linux-kernel - Re: [PATCH] ACPI/PPTT: fix off-by-one error

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: <20250507-aromatic-wooden-leopard-ad23e0@sudeepholla>
Date: Wed, 7 May 2025 17:28:05 +0100
From: Sudeep Holla <sudeep.holla@....com>
To: "Rafael J. Wysocki" <rafael@...nel.org>
Cc: Jeremy Linton <jeremy.linton@....com>,
	"Heyne, Maximilian" <mheyne@...zon.de>,
	Sudeep Holla <sudeep.holla@....com>,
	"stable@...r.kernel.org" <stable@...r.kernel.org>,
	Len Brown <lenb@...nel.org>, Ard Biesheuvel <ardb@...nel.org>,
	Catalin Marinas <catalin.marinas@....com>,
	"linux-acpi@...r.kernel.org" <linux-acpi@...r.kernel.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] ACPI/PPTT: fix off-by-one error

On Wed, May 07, 2025 at 06:12:40PM +0200, Rafael J. Wysocki wrote:
> On Wed, May 7, 2025 at 5:51 PM Jeremy Linton <jeremy.linton@....com> wrote:
> >

[...]
> > So if the bug being fixed is that the length check is validating that
> > the table length is less than the data in the table, that's still a
> > problem because its only validating the processor node without resources.
> 
> Admittedly, it is not my code, but I understand this check as a
> termination condition for the loop: If there's not enough space in the
> table to hold a thing that I'm looking for, I may as well bail out.
> 
> > AKA the return is still potentially returning a pointer to a structure
> > which may not be entirely contained in the table.
> 
> Right, but this check should be made anyway before comparing
> cpu_node->parent to node_entry, when it is known to be a CPU entry
> because otherwise why bother.
> 
> Roughly something like this:
> 
> proc_sz = sizeof(struct acpi_pptt_processor);
> 
> while ((unsigned long)entry + entry->length <= table_end) {

Yes, but in the last/termination run of the loop, entry will be > table_end,
is it safe to access entry->length in that case. That's the point
I was trying to make when I mentioned it is risky to use entry->length
in this check. That location(outside of PPTT) might have a value that may
result in entering the loop. We need to make sure the entry + offset(length)
is within the table_end to access it.

-- 
Regards,
Sudeep