Add support for different number of page table levels dependent
on the highest address used for a process. This will cause a 31 bit
process to use a two level page table instead of the four level page
table that is the default after the pud has been introduced. Likewise
a normal 64 bit process will use three levels instead of four. Only
if a process runs out of the 4 tera bytes which can be addressed with
a three level page table the fourth level is dynamically added. Then
the process can use up to 8 peta byte.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
With the new space saving spinlock_t and a non-debug configuration
the struct page only has 32 bytes for 31 bit s390. The causes an
overflow in the calculation of VMEM_MAX_PHYS which renders the
kernel unbootable.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Remove BUILD_BUG_ON() in vmem code since it causes build failures if
the size of struct page increases. Instead calculate at compile time
the address of the highest physical address that can be added to the
1:1 mapping.
This supposed to fix a build failure with the page owner tracking leak
detector patches as reported by akpm.
page-owner-tracking-leak-detector-broken-on-s390.patch can be removed
from -mm again when this is merged.
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Currently the vmalloc area starts at a dynamic address depending on
the memory size. There was also an 8MB security hole after the
physical memory to catch out-of-bounds accesses.
We can simplify the code by putting the vmalloc area explicitely at
the top of the kernel mapping and setting the vmalloc size to a fixed
value of 128MB/128GB for 31bit/64bit systems. Part of the vmalloc
area will be used for the vmem_map. This leaves an area of 96MB/1GB
for normal vmalloc allocations.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Git commit 3610cce87a (yeah my own :-/)
introduced a bug in regard to pud/pmd table entries.
If the address of the page table refered to by a pud/pmd value happens
to have zeroes in the lower 32 bits, pud_present and pmd_present return
false. The obvious effect is that this triggers the BUG_ON in exit_mmap
because some ptes will not get released on process end. Worse is that
the next fault for memory covered by that pud/pmd will allocate another
pmd/pte table and populate the pud/pmd entry. The old page table
entries hanging below this entry are lost!
The fix is simple, properly check against 0. The check is added for
pud_none/pmd_none as well even if these two functions work because
the invalid bit is in the lower 32 bits.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
- De-confuse the defines for the address-space-control-elements
and the segment/region table entries.
- Create out of line functions for page table allocation / freeing.
- Simplify get_shadow_xxx functions.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
The current tlb flushing code for page table entries violates the
s390 architecture in a small detail. The relevant section from the
principles of operation (SA22-7832-02 page 3-47):
"A valid table entry must not be changed while it is attached
to any CPU and may be used for translation by that CPU except to
(1) invalidate the entry by using INVALIDATE PAGE TABLE ENTRY or
INVALIDATE DAT TABLE ENTRY, (2) alter bits 56-63 of a page-table
entry, or (3) make a change by means of a COMPARE AND SWAP AND
PURGE instruction that purges the TLB."
That means if one thread of a multithreaded applciation uses a vma
while another thread does an unmap on it, the page table entries of
that vma needs to get removed with IPTE, IDTE or CSP. In some strange
and rare situations a cpu could check-stop (die) because a entry has
been pushed out of the TLB that is still needed to complete a
(milli-coded) instruction. I've never seen it happen with the current
code on any of the supported machines, so right now this is a
theoretical problem. But I want to fix it nevertheless, to avoid
headaches in the futures.
To get this implemented correctly without changing common code the
primitives ptep_get_and_clear, ptep_get_and_clear_full and
ptep_set_wrprotect need to use the IPTE instruction to invalidate the
pte before the new pte value gets stored. If IPTE is always used for
the three primitives three important operations will have a performace
hit: fork, mprotect and exit_mmap. Time for some workarounds:
* 1: ptep_get_and_clear_full is used in unmap_vmas to remove page
tables entries in a batched tlb gather operation. If the mmu_gather
context passed to unmap_vmas has been started with full_mm_flush==1
or if only one cpu is online or if the only user of a mm_struct is the
current process then the fullmm indication in the mmu_gather context is
set to one. All TLBs for mm_struct are flushed by the tlb_gather_mmu
call. No new TLBs can be created while the unmap is in progress. In
this case ptep_get_and_clear_full clears the ptes with a simple store.
