[LTS 9.2] netdevsim: Fix memory leak of nsim_dev->fa_cookie #340

pvts-mat · 2025-06-13T20:31:21Z

[LTS 9.2]
CVE-2022-49803
VULN-65792

Problem

https://www.cve.org/CVERecord?id=CVE-2022-49803

In the Linux kernel, the following vulnerability has been resolved

netdevsim: Fix memory leak of nsim_dev->fa_cookie

kmemleak reports this issue:

unreferenced object 0xffff8881bac872d0 (size 8):
comm "sh", pid 58603, jiffies 4481524462 (age 68.065s)
hex dump (first 8 bytes):
04 00 00 00 de ad be ef ……..
backtrace:
[<00000000c80b8577>] __kmalloc+0x49/0x150
[<000000005292b8c6>] nsim_dev_trap_fa_cookie_write+0xc1/0x210 [netdevsim]
[<0000000093d78e77>] full_proxy_write+0xf3/0x180
[<000000005a662c16>] vfs_write+0x1c5/0xaf0
[<000000007aabf84a>] ksys_write+0xed/0x1c0
[<000000005f1d2e47>] do_syscall_64+0x3b/0x90
[<000000006001c6ec>] entry_SYSCALL_64_after_hwframe+0x63/0xcd

Background

The netdevsim is a module simulating a network device in Linux. It's a developer testing tool that can be used to test various control path networking APIs.

The memory leak described in CVE occurs after unloading the module, provided that some data was written through debugfs to the "flow action cookie" metadata storage associated with the simulated device.

This CVE appears to be one of the retroactive CVEs, filed for bugs which were fixed long before (fix commit 064bc73 date 2022-11-16 vs CVE publication date 2025-05-01).

Applicability

The netdevsim module is enabled by the CONFIG_NETDEVSIM option which is m in all configurations of LTS 9.2:

$ grep CONFIG_NETDEVSIM configs/kernel-*.config

configs/kernel-aarch64-64k-debug-rhel.config:CONFIG_NETDEVSIM=m
configs/kernel-aarch64-64k-rhel.config:CONFIG_NETDEVSIM=m
configs/kernel-aarch64-debug-rhel.config:CONFIG_NETDEVSIM=m
configs/kernel-aarch64-rhel.config:CONFIG_NETDEVSIM=m
configs/kernel-ppc64le-debug-rhel.config:CONFIG_NETDEVSIM=m
configs/kernel-ppc64le-rhel.config:CONFIG_NETDEVSIM=m
configs/kernel-s390x-debug-rhel.config:CONFIG_NETDEVSIM=m
configs/kernel-s390x-rhel.config:CONFIG_NETDEVSIM=m
configs/kernel-s390x-zfcpdump-rhel.config:CONFIG_NETDEVSIM=m
configs/kernel-x86_64-debug-rhel.config:CONFIG_NETDEVSIM=m
configs/kernel-x86_64-rhel.config:CONFIG_NETDEVSIM=m

The d3cbb90 commit identified in the fix commit 064bc73 as introducing the bug is present in ciqlts9_2 history, literally

$ git --no-pager branch ciqlts9_2 --contains d3cbb907ae57fe5da314b51d949b617b538bdeae

* ciqlts9_2

The fix commit 064bc73 (1) is not present in the history and neither it's backported.

$ cve-research/git-analysis.xsh \
    histories \
    -C …/kernel-src-tree \
    --file drivers/net/netdevsim/dev.c \
    --log-opts-main='--since=2022-03-18 --date=short' \
    --ref-opts='--no-merges' \
    kernel-mainline ciqlts9_2

kernel-mainline                                                         ciqlts9_2
----------------------------------------------------------------------  ----------------------
a1494d532 2024-10-15 netdevsim: use cond_resched() in nsim_dev_trap_r…
ba5e12721 2024-02-02 netdevsim: avoid potential loop in nsim_dev_trap…
f72207a5c 2023-07-12 netdevsim: fix uninitialized data in nsim_dev_tr…
fb8421a94 2023-01-30 devlink: remove devlink features                
075935f0a 2023-01-27 devlink: protect devlink param list by instance …
82a3aef2e 2023-01-06 netdevsim: move devlink registration under the i…
5c5ea1d09 2023-01-06 netdevsim: rename a label                       
226bf9805 2022-11-30 net: devlink: let the core report the driver nam…
f2fc15e27 2022-11-17 devlink: Allow to set up parent in devl_rate_lea…
064bc7312 2022-11-16 netdevsim: Fix memory leak of nsim_dev->fa_cooki…                         (1)
ac73d4bf2 2022-11-03 net: make drivers to use SET_NETDEV_DEVLINK_PORT…
a6aa8d0ce 2022-10-27 netdevsim: remove dir in nsim_dev_debugfs_init()…
6b1da9f71 2022-10-27 netdevsim: fix memory leak in nsim_drv_probe() w…
5603072e0 2022-09-01 netdevsim: remove redundant variable ret        
f94b60632 2022-08-25 net: devlink: limit flash component name to matc…  ~ ff020891f 2023-01-18
0c1989754 2022-08-25 netdevsim: add version fw.mgmt info info_get() a…  ~ fbb5e9da6 2023-01-18
644a66c60 2022-08-01 net: devlink: convert reload command to take imp…  ~ 31a499f90 2023-01-18
012ec02ae 2022-07-18 netdevsim: convert driver to use unlocked devlin…  ~ 57497ee6e 2023-01-18
14e426bf1 2022-03-21 devlink: hold the instance lock during eswitch_m…  ~ 24b38c8b0 2022-08-22
aff3a9250 2022-03-21 netdevsim: replace vfs_lock with devlink instanc…  ~ 6742a09c4 2022-08-22
76eea6c2e 2022-03-21 netdevsim: replace port_list_lock with devlink i…  ~ 6d57a616e 2022-08-22

