Bob on 7/16/2022 2:28 AM, Bob Pearson wrote: > Li, > > I just sent in an alternative patch that fixes up the error paths. thanks for your patch, i will take a look later. > We not only have a problem with PD but also umem. In my understanding, although the umem is also not set in the same case, it's safe to pass NULL to ib_umem_release() currently. Thanks > I moved the setting > of PD up so it will always be set before cleanup gets called and also > checks if umem is set before freeing it. Please take a look at it.
