[ofa-general] PATCH] IB/core: ignore membership bit when looking for a P_Key in the table

[Or Gerlitz]

Roland Dreier wrote:
> Neither really... just that the double call to ib_find_pkey() in ib_sa
> is not needed after this change.

I am not following you, looking on the 2.6.22 tree, I don't see any 
calls from the ib_sa code to look for a pkey:

> # grep -rn ib_find . | grep pkey
> ./ulp/srp/ib_srp.c:167: ret = ib_find_cached_pkey(target->srp_host->dev->dev,
> ./ulp/ipoib/ipoib_verbs.c:50:   if (ib_find_pkey(priv->ca, priv->port, priv->pkey, &pkey_index)) {
> ./ulp/ipoib/ipoib_cm.c:868:     ret = ib_find_cached_pkey(priv->ca, priv->port, priv->pkey, &qp_attr.pkey_index);
> ./ulp/ipoib/ipoib_ib.c:451:     if (ib_find_pkey(priv->ca, priv->port, priv->pkey, &priv->pkey_index)) {
> ./ulp/ipoib/ipoib_ib.c:491:     if (ib_find_cached_pkey(priv->ca, priv->port, priv->pkey, &pkey_index))
> ./ulp/ipoib/ipoib_ib.c:719:             if (ib_find_pkey(priv->ca, priv->port, priv->pkey, &new_index)) {
> ./core/cm.c:288:        ret = ib_find_cached_pkey(cm_dev->device, port->port_num,
> ./core/cma.c:538:       ret = ib_find_cached_pkey(id_priv->id.device, id_priv->id.port_num,
> ./core/device.c:687: * ib_find_pkey - Returns the PKey table index where a specified
> ./core/device.c:694:int ib_find_pkey(struct ib_device *device,
> ./core/device.c:713:EXPORT_SYMBOL(ib_find_pkey);
> ./core/cache.c:163:int ib_find_cached_pkey(struct ib_device *device,
> ./core/cache.c:193:EXPORT_SYMBOL(ib_find_cached_pkey);


> Never mind, I'll fix it up myself.

thanks. As I said before this has to go to -stable as the bug is present 
in 2.6.22, please let us know if you will do the push or you prefer that 
we will do it.

Or.