Background: I've implemented 1K/2K page tables for s390. These sub-page
page tables are required to properly support the s390 virtualization
instruction with KVM. The SIE instruction requires that the page tables
have 256 page table entries (pte) followed by 256 page status table entries
(pgste). The pgstes are only required if the process is using the SIE
instruction. The pgstes are updated by the hardware and by the hypervisor
for a number of reasons, one of them is dirty and reference bit tracking.
To avoid wasting memory the standard pte table allocation should return
1K/2K (31/64 bit) and 2K/4K if the process is using SIE.
Problem: Page size on s390 is 4K, page table size is 1K or 2K. That means
the s390 version for pte_alloc_one cannot return a pointer to a struct
page. Trouble is that with the CONFIG_HIGHPTE feature on x86 pte_alloc_one
cannot return a pointer to a pte either, since that would require more than
32 bit for the return value of pte_alloc_one (and the pte * would not be
accessible since its not kmapped).
Solution: The only solution I found to this dilemma is a new typedef: a
pgtable_t. For s390 pgtable_t will be a (pte *) - to be introduced with a
later patch. For everybody else it will be a (struct page *). The
additional problem with the initialization of the ptl lock and the
NR_PAGETABLE accounting is solved with a constructor pgtable_page_ctor and
a destructor pgtable_page_dtor. The page table allocation and free
functions need to call these two whenever a page table page is allocated or
freed. pmd_populate will get a pgtable_t instead of a struct page pointer.
To get the pgtable_t back from a pmd entry that has been installed with
pmd_populate a new function pmd_pgtable is added. It replaces the pmd_page
call in free_pte_range and apply_to_pte_range.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: <linux-arch@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
asm/elf.h, asm/page.h and asm/user.h don't export to userspace now, so we can
drop #ifdef __KERNEL__ for them.
[k.shutemov@gmail.com: remove #ifdef __KERNEL_]
Signed-off-by: Kirill A. Shutemov <k.shutemov@gmail.com>
Reviewed-by: David Woodhouse <dwmw2@infradead.org>
Cc: <linux-arch@vger.kernel.org>
Signed-off-by: Kirill A. Shutemov <k.shutemov@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Now that copy_to_user_page()/copy_from_user_page() are wired up, we
can drop the old __copy_xxx() implementations. Now that the page
colouring scheme has changed via kmap_coherent(), we can avoid the
flush in these specific helpers.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
With the kmap_coherent() API in place, this is trivial to implement,
and lets us avoid the cache flush in certain cases.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
As noted by David:
pte_page() is a macro defined as follows;
include/asm-sh/pgtable.h
#define pte_page(x) phys_to_page(pte_val(x)&PTE_PHYS_MASK)
include/asm-sh/page.h
#define phys_to_page(phys) (pfn_to_page(phys >> PAGE_SHIFT))
So as you can see the phys_to_page() macro doesn't wrap the 'phys'
parameter in parentheses so we end up with;
pte_val(x)&PTE_PHYS_MASK >> PAGE_SHIFT
Which is not what we wanted as '>>' has a higher precedence than bitwise
AND. I dug into the git repository and I believe this bug was added with
this commit (104b8deaa5);
2006-03-27 KAMEZAWA Hiroyuki [PATCH] unify pfn_to_page: sh pfn_to_page
-#define phys_to_page(phys) (mem_map + (((phys)-__MEMORY_START) >>
PAGE_SHIFT))
-#define page_to_phys(page) (((page - mem_map) << PAGE_SHIFT) +
__MEMORY_START)
+#define phys_to_page(phys) (pfn_to_page(phys >> PAGE_SHIFT))
+#define page_to_phys(page) (page_to_pfn(page) << PAGE_SHIFT)
Reported-by: David ADDISON <david.addison@st.com>
Reported-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
The extended mode TLB requires both 64-bit PTEs and a 64-bit pgprot,
correspondingly, the PGD also has to be 64-bits, so fix that up.
The kernel and user permission bits really are decoupled in early
cuts of the silicon, which means that we also have to set corresponding
kernel permissions on user pages or we end up with user pages that the
kernel simply can't touch (!).
Finally, with those things corrected, really enable MMUCR.ME and
correct the PTEA value (this simply needs to be the upper 32-bits
of the PTE, with the size and protection bit encoding).
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
This reworks the cache mode configuration in Kconfig, and allows for
explicit selection of write-back/write-through/off configurations.
All of the cache flushing routines are optimized away for the off
case.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Slub currently defaults to 8-byte alignment for the kmalloc
and slab minalign values, where 4 will suffice. In the slab
case BYTES_PER_WORD == 4 already, so defining the minalign
values outright doesn't cause any regressions there either.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
pfn_valid() is already defined in the sparsemem case, so we only
need to define this for CONFIG_FLATMEM.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
This adds in some more __user annotations. These weren't being
handled properly in some of the __get_user and __put_user paths,
so tidy those up.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
This reworks some of the node 0 bootmem initialization in
preparation for discontigmem and sparsemem support.
ARCH_POPULATES_NODE_MAP is switched to as a result of this.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
This adds some preliminary support for the SH-X2 MMU, used by
newer SH-4A parts (particularly SH7785).
This MMU implements a 'compat' mode with SH-X MMUs and an
'extended' mode for SH-X2 extended features. Extended features
include additional page sizes (8kB, 4MB, 64MB), as well as the
addition of page execute permissions.
The extended mode attributes are placed in a second data array,
which requires us to switch to 64-bit PTEs when in X2 mode.
With the addition of the exec perms, we also overhaul the mmap
prots somewhat, now that it's possible to handle them more
intelligently.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Set the SHM alignment at runtime, based off of probed cache desc.
Optimize get_unmapped_area() to only colour align shared mappings.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
This implements initial support for the vsyscall page on SH.
At the moment we leave it configurable due to having nommu
to support from the same code base. We hook it up for the
signal trampoline return at present, with more to be added
later, once uClibc catches up.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
We want to be able to use PAGE_SIZE all over the place,
this is the same approach adopted by other architectures..
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
