, ipend)) {
-		put_unaligned(AF_INET6, &addr->in_addr.ss_family);
-		return 0;
+	/* the input string might not be null terminated, so copy it
+	 * to a null-terminated string
+	 */
+	copy = kstrndup(str, len, GFP_NOFS);
+	if (!copy)
+		return -ENOMEM;
+
+	ret = inet_pton_with_scope(&init_net, AF_UNSPEC, copy, NULL, &stor);
+	kfree(copy);
+
+	if (!ret) {
+		/* ceph_entity_addr might be misaligned, so we have to
+		 * parse to a stack variable first
+		 */
+		memcpy(&addr->in_addr, &stor, sizeof(stor));
+
+		/* return the end of the parsed portion to the caller on success */
+		if (ipend)
+		    *ipend = str + len;
 	}
 
-	return -EINVAL;
+	return ret;
 }
 
 /*
-- 
2.47.3


^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [v2] net/ceph/messenger: support IPv6 Link-Local with scope identifier
  2026-07-06  7:44 [PATCH v2] net/ceph/messenger: support IPv6 Link-Local with scope identifier Max Kellermann
@ 2026-07-06 16:18 ` Alex Markuze
  0 siblings, 0 replies; 2+ messages in thread
From: Alex Markuze @ 2026-07-06 16:18 UTC (permalink / raw)
  To: max.kellermann; +Cc: ceph-devel

Hi Max,

NACK for now. Thanks for the patch, but I have some concerns
that need a v2:

  1. [critical] net/ceph/messenger.c:1241: IPv4 addresses with explicit ports no longer parse correctly
     `ceph_pton()` now sets `*ipend = str + len`, where `len` extends to
     the list delimiter (comma/end-of-string). For
     `10.0.0.1:7001,10.0.0.2:7001`, `ipend` lands on the comma, not the
     colon. The caller `ceph_parse_ips()` (line 1350) checks `*ipend ==
     ':'` to find the port — it never will. Depending on how
     `inet_pton_with_scope()` handles the trailing `:7001` in the
     address string, this either silently defaults every monitor to
     `CEPH_MON_PORT` or fails the parse entirely. Either way, existing
     mount strings with explicit ports regress.
     Suggested fix: Split the address from the port before passing to
     `inet_pton_with_scope()`, or set `*ipend` to where the address
     portion actually ends (before the colon) so the caller's
     port-parsing logic still works.
  2. [major] net/ceph/messenger.c:1241: init_net instead of caller's network namespace
     The patch passes `&init_net` to `inet_pton_with_scope()`. The rest
     of the file uses the caller's namespace: `ceph_messenger_init()`
     saves `current->nsproxy->net_ns` into `msgr->net` (line 1695),
     `sock_create_kern()` uses `read_pnet(&con->msgr->net)` (line 448),
     and `dns_query()` uses `current->nsproxy->net_ns` (line 1271).
     Resolving `%eth0` scope identifiers against `init_net` will
     silently break in containers where the interface only exists in the
     container's namespace — exactly where link-local addresses are most
     commonly used.
     Suggested fix: Use `current->nsproxy->net_ns` instead of
     `&init_net`.
  3. [nit] net/ceph/messenger.c:1252: Spaces instead of tab indentation
     Line is indented with spaces instead of tabs. Kernel coding style
     requires tabs.
     Suggested fix: Replace the 4 spaces with a tab.

New ipend placement breaks IPv4 port parsing (regression); init_net wrong namespace for container use.


-- 
Alex Markuze


^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2026-07-06 16:18 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-07-06  7:44 [PATCH v2] net/ceph/messenger: support IPv6 Link-Local with scope identifier Max Kellermann
2026-07-06 16:18 ` [v2] " Alex Markuze

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox