Reinstate some of "swiotlb: rework "fix info leak with DMA_FROM_DEVICE""
authorLinus Torvalds <torvalds@linux-foundation.org>
Mon, 28 Mar 2022 18:37:05 +0000 (11:37 -0700)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 25 May 2022 07:14:38 +0000 (09:14 +0200)
commitb2f140a9f980806f572d672e1780acea66b9a25c
treed3e5379962dd876e316f5f0bc6ba8b070bbfb50d
parent060f38b1dfb499d0c02d14ea793a16f4e63c51b6
Reinstate some of "swiotlb: rework "fix info leak with DMA_FROM_DEVICE""

commit 901c7280ca0d5e2b4a8929fbe0bfb007ac2a6544 upstream.

Halil Pasic points out [1] that the full revert of that commit (revert
in bddac7c1e02b), and that a partial revert that only reverts the
problematic case, but still keeps some of the cleanups is probably
better.  

And that partial revert [2] had already been verified by Oleksandr
Natalenko to also fix the issue, I had just missed that in the long
discussion.

So let's reinstate the cleanups from commit aa6f8dcbab47 ("swiotlb:
rework "fix info leak with DMA_FROM_DEVICE""), and effectively only
revert the part that caused problems.

Link: https://lore.kernel.org/all/20220328013731.017ae3e3.pasic@linux.ibm.com/
Link: https://lore.kernel.org/all/20220324055732.GB12078@lst.de/
Link: https://lore.kernel.org/all/4386660.LvFx2qVVIh@natalenko.name/
Suggested-by: Halil Pasic <pasic@linux.ibm.com>
Tested-by: Oleksandr Natalenko <oleksandr@natalenko.name>
Cc: Christoph Hellwig" <hch@lst.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[OP: backport to 5.4: adjusted context]
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Documentation/DMA-attributes.txt
include/linux/dma-mapping.h
kernel/dma/swiotlb.c