On Mon, 13 Jul 2009, Giacomo Lozito wrote: > >P.S.: Instead of patching the driver, you should get an equivalent > >result by increasing the "delay_use=" module parameter for usb-storage > >from its default value of 5 seconds up to 15 seconds. Have you tried > >this? > > Tried just now. > It works fine, with no patch applied, > when delay_use is set equal to 10 (or higher than that). > > Out of curiosity... > while they lead to an equivalent result, if I guessed this correctly (not > really sure about how wait_event_freezable_timeout works), the difference > between delay_use and getmaxlun timeout lies in the fact that delay_use always > forces a wait for X seconds (even when un-needed), while the timeout obviously > is not forcefully reached if getmaxlun succeeds. > Is that correct? Just about. There is one other small difference. With the delay_use parameters, the system will wait X seconds before sending the GetMaxLUN request. With the timeout, the system will send the request immediately and then wait up to X seconds to receive a reply. Some devices are sensitive to when the request is sent; they get very confused if they receive it before they are ready. > If that is correct, I guess it would be better to raise the getmaxlun timeout. > If, on the other hand, they both can return before the timeout is reached, > then it doesn't matter which one is raised. > > We could also add a "delay" field in the us_data structure and allow unusual > devs to customize that in their init function, thus raising delay_use (or > getmaxlun timeout, delay could act on any of the two) only when it's known to > be required for a certain device. > Looking at the code a bit, I think I can do something like that. > > Or we could do nothing at all :) > > What do you think it could be the best solution? Raising the GetMaxLUN timeout seems like a reasonable thing to do. Under normal circumstances it won't cause any harm, and it will automatically help other people facing similar problems. As opposed to delay_use, which has to be set manually. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
