linux-kernel - Re: [PATCH V2 3/4] arm64/mm: Consolidate page fault information capture

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]

Date:   Thu, 6 Jun 2019 10:38:11 +0100
From:   Mark Rutland <mark.rutland@....com>
To:     Catalin Marinas <catalin.marinas@....com>
Cc:     Anshuman Khandual <anshuman.khandual@....com>,
        linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
        Will Deacon <will.deacon@....com>,
        James Morse <james.morse@....com>,
        Andrey Konovalov <andreyknvl@...gle.com>
Subject: Re: [PATCH V2 3/4] arm64/mm: Consolidate page fault information
 capture

On Tue, Jun 04, 2019 at 03:42:09PM +0100, Catalin Marinas wrote:
> On Mon, Jun 03, 2019 at 12:11:24PM +0530, Anshuman Khandual wrote:
> > diff --git a/arch/arm64/mm/fault.c b/arch/arm64/mm/fault.c
> > index da02678..4bb65f3 100644
> > --- a/arch/arm64/mm/fault.c
> > +++ b/arch/arm64/mm/fault.c
> > @@ -435,6 +435,14 @@ static bool is_el0_instruction_abort(unsigned int esr)
> >  	return ESR_ELx_EC(esr) == ESR_ELx_EC_IABT_LOW;
> >  }
> >  
> > +/*
> > + * This is applicable only for EL0 write aborts.
> > + */
> > +static bool is_el0_write_abort(unsigned int esr)
> > +{
> > +	return (esr & ESR_ELx_WNR) && !(esr & ESR_ELx_CM);
> > +}
> 
> What makes this EL0 only?

It returns false for EL1 faults caused by DC IVAC, where write
permission is required. EL0 can only issue maintenance that requires
read permission.

For whatever reason, the architecture says that WnR is always 1b1, even
if read permission was sufficient.

How about:

/*
 * Note: not valid for EL1 DC IVAC, but we never use that such that it
 * should fault.
 */
static bool is_write_abort(unsigned int esr)
{
	return (esr & ESR_ELx_WNR) && !(esr & ESR_ELx_CM);
}

... which would also address your concern about EL1 writes to a user
mapping.

Thanks,
Mark.