On Tue, 13 Jan 2009 15:38:41 +0200
Adrian Hunter <ext-adrian.hunter@xxxxxxxxx> wrote:
This also fixes the hidden problem that the controller
was turning off the clock to the card while the card
was busy - the clock was then switched on for the
duration of each Send Status command used for polling,
which is why writes did not lock up altogether.
i.e. really dysfunctional.
Does the driver only keep the clock running as long as there is an
ongoing request? That kind of policy should really be handled in the
core where we have more information and can control it better. E.g. some
cards need the clock to finish up internal housekeeping so we need to
delay things for those.
Rgds