On 2/23/24 11:31, Keith Busch wrote: > On Fri, Feb 23, 2024 at 07:16:30PM +0000, Chaitanya Kulkarni wrote: >> On 2/23/24 07:59, Keith Busch wrote: >>> From: Keith Busch <kbusch@xxxxxxxxxx> >>> >>> Changes from v3: >>> >>> Added reviewed and tested by tags >>> >>> More formatting cleanups in patch 2 (Christoph) >>> >>> A more descriptive name for the bio chain wait helper (Christoph) >>> >>> Don't fallback to the zero page on fatal signal error (Nilay) >>> >> we need a blktests for this, for whole series :- > How would you know know if the device completed the operation before the > fatal signal or not? This is a quick test script off the top of my head, > but whether or not it shows a difference with this patch series depends > on your device. > > # time sh -c "blkdiscard -z /dev/nvme0n1 & sleep 1 && kill %1 && wait" something like that :- # record the start time start_time=get_time_sec # Use ginormous value for the device size without membacked for null_blk # such that it will not finish before 20 min modprobe -r null_blk modprobe null_blk gb=2024000000 # this should not finish within 20 min # kill the process kill -9 `ps aux | grep blkdiscard | grep -v grep | tr -s ' ' ' ' | cut -f 2 -d ' '` # record total time total_time=get_time_sec - start_time # make sure process doesn't exists ps aux | grep -v grep | grep blkdiscard if [ $? -eq 0]; then echo "Test fail" fi # make sure total_time < 60 sec if [ $total_time -gt 60 ]; then echo "Fail" fi echo "success" WDYT ? -ck