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  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date:   Fri, 8 Oct 2021 10:11:22 +0800
From:   Jisheng Zhang <Jisheng.Zhang@...aptics.com>
To:     Catalin Marinas <catalin.marinas@....com>
Cc:     Will Deacon <will@...nel.org>, Jonathan Corbet <corbet@....net>,
        linux-arm-kernel@...ts.infradead.org, linux-doc@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] Documentation/arm64: add memory layout with 4KB pages +
 VA39-bit

On Wed, 6 Oct 2021 18:14:38 +0100
Catalin Marinas <catalin.marinas@....com> wrote:


> 
> 
> On Thu, Sep 30, 2021 at 06:50:26PM +0800, Jisheng Zhang wrote:
> > The 4KB pages + 3 levels (39-bit) combination is also widely used in
> > arm64 world, add the memory layout description for this combination.
> >
> > Signed-off-by: Jisheng Zhang <Jisheng.Zhang@...aptics.com>
> > ---
> >  Documentation/arm64/memory.rst | 17 +++++++++++++++++
> >  1 file changed, 17 insertions(+)
> >
> > diff --git a/Documentation/arm64/memory.rst b/Documentation/arm64/memory.rst
> > index 901cd094f4ec..d1745b570f0c 100644
> > --- a/Documentation/arm64/memory.rst
> > +++ b/Documentation/arm64/memory.rst
> > @@ -26,6 +26,23 @@ The swapper_pg_dir address is written to TTBR1 and never written to
> >  TTBR0.
> >
> >
> > +AArch64 Linux memory layout with 4KB pages + 3 levels (39-bit)::
> > +  Start                      End                     Size            Use
> > +  -----------------------------------------------------------------------
> > +  0000000000000000   0000007fffffffff         512GB          user
> > +  ffffff8000000000   ffffffbfffffffff         256GB          kernel logical memory map
> > + [ffffffb000000000   ffffffbfffffffff]         64GB          [kasan shadow region]
> > +  ffffffc000000000   ffffffc007ffffff         128MB          bpf jit region
> > +  ffffffc008000000   ffffffc00fffffff         128MB          modules
> > +  ffffffc010000000   fffffffdefffffff      253440MB          vmalloc
> > +  fffffffdf0000000   fffffffdfdffffff         224MB          fixed mappings (top down)
> > +  fffffffdfe000000   fffffffdfe7fffff           8MB          [guard region]
> > +  fffffffdfe800000   fffffffdff7fffff          16MB          PCI I/O space
> > +  fffffffdff800000   fffffffdffffffff           8MB          [guard region]
> > +  fffffffe00000000   ffffffffefffffff           4GB          vmemmap
> > +  ffffffff00000000   ffffffffffffffff           4GB          [guard region]  
> 
> I wouldn't bother maintaining these. There are other combinations that
> people may use. The 4KB + 48-bit VA is defconfig while 64KB + 52-bit was
> more interesting, so we thought it would be useful.

If kernel config file isn't based on the defconfig, the 4KB + 39bit VA is the
default option in Kconfig:

...
        prompt "Page size"
        default ARM64_4K_PAGES
        help
          Page size (translation granule) configuration.

...

choice
        prompt "Virtual address space size"
        default ARM64_VA_BITS_39 if ARM64_4K_PAGES
        default ARM64_VA_BITS_47 if ARM64_16K_PAGES
        default ARM64_VA_BITS_42 if ARM64_64K_PAGES
        help
          Allows choosing one of multiple possible virtual address


I believe this combination is widely used if 48bit VA isn't necessary. In
fact, I think the default android GKI image is built with 4KB + 39bit VA. This
is the reason why I added description for 4K + VA39 but leave other
combinations.

> 
> I'm more inclined to remove them altogether and maybe just add some
> high-level ascii art as per the log of commit f4693c2716b3 ("arm64: mm:
> extend linear region for 52-bit VA configurations").

The detailed memory layout is useful, for example, I can immediately know
which region the virtual address belongs to when checking kernel panic logs.
If changed to high level ascii, I need to do some calculation, sometimes I
need to check source code to get the value of some MACROS defitions.

Powered by blists - more mailing lists