On 06/09, Paulo Alcantara wrote:
Enzo Matsumiya <ematsumiya@xxxxxxx> writes:The initial value is the same as before (SMB_DNS_RESOLVE_INTERVAL_DEFAULT, 600s). The user don't need to know a value to be used, unless they know the value the server uses (either manually set by themselves or by some external script) and then they can use that same value for this new dns_interval setting. A very simple example on how one could do it, if there's no desire to use the default 600s: # TTL=$(dig +noall +answer my.server.domain | awk '{ print $2 }') # echo $TTL > /proc/fs/cifs/dns_intervalThat's not what I meant. Sorry if I was unclear. If the upcalled program sets the record TTL (key's expire time), then any values set in /proc/fs/cifs/dns_interval will not work. The user might expect it to work, though.
Exactly. Currently, key.dns_resolver uses getaddrinfo() to resolve names, which doesn't contain the TTL for the record, hence *always* returns 0 to cifs.ko. This patch is just a way to provide some flexibility to the user, in case they don't want to use the currently-always-fixed 600s. As we discussed, this is supposed to be fixed in key.dns_resolver, but my proposed patch for that never got ack'd, or even replied to. I might try again at some point. Cheers, Enzo
