Re: [PATCH] Documentation: DocBook: kgdb: update CONFIG_STRICT_KERNEL_RWX info

[Daniel Thompson]

On 21/04/17 03:26, Li Qiang wrote:
> @Daniel
>
> 2017-04-20 23:28 GMT+08:00 Daniel Thompson <daniel.thompson@xxxxxxxxxx
> <mailto:daniel.thompson@xxxxxxxxxx>>:
>
>     On 19/04/17 02:58, Li Qiang wrote:
>
>         CONFIG_STRICT_KERNEL_RWX is no longer selectable on most
>         architectures.
>         Update this info to thedocumentation.
>
>
>     "git grep STRICT_KERNEL_RWX" comes up with nothing.
>
>
> It was introduced in commit 0f5bf6d0afe4be6e1391908ff2d6dc9730e91550.

Oops. I did the grep on the wrong machine :-( and therefore on an older 
kernel than I thought...


>     It is selectable on any architecture? If not we should remove it
>     entirely!
>
> The 'STRICT_KERNEL_RWX' is renamed from 'CONFIG_DEBUG_RODATA
> '. The original option is selectable.
>
> I'm not sure is this selectable on any architecture.

So... having found the right kernel, it looks to me like only arm, 
arm64, parisc, s390 and x86 define ARCH_HAS_STRICT_KERNEL_RWX. Of these 
five, only arm defines ARCH_OPTIONAL_KERNEL_RWX and makes it user 
selectable.


> @Jonathan
>
>     On Tue, 18 Apr 2017 18:58:45 -0700
>     Li Qiang <liq3ea@xxxxxxxxx <mailto:liq3ea@xxxxxxxxx>> wrote:
>
>     > CONFIG_STRICT_KERNEL_RWX is no longer selectable on most architectures.
>     > Update this info to the documentation.
>     >
>     > Signed-off-by: Li Qiang <liqiang6-s@xxxxxx <mailto:liqiang6-s@xxxxxx>>
>     > ---
>     >  Documentation/DocBook/kgdb.tmpl | 4 +++-
>     >  1 file changed, 3 insertions(+), 1 deletion(-)
>     >
>     > diff --git a/Documentation/DocBook/kgdb.tmpl b/Documentation/DocBook/kgdb.tmpl
>     > index 856ac20..ef0b67b 100644
>     > --- a/Documentation/DocBook/kgdb.tmpl
>     > +++ b/Documentation/DocBook/kgdb.tmpl
>     > @@ -121,7 +121,9 @@
>     >      If kgdb supports it for the architecture you are using, you can
>     >      use hardware breakpoints if you desire to run with the
>     >      CONFIG_STRICT_KERNEL_RWX option turned on, else you need to turn off
>     > -    this option.
>     > +    this option. In most architectures, this option is not selectable.
>     > +    For this situation, it can be turned off by adding a runtime parameter
>     > +    'rodata=off'.
>
>     So this is an improvement, I guess, though the paragraph remains kind of
>     confusing.  Is there any chance we could actually just say which
>     architectures can use hardware breakpoints, and which should boot with
>     rodata=off?
>
>
> I think this is unnecessary as it is not common to change the
> default CONFIG_STRICT_KERNEL_RWX /add rodata=off.
> We here give this hint because CONFIG_STRICT_KERNEL_RWX is renamed
> from CONFIG_DEBUG_RODATA.
> And the latter is selectable, this can help the peoples who
> think CONFIG_STRICT_KERNEL_RWX  is also selectable.

Having looked at the earlier part of the paragraph I think the info 
about rodata actually needs to be introduced slightly earlier (and 
rodata should be presented as the primary way to do it because 4 of the 
5 architectures don't make STRICT_KERNEL_RWX optional).

Something like:

  If the architecture that you are using supports making the text
  section read-only (CONFIG_STRICT_KERNEL_RWX), you should consider
  turning it off by adding 'rodata=off' to the kernel commandline or,
  if your architecture makes CONFIG_STRICT_KERNEL_RWX optional, by
  disabling this config option. Alternatively, if your architecture
  supports hardware breakpoints, these can be used to provide limited
  breakpoint support if you desire to run with a read-only text section.


Daniel.