2017-12-12 02:20:40

by Jia-Ju Bai

[permalink] [raw]
Subject: [BUG] drivers/scsi/dpt_i2o: a possible sleep-in-atomic bug in adpt_i2o_status_get

According to drivers/scsi/dpt_i2o.c, the kernel module may sleep under a
spinlock.
The function call paths are:
adpt_reset (acquire the spinlock)
__adpt_reset
adpt_hba_reset
adpt_i2o_activate_hba
adpt_i2o_status_get
dma_alloc_coherent(GFP_KERNEL) --> may sleep
schedule_timeout_uninterruptible --> may sleep

A possible fixing is to replace "schedule_timeout_uninterruptible" with
"mdelay" and replace "GFP_KERNEL" with "GFP_ATOMIC".
If this fixing is correct, I can send a patch.
This possible bug is found by my static analysis tool (DSAC) and checked
by my code review.


Thanks,
Jia-Ju Bai