On 13 June 2017 at 16:54, axie <axie at amd.com> wrote: > > On 2017-06-12 06:47 PM, Christian König wrote: >> >> Am 12.06.2017 um 22:31 schrieb Alex Xie: >>> >>> Make the critical section smaller. There is no >>> need to protect the bo_list, because there is >>> no other task can access the newly created BO >>> list yet. >> >> >> NAK, a task can guess the next id number so could get the kernel to use >> the structure before it is initialized. >> >> Christian. >> > How did you find such an extreme corner case? I am fine with this comment. > Tuesday/Wednesday I will address it with next version of patch set. > > Currently, there are 2 options: > Option 1: I may use a write_lock in the create function. And restore the > original code for the creation of BO list. > Option 2: I may move the function call of idr_alloc to the end of the > creation of BO list ioctl. This is more efficient but > the code look dirty. You can idr_alloc and idr_replace later btw. We do this in a few places in the drm. Dave.
