Re: [PATCH] kernel32: Don't limit physical memory size even if not large address aware.

[Sebastian Lackner]

On 07.06.2017 19:39, Zebediah Figura wrote:
> From testing on Windows 7, it appears Windows doesn't do this either. Also,
> the virtual size is already limited in ntdll, so that check is unnecessary
> as well.
>
> This fixes a bug with sndrec32.exe giving OOM errors when recording.
> Signed-off-by: Zebediah Figura
>
> dlls/kernel32/heap.c | 10 ----------
>  1 file changed, 10 deletions(-)

Are you sure you tested it correctly (with a binary that does not have
IMAGE_FILE_LARGE_ADDRESS_AWARE set)? And what about Adobe Illustrator 8
which depends on this, according to the comment in the source code?

Regards,
Sebastian

[PATCH] kernel32: Don't limit physical memory size even if not large address aware.

[Zebediah Figura]

On 06/07/2017 01:19 PM, Sebastian Lackner wrote:
> On 07.06.2017 19:39, Zebediah Figura wrote:
> > From testing on Windows 7, it appears Windows doesn't do this either. Also,
> > the virtual size is already limited in ntdll, so that check is unnecessary
> > as well.
> >
> > This fixes a bug with sndrec32.exe giving OOM errors when recording.
> > Signed-off-by: Zebediah Figura
> >
> > dlls/kernel32/heap.c | 10 ----------
> >   1 file changed, 10 deletions(-)
>
> Are you sure you tested it correctly (with a binary that does not have
> IMAGE_FILE_LARGE_ADDRESS_AWARE set)? And what about Adobe Illustrator 8
> which depends on this, according to the comment in the source code?
>
> Regards,
> Sebastian

I tested the same binary with IMAGE_FILE_LARGE_ADDRESS_AWARE set and 
unset; the only values that differed were the size of the virtual 
memory. The comment would seem to imply that Adobe Illustrator 8 
requires only that the page size not be modified, unless I'm misreading it.