[PATCH 5.15 105/279] bpf: Fix inner map state pruning regression.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



From: Alexei Starovoitov <ast@xxxxxxxxxx>

[ Upstream commit 34d11a440c6167133201b7374065b59f259730d7 ]

Introduction of map_uid made two lookups from outer map to be distinct.
That distinction is only necessary when inner map has an embedded timer.
Otherwise it will make the verifier state pruning to be conservative
which will cause complex programs to hit 1M insn_processed limit.
Tighten map_uid logic to apply to inner maps with timers only.

Fixes: 3e8ce29850f1 ("bpf: Prevent pointer mismatch in bpf_timer_init.")
Reported-by: Lorenz Bauer <lmb@xxxxxxxxxxxxxx>
Signed-off-by: Alexei Starovoitov <ast@xxxxxxxxxx>
Signed-off-by: Daniel Borkmann <daniel@xxxxxxxxxxxxx>
Tested-by: Lorenz Bauer <lmb@xxxxxxxxxxxxxx>
Link: https://lore.kernel.org/bpf/CACAyw99hVEJFoiBH_ZGyy=+oO-jyydoz6v1DeKPKs2HVsUH28w@xxxxxxxxxxxxxx
Link: https://lore.kernel.org/bpf/20211110172556.20754-1-alexei.starovoitov@xxxxxxxxx
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
---
 kernel/bpf/verifier.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index ddba80554fef3..cba37d83451eb 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -1143,7 +1143,8 @@ static void mark_ptr_not_null_reg(struct bpf_reg_state *reg)
 			/* transfer reg's id which is unique for every map_lookup_elem
 			 * as UID of the inner map.
 			 */
-			reg->map_uid = reg->id;
+			if (map_value_has_timer(map->inner_map_meta))
+				reg->map_uid = reg->id;
 		} else if (map->map_type == BPF_MAP_TYPE_XSKMAP) {
 			reg->type = PTR_TO_XDP_SOCK;
 		} else if (map->map_type == BPF_MAP_TYPE_SOCKMAP ||
-- 
2.33.0






[Index of Archives]     [Linux Kernel]     [Kernel Development Newbies]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite Hiking]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux