Open leeymcj opened 6 years ago
Thanks for noting this. I think I have run into the same issue, but with kernel 3.17. Do you have a fix that handles this?
This appears to be reproducible with 3.13 even (centos 7.5, kernel 3.13.0, MLNX OFED 3.4).
Below is the kernel oops right after swapon /dev/infiniswap0. Sometimes this error doesn't happen until the swap device is being used, but always very quickly after 10-100MB have been swapped out.
[ 728.619403] In IS_session_create() with portal: rdma://1,10.10.10.4:9400,
[ 728.626210] rdma://1,10.10.10.4:9400,
[ 728.629876] portal: 10.10.10.4, 9400
[ 733.985157] IS_register_block_device, dev_name infiniswap0
[ 733.990649] IS: init done
[ 733.993432] stackbd: init done
[ 733.996501] Opened /dev/loop0
[ 733.999476] stackbd: Device real capacity: 104857600
[ 734.004440] stackbd: Max sectors: 8
[ 734.007965] stackbd: done initializing successfully
[ 748.056376] evict_handler, waiting for STOP msg
[ 752.832599] BUG: unable to handle kernel NULL pointer dereference at 0000000000000068
[ 752.840455] IP: [
If I check for (req->bio == NULL) before calling bio_clone(), I get a hit, but kernel module still fails because the request is not handled.
New kernel (>4.x) generates requests with NULL bio Stackbd tries to clone it but panicking with NULL pointer exception.