Pull infiniband fixes from Roland Dreier:
"Fix a regression in the /sys/class/infiniband/.../rate attribute --
old kernels used to just return something, even if the underlying
value was out-of-bounds, while 3.4-rc1 returned EINVAL to userspace.
This breaks some applications that check for the error, so go back to
the old behavior."
* tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
IB/core: Don't return EINVAL from sysfs rate attribute for invalid speeds
IB/mlx4: Don't return an invalid speed when a port is down
{
struct ib_port_attr attr;
char *speed = "";
- int rate = -1; /* in deci-Gb/sec */
+ int rate; /* in deci-Gb/sec */
ssize_t ret;
ret = ib_query_port(p->ibdev, p->port_num, &attr);
return ret;
switch (attr.active_speed) {
- case IB_SPEED_SDR:
- rate = 25;
- break;
case IB_SPEED_DDR:
speed = " DDR";
rate = 50;
speed = " EDR";
rate = 250;
break;
+ case IB_SPEED_SDR:
+ default: /* default to SDR for invalid rates */
+ rate = 25;
+ break;
}
rate *= ib_width_enum_to_int(attr.active_width);
if (out_mad->data[15] & 0x1)
props->active_speed = IB_SPEED_FDR10;
}
+
+ /* Avoid wrong speed value returned by FW if the IB link is down. */
+ if (props->state == IB_PORT_DOWN)
+ props->active_speed = IB_SPEED_SDR;
+
out:
kfree(in_mad);
kfree(out_mad);