Re: [-next] scsi: iscsi: fix possible memory leak in iscsi_register_transport

[Zhouguanghui <zhouguanghui1@huawei.com>]

On 2022/11/10 3:36, Mike Christie wrote:
> On 11/9/22 2:19 AM, Zhou Guanghui wrote:
>> "unreferenced object 0xffff888117908420 (size 16):
>>    comm ""modprobe"", pid 18125, jiffies 4319017437 (age 73.039s)
>>    hex dump (first 16 bytes):
>>      62 65 32 69 73 63 73 69 00 84 90 17 81 88 ff ff  be2iscsi........
>>    backtrace:
>>      [<00000000f78a13b3>] __kmem_cache_alloc_node+0x157/0x220
>>      [<00000000200a51a4>] __kmalloc_node_track_caller+0x44/0x1b0
>>      [<0000000033ea4d64>] kstrdup+0x3a/0x70
>>      [<00000000ec6d2980>] kstrdup_const+0x41/0x60
>>      [<0000000055015f6f>] kvasprintf_const+0xf5/0x180
>>      [<000000009dd443d2>] kobject_set_name_vargs+0x56/0x150
>>      [<00000000f3448e98>] dev_set_name+0xab/0xe0
>>      [<0000000080ab8992>] iscsi_register_transport+0x1f8/0x610 [scsi_transport_iscsi]
>>      [<000000005e2c324d>] 0xffffffffc1260012
>>      [<00000000df6e6a36>] do_one_initcall+0xcb/0x4d0
>>      [<00000000181109df>] do_init_module+0x1ca/0x5f0
>>      [<00000000b3c4fec8>] load_module+0x6133/0x70f0
>>      [<00000000feb08394>] __do_sys_finit_module+0x12f/0x1c0
>>      [<00000000ca6af44d>] do_syscall_64+0x37/0x90
>>      [<00000000132e1a8b>] entry_SYSCALL_64_after_hwframe+0x63/0xcd"
>>
>> If device_register() returns error in iscsi_register_transport(),
>> the name allocated by the dev_set_name() need be freed.
>>
>> Fix this by calling put_device(), the name will be freed in the
>> kobject_cleanup(), and the priv will be freed in
>> iscsi_transport_release.
>>
>> Signed-off-by: Zhou Guanghui <zhouguanghui1@huawei.com>
>> ---
>>   drivers/scsi/scsi_transport_iscsi.c | 6 +++---
>>   1 file changed, 3 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/scsi/scsi_transport_iscsi.c b/drivers/scsi/scsi_transport_iscsi.c
>> index cd3db9684e52..51e2c0f5e2d0 100644
>> --- a/drivers/scsi/scsi_transport_iscsi.c
>> +++ b/drivers/scsi/scsi_transport_iscsi.c
>> @@ -4815,7 +4815,7 @@ iscsi_register_transport(struct iscsi_transport *tt)
>>   	dev_set_name(&priv->dev, "%s", tt->name);
>>   	err = device_register(&priv->dev);
>>   	if (err)
>> -		goto free_priv;
>> +		goto put_dev;
>>   
>>   	err = sysfs_create_group(&priv->dev.kobj, &iscsi_transport_group);
>>   	if (err)
>> @@ -4850,8 +4850,8 @@ iscsi_register_transport(struct iscsi_transport *tt)
>>   unregister_dev:
>>   	device_unregister(&priv->dev);
>>   	return NULL;
>> -free_priv:
>> -	kfree(priv);
>> +put_dev:
>> +	put_device(&priv->dev);
>>   	return NULL;
>>   }
>>   EXPORT_SYMBOL_GPL(iscsi_register_transport);
> 
> Reviewed-by: Mike Christie <michael.christie@oracle.com>
> 
> Shoot, I see the comment about using put_device in device_add.
> I'm not sure what happened, but I made the same mistake above
> in 4 other places.
> 
> Do you want to send patches for the other ones? If not, I'll do
> it.
> 

Mike，thanks.

I also found 4 other places that have the same mistake. I'll be sending 
a patch v2 soon.