2011/7/23 Malcolm Priestley <tvboxspy@xxxxxxxxx>: > On Sat, 2011-07-23 at 01:47 +0200, HoP wrote: >> 2011/7/23 Antti Palosaari <crope@xxxxxx>: >> > On 07/23/2011 02:31 AM, Antti Palosaari wrote: >> >> >> >> On 07/23/2011 02:01 AM, HoP wrote: >> >>> >> >>> 2011/7/23 Antti Palosaari<crope@xxxxxx>: >> >>>> >> >>>> But now I see what you mean. msg2[1] is set as garbage fields in case of >> >>>> incoming msg len is 1. True, but it does not harm since it is not >> >>>> used in >> >>>> that case. >> >>> >> >>> In case of write, cxd2820r_tuner_i2c_xfer() gets msg[] parameter >> >>> with only one element, true? If so, then my patch is correct. >> >> >> >> Yes it is true but nonsense. It is also wrong to make always msg2 as two >> >> element array too, but those are just simpler and generates most likely >> >> some code less. Could you see it can cause problem in some case? >> > >> > Now I thought it more, could it crash if it point out of memory area? > Arrays are not fussy they will read anything, just don't poke them :-) Are you sure about not crashing? On every architecture on which linux can run? Even if it not crash kernel, I hope we can agree that it is incorrect and need to be fixed. >> >> I see you finally understood what I wanted to do :-) >> >> I'm surprised that it not crashed already. I thought I have to missed something. > > It does not crash because num is constant throughout, when the number of > messages is one the second element isn't transferred. Sure, that is evident. My fix was about not do read access outside of input array msg[]. I still don't understand the NACK. /Honza -- 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
