Hi, Recently I was doing I/O stress test on my raid5, however, I stepped into a BUG_ON and I had no idea why would that happen, the BUG_ON(test_bit(R5_Wantread, &dev->flags)) is in fetch_block function in raid5.c. After looking the source code, I found that the bit R5_Wantread would be cleared in ops_run_io if the read operation is about to be conducted, the only reason the read was not handled is that the write is handled at the same time, is there any possible reason that could happen? or under what condition BUG_ON(test_bit(R5_Wantread, &dev->flags)) would happen? thanks,
