On Tue, Jul 05, 2016 at 11:10:22PM +0200, Linus Walleij wrote: > On Tue, Jul 5, 2016 at 4:54 PM, Vinod Koul <vinod.koul@xxxxxxxxx> wrote: > > > dmaengine device should explicitly call devm_free_irq() when using > > devm_reqister_irq(). > > > > The irq is still ON when devices remove is executed and irq should be > > quiesced before remove is completed. > > > > Signed-off-by: Vinod Koul <vinod.koul@xxxxxxxxx> > > Cc: Linus Walleij <linus.walleij@xxxxxxxxxx> > > Acked-by: Linus Walleij <linus.walleij@xxxxxxxxxx> > > This patch makes me very uneasy. A spurious IRQ can always happen, so > Isn't this problem existing on more or less the .remove() path of 95% > if all drivers > using devm_request_irq() in the entire kernel? Yes it is indeed a problem. IMHO we should never have done devm variant for irq APIs. I have been pushing back on drivers not freeing up irqs in .remove. Maybe we should have devm variants made deprecated and discourage the use of these.. Thanks -- ~Vinod -- 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