From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: DKIM-Filter: OpenDKIM Filter v2.11.0 smtp3.osuosl.org 303A660C28 DKIM-Filter: OpenDKIM Filter v2.11.0 smtp3.osuosl.org 39FB960D71 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=blackwall-org.20221208.gappssmtp.com; s=20221208; t=1684226293; x=1686818293; h=content-transfer-encoding:in-reply-to:from:references:cc:to :content-language:subject:user-agent:mime-version:date:message-id :from:to:cc:subject:date:message-id:reply-to; bh=qoPnstjPrS2tzxEaPLFxVKxPj2WD+daPAFjbj9WTzeA=; b=x+AYYlvmcvoPE6XEGFZ+zXR1dy97/Gjk9/JFLJgq598TT4S2KgCmm7js1PYCuIu2aa BgK6e/PbU25N55S47oUFbt8DQ4RJSpuOYyYOf1QTi3BuRFdSSSrcJnkgzDDzV3lX7Uo4 ZuCRK7bEeBkio/9mz3kpnivYgynTlowWBwmnd6Z0jR2frjS+euEznXRhH7wr1/TJ6QLq VDdyZKJttmdjpHvDAzqdFAroThW0zhwWEcLH4NRdbcEGY1FbzjnHSzc8Rm9Lp/A+flMK 2enzAg2BLSXp1asCYWGYxX6zQ2eupJXgV0c/Xooa8maBNLaUOO0FajLMXLjLcnXgTGpl o7dA== Message-ID: Date: Tue, 16 May 2023 11:38:11 +0300 MIME-Version: 1.0 Content-Language: en-US References: <20230515085046.4457-1-jnixdorf-oss@avm.de> From: Nikolay Aleksandrov In-Reply-To: <20230515085046.4457-1-jnixdorf-oss@avm.de> Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 8bit Subject: Re: [Bridge] [PATCH net-next 1/2] bridge: Add a limit on FDB entries List-Id: Linux Ethernet Bridging List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Johannes Nixdorf , netdev@vger.kernel.org Cc: bridge@lists.linux-foundation.org, Eric Dumazet , Roopa Prabhu , Jakub Kicinski , Paolo Abeni , "David S. Miller" On 15/05/2023 11:50, Johannes Nixdorf wrote: > A malicious actor behind one bridge port may spam the kernel with packets > with a random source MAC address, each of which will create an FDB entry, > each of which is a dynamic allocation in the kernel. > > There are roughly 2^48 different MAC addresses, further limited by the > rhashtable they are stored in to 2^31. Each entry is of the type struct > net_bridge_fdb_entry, which is currently 128 bytes big. This means the > maximum amount of memory allocated for FDB entries is 2^31 * 128B = > 256GiB, which is too much for most computers. > > Mitigate this by adding a bridge netlink setting IFLA_BR_FDB_MAX_ENTRIES, > which, if nonzero, limits the amount of entries to a user specified > maximum. > > For backwards compatibility the default setting of 0 disables the limit. > > All changes to fdb_n_entries are under br->hash_lock, which means we do > not need additional locking. The call paths are (✓ denotes that > br->hash_lock is taken around the next call): > > - fdb_delete <-+- fdb_delete_local <-+- br_fdb_changeaddr ✓ > | +- br_fdb_change_mac_address ✓ > | +- br_fdb_delete_by_port ✓ > +- br_fdb_find_delete_local ✓ > +- fdb_add_local <-+- br_fdb_changeaddr ✓ > | +- br_fdb_change_mac_address ✓ > | +- br_fdb_add_local ✓ > +- br_fdb_cleanup ✓ > +- br_fdb_flush ✓ > +- br_fdb_delete_by_port ✓ > +- fdb_delete_by_addr_and_port <--- __br_fdb_delete ✓ > +- br_fdb_external_learn_del ✓ > - fdb_create <-+- fdb_add_local <-+- br_fdb_changeaddr ✓ > | +- br_fdb_change_mac_address ✓ > | +- br_fdb_add_local ✓ > +- br_fdb_update ✓ > +- fdb_add_entry <--- __br_fdb_add ✓ > +- br_fdb_external_learn_add ✓ > > Signed-off-by: Johannes Nixdorf > --- > include/uapi/linux/if_link.h | 1 + > net/bridge/br_device.c | 2 ++ > net/bridge/br_fdb.c | 6 ++++++ > net/bridge/br_netlink.c | 9 ++++++++- > net/bridge/br_private.h | 2 ++ > 5 files changed, 19 insertions(+), 1 deletion(-) > I completely missed the fact that you don't deal with the situation where you already have fdbs created and a limit is set later, then it would be useless because it will start counting from 0 even though there are already entries. Also another issue that came to mind is that you don't deal with fdb_create() for "special" entries, i.e. when adding a port. Currently it will print an error, but you should revisit all callers and see where it might be a problem.