| Message ID | 1668234485-27635-1-git-send-email-wangyufen@huawei.com (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | netdevsim: Fix memory leak of nsim_dev->fa_cookie | expand |
On Sat, 12 Nov 2022 14:28:05 +0800 Wang Yufen wrote: > nsim_dev_trap_fa_cookie_write() > kmalloc() fa_cookie > nsim_dev->fa_cookie = fa_cookie > .. > nsim_drv_remove() > > nsim_dev->fa_cookie alloced, but the nsim_dev_trap_report_work() > job has not been done, the flow action cookie has not been assigned > to the metadata. To fix, add kfree(nsim_dev->fa_cookie) to > nsim_drv_remove(). I don't see the path thru nsim_dev_trap_report_work() which would free the fa_cookie. The fix looks right, but the commit message seems incorrect. Isn't the leak always there, without any race?
在 2022/11/15 10:50, Jakub Kicinski 写道: > On Sat, 12 Nov 2022 14:28:05 +0800 Wang Yufen wrote: >> nsim_dev_trap_fa_cookie_write() >> kmalloc() fa_cookie >> nsim_dev->fa_cookie = fa_cookie >> .. >> nsim_drv_remove() >> >> nsim_dev->fa_cookie alloced, but the nsim_dev_trap_report_work() >> job has not been done, the flow action cookie has not been assigned >> to the metadata. To fix, add kfree(nsim_dev->fa_cookie) to >> nsim_drv_remove(). > I don't see the path thru nsim_dev_trap_report_work() which would free > the fa_cookie. > > The fix looks right, but the commit message seems incorrect. Isn't the > leak always there, without any race? Sorry, I didn't make it clear. The detailed process of nsim_dev_trap_report_work() is as follows: nsim_dev_trap_report_work() nsim_dev_trap_report_work() ... devlink_trap_report() devlink_trap_report_metadata_set() <-- fa_cookie is assigned to metadata->fa_cookie here， and will be freed in net_dm_hw_metadata_free()
On Tue, 15 Nov 2022 11:38:26 +0800 wangyufen wrote: > Sorry, I didn't make it clear. > > The detailed process of nsim_dev_trap_report_work() is as follows: > > nsim_dev_trap_report_work() > nsim_dev_trap_report_work() > ... > devlink_trap_report() > devlink_trap_report_metadata_set() > <-- fa_cookie is assigned to metadata->fa_cookie here， and will be freed in net_dm_hw_metadata_free() What's assigned here and freed in net_dm_hw_metadata_free() is a copy made with net_dm_hw_metadata_copy(), no? Could you double check the whole path?
diff --git a/drivers/net/netdevsim/dev.c b/drivers/net/netdevsim/dev.c index a7880c7..68e56e4 100644 --- a/drivers/net/netdevsim/dev.c +++ b/drivers/net/netdevsim/dev.c @@ -1683,6 +1683,7 @@ void nsim_drv_remove(struct nsim_bus_dev *nsim_bus_dev) ARRAY_SIZE(nsim_devlink_params)); devl_resources_unregister(devlink); kfree(nsim_dev->vfconfigs); + kfree(nsim_dev->fa_cookie); devl_unlock(devlink); devlink_free(devlink); dev_set_drvdata(&nsim_bus_dev->dev, NULL);
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() nsim_dev->fa_cookie alloced, but the nsim_dev_trap_report_work() job has not been done, the flow action cookie has not been assigned to the metadata. To fix, add kfree(nsim_dev->fa_cookie) to nsim_drv_remove(). Fixes: d3cbb907ae57 ("netdevsim: add ACL trap reporting cookie as a metadata") Signed-off-by: Wang Yufen <wangyufen@huawei.com> Cc: Jiri Pirko <jiri@mellanox.com> --- drivers/net/netdevsim/dev.c | 1 + 1 file changed, 1 insertion(+)