* 2: ptep_get_and_clear is used in change_protection to clear the
ptes from the page tables before they are reentered with the new
access flags. At the end of the update flush_tlb_range clears the
remaining TLBs. In general the ptep_get_and_clear has to issue IPTE
for each pte and flush_tlb_range is a nop. But if there is only one
user of the mm_struct then ptep_get_and_clear uses simple stores
to do the update and flush_tlb_range will flush the TLBs.
* 3: Similar to 2, ptep_set_wrprotect is used in copy_page_range
for a fork to make all ptes of a cow mapping read-only. At the end of
of copy_page_range dup_mmap will flush the TLBs with a call to
flush_tlb_mm. Check for mm->mm_users and if there is only one user
avoid using IPTE in ptep_set_wrprotect and let flush_tlb_mm clear the
TLBs.
Overall for single threaded programs the tlb flush code now performs
better, for multi threaded programs it is slightly worse. In particular
exit_mmap() now does a single IDTE for the mm and then just frees every
page cache reference and every page table page directly without a delay
over the mmu_gather structure.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Prevent that modules get loaded at addresses below 4GB to
prevent exchanging system call table entries.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Nobody is using ptep_test_and_clear_dirty and ptep_clear_flush_dirty. Remove
the functions from all architectures.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The last user of ptep_establish in mm/ is long gone. Remove the architecture
primitive as well.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Kill pte_rdprotect(), pte_exprotect(), pte_mkread(), pte_mkexec(), pte_read(),
pte_exec(), and pte_user() except where arch-specific code is making use of
them.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Some changes done a while ago to avoid pounding on ptep_set_access_flags and
update_mmu_cache in some race situations break sun4c which requires
update_mmu_cache() to always be called on minor faults.
This patch reworks ptep_set_access_flags() semantics, implementations and
callers so that it's now responsible for returning whether an update is
necessary or not (basically whether the PTE actually changed). This allow
fixing the sparc implementation to always return 1 on sun4c.
[akpm@linux-foundation.org: fixes, cleanups]
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: David Miller <davem@davemloft.net>
Cc: Mark Fortescue <mark@mtfhpc.demon.co.uk>
Acked-by: William Lee Irwin III <wli@holomorphy.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The page_test_and_clear_dirty primitive really consists of two
operations, page_test_dirty and the page_clear_dirty. The combination
of the two is not an atomic operation, so it makes more sense to have
two separate operations instead of one.
In addition to the improved readability of the s390 version of
SetPageUptodate, it now avoids the page_test_dirty operation which is
an insert-storage-key-extended (iske) instruction which is an expensive
operation.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
This provides a noexec protection on s390 hardware. Our hardware does
not have any bits left in the pte for a hw noexec bit, so this is a
different approach using shadow page tables and a special addressing
mode that allows separate address spaces for code and data.
As a special feature of our "secondary-space" addressing mode, separate
page tables can be specified for the translation of data addresses
(storage operands) and instruction addresses. The shadow page table is
used for the instruction addresses and the standard page table for the
data addresses.
The shadow page table is linked to the standard page table by a pointer
in page->lru.next of the struct page corresponding to the page that
contains the standard page table (since page->private is not really
private with the pte_lock and the page table pages are not in the LRU
list).
Depending on the software bits of a pte, it is either inserted into
both page tables or just into the standard (data) page table. Pages of
a vma that does not have the VM_EXEC bit set get mapped only in the
data address space. Any try to execute code on such a page will cause a
page translation exception. The standard reaction to this is a SIGSEGV
with two exceptions: the two system call opcodes 0x0a77 (sys_sigreturn)
and 0x0aad (sys_rt_sigreturn) are allowed. They are stored by the
kernel to the signal stack frame. Unfortunately, the signal return
mechanism cannot be modified to use an SA_RESTORER because the
exception unwinding code depends on the system call opcode stored
behind the signal stack frame.
