mman.h patch

mman.h patch

[Vineet Gupta]

On 02/01/2018 11:43 AM, Waldemar Brodkorb wrote:
> Hi,
>
> I would like to push attached patch next week.
> Any veto?

So it seems there is code consolidation and additionally there's seems to be a 
semantical change to ldso mmap().

Assuming __ARCH_USE_MMU__ is true for say ARC, now we have _MAP_UNINITIALIZED == 0 
vs. before MAP_UNINITIALIZED == 0x4000000.

So we won't be zero init ldso anon mmaps - which could be functionally OK, but I 
think better to split it out as a seperate patch just to call this out !

-Vineet

mman.h patch

[Waldemar Brodkorb]

Hi,
Vineet Gupta wrote,

> On 02/01/2018 11:43 AM, Waldemar Brodkorb wrote:
> >Hi,
> >
> >I would like to push attached patch next week.
> >Any veto?
> 
> So it seems there is code consolidation and additionally there's seems to be
> a semantical change to ldso mmap().
> 
> Assuming __ARCH_USE_MMU__ is true for say ARC, now we have
> _MAP_UNINITIALIZED == 0 vs. before MAP_UNINITIALIZED == 0x4000000.
> 
> So we won't be zero init ldso anon mmaps - which could be functionally OK,
> but I think better to split it out as a seperate patch just to call this out
> !

I believe MAP_UNINITIALIZED is only used by the Linux kernel for
noMMU cases, so the ARC definition was wrong.
Is it really a semantically change, if Linux kernel ignores it for
MMU targets?
May be a better commit log like the attached clarifies it?

best regards
 Waldemar
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0006-bits-mman.h-consolidate-header-file.patch
Type: text/x-diff
Size: 70047 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-snps-arc/attachments/20180202/d9a9e740/attachment-0001.bin>