CVE-2026-93231: lockd: fix swapped arguments in nlmsvc_match_ip()

In the Linux kernel, the following vulnerability has been resolved:

lockd: fix swapped arguments in nlmsvcmatchip()

When releasing locks by server IP address via /proc/fs/nfsd/unlockip, nlmsvcunlockallbyip() calls nlmtraversefiles() with the server sockaddr as the opaque @data argument:

nlmtraversefiles(serveraddr, nlmsvcmatchip, NULL);

The match callback is later invoked from nlmtraverselocks() as:

match(lockhost, host);

where the first argument is the nlmhost that owns the lock, and the second argument is the @data that was originally passed down (here the server sockaddr). This is the convention every other match callback relies on (nlmsvcmarkhost(), nlmsvcsamehost(), nlmsvcisclient()): arg1 is the real nlmhost, arg2 is the caller-supplied reference value.

nlmsvcmatchip() has had these two arguments reversed ever since the unlock-by-IP feature was introduced in commit 4373ea84c84d ("lockd: unlock lockd locks associated with a given server ip"):

return rpccmpaddr(nlmsrcaddr(host), datap);

Here @host is actually the server sockaddr, so nlmsrcaddr(host) dereferences a struct sockaddr as a struct nlmhost and reads garbage at the offset of hsrcaddr; meanwhile @datap is actually the lock owner's nlmhost but is compared as a sockaddr. As a result the comparison practically never matches and locks are not released for the requested IP.

Swap the arguments so the lock owner's source address is compared against the requested server address:

return rpccmpaddr(nlmsrcaddr(datap), (struct sockaddr )host);

[ cel: fix the misleading typedef parameter names too ]