]> git.baikalelectronics.ru Git - kernel.git/commit
net: ipv4: really enforce backoff for redirects
authorPaolo Abeni <pabeni@redhat.com>
Fri, 8 May 2020 17:28:34 +0000 (19:28 +0200)
committerJakub Kicinski <kuba@kernel.org>
Sat, 9 May 2020 05:52:57 +0000 (22:52 -0700)
commitb248be9b0f58bbfa891585588a4b23cbd4aaca6a
tree13cdf95f875974db0c04606c580fa9bd999edfd8
parentb5f9559b1fd1b660eac43ed1975615ca2235721a
net: ipv4: really enforce backoff for redirects

In commit 26a77052e863 ("net: ipv4: avoid mixed n_redirects and
rate_tokens usage") I missed the fact that a 0 'rate_tokens' will
bypass the backoff algorithm.

Since rate_tokens is cleared after a redirect silence, and never
incremented on redirects, if the host keeps receiving packets
requiring redirect it will reply ignoring the backoff.

Additionally, the 'rate_last' field will be updated with the
cadence of the ingress packet requiring redirect. If that rate is
high enough, that will prevent the host from generating any
other kind of ICMP messages

The check for a zero 'rate_tokens' value was likely a shortcut
to avoid the more complex backoff algorithm after a redirect
silence period. Address the issue checking for 'n_redirects'
instead, which is incremented on successful redirect, and
does not interfere with other ICMP replies.

Fixes: 26a77052e863 ("net: ipv4: avoid mixed n_redirects and rate_tokens usage")
Reported-and-tested-by: Colin Walters <walters@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
net/ipv4/route.c