Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <20200217083223.2011-5-zong.li@sifive.com> <mhng-6a94c49b-419b-4b5a-a11d-dda1fb0aa896@palmerdabbelt-glaptop1>
In-Reply-To: <mhng-6a94c49b-419b-4b5a-a11d-dda1fb0aa896@palmerdabbelt-glaptop1>
From:   Zong Li <zongbox@gmail.com>
Date:   Thu, 5 Mar 2020 12:01:04 +0800
Message-ID: <CA+ZOyai1rUVttopEOGMjug8fM6YkV3c+WTstxD6YojdwMCmHzw@mail.gmail.com>
Subject: Re: [PATCH 4/8] riscv: move exception table immediately after RO_DATA
To:     Palmer Dabbelt <palmer@dabbelt.com>
Cc:     Zong Li <zong.li@sifive.com>,
        linux-riscv <linux-riscv@lists.infradead.org>,
        Albert Ou <aou@eecs.berkeley.edu>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Paul Walmsley <paul.walmsley@sifive.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Palmer Dabbelt <palmer@dabbelt.com> =E6=96=BC 2020=E5=B9=B43=E6=9C=885=E6=
=97=A5 =E9=80=B1=E5=9B=9B =E4=B8=8A=E5=8D=888:58=E5=AF=AB=E9=81=93=EF=BC=9A
>
> On Mon, 17 Feb 2020 00:32:19 PST (-0800), zong.li@sifive.com wrote:
> > Move EXCEPTION_TABLE immediately after RO_DATA. Make it easy to set the
> > attribution of the sections which should be read-only at a time.
> > Move .sdata to indicate the start of data section with write permission=
.
> > This patch is prepared for STRICT_KERNEL_RWX support.
> >
> > Signed-off-by: Zong Li <zong.li@sifive.com>
> > ---
> >  arch/riscv/kernel/vmlinux.lds.S | 8 +++++---
> >  1 file changed, 5 insertions(+), 3 deletions(-)
> >
> > diff --git a/arch/riscv/kernel/vmlinux.lds.S b/arch/riscv/kernel/vmlinu=
x.lds.S
> > index 1e0193ded420..4ba8a5397e8b 100644
> > --- a/arch/riscv/kernel/vmlinux.lds.S
> > +++ b/arch/riscv/kernel/vmlinux.lds.S
> > @@ -9,6 +9,7 @@
> >  #include <asm/page.h>
> >  #include <asm/cache.h>
> >  #include <asm/thread_info.h>
> > +#include <asm/set_memory.h>
> >
> >  OUTPUT_ARCH(riscv)
> >  ENTRY(_start)
> > @@ -52,12 +53,15 @@ SECTIONS
> >       }
> >
> >       /* Start of data section */
> > -     _sdata =3D .;
> >       RO_DATA(L1_CACHE_BYTES)
> >       .srodata : {
> >               *(.srodata*)
> >       }
> >
> > +     EXCEPTION_TABLE(0x10)
> > +
> > +     _sdata =3D .;
> > +
> >       RW_DATA(L1_CACHE_BYTES, PAGE_SIZE, THREAD_SIZE)
> >       .sdata : {
> >               __global_pointer$ =3D . + 0x800;
> > @@ -69,8 +73,6 @@ SECTIONS
> >
> >       BSS_SECTION(PAGE_SIZE, PAGE_SIZE, 0)
> >
> > -     EXCEPTION_TABLE(0x10)
> > -
> >       .rel.dyn : {
> >               *(.rel.dyn*)
> >       }
>
> As far as I can tell this is OK: core_kernel_data() explicitly says that =
RODATA
> may or may not be between _sdata and _edata.  That said, I think we shoul=
d add
> __start_rodata and __end_rodata atomicly with this change (around RO_DATA=
 and
> .srodata).
>

OK, I'll move _sdata back. Actually, here I need a symbol to specify
the start address at writable data (RW_DATA), thus, I could remove the
executable permission of .data section (from this symbol), and make
.rodata, .srodata and __ex_table read-only at a time (from
__start_rodata to this symbol). So even if we use __end_rodata to wrap
.srodata together with .rodata, exception table still be excluded, and
we have no idea where is the .data section start address. Do you think
it would be OK if we use _data to specify the start address at
writable data? If it's OK, whether we still need to add __end_rodata
after .srodata?