Re: [2/2] ntdll: Allow reading from a VPROT_WRITECOPY protected memory

[Dmitry Timoshkov]

André Hentschel wrote:
> As per tests in kernel32:loader it seems linux&co allows reads on PROT_WRITE areas, but solaris doesn't and crashes.
> This fixes this issue.
> -        if (vprot & VPROT_WRITECOPY) prot |= PROT_WRITE;
> +        if (vprot & VPROT_WRITECOPY) prot |= PROT_READ | PROT_WRITE;

Do you mean that the recently added test for resource section access
mapping crashes on Solaris? If yes, then the fix should be done in different
place.

-- 
Dmitry.

[2/2] ntdll: Allow reading from a VPROT_WRITECOPY protected memory

[Francois Gouget]

On Tue, 6 Sep 2011, Dmitry Timoshkov wrote:
> André Hentschel wrote:
> > As per tests in kernel32:loader it seems linux&co allows reads on PROT_WRITE
> > areas, but solaris doesn't and crashes.
> > This fixes this issue.
> > -        if (vprot & VPROT_WRITECOPY) prot |= PROT_WRITE;
> > +        if (vprot & VPROT_WRITECOPY) prot |= PROT_READ | PROT_WRITE;
>
> Do you mean that the recently added test for resource section access
> mapping crashes on Solaris?

The kernel32:loader test was crashing on line 660:

    ok(!memcmp((const char *)info.BaseAddress, section_data, 
       section.SizeOfRawData), "wrong section data\n");

Further that was just for the following entries:
        { IMAGE_SCN_MEM_WRITE, PAGE_WRITECOPY },
and
        { IMAGE_SCN_MEM_WRITE | IMAGE_SCN_MEM_EXECUTE, PAGE_EXECUTE_WRITECOPY },
> If yes, then the fix should be done in different place.

What do you suggest?


-- 
Francois Gouget <fgouget(a)free.fr>              http://fgouget.free.fr/
A particle is an irreducible representation of the Poincaré Group - Eugene Wigner

[2/2] ntdll: Allow reading from a VPROT_WRITECOPY protected memory

[Dmitry Timoshkov]

Francois Gouget  wrote:
> The kernel32:loader test was crashing on line 660:
>
> ok(!memcmp((const char *)info.BaseAddress, section_data, 
>        section.SizeOfRawData), "wrong section data\n");
>
> Further that was just for the following entries:
>         { IMAGE_SCN_MEM_WRITE, PAGE_WRITECOPY },
> and
>         { IMAGE_SCN_MEM_WRITE | IMAGE_SCN_MEM_EXECUTE, PAGE_EXECUTE_WRITECOPY },
> > If yes, then the fix should be done in different place.
>
> What do you suggest?

The relevant part of
http://source.winehq.org/git/wine.git/commitdiff/3d81e97c753a73646081b1084ad...

is

- if (sec->Characteristics & IMAGE_SCN_MEM_WRITE) vprot |= VPROT_READ|VPROT_WRITE;
+ if (sec->Characteristics & IMAGE_SCN_MEM_WRITE) vprot |= VPROT_WRITECOPY;

So, probably restore adding VPROT_READ. I should note, that VPROT_READ
not always was there, I added it long time ago for a broken DLL which was
crashing on Linux. Perhaps Linux has been fixed since then.

-- 
Dmitry.

[2/2] ntdll: Allow reading from a VPROT_WRITECOPY protected memory

[Francois Gouget]

On Tue, 6 Sep 2011, Dmitry Timoshkov wrote:
[...]
> - if (sec->Characteristics & IMAGE_SCN_MEM_WRITE) vprot |=
> VPROT_READ|VPROT_WRITE;
> + if (sec->Characteristics & IMAGE_SCN_MEM_WRITE) vprot |= VPROT_WRITECOPY;
>
> So, probably restore adding VPROT_READ.

Yep, that works.
André, I'll let you do the honors for the patch.


-- 
Francois Gouget <fgouget(a)free.fr>              http://fgouget.free.fr/
                            $live{free} || die "";