Hi Maxime, On Friday 17 October 2014 13:23:56 Maxime Ripard wrote: > Hi Laurent, > > Just getting back on something... > > On Mon, Oct 06, 2014 at 03:09:48PM +0300, Laurent Pinchart wrote: > > > + * device_prep_dma_* > > > + - These functions are matching the capabilities you registered > > > + previously. > > > + - These functions all take the buffer or the scatterlist relevant > > > + for the transfer being prepared, and should create a hardware > > > + descriptor or a list of descriptors from it > > > + - These functions can be called from an interrupt context > > > + - Any allocation you might do should be using the GFP_NOWAIT > > > + flag, in order not to potentially sleep, but without depleting > > > + the emergency pool either. > > > > You could add "Drivers should try to preallocate the data structures they > > require to prepare a transfer." > > Isn't that obvious? > > I mean, if we're in this function, we're already preparing a > transfer... And I would expect any programmer that followed CS101 to > be able to allocate the memory it needs :) I meant that memory should be pre-allocated earlier (at probe time or channel alloc time for instance) to avoid putting pressure on the nowait memory pool. > The rest of the issues have been fixed, thanks! > Maxime -- Regards, Laurent Pinchart -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
