On Fri, Jan 22, 2010 at 4:03 PM, Mauro Carvalho Chehab <mchehab@xxxxxxxxxxxxx> wrote: > I've checked the core and it does it on a balanced way. Yet, it is risky > to assume that this will always happen, and having a bad, non-interrupt > mutex there can lead to machine hangups. The core does balance the calls (I actually submitted a patch last year fixing a bug where it wasn't). However, if the tuner driver itself calls the i2c_gate_ctrl() function [and some tuners do do this] then you would hit a deadlock if it were used with Manu's frontend driver. >> Today, there is no real problem if a particular call path attempts to >> enable the gate if it is already open (or disable if it is already >> closed). With your proposed change, you will result in a deadlock. > > Precisely. Of course, both of these concerns will be no longer be relevant now that he is removing it from the dvb_frontend_ops (which means neither the tuner nor the frontend will be able to manipulate the gate). Devin -- Devin J. Heitmueller - Kernel Labs http://www.kernellabs.com -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
