Currently driver returns EAGAIN as error code if vdev transition is
not possible, driver is not able to process this vdev transition
because some other PSOC/PDEV transition is going on.
On receiving EAGAIN error code userspace retries the same operation
immediately which again leads to failure as existing transition is
still in progress.
In current scenario, for STA+P2P case, if station is connected to
some ap and for p2p if DUT becomes p2p go, now if SSR is triggered,
host driver gets interface down for sta for change mac address as
station gets disconnected as part of SSR and this interface down
is cached as SSR is in progress. Once reinit gets complete, driver
processes this cached interface down as part of SSR and in this
process driver gets the change interface for p2p go, which it
rejects the error code as EAGAIN from __dsc_vdev_can_trans as
SSR is going on and change interface fails to get dsc op.
When user space gets EAGAIN as error code it immediately tries
the change interface again which driver again rejects as SSR is
still in progress. Also, on receiving EAGAIN error code, userspace
retries the change iface operation only once.
To address above issue, return error code as EBUSY, on receiving
EBUSY as erorr code, user space addds some wait before invoking
the same operation again and also it tries this operation 10 times
if it gets EBUSY as error code, this gives sufficient time to
complete the ongoing transition at driver.
Change-Id: I96ec94432e7624546363cda190abfc9970ab1eb9
CRs-Fixed: 3126558
ba7faf192a