The bug's replication on ciqlts9_2 confirms the applicability.

Solution

As mentioned before, the mainline fix is in the 064bc73 commit. It applies with cherry-pick to ciqlts9_2 without modifications.

kABI check: passed

$ CVE=CVE-2022-49803 ./ninja.sh _kabi_checked__x86_64--test--ciqlts9_2-CVE-2022-49803

…
[4/5] Check ABI of kernel [ciqlts9_2-CVE-2022-49803]
++ uname -m
+ python3 /data/src/ctrliq-github/kernel-dist-git-el-9.2/SOURCES/check-kabi -k /data/src/ctrliq-github/kernel-dist-git-el-9.2/SOURCES/Module.kabi_x86_64 -s vms/x86_64--build--ciqlts9_2/build_files/kernel-src-tree-ciqlts9_2-CVE-2022-49803/Module.symvers
kABI check passed
+ touch state/kernels/ciqlts9_2-CVE-2022-49803/x86_64/kabi_checked

Boot test: passed

boot-test.log

Kselftests: passed relative

Coverage

All net-related tests (except the unstable ones):

net/forwarding (except mirror_gre_bridge_1d_vlan.sh, ipip_hier_gre_keys.sh, sch_ets.sh, sch_tbf_prio.sh, sch_tbf_ets.sh, q_in_vni.sh, tc_police.sh, mirror_gre_vlan_bridge_1q.sh, dual_vxlan_bridge.sh, tc_actions.sh, sch_tbf_root.sh, sch_red.sh, vxlan_bridge_1d_ipv6.sh), net/mptcp (except simult_flows.sh, userspace_pm.sh), net (except gro.sh, txtimestamp.sh, xfrm_policy.sh, udpgro_fwd.sh, fib_nexthops.sh, udpgso_bench.sh, reuseport_addr_any.sh, reuseaddr_conflict, ip_defrag.sh), netfilter (except nft_trans_stress.sh)

Reference

kselftests–ciqlts9_2–run1.log
kselftests–ciqlts9_2–run2.log
kselftests–ciqlts9_2–run3.log

Patch

kselftests–ciqlts9_2-CVE-2022-49803–run1.log
kselftests–ciqlts9_2-CVE-2022-49803–run2.log
kselftests–ciqlts9_2-CVE-2022-49803–run3.log

Comparison

The results for the reference and patched kernel are the same.

$ ktests.xsh diff -d kselftests*.log

Column    File
--------  ----------------------------------------------
Status0   kselftests--ciqlts9_2--run1.log
Status1   kselftests--ciqlts9_2--run2.log
Status2   kselftests--ciqlts9_2--run3.log
Status3   kselftests--ciqlts9_2-CVE-2022-49803--run1.log
Status4   kselftests--ciqlts9_2-CVE-2022-49803--run2.log
Status5   kselftests--ciqlts9_2-CVE-2022-49803--run3.log

Specific tests: passed

The bug replication steps are hinted in the fix commit 064bc73 message:

The issue occurs in the following scenarios:

nsim_dev_trap_fa_cookie_write()
  kmalloc() fa_cookie
  nsim_dev->fa_cookie = fa_cookie
..
nsim_drv_remove()

Reference

The bug was successfully replicated on ciqlts9_2 compiled with the CONFIG_DEBUG_KMEMLEAK option set. The steps:

Load the netdevsim module
```
# modprobe netdevsim
```

Create a new netdevsim device with index 0

# echo "0 1 1" > /sys/bus/netdevsim/new_device

Write "abc" encoded in hex to the flow action cookie

# echo "616263" > /sys/kernel/debug/netdevsim/netdevsim0/trap_flow_action_cookie

Remove the device

# echo 0 > /sys/bus/netdevsim/del_device

Unload the netdevsim module
```
# rmmod netdevsim
```

Trigger memory leaks scan

# echo scan > /sys/kernel/debug/kmemleak

Check the leaks report

# cat /sys/kernel/debug/kmemleak

unreferenced object 0xffff954a00cd54e8 (size 8):
  comm "bash", pid 2044, jiffies 4354161471 (age 215.438s)
  hex dump (first 8 bytes):
    03 00 00 00 61 62 63 d3                          ....abc.
  backtrace:
    [<00000000f3b0c80b>] nsim_dev_trap_fa_cookie_write+0x83/0x130 [netdevsim]
    [<00000000cdd405a5>] full_proxy_write+0x50/0x80
    [<00000000349ec0c3>] vfs_write+0xb2/0x280
    [<000000003290ff1e>] ksys_write+0x5f/0xe0
    [<000000005b589185>] do_syscall_64+0x59/0x90
    [<000000009081b09f>] entry_SYSCALL_64_after_hwframe+0x63/0xcd

Patch

The same steps as done on the reference kernel produced no memory leak logs on the patched kernel.

