System crash on NULL pointer dereference in scsi_send_eh_cmnd() function of SCSI error handler kernel thread code path

Issue

• System crashes in SCSI error handling code during an issue with storage.
• Kernel crashes with null point dereference in function scsi_send_eh_cmnd+0x63 as follows :

BUG: unable to handle kernel NULL pointer dereference at (null)
IP: [<ffffffff813860e3>] scsi_send_eh_cmnd+0x63/0x530

• Or similar traces on PowerPC platform:

ses 4:0:2:0: [sg78] Aborting command: 12
Unable to handle kernel paging request for data at address 0x00000000
Faulting instruction address: 0xc0000000003ed230
Oops: Kernel access of bad area, sig: 11 [#1]
...
TASK = c00000073f4e4610[773] 'scsi_eh_4' THREAD: c00000073e178000 CPU: 25
...
NIP [c0000000003ed230] .scsi_send_eh_cmnd+0x80/0x5b0
LR [c0000000003ed874] .scsi_eh_tur+0x54/0x110
Call Trace:
[c00000073e17bac0] [c00000073e17bb80] 0xc00000073e17bb80 (unreliable)
[c00000073e17bc40] [c0000000003ed874] .scsi_eh_tur+0x54/0x110
[c00000073e17bce0] [c0000000003ed9d0] .scsi_eh_test_devices+0xa0/0x2d0
[c00000073e17bd80] [c0000000003ef110] .scsi_error_handler+0x6c0/0x8d0
[c00000073e17bed0] [c0000000000bf03c] .kthread+0xbc/0xd0
[c00000073e17bf90] [c000000000033af4] .kernel_thread+0x54/0x70