linux-kernel - Re: [PATCH 3/3] arm64/hw_breakpoint: Enable FEAT

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: <Zx-H8IU2wZmyowGe@J2N7QTR9R3>
Date: Mon, 28 Oct 2024 12:47:44 +0000
From: Mark Rutland <mark.rutland@....com>
To: Anshuman Khandual <anshuman.khandual@....com>
Cc: linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
	Jonathan Corbet <corbet@....net>, Marc Zyngier <maz@...nel.org>,
	Oliver Upton <oliver.upton@...ux.dev>,
	James Morse <james.morse@....com>,
	Suzuki K Poulose <suzuki.poulose@....com>,
	Catalin Marinas <catalin.marinas@....com>,
	Will Deacon <will@...nel.org>, Mark Brown <broonie@...nel.org>,
	kvmarm@...ts.linux.dev
Subject: Re: [PATCH 3/3] arm64/hw_breakpoint: Enable FEAT_Debugv8p9

On Wed, Oct 23, 2024 at 01:01:52PM +0530, Anshuman Khandual wrote:
> 
> 
> On 10/22/24 21:04, Mark Rutland wrote:
> > On Tue, Oct 01, 2024 at 10:06:02AM +0530, Anshuman Khandual wrote:

[...]

> > Wherever this lives it needs a comment explaining what it is doing and
> > why. I assume this is intended to protect the bank in sequences like:
> > 
> > 	MSR	MDSELR, <...>
> > 	ISB
> > 	MRS	<..._, BANKED_REGISTER
> 
> Correct, it is protecting the above sequence.
> 
> > 
> > ... but is theat suffucient for mutual exclusion against
> > exception handlers, or does that come from somewhere else?
> 
> Looking at all existing use cases for breakpoint/watchpoints, it should
> be sufficient to protect against mutual exclusion. But thinking, do you
> have a particular exception handler scenario in mind where this might
> still be problematic ? Will keep looking into it.

Where does the mutual exclusion come from for the existing sequences?
We should be able to descrive should be able to describe that in the
commit message or in a comment somewhere (or better, with some
assertions that get tested).

For example, what prevents watchpoint_handler() from firing in the
middle of arch_install_hw_breakpoint() or
arch_uninstall_hw_breakpoint()?

Is the existing code correct?

Mark.