| Message ID | 1368442894.537.59.camel@zakaz.uk.xensource.com |
|---|---|
| State | New |
| Headers | show |
>>> On 13.05.13 at 13:01, Ian Campbell <Ian.Campbell@citrix.com> wrote: > On Fri, 2013-05-10 at 16:11 +0100, Julien Grall wrote: >> ioremap function can unlikely return an unaligned virtual address if >> the physical address itself is unaligned on a page size. > > On x86 this causes: > In file included from > /local/scratch/ianc/devel/committer.git/xen/include/asm/io.h:5, > from > /local/scratch/ianc/devel/committer.git/xen/include/asm/mc146818rtc.h:8, > from rtc.c:26: > /local/scratch/ianc/devel/committer.git/xen/include/xen/vmap.h: In function > ‘iounmap’: > /local/scratch/ianc/devel/committer.git/xen/include/xen/vmap.h:20: error: > ‘PAGE_MASK’ undeclared (first use in this function) > /local/scratch/ianc/devel/committer.git/xen/include/xen/vmap.h:20: error: > (Each undeclared identifier is reported only once > /local/scratch/ianc/devel/committer.git/xen/include/xen/vmap.h:20: error: > for each function it appears in.) > make[5]: *** [rtc.o] Error 1 > make[5]: *** Waiting for unfinished jobs.... With this and the other two bugs that the recent "ARM-sharing- more-code-with-x86" effort introduced, I wonder whether it's really appropriate for doing this sort of stuff - without proper testing on the x86 side - during the RC phase. Jan
On Mon, 2013-05-13 at 12:09 +0100, Jan Beulich wrote: > >>> On 13.05.13 at 13:01, Ian Campbell <Ian.Campbell@citrix.com> wrote: > > On Fri, 2013-05-10 at 16:11 +0100, Julien Grall wrote: > >> ioremap function can unlikely return an unaligned virtual address if > >> the physical address itself is unaligned on a page size. > > > > On x86 this causes: > > In file included from > > /local/scratch/ianc/devel/committer.git/xen/include/asm/io.h:5, > > from > > /local/scratch/ianc/devel/committer.git/xen/include/asm/mc146818rtc.h:8, > > from rtc.c:26: > > /local/scratch/ianc/devel/committer.git/xen/include/xen/vmap.h: In function > > ‘iounmap’: > > /local/scratch/ianc/devel/committer.git/xen/include/xen/vmap.h:20: error: > > ‘PAGE_MASK’ undeclared (first use in this function) > > /local/scratch/ianc/devel/committer.git/xen/include/xen/vmap.h:20: error: > > (Each undeclared identifier is reported only once > > /local/scratch/ianc/devel/committer.git/xen/include/xen/vmap.h:20: error: > > for each function it appears in.) > > make[5]: *** [rtc.o] Error 1 > > make[5]: *** Waiting for unfinished jobs.... > > With this and the other two bugs that the recent "ARM-sharing- > more-code-with-x86" effort introduced, I wonder whether it's really > appropriate for doing this sort of stuff - without proper testing on > the x86 side - during the RC phase. That's a good question, which I'm afraid I didn't see until after I had hit push. Sorry. We could revert but I'd rather not. If it is any consolation I think this was the last major bit of ARM functionality which was to be given such a broad freeze exception. (I hope Stefano or Julien will correct me ASAP if this is wrong) Ian.
diff --git a/xen/include/xen/vmap.h b/xen/include/xen/vmap.h index daee7f5..b1923dd 100644 --- a/xen/include/xen/vmap.h +++ b/xen/include/xen/vmap.h @@ -2,6 +2,7 @@ #define __XEN_VMAP_H__ #include <xen/types.h> +#include <asm/page.h> void *vm_alloc(unsigned int nr, unsigned int align); void vm_free(const void *);