Re: [PATCH v2 2/3] net: add support for nvmem to eth_platform_get_mac_address()

From: Russell King - ARM Linux
Date: Thu Jul 19 2018 - 08:24:19 EST


On Thu, Jul 19, 2018 at 11:48:37AM +0300, Dan Carpenter wrote:
> On Thu, Jul 19, 2018 at 10:20:27AM +0200, Bartosz Golaszewski wrote:
> > From: Bartosz Golaszewski <bgolaszewski@xxxxxxxxxxxx>
> >
> > Many non-DT platforms read the MAC address from EEPROM. Usually it's
> > either done with callbacks defined in board files or from SoC-specific
> > ethernet drivers.
> >
> > In order to generalize this, try to read the MAC from nvmem in
> > eth_platform_get_mac_address() using a standard lookup name:
> > "mac-address".
> >
> > Signed-off-by: Bartosz Golaszewski <bgolaszewski@xxxxxxxxxxxx>
> > ---
> > net/ethernet/eth.c | 27 +++++++++++++++++++++++++++
> > 1 file changed, 27 insertions(+)
> >
> > diff --git a/net/ethernet/eth.c b/net/ethernet/eth.c
> > index 39af03894598..af3b4b1b77eb 100644
> > --- a/net/ethernet/eth.c
> > +++ b/net/ethernet/eth.c
> > @@ -54,6 +54,7 @@
> > #include <linux/if_ether.h>
> > #include <linux/of_net.h>
> > #include <linux/pci.h>
> > +#include <linux/nvmem-consumer.h>
> > #include <net/dst.h>
> > #include <net/arp.h>
> > #include <net/sock.h>
> > @@ -527,8 +528,11 @@ unsigned char * __weak arch_get_platform_mac_address(void)
> >
> > int eth_platform_get_mac_address(struct device *dev, u8 *mac_addr)
> > {
> > + unsigned char addrbuf[ETH_ALEN];
> > const unsigned char *addr;
> > + struct nvmem_cell *nvmem;
> > struct device_node *dp;
> > + size_t alen;
> >
> > if (dev_is_pci(dev))
> > dp = pci_device_to_OF_node(to_pci_dev(dev));
> > @@ -541,6 +545,29 @@ int eth_platform_get_mac_address(struct device *dev, u8 *mac_addr)
> > if (!addr)
> > addr = arch_get_platform_mac_address();
> >
> > + if (!addr) {
> > + nvmem = nvmem_cell_get(dev, "mac-address");
> > + if (IS_ERR(nvmem) && PTR_ERR(nvmem) == -EPROBE_DEFER)
> > + /* We may have a lookup registered for MAC address but
> > + * the corresponding nvmem provider hasn't been
> > + * registered yet.
> > + */
> > + return -EPROBE_DEFER;
> > +
> > + if (!IS_ERR(nvmem)) {
> > + addr = nvmem_cell_read(nvmem, &alen);
> > + if (!IS_ERR(addr)) {
> ^^^^
> Never do success handling. Always error handling. Otherwise the code
> is indent a lot and the error handling is far from the call.
>
> > + if (alen == ETH_ALEN)
> > + ether_addr_copy(addrbuf, addr);
> > +
> > + kfree(addr);
> > + addr = alen == ETH_ALEN ? addrbuf : NULL;
> > + }
> > +
> > + nvmem_cell_put(nvmem);
> > + }
> > + }
> > +
> > if (!addr || !is_valid_ether_addr(addr))
> ^^^^
> Instead of handling the error we dereference the error pointer here.
>
> *frowny face*
>
> > return -ENODEV;
> >
> > --
>
> Maybe this?
>
> if (!addr) {
> nvmem = nvmem_cell_get(dev, "mac-address");
> if (PTR_ERR(nvmem) == -EPROBE_DEFER)
> return -EPROBE_DEFER;
> if (IS_ERR(nvmem))
> return -ENODEV;
> addr = nvmem_cell_read(nvmem, &alen);
> if (IS_ERR(addr))
> return PTR_ERR(addr);

The problem with doing it this way is... error handling is Hard(tm).
You missed the call to nvmem_cell_put() here.

> if (alen != ETH_ALEN) {
> kfree(addr);

and again here.

> return -ENODEV;
> }
> ether_addr_copy(addrbuf, addr);
> kfree(addr);
> addr = addrbuf;

and here.

> }
> if (!is_valid_ether_addr(addr))
> return -ENODEV;
> ether_addr_copy(mac_addr, addr);
> return 0;

Without checking the semantics, a possible solution to that could be:

if (!addr) {
nvmem = nvmem_cell_get(dev, "mac-address");
if (PTR_ERR(nvmem) == -EPROBE_DEFER)
return -EPROBE_DEFER;
if (IS_ERR(nvmem))
return -ENODEV;
addr = nvmem_cell_read(nvmem, &alen);
+ nvmem_cell_put(nvmem);
if (IS_ERR(addr))
return PTR_ERR(addr);
if (alen != ETH_ALEN) {
kfree(addr);
return -ENODEV;
}
ether_addr_copy(addrbuf, addr);
kfree(addr);
addr = addrbuf;
}
if (!is_valid_ether_addr(addr))
return -ENODEV;
ether_addr_copy(mac_addr, addr);
return 0;

A potential better solution would be to put this code in a separate
function, and then do:

if (!addr)
addr = eth_platform_get_nvmem_mac_address(dev, addrbuf);

which returns either NULL or addrbuf depending on whether it failed or
succeeded, which would probably end up with cleaner code.

--
RMK's Patch system: http://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line in suburbia: sync at 13.8Mbps down 630kbps up
According to speedtest.net: 13Mbps down 490kbps up