This feature requires that user space is executed in secondary-space
mode and the kernel in home-space mode, which means that the addressing
modes need to be switched and that the noexec protection only works
for user space.
After switching the addressing modes, we cannot use the mvcp/mvcs
instructions anymore to copy between kernel and user space. A new
mvcos instruction has been added to the z9 EC/BC hardware which allows
to copy between arbitrary address spaces, but on older hardware the
page tables need to be walked manually.
Signed-off-by: Gerald Schaefer <geraldsc@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Virtual memmap support for s390. Inspired by the ia64 implementation.
Unlike ia64 we need a mechanism which allows us to dynamically attach
shared memory regions.
These memory regions are accessed via the dcss device driver. dcss
implements the 'direct_access' operation, which requires struct pages
for every single shared page.
Therefore this implementation provides an interface to attach/detach
shared memory:
int add_shared_memory(unsigned long start, unsigned long size);
int remove_shared_memory(unsigned long start, unsigned long size);
The purpose of the add_shared_memory function is to add the given
memory range to the 1:1 mapping and to make sure that the
corresponding range in the vmemmap is backed with physical pages.
It also initialises the new struct pages.
remove_shared_memory in turn only invalidates the page table
entries in the 1:1 mapping. The page tables and the memory used for
struct pages in the vmemmap are currently not freed. They will be
reused when the next segment will be attached.
Given that the maximum size of a shared memory region is 2GB and
in addition all regions must reside below 2GB this is not too much of
a restriction, but there is room for improvement.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
VMALLOC_END on 31bit should be 0x8000000UL instead of 0x7fffffffL.
The page mask which is used to make sure memory_end is on 4MB/2MB
boundary is wrong and not needed. Therefore remove it.
Make sure a vmalloc area does also exist and work on (future)
machines with 4TB and more memory.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
handle_pte_fault uses pte_present, pte_none and pte_file to find out
the type of a pte. That is done without holding the page table lock.
This clashes with the way how ptep_clear_flush removes active page
table entries from the system. First the ipte instruction is used
to invalidate the pte and remove all plt entries for the page. The
ipte sets the hardware invalid bit without changing any other bit.
After the ipte finished the pte is cleared. A concurrent fault can
observe the the previously valid pte with the invalid bit set. With
the current encoding of the different pte types an invalidated
read-only pte can be misinterpreted as a swap-pte.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Convert s390 page handling macros to functions. In particular this fixes a
problem with s390's SetPageUptodate macro which uses its input parameter
twice which again can cause subtle bugs.
[akpm@osdl.org: build fix]
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Major cleanup of all s390 inline assemblies. They now have a common
coding style. Quite a few have been shortened, mainly by using register
asm variables. Use of the EX_TABLE macro helps as well. The atomic ops,
bit ops and locking inlines new use the Q-constraint if a newer gcc
is used. That results in slightly better code.
Thanks to Christian Borntraeger for proof reading the changes.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
One of the changes necessary for shared page tables is to standardize the
pxx_page macros. pte_page and pmd_page have always returned the struct
page associated with their entry, while pte_page_kernel and pmd_page_kernel
have returned the kernel virtual address. pud_page and pgd_page, on the
other hand, return the kernel virtual address.
Shared page tables needs pud_page and pgd_page to return the actual page
structures. There are very few actual users of these functions, so it is
simple to standardize their usage.
Since this is basic cleanup, I am submitting these changes as a standalone
patch. Per Hugh Dickins' comments about it, I am also changing the
pxx_page_kernel macros to pxx_page_vaddr to clarify their meaning.
Signed-off-by: Dave McCracken <dmccr@us.ibm.com>
Cc: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Changed and simplified some page table related #defines and code.
