Re: [Qemu-devel] [PATCH v15 1/2] util/mmap-alloc: support MAP_SYNC in qemu_ram_mmap()

[Eduardo Habkost]

On Tue, Apr 30, 2019 at 07:38:59PM -0400, Michael S. Tsirkin wrote:
> On Wed, May 01, 2019 at 01:11:34AM +0200, Paolo Bonzini wrote:
> > On 01/05/19 00:50, Michael S. Tsirkin wrote:
> > >> Stefan, did you hit a build failure, or it was just theoretical?
> > >>
> > >> linux-headers/*/mman.h is updated by "linux-headers: add
> > >> linux/mman.h" (commit 8cf108c5d159).  If the build really fails,
> > >> something else is broken in our build system.
> > > I think it's for non-linux hosts. linux-headers/ is only used
> > > on linux hosts.
> > 
> > Yes, it is.  Maybe the #ifndef/#define  part should be only used for
> > non-Linux.
> > 
> > Paolo
> 
> Makes sense. We'd rather have an error on linux than stub it out as 0.


I'm confused by these replies.  When exactly do you expect an
error on Linux?

For context, I'm talking about the code in v14 of the series (which was already
merged):

#ifdef CONFIG_LINUX
#include <linux/mman.h>
#else  /* !CONFIG_LINUX */
#define MAP_SYNC              0x0
#define MAP_SHARED_VALIDATE   0x0
#endif /* CONFIG_LINUX */

I fail to see any case where the build would fail in v14.

-- 
Eduardo