NULL Pointer Dereference Affecting perf package, versions <0:4.18.0-477.10.1.el8_8
Threat Intelligence
Do your applications use this vulnerable package?
In a few clicks we can analyze your entire application and see what components are vulnerable in your application, and suggest you quick fixes.
Test your applications- Snyk ID SNYK-RHEL8-PERF-8167538
- published 8 Oct 2024
- disclosed 19 Jun 2024
Introduced: 19 Jun 2024
CVE-2021-47592 Open this link in a new tabHow to fix?
Upgrade RHEL:8
perf
to version 0:4.18.0-477.10.1.el8_8 or higher.
This issue was patched in RHSA-2023:2951
.
NVD Description
Note: Versions mentioned in the description apply only to the upstream perf
package and not the perf
package as distributed by RHEL
.
See How to fix?
for RHEL:8
relevant fixed versions and status.
In the Linux kernel, the following vulnerability has been resolved:
net: stmmac: fix tc flower deletion for VLAN priority Rx steering
To replicate the issue:-
Add 1 flower filter for VLAN Priority based frame steering:- $ IFDEVNAME=eth0 $ tc qdisc add dev $IFDEVNAME ingress $ tc qdisc add dev $IFDEVNAME root mqprio num_tc 8
map 0 1 2 3 4 5 6 7 0 0 0 0 0 0 0 0
queues 1@0 1@1 1@2 1@3 1@4 1@5 1@6 1@7 hw 0 $ tc filter add dev $IFDEVNAME parent ffff: protocol 802.1Q
flower vlan_prio 0 hw_tc 0Get the 'pref' id $ tc filter show dev $IFDEVNAME ingress
Delete a specific tc flower record (say pref 49151) $ tc filter del dev $IFDEVNAME parent ffff: pref 49151
From dmesg, we will observe kernel NULL pointer ooops
[ 197.170464] BUG: kernel NULL pointer dereference, address: 0000000000000000 [ 197.171367] #PF: supervisor read access in kernel mode [ 197.171367] #PF: error_code(0x0000) - not-present page [ 197.171367] PGD 0 P4D 0 [ 197.171367] Oops: 0000 [#1] PREEMPT SMP NOPTI
<snip>
[ 197.171367] RIP: 0010:tc_setup_cls+0x20b/0x4a0 [stmmac]
<snip>
[ 197.171367] Call Trace: [ 197.171367] <TASK> [ 197.171367] ? __stmmac_disable_all_queues+0xa8/0xe0 [stmmac] [ 197.171367] stmmac_setup_tc_block_cb+0x70/0x110 [stmmac] [ 197.171367] tc_setup_cb_destroy+0xb3/0x180 [ 197.171367] fl_hw_destroy_filter+0x94/0xc0 [cls_flower]
The above issue is due to previous incorrect implementation of tc_del_vlan_flow(), shown below, that uses flow_cls_offload_flow_rule() to get struct flow_rule *rule which is no longer valid for tc filter delete operation.
struct flow_rule *rule = flow_cls_offload_flow_rule(cls); struct flow_dissector *dissector = rule->match.dissector;
So, to ensure tc_del_vlan_flow() deletes the right VLAN cls record for earlier configured RX queue (configured by hw_tc) in tc_add_vlan_flow(), this patch introduces stmmac_rfs_entry as driver-side flow_cls_offload record for 'RX frame steering' tc flower, currently used for VLAN priority. The implementation has taken consideration for future extension to include other type RX frame steering such as EtherType based.
v2:
- Clean up overly extensive backtrace and rewrite git message to better explain the kernel NULL pointer issue.