Hi Richard,
richardretanubun wrote:
Allow uec_init to run more than once, based on the netretry environment variable. This allows for manual (back and forth) switching between network interfaces.
My general issue with this patch is that if this functionality is really useful, it should be in the main device control flow (net/eth.c), not in a single driver.
Signed-off-by: Richard Retanubun <RichardRetanubun_at_ruggedcom.com>
diff --git a/drivers/qe/uec.c b/drivers/qe/uec.c index 344c649..88402ca 100644 --- a/drivers/qe/uec.c +++ b/drivers/qe/uec.c @@ -1192,9 +1192,17 @@ static int uec_init(struct eth_device* dev, bd_t *bd) uec_private_t *uec; int err, i; struct phy_info *curphy;
char *netretry = NULL;
uec = (uec_private_t *)dev->priv;
/* Allow for net re-initialization based on netretry environment
setting */
- netretry = getenv("netretry");
- if (netretry != NULL && (strcmp(netretry, "yes") == 0)) {
uec->the_first_run = 0;- }
I'm sure you know that "netretry" already exists, and has a different meaning. It's not a good idea for an environment variable to do different things depending on context.
if (uec->the_first_run == 0) { err = init_phy(dev); if (err) {@@ -1223,12 +1231,16 @@ static int uec_init(struct eth_device* dev, bd_t *bd) if (err || i <= 0) printf("warning: %s: timeout on PHY link\n", dev->name);
uec->the_first_run = 1;
/* If netretry is not set, or not set to yes, assume no retry */netretry = getenv("netretry");if (netretry == NULL || (strcmp(netretry, "yes") != 0)) {uec->the_first_run = 1;}}
/* Set up the MAC address */ if (dev->enetaddr[0] & 0x01) {
printf("%s: MacAddress is multcast address\n",
printf("%s: MacAddress is multicast address\n", __FUNCTION__);
OK, good catch.
Sorry again for taking so long to respond properly.
regards, Ben