Date:   Fri, 22 Sep 2017 11:04:04 +0530
From:   Santosh Sivaraj <santosh@...six.org>
To:     Sergey Senozhatsky <sergey.senozhatsky@...il.com>
Cc:     Tony Luck <tony.luck@...el.com>, Fenghua Yu <fenghua.yu@...el.com>,
        Benjamin Herrenschmidt <benh@...nel.crashing.org>,
        Paul Mackerras <paulus@...ba.org>,
        Michael Ellerman <mpe@...erman.id.au>,
        James Bottomley <jejb@...isc-linux.org>,
        Helge Deller <deller@....de>, Petr Mladek <pmladek@...e.com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Jessica Yu <jeyu@...nel.org>,
        Alexei Starovoitov <ast@...nel.org>,
        linux-ia64@...r.kernel.org, linux-parisc@...r.kernel.org,
        linuxppc-dev@...ts.ozlabs.org, linux-kernel@...r.kernel.org
Subject: Re: [RFC][PATCH v2 0/7] printk/ia64/ppc64/parisc64: let's deprecate
 %pF/%pf printk specifiers

* Sergey Senozhatsky <sergey.senozhatsky@...il.com> wrote (on 2017-09-20 16:29:02 +0000):

>         Hello
> 
>         RFC
> 
>         On some arches C function pointers are indirect and point to
> a function descriptor, which contains the actual pointer to the code.
> This mostly doesn't matter, except for cases when people want to print
> out function pointers in symbolic format, because the usual '%pS/%ps'
> does not work on those arches as expected. That's the reason why we
> have '%pF/%pf', but since it's here because of a subtle ABI detail
> specific to some arches (ppc64/ia64/parisc64) it's easy to misuse
> '%pF/%pf' and '%pS/%ps' (see [1], for example).
> 
>         This patch set attempts to move ia64/ppc64/parisc64 C function
> pointer ABI details out of printk() to arch code. Function dereference
> code now checks if a pointer belongs to a .opd ELF section and dereferences
> that pointer only if it does. The kernel and modules have their own .opd
> sections that's why I use two different ARCH functions: for kernel and
> for module pointer dereference.
> 
>         I planned to remove dereference_function_descriptor() entirely,
> but then I discovered a bunch other uses cases (kgdbts, init/main.c,
> extable, etc.), so I decided to keep dereference_function_descriptor()
> around because the main point of this patch set is to deprecate %pF/%pf.
> But at the same time, I think I can go further and handle both kernel
> and module descriptor dereference in dereference_function_descriptor().
> We need a module pointer for module .opd check, so that will come at an
> extra cost of module lookup (may be there will some other issues along
> the way, haven't checked it).
> 
> Right now we've got:
> 
> - dereference_function_descriptor(addr)
>         a generic (old) function. it simply attempts to dereference
>         whatever pointer we give it.
> 
> - dereference_kernel_function_descriptor(addr)
>         dereferences a kernel pointer if it's within the kernel's .opd
>         section.
> 
> - dereference_module_function_descriptor(module, addr)
>         dereference a module pointer if it's within the module's .opd
>         section.
> 
> 
> *** A BIG NOTE ***
>         I don't own ia64/ppc64/parisc64 hardware, so the patches are not
>         tested. Sorry about that!

Tested patch series on ppc64 sucessfully.

You may add tested by to the series.

Tested-by: Santosh Sivaraj <santosh@...six.org>

Thanks,
Santosh

> 
> Another note:
>         I need to check what is BPF symbol lookup and do we need to
>         do any dereference there.
> 
> v2:
> -- convert dereference_function_descriptor() to unsigned long
> -- fix kernel descriptor range checks (Helge)
> -- fix parisc module descriptor range check (Helge)
> -- fix ppc64 module range check
> -- add checkpatch patch
> 
> 
> Sergey Senozhatsky (7):
>   switch dereference_function_descriptor() to `unsigned long'
>   sections: split dereference_function_descriptor()
>   ia64: Add .opd based function descriptor dereference
>   powerpc64: Add .opd based function descriptor dereference
>   parisc64: Add .opd based function descriptor dereference
>   symbol lookup: use new kernel and module dereference functions
>   checkpatch: add pF/pf deprecation warning
> 
>  Documentation/printk-formats.txt          | 15 +++++----------
>  arch/ia64/include/asm/sections.h          | 16 ++++++++++++----
>  arch/ia64/kernel/module.c                 | 13 +++++++++++++
>  arch/ia64/kernel/vmlinux.lds.S            |  2 ++
>  arch/parisc/boot/compressed/vmlinux.lds.S |  2 ++
>  arch/parisc/include/asm/sections.h        |  4 +++-
>  arch/parisc/kernel/module.c               | 17 +++++++++++++++++
>  arch/parisc/kernel/process.c              | 15 ++++++++++++---
>  arch/parisc/kernel/vmlinux.lds.S          |  2 ++
>  arch/parisc/mm/init.c                     |  4 ++--
>  arch/powerpc/include/asm/module.h         |  3 +++
>  arch/powerpc/include/asm/sections.h       | 17 ++++++++++++++---
>  arch/powerpc/kernel/module_64.c           | 16 ++++++++++++++++
>  arch/powerpc/kernel/vmlinux.lds.S         |  2 ++
>  drivers/misc/kgdbts.c                     |  2 +-
>  include/asm-generic/sections.h            |  8 ++++++--
>  include/linux/moduleloader.h              |  4 ++++
>  init/main.c                               |  2 +-
>  kernel/extable.c                          |  2 +-
>  kernel/kallsyms.c                         |  1 +
>  kernel/module.c                           |  9 ++++++++-
>  lib/vsprintf.c                            |  5 +----
>  scripts/checkpatch.pl                     |  6 ++++--
>  23 files changed, 132 insertions(+), 35 deletions(-)

-- 

Download attachment "signature.asc" of type "application/pgp-signature" (863 bytes)

Powered by blists - more mailing lists