Re: [PATCH net-next 3/3] net: ethernet: ti: cpsw: add XDP support

[Dan Carpenter]

Hi Ivan,

Thank you for the patch! Perhaps something to improve:

url:    https://github.com/0day-ci/linux/commits/Ivan-Khoronzhuk/net-ethernet-ti-cpsw-Add-XDP-support/20190524-114123

If you fix the issue, kindly add following tag
Reported-by: kbuild test robot <lkp@xxxxxxxxx>
Reported-by: Dan Carpenter <dan.carpenter@xxxxxxxxxx>

smatch warnings:
drivers/net/ethernet/ti/cpsw_ethtool.c:564 cpsw_xdp_rxq_reg() error: uninitialized symbol 'ret'.

# https://github.com/0day-ci/linux/commit/3cf4eb125ed19d18340fd3b0c4d7eb2f1ebdfb28
git remote add linux-review https://github.com/0day-ci/linux
git remote update linux-review
git checkout 3cf4eb125ed19d18340fd3b0c4d7eb2f1ebdfb28
vim +/ret +564 drivers/net/ethernet/ti/cpsw_ethtool.c

c24eef28 Grygorii Strashko 2019-04-26  534  
3cf4eb12 Ivan Khoronzhuk   2019-05-23  535  static int cpsw_xdp_rxq_reg(struct cpsw_common *cpsw, int ch)
3cf4eb12 Ivan Khoronzhuk   2019-05-23  536  {
3cf4eb12 Ivan Khoronzhuk   2019-05-23  537  	struct cpsw_slave *slave;
3cf4eb12 Ivan Khoronzhuk   2019-05-23  538  	struct cpsw_priv *priv;
3cf4eb12 Ivan Khoronzhuk   2019-05-23  539  	int i, ret;
3cf4eb12 Ivan Khoronzhuk   2019-05-23  540  
3cf4eb12 Ivan Khoronzhuk   2019-05-23  541  	/* As channels are common for both ports sharing same queues, xdp_rxq
3cf4eb12 Ivan Khoronzhuk   2019-05-23  542  	 * information also becomes shared and used by every packet on this
3cf4eb12 Ivan Khoronzhuk   2019-05-23  543  	 * channel. But exch xdp_rxq holds link on netdev, which by the theory
3cf4eb12 Ivan Khoronzhuk   2019-05-23  544  	 * can have different memory model and so, network device must hold it's
3cf4eb12 Ivan Khoronzhuk   2019-05-23  545  	 * own set of rxq and thus both netdevs should be prepared
3cf4eb12 Ivan Khoronzhuk   2019-05-23  546  	 */
3cf4eb12 Ivan Khoronzhuk   2019-05-23  547  	for (i = cpsw->data.slaves, slave = cpsw->slaves; i; i--, slave++) {
3cf4eb12 Ivan Khoronzhuk   2019-05-23  548  		if (!slave->ndev)
3cf4eb12 Ivan Khoronzhuk   2019-05-23  549  			continue;

Smatch always complains that every loop iteration could continue.  Or
that cpsw->data.slaves might be zero at the start...  It seems
implausible.

3cf4eb12 Ivan Khoronzhuk   2019-05-23  550  
3cf4eb12 Ivan Khoronzhuk   2019-05-23  551  		priv = netdev_priv(slave->ndev);
3cf4eb12 Ivan Khoronzhuk   2019-05-23  552  
3cf4eb12 Ivan Khoronzhuk   2019-05-23  553  		ret = xdp_rxq_info_reg(&priv->xdp_rxq[ch], priv->ndev, ch);
3cf4eb12 Ivan Khoronzhuk   2019-05-23  554  		if (ret)
3cf4eb12 Ivan Khoronzhuk   2019-05-23  555  			goto err;
3cf4eb12 Ivan Khoronzhuk   2019-05-23  556  
3cf4eb12 Ivan Khoronzhuk   2019-05-23  557  		ret = xdp_rxq_info_reg_mem_model(&priv->xdp_rxq[ch],
3cf4eb12 Ivan Khoronzhuk   2019-05-23  558  						 MEM_TYPE_PAGE_POOL,
3cf4eb12 Ivan Khoronzhuk   2019-05-23  559  						 cpsw->rx_page_pool);
3cf4eb12 Ivan Khoronzhuk   2019-05-23  560  		if (ret)
3cf4eb12 Ivan Khoronzhuk   2019-05-23  561  			goto err;
3cf4eb12 Ivan Khoronzhuk   2019-05-23  562  	}
3cf4eb12 Ivan Khoronzhuk   2019-05-23  563  
3cf4eb12 Ivan Khoronzhuk   2019-05-23 @564  	return ret;

This would be more readable as "return 0;" anyway.

3cf4eb12 Ivan Khoronzhuk   2019-05-23  565  
3cf4eb12 Ivan Khoronzhuk   2019-05-23  566  err:
3cf4eb12 Ivan Khoronzhuk   2019-05-23  567  	cpsw_xdp_rxq_unreg(cpsw, ch);
3cf4eb12 Ivan Khoronzhuk   2019-05-23  568  	return ret;
3cf4eb12 Ivan Khoronzhuk   2019-05-23  569  }

---
0-DAY kernel test infrastructure                Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all                   Intel Corporation