[root@ciqlts-9-2 pvts]# modprobe netdevsim
[root@ciqlts-9-2 pvts]# echo "0 1 1" > /sys/bus/netdevsim/new_device
[root@ciqlts-9-2 pvts]# echo "616263" > /sys/kernel/debug/netdevsim/netdevsim0/trap_flow_action_cookie
[root@ciqlts-9-2 pvts]# echo 0 > /sys/bus/netdevsim/del_device
[root@ciqlts-9-2 pvts]# rmmod netdevsim
[root@ciqlts-9-2 pvts]# echo scan > /sys/kernel/debug/kmemleak
[root@ciqlts-9-2 pvts]# cat /sys/kernel/debug/kmemleak
[root@ciqlts-9-2 pvts]#

jira VULN-65792 cve CVE-2022-49803 commit-author Wang Yufen <wangyufen@huawei.com> commit 064bc73 kmemleak reports this issue: unreferenced object 0xffff8881bac872d0 (size 8): comm "sh", pid 58603, jiffies 4481524462 (age 68.065s) hex dump (first 8 bytes): 04 00 00 00 de ad be ef ........ backtrace: [<00000000c80b8577>] __kmalloc+0x49/0x150 [<000000005292b8c6>] nsim_dev_trap_fa_cookie_write+0xc1/0x210 [netdevsim] [<0000000093d78e77>] full_proxy_write+0xf3/0x180 [<000000005a662c16>] vfs_write+0x1c5/0xaf0 [<000000007aabf84a>] ksys_write+0xed/0x1c0 [<000000005f1d2e47>] do_syscall_64+0x3b/0x90 [<000000006001c6ec>] entry_SYSCALL_64_after_hwframe+0x63/0xcd The issue occurs in the following scenarios: nsim_dev_trap_fa_cookie_write() kmalloc() fa_cookie nsim_dev->fa_cookie = fa_cookie .. nsim_drv_remove() The fa_cookie allocked in nsim_dev_trap_fa_cookie_write() is not freed. To fix, add kfree(nsim_dev->fa_cookie) to nsim_drv_remove(). Fixes: d3cbb90 ("netdevsim: add ACL trap reporting cookie as a metadata") Signed-off-by: Wang Yufen <wangyufen@huawei.com> Cc: Jiri Pirko <jiri@mellanox.com> Link: https://lore.kernel.org/r/1668504625-14698-1-git-send-email-wangyufen@huawei.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> (cherry picked from commit 064bc73) Signed-off-by: Marcin Wcisło <marcin.wcislo@conclusive.pl>

bmastbergen

🥌

thefossguy-ciq

🚤

pvts-mat changed the title ~~netdevsim: Fix memory leak of nsim_dev->fa_cookie~~ [LTS 9.2] netdevsim: Fix memory leak of nsim_dev->fa_cookie Jun 13, 2025

This was referenced Jun 13, 2025

[LTS 8.8] netdevsim: Fix memory leak of nsim_dev->fa_cookie #341

Open

[LTS 8.6] netdevsim: Fix memory leak of nsim_dev->fa_cookie #342

Open

bmastbergen requested review from kerneltoast, jainanmol84, PlaidCat, bmastbergen and thefossguy-ciq June 16, 2025 15:31

bmastbergen approved these changes Jun 16, 2025

View reviewed changes

thefossguy-ciq approved these changes Jun 16, 2025

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[LTS 9.2] netdevsim: Fix memory leak of nsim_dev->fa_cookie #340

[LTS 9.2] netdevsim: Fix memory leak of nsim_dev->fa_cookie #340

Uh oh!

pvts-mat commented Jun 13, 2025 •

edited

Loading

Uh oh!

bmastbergen left a comment

Uh oh!

thefossguy-ciq left a comment

Uh oh!

Uh oh!

[LTS 9.2] netdevsim: Fix memory leak of nsim_dev->fa_cookie #340

Are you sure you want to change the base?

[LTS 9.2] netdevsim: Fix memory leak of nsim_dev->fa_cookie #340

Uh oh!

Conversation

pvts-mat commented Jun 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Problem

Background

Applicability

Solution

kABI check: passed

Boot test: passed

Kselftests: passed relative

Coverage

Reference

Patch

Comparison

Specific tests: passed

Reference

Patch

Uh oh!

bmastbergen left a comment

Choose a reason for hiding this comment

Uh oh!

thefossguy-ciq left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

pvts-mat commented Jun 13, 2025 •

edited

Loading