On 9/1/23 09:31, Mathieu Desnoyers wrote:
On 9/1/23 05:53, Denis Arefev wrote:
The value of an arithmetic expression 1 << (cpu - sdp->mynode->grplo)
is subject to overflow due to a failure to cast operands to a larger
data type before performing arithmetic
Found by Linux Verification Center (linuxtesting.org) with SVACE.
Signed-off-by: Denis Arefev <arefev@xxxxxxxxx>
---
kernel/rcu/srcutree.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/kernel/rcu/srcutree.c b/kernel/rcu/srcutree.c
index 20d7a238d675..e14b74fb1ba0 100644
--- a/kernel/rcu/srcutree.c
+++ b/kernel/rcu/srcutree.c
@@ -223,7 +223,7 @@ static bool init_srcu_struct_nodes(struct
srcu_struct *ssp, gfp_t gfp_flags)
snp->grplo = cpu;
snp->grphi = cpu;
}
- sdp->grpmask = 1 << (cpu - sdp->mynode->grplo);
+ sdp->grpmask = 1UL << (cpu - sdp->mynode->grplo);
What possible values of cpus supported by the Linux kernel and grplo can
cause this to overflow on 64-bit architectures ? I suspect the maximum
result of this subtraction is defined by the RCU_FANOUT or other srcu
level-spread values assigned by rcu_init_levelspread(), which can indeed
cause the signed 32-bit integer literal ("1") to overflow when shifted
by any value greater than 31. This analysis should be added to the
commit message so the impact of the issue can be understood.
Another effect that might be worth mentioning in the commit message is
what happens with the sign-carry for "1 << 31" when it is promoted from
signed 32-bit integer to unsigned long on a 64-bit kernel: it translates
to 0xffffffff80000000, which is certainly not what is expected here.
Thanks,
Mathieu
I also notice this in the same file:
srcu_schedule_cbs_snp():
for (cpu = snp->grplo; cpu <= snp->grphi; cpu++) {
if (!(mask & (1 << (cpu - snp->grplo))))
continue;
Which should be fixed at the same time.
Thanks,
Mathieu
}
smp_store_release(&ssp->srcu_sup->srcu_size_state,
SRCU_SIZE_WAIT_BARRIER);
return true;
--
Mathieu Desnoyers
EfficiOS Inc.
https://www.efficios.com
