[PATCH 5.15 536/917] bpf: Fixes possible race in update_prog_stats() for 32bit arches

From: Greg Kroah-Hartman
Date: Mon Nov 15 2021 - 18:43:39 EST


From: Eric Dumazet <edumazet@xxxxxxxxxx>

[ Upstream commit d979617aa84d96acca44c2f5778892b4565e322f ]

It seems update_prog_stats() suffers from same issue fixed
in the prior patch:

As it can run while interrupts are enabled, it could
be re-entered and the u64_stats syncp could be mangled.

Fixes: fec56f5890d9 ("bpf: Introduce BPF trampoline")
Signed-off-by: Eric Dumazet <edumazet@xxxxxxxxxx>
Signed-off-by: Alexei Starovoitov <ast@xxxxxxxxxx>
Link: https://lore.kernel.org/bpf/20211026214133.3114279-3-eric.dumazet@xxxxxxxxx
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
---
kernel/bpf/trampoline.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
index fe1e857324e66..d3a307a8c42b9 100644
--- a/kernel/bpf/trampoline.c
+++ b/kernel/bpf/trampoline.c
@@ -585,11 +585,13 @@ static void notrace update_prog_stats(struct bpf_prog *prog,
* Hence check that 'start' is valid.
*/
start > NO_START_TIME) {
+ unsigned long flags;
+
stats = this_cpu_ptr(prog->stats);
- u64_stats_update_begin(&stats->syncp);
+ flags = u64_stats_update_begin_irqsave(&stats->syncp);
stats->cnt++;
stats->nsecs += sched_clock() - start;
- u64_stats_update_end(&stats->syncp);
+ u64_stats_update_end_irqrestore(&stats->syncp, flags);
}
}

--
2.33.0