Skip to contentSkip to conversation

kernel: kcov: a possible sleep-in-atomic-context bug in kcov_ioctl()

BUG Jia-Ju Bai 2 participants 3 messages Jun 21, 2018

lore
Share & tools
lore

Jia-Ju Bai

author
#0 lore raw

The kernel may sleep with holding a spinlock.
The function call path (from bottom to top) in Linux-4.16.7 is:

[FUNC] vfree --> can sleep
kernel/kcov.c, 237: vfree in kcov_put kernel/kcov.c, 396: kcov_put in kcov_ioctl_locked kernel/kcov.c, 410: kcov_ioctl_locked in kcov_ioctl kernel/kcov.c, 409: spin_lock in kcov_ioctl

This bug is found by my static analysis tool (DSAC-2) and checked by my code review.

I do not know how to correctly fix this bug, so I just report them.

Best wishes,
Jia-Ju Bai

#1 lore raw
Quoted 9 lines · Jia-Ju Bai

The kernel may sleep with holding a spinlock.
The function call path (from bottom to top) in Linux-4.16.7 is:

[FUNC] vfree --> can sleep
kernel/kcov.c, 237: vfree in kcov_put kernel/kcov.c, 396: kcov_put in kcov_ioctl_locked kernel/kcov.c, 410: kcov_ioctl_locked in kcov_ioctl kernel/kcov.c, 409: spin_lock in kcov_ioctl

Quoted 4 lines

This bug is found by my static analysis tool (DSAC-2) and checked by my code review.

I do not know how to correctly fix this bug, so I just report them.

Assuming it's a bug in the first place, that is. Note that
* we never modify task->kcov for task != current
* task->kcov contributes to refcount
* opened file contributes to refcount
* that kcov_put() of yours happens from ->ioctl() and removes
current->kcov reference; it *can't* be the last reference - the one held by struct file used to call ->ioctl() is also there.

IOW, it's a false positive.

author
#2 lore raw
Quoted 22 lines · Al Viro
Quoted 9 lines · Jia-Ju Bai

The kernel may sleep with holding a spinlock.
The function call path (from bottom to top) in Linux-4.16.7 is:

[FUNC] vfree --> can sleep
kernel/kcov.c, 237: vfree in kcov_put kernel/kcov.c, 396: kcov_put in kcov_ioctl_locked kernel/kcov.c, 410: kcov_ioctl_locked in kcov_ioctl kernel/kcov.c, 409: spin_lock in kcov_ioctl

Quoted 4 lines

This bug is found by my static analysis tool (DSAC-2) and checked by my code review.

I do not know how to correctly fix this bug, so I just report them.

Assuming it's a bug in the first place, that is. Note that
* we never modify task->kcov for task != current
* task->kcov contributes to refcount
* opened file contributes to refcount
* that kcov_put() of yours happens from ->ioctl() and removes
current->kcov reference; it *can't* be the last reference - the one held by struct file used to call ->ioctl() is also there.

Quoted 1 line

IOW, it's a false positive.

Okay, thanks for your reply and explanation.
I agree that my report is false.

Best wishes,
Jia-Ju Bai