Re: dcache: NULL ptr deref in dentry_kill

[Al Viro <viro@ZenIV.linux.org.uk>]

On Sun, Oct 05, 2014 at 08:27:47PM -0400, Sasha Levin wrote:

> [  434.580818] BUG: unable to handle kernel NULL pointer dereference at 0000000000000090
> [  434.582208] IP: do_raw_spin_trylock (./arch/x86/include/asm/spinlock.h:108 kernel/locking/spinlock_debug.c:143)
[snip]
spin_lock((void *)0x90)
> [  434.590025] ? _raw_spin_trylock (include/linux/spinlock_api_smp.h:89 kernel/locking/spinlock.c:135)
> [  434.590025] ? lockref_put_or_lock (lib/lockref.c:131)
> [  434.590025] dput (fs/dcache.c:513 fs/dcache.c:616)

ummm...  lockref_put_or_lock(&dentry->d_lockref) ending up with 0x90 passed
to lockref_put_or_lock()...  What offset does d_lockref have on your build?
On my config it's 0x80 (i.e. that would mean dput((void *)0x10)); could you
check it on yours?  Do you have CONFIG_DEBUG_LOCK_ALLOC and CONFIG_LOCK_STAT?

> [  434.590025] __fput (fs/file_table.c:235)

... and here we have a struct file with ->f_path.dentry being a small positive
number.

Might make sense to slap BUG_ON((unsigned long)dentry < PAGE_SIZE); into
__fput() and try to reproduce.  And throw
	BUG_ON((unsigned long)file->f_path.dentry < PAGE_SIZE);
into fput(), while we are at it, to see if that's bogus ->f_path.dentry at
fput() time or if it's getting corrupted between task_work_add() and actual
execution of __fput().

And the value of offsetof(struct dentry, d_lock) on your build would be
interesting to see - that would tell us which value are we seeing passed
to dput().