Re: [PATCH net v2 1/1] ipv4: fib: bound automatic table ID allocation

On Tue, Sep 1, 2026 at 5:21 PM Ido Schimmel [off-list ref] wrote:

On Tue, Sep 01, 2026 at 03:00:01AM +0000, Zihan Xi wrote:

quoted

fib_empty_table() probes every table ID from 1 until it finds a
free one.  IPv4 tables are stored in a 256-bucket hash table, so a
dense set of IDs makes each probe walk a growing hash chain while
RTNL is held.

Automatic table assignment ("ip rule ... table 0") is an IPv4-only
legacy path.  Bound the automatically allocated ID to 4096 so the
RTNL hold stays bounded, without changing lookups of explicitly
specified table IDs.

Fixes: b801f54917b7 ("[NET]: Increate RT_TABLE_MAX to 2^32")
Cc: stable@vger.kernel.org
Reported-by: Vega <redacted>
Suggested-by: Ido Schimmel <idosch@nvidia.com>
Assisted-by: Codex:gpt-5.4
Signed-off-by: Zihan Xi <redacted>

Reviewed-by: Ido Schimmel <idosch@nvidia.com>

The bots will likely complain about the change in behavior because the
commit message doesn't spell it out. I explained why I think it's safe
in [1]. In addition, I checked, and both NetworkManager [2] and systemd
[3][4] explicitly forbid "table 0".

[1] https://lore.kernel.org/all/20260831110647.GA3202627@shredder/ (local)
[2] https://github.com/NetworkManager/NetworkManager/blob/2e83a4082b223499c4f1bba156998d0171071967/src/libnm-core-impl/nm-setting-ip-config.c#L2747-L2758
[3] https://github.com/systemd/systemd/blob/9b62fd169e06d14cbab801c66f2f9c2006edb75c/src/network/networkd-route-util.c#L456-L457
[4] https://www.freedesktop.org/software/systemd/man/latest/systemd.network.html#Table=

Thanks for the review. I will send a v3 that spells out the user-visible
change: table-0 auto-assignment now fails with ENOBUFS if IDs 1..4096
are occupied, while explicit table IDs are unchanged.

I will also keep your Reviewed-by.