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>
Proposed fix for ptep_get_and_clear_full PAE bug. Pte_clear had the same bug,
so use the same fix for both. Turns out pmd_clear had it as well, but pgds
are not affected.
The problem is rather intricate. Page table entries in PAE mode are 64-bits
wide, but the only atomic 8-byte write operation available in 32-bit mode is
cmpxchg8b, which is expensive (at least on P4), and thus avoided. But it can
happen that the processor may prefetch entries into the TLB in the middle of an
operation which clears a page table entry. So one must always clear the P-bit
in the low word of the page table entry first when clearing it.
Since the sequence *ptep = __pte(0) leaves the order of the write dependent on
the compiler, it must be coded explicitly as a clear of the low word followed
by a clear of the high word. Further, there must be a write memory barrier
here to enforce proper ordering by the compiler (and, in the future, by the
processor as well).
On > 4GB memory machines, the implementation of pte_clear for PAE was clearly
deficient, as it could leave virtual mappings of physical memory above 4GB
aliased to memory below 4GB in the TLB. The implementation of
ptep_get_and_clear_full has a similar bug, although not nearly as likely to
occur, since the mappings being cleared are in the process of being destroyed,
and should never be dereferenced again.
But, as luck would have it, it is possible to trigger bugs even without ever
dereferencing these bogus TLB mappings, even if the clear is followed fairly
soon after with a TLB flush or invalidation. The problem is that memory above
4GB may now be aliased into the first 4GB of memory, and in fact, may hit a
region of memory with non-memory semantics. These regions include AGP and PCI
space. As such, these memory regions are not cached by the processor. This
introduces the bug.
The processor can speculate memory operations, including memory writes, as long
as they are committed with the proper ordering. Speculating a memory write to
a linear address that has a bogus TLB mapping is possible. Normally, the
speculation is harmless. But for cached memory, it does leave the falsely
speculated cacheline unmodified, but in a dirty state. This cache line will be
eventually written back. If this cacheline happens to intersect a region of
memory that is not protected by the cache coherency protocol, it can corrupt
data in I/O memory, which is generally a very bad thing to do, and can cause
total system failure or just plain undefined behavior.
These bugs are extremely unlikely, but the severity is of such magnitude, and
the fix so simple that I think fixing them immediately is justified. Also,
they are nearly impossible to debug.
Signed-off-by: Zachary Amsden <zach@vmware.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Registering a callback handler through register_die_notifier() is obviously
primarily intended for use by modules. However, the way these currently
get called it is basically impossible for them to actually be used by
modules, as there is, on non-PAE configurationes, a good chance (the larger
the module, the better) for the system to crash as a result.
This is because the callback gets invoked
(a) in the page fault path before the top level page table propagation
gets carried out (hence a fault to propagate the top level page table
entry/entries mapping to module's code/data would nest infinitly) and
(b) in the NMI path, where nested faults must absolutely not happen,
since otherwise the IRET from the nested fault re-enables NMIs,
potentially resulting in nested NMI occurences.
Besides the modular aspect, similar problems would even arise for in-
kernel consumers of the API if they touched ioremap()ed or vmalloc()ed
memory inside their handlers.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Join together some common functions (pmd_page{,_kernel}) over 2level and
3level pages.
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
Acked-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Also, setting PDPEs in PAE mode does not require atomic operations, since the
PDPEs are cached by the processor, and only reloaded on an explicit or
implicit reload of CR3.
Since the four PDPEs must always be present in an active root, and the kernel
PDPE is never updated, we are safe even from SMIs and interrupts / NMIs using
task gates (which reload CR3). Actually, much of this is moot, since the user
PDPEs are never updated either, and the only usage of task gates is by the
doublefault handler. It appears the only place PGDs get updated in PAE mode
is in init_low_mappings() / zap_low_mapping() for initial page table creation
and recovery from ACPI sleep state, and these sites are safe by inspection.
Getting rid of the cmpxchg8b saves code space and 720 cycles in pgd_alloc on
P4.
Signed-off-by: Zachary Amsden <zach@vmware.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!
