[PATCH] Enforce physical alignment based upon size for the SPARC32 romvec obp_dumb_memalloc() function.

[Mark Cave-Ayland]

It appears that Solaris assumes that the resulting memory is physically aligned based upon the
value of size. Make sure we also respect this alignment, so that memory allocated using this
interface can be correctly used for the Solaris IOMMU page tables.

Signed-off-by: Mark Cave-Ayland 
---
 openbios-devel/arch/sparc32/lib.c |   14 ++++++++++++--
 1 files changed, 12 insertions(+), 2 deletions(-)

diff --git a/openbios-devel/arch/sparc32/lib.c b/openbios-devel/arch/sparc32/lib.c
index 5b41dff..0219a19 100644
--- a/openbios-devel/arch/sparc32/lib.c
+++ b/openbios-devel/arch/sparc32/lib.c
@@ -351,9 +351,19 @@ char *obp_dumb_memalloc(char *va, unsigned int size)
 {
     phys_addr_t phys;
     ucell virt;
-
+    unsigned int align;
+    
+    /* Solaris seems to assume that the returned value is physically aligned to size. For
+       example, not having this here causes the Solaris 8 kernel to fault because the 
+       IOMMU page table base address is calculated incorrectly. */
+
+    /* Enforce a minimum alignment of CONFIG_OFMEM_MALLOC_ALIGN, and also ensure that
+       only the MSB is set for the alignment value. This prevents spurious
+       "bad alignment" warnings from OFMEM when running with CONFIG_DEBUG_OFMEM. */
+    align = (size + CONFIG_OFMEM_MALLOC_ALIGN - 1) & ~(size - 1);
+    
     /* Claim physical memory */
-    phys = ofmem_claim_phys(-1, size, CONFIG_OFMEM_MALLOC_ALIGN);
+    phys = ofmem_claim_phys(-1, size, align);

/* Claim virtual memory */
     virt = ofmem_claim_virt(pointer2cell(va), size, 0);

[PATCH] Enforce physical alignment based upon size for the SPARC32 romvec obp_dumb_memalloc() function.

[Blue Swirl]

On Mon, Apr 4, 2011 at 10:44 PM, Mark Cave-Ayland
 wrote:
> It appears that Solaris assumes that the resulting memory is physically aligned based upon the
> value of size. Make sure we also respect this alignment, so that memory allocated using this
> interface can be correctly used for the Solaris IOMMU page tables.
> Signed-off-by: Mark Cave-Ayland
>
> openbios-devel/arch/sparc32/lib.c |   14 ++++++++++++--
>  1 files changed, 12 insertions(+), 2 deletions(-)
>
> diff --git a/openbios-devel/arch/sparc32/lib.c b/openbios-devel/arch/sparc32/lib.c
> index 5b41dff..0219a19 100644
> --- a/openbios-devel/arch/sparc32/lib.c
> +++ b/openbios-devel/arch/sparc32/lib.c
> @@ -351,9 +351,19 @@ char *obp_dumb_memalloc(char *va, unsigned int size)
>  {
>     phys_addr_t phys;
>     ucell virt;
> unsigned int align;
> /* Solaris seems to assume that the returned value is physically aligned to size. For
> example, not having this here causes the Solaris 8 kernel to fault because the
> IOMMU page table base address is calculated incorrectly. */

Awesome work!
> /* Enforce a minimum alignment of CONFIG_OFMEM_MALLOC_ALIGN, and also ensure that
> only the MSB is set for the alignment value. This prevents spurious
> "bad alignment" warnings from OFMEM when running with CONFIG_DEBUG_OFMEM. */
> align = (size + CONFIG_OFMEM_MALLOC_ALIGN - 1) & ~(size - 1);

What happens if size is not a power of 2?

[PATCH] Enforce physical alignment based upon size for the SPARC32 romvec obp_dumb_memalloc() function.

[Mark Cave-Ayland]

On 04/04/11 22:15, Blue Swirl wrote:
> > /* Enforce a minimum alignment of CONFIG_OFMEM_MALLOC_ALIGN, and also ensure that
> >   only the MSB is set for the alignment value. This prevents spurious
> >   "bad alignment" warnings from OFMEM when running with CONFIG_DEBUG_OFMEM. */
> > align = (size + CONFIG_OFMEM_MALLOC_ALIGN - 1)&  ~(size - 1);
>
> What happens if size is not a power of 2?

Hmmmm that's a good point actually. I've just done a quick test of a 
revised patch that uses the standard bit-rotation algorithm to round up 
to the next power of 2, and that seems to work in the same way while 
making things more robust. Patch to follow shortly.

ATB,

Mark.