Chuandong Li
Chuandong Li
@cchanging I need your input on this issue.
> ld you double check the correctness of `Condvar` under concurrency? Also, whether the implementation can be further simplified. T OK. I will take a look.
I have no objections regarding the current correctness as all counter modifications are protected under a `SpinLock`. Additionally, I think the current implementation indeed meets the description provided in the...
> I think we should implement the `PageCacheBackend` for the metadata of `RamInode`, which can fix the deadlock issue. > > This fix cannot guarantee the concurrency of `resize`.May result...
I will take a look at this.
I've tracked the call process where 437.leslie3d hangs, and it actually results in a deadlock, as shown below: Initially, a read lock is obtained in `RamInode::resize()` method and is then...