Maxim Syrchin schrieb: > Denis Oliver Kropp wrote: > >> Maxim Syrchin schrieb: >> >>> Hello all. >>> >>> fusion_entry_destroy() function removes entry while it being locked. >>> It can have a lot of bad consequences (e.g. in case of deadlock >>> detection code in kernel it causes segmentation fault). Possible fix >>> for this case is attached, please take a look on it. >> >> >> Thanks! I'll add it, though I think it should be allowed to destroy a >> locked mutex. It's destroyed anyways, so why do the unlock? >> > Why do the unlock ? IMHO it is some kind of clean up :)) I agree - > nobody uses removed mutex from fusion. > But actually we do not destroy mutex. We do kfree on "struct entry" > which holds mutex variable "lock". So, mutex remains locked, but it's > variable is lost and can be allocated/rewrited by other application or > kernel driver. So, if some code monitor mutexes (such as deadlock > detection code or code for gathering preemption statistic) it will work > incorrectly or cause segfault.
Thanks for the explanation, I saw it already, see other mail :) -- Best regards, Denis Oliver Kropp .------------------------------------------. | DirectFB - Hardware accelerated graphics | | http://www.directfb.org/ | "------------------------------------------" _______________________________________________ directfb-dev mailing list [email protected] http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-dev
