discard requests using fstrim command are slow when performed on filesystem using underlying device as mdadm raid10 array

Issue

• discard requests using fstrim command are slow when performed on filesystem using underlying device as mdadm raid10 array.
  For example, with the device and filesystem layout such as

#lsblk
NAME                                              MAJ:MIN RM   SIZE RO TYPE   MOUNTPOINT
nvme0n1                                           259:0    0   5.8T  0 disk   
`-nvme0n1p1                                       259:1    0   5.8T  0 part   
  `-md125                                           9:125  0  11.7T  0 raid10 
    `-vg2-lv_data                                 253:4    0  11.7T  0 lvm    
      `-luks-data                                 253:6    0  11.7T  0 crypt  /data
nvme1n1                                           259:2    0   5.8T  0 disk   
`-nvme1n1p1                                       259:3    0   5.8T  0 part   
  `-md125                                           9:125  0  11.7T  0 raid10 
    `-vg2-lv_data                                 253:4    0  11.7T  0 lvm    
      `-luks-data                                 253:6    0  11.7T  0 crypt  /data
nvme2n1                                           259:4    0   5.8T  0 disk   
`-nvme2n1p1                                       259:5    0   5.8T  0 part   
  `-md125                                           9:125  0  11.7T  0 raid10 
    `-vg2-lv_data                                 253:4    0  11.7T  0 lvm    
      `-luks-data                                 253:6    0  11.7T  0 crypt  /data
nvme3n1                                           259:6    0   5.8T  0 disk   
`-nvme3n1p1                                       259:7    0   5.8T  0 part   
  `-md125                                           9:125  0  11.7T  0 raid10 
    `-vg2-lv_data                                 253:4    0  11.7T  0 lvm    
      `-luks-data                                 253:6    0  11.7T  0 crypt  /data