linux-kernel - Re: [PATCH 1/2] drm/nouveau/firmware: Fix SG_DEBUG error with nvkm_firmware

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <a2c4681b900796f9ac47b61b802d9134f9a915b3.camel@nvidia.com>
Date: Sun, 28 Apr 2024 18:19:55 +0000
From: Timur Tabi <ttabi@...dia.com>
To: "nouveau@...ts.freedesktop.org" <nouveau@...ts.freedesktop.org>,
	"lyude@...hat.com" <lyude@...hat.com>
CC: "bskeggs@...hat.com" <bskeggs@...hat.com>, "kherbst@...hat.com"
	<kherbst@...hat.com>, "airlied@...hat.com" <airlied@...hat.com>,
	"justinstitt@...gle.com" <justinstitt@...gle.com>,
	"dri-devel@...ts.freedesktop.org" <dri-devel@...ts.freedesktop.org>,
	"keescook@...omium.org" <keescook@...omium.org>, "airlied@...il.com"
	<airlied@...il.com>, "dakr@...hat.com" <dakr@...hat.com>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"daniel@...ll.ch" <daniel@...ll.ch>
Subject: Re: [PATCH 1/2] drm/nouveau/firmware: Fix SG_DEBUG error with
 nvkm_firmware_ctor()

On Sun, 2024-04-28 at 11:52 -0400, Lyude Paul wrote:
> But we can't get a page pointer from an allocation made by
> dma_alloc_coherent() - but we can from vmalloc(). I'll fix the patch
> explanation in the next version, I have to send out another version
> anyhow since I realized that patch #2 still needs one more check to
> work properly

Glad my confusion was helpful!  You wouldn't happent to know why this
statement is true?

	DMA allocations can never be turned back into a page pointer

I'm guessing DMA allocations are taken from some memory pool that isn't
tracked using page pointers, but I don't understand why.  It seems weird that
you can have a virtual address to the buffer, therefore it is surely mapped on
a page-by-page basis, so why are there no page pointers?