Re: [PATCH] scsi: ufs: core: Fix memory crash in case arpmb command failed

[Bean Huo <huobean@gmail.com>]

On Mon, 2025-02-17 at 18:43 +0200, Arthur Simchaev wrote:
> In case the device doesn't support arpmb, the kernel get memory crash
> due to copy user data in bsg_transport_sg_io_fn level. So in case
> ufshcd_send_bsg_uic_cmd returned error, do not change the job's
> reply_len.
> 
> Memory crash backtrace:
> 3,1290,531166405,-;ufshcd 0000:00:12.5: ARPMB OP failed: error code -
> 22
> 
> 4,1308,531166555,-;Call Trace:
> 
> 4,1309,531166559,-; <TASK>
> 
> 4,1310,531166565,-; ? show_regs+0x6d/0x80
> 
> 4,1311,531166575,-; ? die+0x37/0xa0
> 
> 4,1312,531166583,-; ? do_trap+0xd4/0xf0
> 
> 4,1313,531166593,-; ? do_error_trap+0x71/0xb0
> 
> 4,1314,531166601,-; ? usercopy_abort+0x6c/0x80
> 
> 4,1315,531166610,-; ? exc_invalid_op+0x52/0x80
> 
> 4,1316,531166622,-; ? usercopy_abort+0x6c/0x80
> 
> 4,1317,531166630,-; ? asm_exc_invalid_op+0x1b/0x20
> 
> 4,1318,531166643,-; ? usercopy_abort+0x6c/0x80
> 
> 4,1319,531166652,-; __check_heap_object+0xe3/0x120
> 
> 4,1320,531166661,-; check_heap_object+0x185/0x1d0
> 
> 4,1321,531166670,-; __check_object_size.part.0+0x72/0x150
> 
> 4,1322,531166679,-; __check_object_size+0x23/0x30
> 
> 4,1323,531166688,-; bsg_transport_sg_io_fn+0x314/0x3b0
> 
> Signed-off-by: Arthur Simchaev <arthur.simchaev@sandisk.com>
> ---
>  drivers/ufs/core/ufs_bsg.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/ufs/core/ufs_bsg.c b/drivers/ufs/core/ufs_bsg.c
> index 8d4ad0a3f2cf..a8ed9bc6e4f1 100644
> --- a/drivers/ufs/core/ufs_bsg.c
> +++ b/drivers/ufs/core/ufs_bsg.c
> @@ -194,10 +194,12 @@ static int ufs_bsg_request(struct bsg_job *job)
>         ufshcd_rpm_put_sync(hba);
>         kfree(buff);
>         bsg_reply->result = ret;
> -       job->reply_len = !rpmb ? sizeof(struct ufs_bsg_reply) :
> sizeof(struct ufs_rpmb_reply);
>         /* complete the job here only if no error */
> -       if (ret == 0)
> +       if (ret == 0) {
> +               job->reply_len = !rpmb ? sizeof(struct ufs_bsg_reply)
> :
> +                                        sizeof(struct
> ufs_rpmb_reply);
>                 bsg_job_done(job, ret, bsg_reply-
> >reply_payload_rcv_len);
> +       }
>  
>         return ret;
>  }


Arthur, 

The change appears logical because we only need to copy the payload
when the operation is successful. 

However, I don't fully understand how the memory crash could occur. If
the function in question is `ufshcd_send_bsg_uic_cmd`, it wouldn't
involve RPMB access, meaning `rpmb` would be `false`. In that case, the
size used would be `sizeof(struct ufs_bsg_reply)`, which has no
connection to the advanced RPMB functionality. 

Kind regards,
Bean