Greeting, FYI, we noticed the following commit (built with gcc-9): commit: ae3c57b5ca47665dc3416447a5534a9796096d86 ("nfsd: Cache R, RW, and W opens separately") https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master in testcase: fsmark version: fsmark-x86_64-3.3-1_20201110 with following parameters: iterations: 1x nr_threads: 32t disk: 1HDD fs: f2fs fs2: nfsv4 filesize: 16MB test_size: 60G sync_method: fsyncBeforeClose nr_directories: 16d nr_files_per_directory: 256fpd cpufreq_governor: performance ucode: 0x5002f01 test-description: The fsmark is a file system benchmark to test synchronous write workloads, for example, mail servers workload. test-url: https://sourceforge.net/projects/fsmark/ on test machine: 96 threads Intel(R) Xeon(R) Gold 6252 CPU @ 2.10GHz with 512G memory caused below changes (please refer to attached dmesg/kmsg for entire log/backtrace): If you fix the issue, kindly add following tag Reported-by: kernel test robot 2020-12-01 07:16:05 fs_mark -d /nfs/sda1/1 -d /nfs/sda1/2 -d /nfs/sda1/3 -d /nfs/sda1/4 -d /nfs/sda1/5 -d /nfs/sda1/6 -d /nfs/sda1/7 -d /nfs/sda1/8 -d /nfs/sda1/9 -d /nfs/sda1/10 -d /nfs/sda1/11 -d /nfs/sda1/12 -d /nfs/sda1/13 -d /nfs/sda1/14 -d /nfs/sda1/15 -d /nfs/sda1/16 -d /nfs/sda1/17 -d /nfs/sda1/18 -d /nfs/sda1/19 -d /nfs/sda1/20 -d /nfs/sda1/21 -d /nfs/sda1/22 -d /nfs/sda1/23 -d /nfs/sda1/24 -d /nfs/sda1/25 -d /nfs/sda1/26 -d /nfs/sda1/27 -d /nfs/sda1/28 -d /nfs/sda1/29 -d /nfs/sda1/30 -d /nfs/sda1/31 -d /nfs/sda1/32 -D 16 -N 256 -n 120 -L 1 -S 1 -s 16777216 # fs_mark -d /nfs/sda1/1 -d /nfs/sda1/2 -d /nfs/sda1/3 -d /nfs/sda1/4 -d /nfs/sda1/5 -d /nfs/sda1/6 -d /nfs/sda1/7 -d /nfs/sda1/8 -d /nfs/sda1/9 -d /nfs/sda1/10 -d /nfs/sda1/11 -d /nfs/sda1/12 -d /nfs/sda1/13 -d /nfs/sda1/14 -d /nfs/sda1/15 -d /nfs/sda1/16 -d /nfs/sda1/17 -d /nfs/sda1/18 -d /nfs/sda1/19 -d /nfs/sda1/20 -d /nfs/sda1/21 -d /nfs/sda1/22 -d /nfs/sda1/23 -d /nfs/sda1/24 -d /nfs/sda1/25 -d /nfs/sda1/26 -d /nfs/sda1/27 -d /nfs/sda1/28 -d /nfs/sda1/29 -d /nfs/sda1/30 -d /nfs/sda1/31 -d /nfs/sda1/32 -D 16 -N 256 -n 120 -L 1 -S 1 -s 16777216 # Version 3.3, 32 thread(s) starting at Tue Dec 1 07:16:05 2020 # Sync method: INBAND FSYNC: fsync() per file in write loop. # Directories: Round Robin between directories across 16 subdirectories with 256 files per subdirectory. # File names: 40 bytes long, (16 initial bytes of time stamp with 24 random bytes at end of name) # Files info: size 16777216 bytes, written with an IO size of 16384 bytes per write # App overhead is time in microseconds spent in the test not doing file writing related system calls. FSUse% Count Size Files/sec App Overhead in attached dmesg [ 2646.633157] nfs: server localhost not responding, still trying [ 2646.633196] nfs: server localhost not responding, still trying [ 2646.633199] nfs: server localhost not responding, still trying [ 2646.633209] nfs: server localhost not responding, still trying [ 2646.633212] nfs: server localhost not responding, still trying [ 2646.633213] nfs: server localhost not responding, still trying [ 2646.633229] nfs: server localhost not responding, still trying [ 2646.633230] nfs: server localhost not responding, still trying [ 2646.633231] nfs: server localhost not responding, still trying [ 2646.633242] nfs: server localhost not responding, still trying [ 2646.649307] nfs: server localhost OK [ 2646.649497] nfs: server localhost OK [ 2646.649507] nfs: server localhost OK [ 2646.649571] nfs: server localhost OK [ 2646.649590] nfs: server localhost OK [ 2646.649652] nfs: server localhost OK [ 2646.649663] nfs: server localhost OK [ 2646.649673] nfs: server localhost OK [ 2646.649734] nfs: server localhost OK [ 2646.650546] nfs: server localhost OK [ 3639.661587] Wed Dec 2 22:15:18 UTC 2020 detected soft_timeout [ 3639.661590] [ 3639.704756] Terminated [ 3639.704758] [ 3639.705909] kill 2538 /usr/bin/time -v -o /tmp/lkp/fsmark.time /lkp/lkp/src/tests/fsmark To reproduce: git clone https://github.com/intel/lkp-tests.git cd lkp-tests bin/lkp install job.yaml # job file is attached in this email bin/lkp run job.yaml Thanks, Oliver Sang