kernel/git/stable/linux.git/net/batman-adv/hard-interface.c, branch master

treewide: refresh kmalloc_obj() conversions

2026-09-05T04:37:00+00:00

This is another run of the Coccinelle script for converting kmalloc() family of allocations to kmalloc_obj() via the existing rules in scripts/coccinelle/api/kmalloc_objs.cocci This catches both the set of kmalloc() uses added since the first kmalloc_obj() conversions in v7.0 and adds a large group missed in the first pass due to Coccinelle not interacting well with the cleanup.h scoped_...() family of macros[1]. I worked around this with spatch's "--macro-file" argument to a file with all the scoped_...() macros mapped to Coccinelle's YACFE_ITERATOR[2] as that was the closest viable control flow indicator I could find. Build tested allmodconfig on x86, arm64, arm, loongarch, mips, powerpc, riscv, and s390 with no new warnings. Link: https://lore.kernel.org/lkml/202609021314.8A9C0B8@keescook/ [1] Link: https://github.com/coccinelle/coccinelle/blob/master/standard.h [2] Signed-off-by: Kees Cook

batman-adv: switch var declarations to reverse x-mas tree order

2026-08-04T15:33:05+00:00

The network related code should use for local variable declarations an ordering scheme which orders lines longest to shortest. Initializations should only be kept in the declarations when the dependencies between them are not preventing the reverse x-mas tree order. Many functions are already using this order. The remaining ones were supposed to slowly convert to the x-mas tree order when working on them. But this never happened because the patches tried to only modify the relevant lines. Instead of getting better, the order often just became worse. Just fix the remaining offending functions to finally solve this coding style (minor) problem. The anonymous dhcp structs were only extracted to have a clean reverse x-mas tree in functions and are not yet meant as an opportunity for further cleanups. Signed-off-by: Sven Eckelmann

batman-adv: add missing kernel-doc comments

2026-08-04T15:33:04+00:00

batman-adv requires kernel-doc for all functions and data types visible outside their own translation unit. Promoting a function from static to module-wide visibility currently requires adding documentation from scratch. However, this burden falls on whoever promotes a function from static to module-wide visibility, rather than its original implementer. Add the missing kernel-doc comments for the remaining undocumented functions and data types to reduce the complexity for new contributions. Signed-off-by: Sven Eckelmann

batman-adv: drop NULL check for immutable hardif->mesh_iface

2026-06-28T19:58:36+00:00

The batadv_hard_iface->mesh_iface became immutable after the global batadv_hardif_list was removed and batadv_hard_iface only exists when it is assigned to an mesh_iface. This member can never become NULL and thus a check is now unnecessary. Signed-off-by: Sven Eckelmann

batman-adv: drop unneeded goto and initialization from batadv_hardif_disable_interface()

2026-06-28T19:45:29+00:00

The only use of the label was too early for primary_if to be set anyways. Also move the put of primary_if further up to hold the reference only as long as necessary, hopefully avoiding the need to re-introduce the goto label with future code changes. Signed-off-by: Nora Schiffer Signed-off-by: Sven Eckelmann

batman-adv: move hardif generation counter into batadv_priv

2026-06-28T19:45:28+00:00

The counter doesn't need to be global. Signed-off-by: Nora Schiffer Signed-off-by: Sven Eckelmann

batman-adv: remove BATADV_IF_NOT_IN_USE hardif state

2026-06-28T19:45:28+00:00

With hardifs only existing while an interface is part of a mesh, the BATADV_IF_NOT_IN_USE state has become redundant. Signed-off-by: Nora Schiffer Signed-off-by: Sven Eckelmann

batman-adv: make hard_iface->mesh_iface immutable

2026-06-28T19:45:28+00:00

With the hard_iface now being created for a specific mesh_iface, it is beneficial not to set mesh_iface to NULL when the interface is disabled, but instead keeping it immutable after the initial setup of the hard_iface. By also holding the reference to the mesh_iface until the hard_iface is released, hard_ifaces iterated over under RCU will always point to a valid mesh_iface. Co-developed-by: Nora Schiffer Signed-off-by: Nora Schiffer Signed-off-by: Sven Eckelmann

batman-adv: remove global hardif list

2026-06-28T19:45:28+00:00

With interfaces being kept track of as iflink private data, there is no need for the global list anymore. batadv_hardif_get_by_netdev() can now use netdev_master_upper_dev_get()+netdev_lower_dev_get_private() to find the hardif corresponding to a netdev. Signed-off-by: Nora Schiffer Signed-off-by: Sven Eckelmann

batman-adv: create hardif only for netdevs that are part of a mesh

2026-06-28T19:45:28+00:00

batman-adv is using netdev notifiers to create a hard_iface struct for every Ethernet-like netdev in the system. These hardifs are tracked in a global linked list, which results in a few performance issues: Lookups in this list are O(n) in the total number of netdevs. As a hardif is looked up when a netdev is removed, this also takes O(n) in the number of netdevs, and removing n netdevs may take O(n^2). This slowdown will always happen when the batman-adv module is loaded, no mesh needs to be active. With the hardif being referenced as iflink private data, the global list is only needed for hardifs that are *not* part of a mesh (that is, the hardif is unused). To prepare for removing the global list, only create a hardif struct when an interface is added to a mesh and destroy it on removal. As adding/removing and enabling/disabling a hardif become one and the same, batadv_hardif_add_interface() is merged into batadv_hardif_enable_interface(), and batadv_hardif_remove_interface() can be dropped altogether. Signed-off-by: Nora Schiffer Signed-off-by: Sven Eckelmann