Signed-off-by: Gerald Schaefer <geraldsc@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
NOTE: ZVC are *not* the lightweight event counters. ZVCs are reliable whereas
event counters do not need to be.
Zone based VM statistics are necessary to be able to determine what the state
of memory in one zone is. In a NUMA system this can be helpful for local
reclaim and other memory optimizations that may be able to shift VM load in
order to get more balanced memory use.
It is also useful to know how the computing load affects the memory
allocations on various zones. This patchset allows the retrieval of that data
from userspace.
The patchset introduces a framework for counters that is a cross between the
existing page_stats --which are simply global counters split per cpu-- and the
approach of deferred incremental updates implemented for nr_pagecache.
Small per cpu 8 bit counters are added to struct zone. If the counter exceeds
certain thresholds then the counters are accumulated in an array of
atomic_long in the zone and in a global array that sums up all zone values.
The small 8 bit counters are next to the per cpu page pointers and so they
will be in high in the cpu cache when pages are allocated and freed.
Access to VM counter information for a zone and for the whole machine is then
possible by simply indexing an array (Thanks to Nick Piggin for pointing out
that approach). The access to the total number of pages of various types does
no longer require the summing up of all per cpu counters.
Benefits of this patchset right now:
- Ability for UP and SMP configuration to determine how memory
is balanced between the DMA, NORMAL and HIGHMEM zones.
- loops over all processors are avoided in writeback and
reclaim paths. We can avoid caching the writeback information
because the needed information is directly accessible.
- Special handling for nr_pagecache removed.
- zone_reclaim_interval vanishes since VM stats can now determine
when it is worth to do local reclaim.
- Fast inline per node page state determination.
- Accurate counters in /sys/devices/system/node/node*/meminfo. Current
counters are counting simply which processor allocated a page somewhere
and guestimate based on that. So the counters were not useful to show
the actual distribution of page use on a specific zone.
- The swap_prefetch patch requires per node statistics in order to
figure out when processors of a node can prefetch. This patch provides
some of the needed numbers.
- Detailed VM counters available in more /proc and /sys status files.
References to earlier discussions:
V1 http://marc.theaimsgroup.com/?l=linux-kernel&m=113511649910826&w=2
V2 http://marc.theaimsgroup.com/?l=linux-kernel&m=114980851924230&w=2
V3 http://marc.theaimsgroup.com/?l=linux-kernel&m=115014697910351&w=2
V4 http://marc.theaimsgroup.com/?l=linux-kernel&m=115024767318740&w=2
Performance tests with AIM7 did not show any regressions. Seems to be a tad
faster even. Tested on ia64/NUMA. Builds fine on i386, SMP / UP. Includes
fixes for s390/arm/uml arch code.
This patch:
Move counter code from page_alloc.c/page-flags.h to vmstat.c/h.
Create vmstat.c/vmstat.h by separating the counter code and the proc
functions.
Move the vm_stat_text array before zoneinfo_show.
[akpm@osdl.org: s390 build fix]
[akpm@osdl.org: HOTPLUG_CPU build fix]
Signed-off-by: Christoph Lameter <clameter@sgi.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Fix more include file problems that surfaced since I submitted the previous
fix-missing-includes.patch. This should now allow not to include sched.h
from module.h, which is done by a followup patch.
Signed-off-by: Tim Schmielau <tim@physik3.uni-rostock.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Replace misleading definition of FIRST_USER_PGD_NR 0 by definition of
FIRST_USER_ADDRESS 0 in all the MMU architectures beyond arm and arm26.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.
Let it rip!
Martin Schwidefsky
Heiko Carstens
Christian Borntraeger
Jan Beulich
Benjamin Herrenschmidt
Gerald Schaefer
Dave McCracken
Christoph Lameter
Adrian Bunk
Tim Schmielau
Hugh Dickins
Linus Torvalds