On Friday 26 August 2005 21:11, Hugh Dickins wrote:
> On Fri, 26 Aug 2005, Blaisorblade wrote:
> > This is a followup to my post of last week (Aug 12) about
> > remap_file_pages protection support. I've improved and consolidated the
> > patches and updated them against 2.6.13-rc6/rc7 (the same patches apply
> > against both versions). I'm sending the full patch series only to akpm,
> > mingo and LKML.
> >
> > I've also reduced them to only 18, and made the splitting more
> > significant. I'm not resending all the patches for foreign architectures,
> > because they're almost unchanged since last time (there's just a trivial
> > reject from ppc32, because one change has already been done after -rc4).
> >
> > I'm working on this to provide support for UML, which currently easily
> > creates more than 64K (the default limit) vma's for a single process.
> > Actually, it needs one VMA per each page. So, with this patch and
> > specific UML support, which Ingo wrote and which I'm porting to recent
> > UMLs.
>
> I'll try to take a look sometime next week - or, if I wait until
> next Friday, can we expect it to have come down to 9 patches ;-?
:-) Don't think so, unless you want just me to join patches together. However,
there are still some oneliners, so the patchset is not so huge.
Well, diffstat seems to contradict me (on the joined-up patch):
Documentation/feature-removal-schedule.txt | 12 +
arch/i386/mm/fault.c | 19 ++
arch/um/kernel/trap_kern.c | 52 ++++++-
arch/x86_64/mm/fault.c | 6
include/asm-i386/mman.h | 1
include/asm-i386/pgtable-2level.h | 15 +-
include/asm-i386/pgtable-3level.h | 11 +
include/asm-i386/pgtable.h | 3
include/asm-ia64/mman.h | 1
include/asm-ppc/mman.h | 1
include/asm-ppc64/mman.h | 1
include/asm-s390/mman.h | 1
include/asm-um/pgtable-2level.h | 15 +-
include/asm-um/pgtable-3level.h | 21 ++-
include/asm-um/pgtable.h | 3
include/asm-x86_64/mman.h | 1
include/asm-x86_64/pgtable.h | 12 +
include/linux/mm.h | 40 ++++--
include/linux/pagemap.h | 32 ++++
mm/filemap.c | 18 ++
mm/fremap.c | 135 +++++++++++++++-----
mm/madvise.c | 2
mm/memory.c | 193
++++++++++++++++++++++-------
mm/mmap.c | 14 +-
mm/mprotect.c | 3
mm/rmap.c | 6
mm/shmem.c | 13 +
> I should say, my initial reaction is very much like Andi's last week.
> sys_remap_file_pages solves a real problem, but it does so by breaking
> lots of rules. For more than a year after it came in, almost every
> development we tried in mm would come up against "but then what do we
> do about the nonlinear mappings?".
Nonuniform mappings are much less of a problem. Really. The reason nonlinear
mappings reached mainline before nonuniform ones (and I don't know if they
willl ever) is not simplicity, but the miss of uses until now. And also the
fact that Ingo hadn't the time to finish it.
In fact, I've been playing a lot with the GIT history during this month of
development, particularly with objrmap, so I've come across those problems
quite a lot, but what I noticed is that you mostly don't care about the VMA
to be uniform, just it to be linear or not (because nonlinear VMAs break
objrmap).
This patch, in comparison, is just:
*) check permissions in the generic VM when faulting in pages, if and only if
the vma is nonuniform (yes, nontrivial at all).
*) be anally picky to save the PTE protections together with the rest, and do
it *everywhere*; but if you say "nonuniform implies linear", you lose this
problem almost completely.
The only exception is that when you remap an address range with PROT_NONE
(thus effectively unmapping those addressed), you can't clear the PTEs but
you must use pfn_pte(0, __S000) to fill them in (this is done in the
optimization-fixup patch #15).
> That has settled down now, but I don't look forward to extending it.
> On the other hand, UML does deserve better support.
> Hugh
--
Inform me of my mistakes, so I can keep imitating Homer Simpson's "Doh!".
Paolo Giarrusso, aka Blaisorblade (Skype ID "PaoloGiarrusso", ICQ 215621894)
http://www.user-mode-linux.org/~blaisorblade
___________________________________
Yahoo! Mail: gratis 1GB per i messaggi e allegati da 10MB
http://mail.yahoo.it
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[Index of Archives]
[Kernel Newbies]
[Netfilter]
[Bugtraq]
[Photo]
[Gimp]
[Yosemite News]
[MIPS Linux]
[ARM Linux]
[Linux Security]
[Linux RAID]
[Video 4 Linux]
[Linux for the blind]
| |
 |