linux-kernel - Re: [PATCH] kernel: dma: Make CMA boot parameters __ro_after

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: <CAOfkYf5wagQzj0UboBdBh6iDq1ox=TN7inpatuhitw+Gsak1GQ@mail.gmail.com>
Date:   Wed, 16 Oct 2019 15:41:39 +0530
From:   Shyam Saini <mayhs11saini@...il.com>
To:     Nathan Chancellor <natechancellor@...il.com>
Cc:     Kernel Hardening <kernel-hardening@...ts.openwall.com>,
        iommu@...ts.linux-foundation.org,
        linux-kernel <linux-kernel@...r.kernel.org>,
        linux-mm <linux-mm@...ck.org>, Christoph Hellwig <hch@....de>,
        Marek Szyprowski <m.szyprowski@...sung.com>,
        Robin Murphy <robin.murphy@....com>,
        Matthew Wilcox <willy@...radead.org>,
        Christopher Lameter <cl@...ux.com>,
        Kees Cook <keescook@...omium.org>,
        clang-built-linux@...glegroups.com
Subject: Re: [PATCH] kernel: dma: Make CMA boot parameters __ro_after_init

Hi Nathan,

On Mon, Oct 14, 2019 at 7:55 AM Nathan Chancellor
<natechancellor@...il.com> wrote:
>
> On Sat, Oct 12, 2019 at 05:59:18PM +0530, Shyam Saini wrote:
> > This parameters are not changed after early boot.
> > By making them __ro_after_init will reduce any attack surface in the
> > kernel.
> >
> > Link: https://lwn.net/Articles/676145/
> > Cc: Christoph Hellwig <hch@....de>
> > Cc: Marek Szyprowski <m.szyprowski@...sung.com>
> > Cc: Robin Murphy <robin.murphy@....com>
> > Cc: Matthew Wilcox <willy@...radead.org>
> > Cc: Christopher Lameter <cl@...ux.com>
> > Cc: Kees Cook <keescook@...omium.org>
> > Signed-off-by: Shyam Saini <mayhs11saini@...il.com>
> > ---
> >  kernel/dma/contiguous.c | 8 ++++----
> >  1 file changed, 4 insertions(+), 4 deletions(-)
> >
> > diff --git a/kernel/dma/contiguous.c b/kernel/dma/contiguous.c
> > index 69cfb4345388..1b689b1303cd 100644
> > --- a/kernel/dma/contiguous.c
> > +++ b/kernel/dma/contiguous.c
> > @@ -42,10 +42,10 @@ struct cma *dma_contiguous_default_area;
> >   * Users, who want to set the size of global CMA area for their system
> >   * should use cma= kernel parameter.
> >   */
> > -static const phys_addr_t size_bytes = (phys_addr_t)CMA_SIZE_MBYTES * SZ_1M;
> > -static phys_addr_t size_cmdline = -1;
> > -static phys_addr_t base_cmdline;
> > -static phys_addr_t limit_cmdline;
> > +static const phys_addr_t __ro_after_init size_bytes = (phys_addr_t)CMA_SIZE_MBYTES * SZ_1M;
>
> The 0day bot reported an issue with this change with clang:
>
> https://groups.google.com/d/msgid/clang-built-linux/201910140334.nhultlt8%25lkp%40intel.com
>
> kernel/dma/contiguous.c:46:36: error: 'size_cmdline' causes a section type conflict with 'size_bytes'
> static phys_addr_t __ro_after_init size_cmdline = -1;
>                                    ^
> kernel/dma/contiguous.c:45:42: note: declared here
> static const phys_addr_t __ro_after_init size_bytes = (phys_addr_t)CMA_SIZE_MBYTES * SZ_1M;
>                                          ^
> kernel/dma/contiguous.c:47:36: error: 'base_cmdline' causes a section type conflict with 'size_bytes'
> static phys_addr_t __ro_after_init base_cmdline;
>                                    ^
> kernel/dma/contiguous.c:45:42: note: declared here
> static const phys_addr_t __ro_after_init size_bytes = (phys_addr_t)CMA_SIZE_MBYTES * SZ_1M;
>                                          ^
> kernel/dma/contiguous.c:48:36: error: 'limit_cmdline' causes a section type conflict with 'size_bytes'
> static phys_addr_t __ro_after_init limit_cmdline;
>                                    ^
> kernel/dma/contiguous.c:45:42: note: declared here
> static const phys_addr_t __ro_after_init size_bytes = (phys_addr_t)CMA_SIZE_MBYTES * SZ_1M;
>                                          ^
> 3 errors generated.

Thanks for your feedback and reporting this error.

> The errors seem kind of cryptic at first but something that is const
> should automatically be in the read only section, this part of the
> commit seems unnecessary. Removing that part of the change fixes the error.

I have overlooked size_bytes variable
It shouldn't be const if it is declared as __ro_after_init.

I will fix and resend it.