[dm-devel] dm_set_target_max_io_len

Mikulas Patocka mpatocka at redhat.com
Thu Mar 21 10:33:32 UTC 2019


Hi

I'm looking at the patch 8f50e358153dd68182c714626be4a90b64179cf4. The 
patch claims that dm-crypt can't deal with large bios - but I fixed this 
problem with dm-crypt a year before (see the patch 
4e870e948fbabf62b78e8410f04c67703e7c816b). Was my fix for dm-crypt 
inefficient? Can you provide some test where dm-crypt fails with big bios?

Should 8f50e358153dd68182c714626be4a90b64179cf4 be dropped? The variable 
ti->max_io_len is really split boundary (although it is incorrectly called 
len), so if we set it, se are splitting even small bios that cross this 
boundary.

Mikulas




More information about the dm-devel mailing list