linux-kernel - Re: [PATCH] arm64/trap: fix broken ct->nmi_nesting when die() is called in a kthread

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 for Android: free password hash cracker in your pocket

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

Message-ID: <aD78oNHA5SlaH50z@J2N7QTR9R3.cambridge.arm.com>
Date: Tue, 3 Jun 2025 14:46:08 +0100
From: Mark Rutland <mark.rutland@....com>
To: Yeoreum Yun <yeoreum.yun@....com>
Cc: Will Deacon <will@...nel.org>, catalin.marinas@....com,
	geert@...ux-m68k.org, broonie@...nel.org, mcgrof@...nel.org,
	joey.gouly@....com, kristina.martsenko@....com, rppt@...nel.org,
	pcc@...gle.com, bigeasy@...utronix.de, ptosi@...gle.com,
	james.morse@....com, linux-arm-kernel@...ts.infradead.org,
	linux-kernel@...r.kernel.org, ada.coupriediaz@....com
Subject: Re: [PATCH] arm64/trap: fix broken ct->nmi_nesting when die() is
 called in a kthread

On Tue, Jun 03, 2025 at 12:14:18PM +0100, Yeoreum Yun wrote:
> > On Mon, Jun 02, 2025 at 06:50:53PM +0100, Yeoreum Yun wrote:
> > > So, what I think:
> > >   1. arm64_enter_el1_dbg() should ct_nmi_enter() as it is.
> > >   2. in bug_handler() while handling BUG_TYPE, add above ct_nmi_exit()
> > >      conditional call.
> > >   3. DAIF.D and DAIF.A handling.
> >
> > No, that is not safe. In step 2, calling ct_nmi_exit() would undo *all*
> > of the ct_nmi_enter() logic, and may stop RCU from watching if the
> > exception was entered from some intermediate/inconsistent state.
> 
> Yes if call ct_nmi_enter() without condition.
> But I imply with the condition check what I posted.
> if CT_NESTING_IRQ_NONIDLE,
> it wouldn't need call and that cpu can be watched by RCU.

I am not keen on conditionally calling ct_nmi_exit(), and would strongly
prefer to avoid that, regardless of where that lives in the flow.

I suspect that it would be bettter to triage the interrupted context
earlier, and rethink the way entry/exit works, but that's a much larger
bit of work and will take more thinking.

Mark.