Linus Torvalds [Tue, 23 Oct 2007 02:11:06 +0000 (19:11 -0700)]
Merge branch 'sg' of git://git.kernel.dk/linux-2.6-block
* 'sg' of git://git.kernel.dk/linux-2.6-block:
Add CONFIG_DEBUG_SG sg validation
Change table chaining layout
Update arch/ to use sg helpers
Update swiotlb to use sg helpers
Update net/ to use sg helpers
Update fs/ to use sg helpers
[SG] Update drivers to use sg helpers
[SG] Update crypto/ to sg helpers
[SG] Update block layer to use sg helpers
[SG] Add helpers for manipulating SG entries
Jens Axboe [Mon, 22 Oct 2007 17:57:20 +0000 (19:57 +0200)]
Change table chaining layout
Change the page member of the scatterlist structure to be an unsigned
long, and encode more stuff in the lower bits:
- Bits 0 and 1 zero: this is a normal sg entry. Next sg entry is located
at sg + 1.
- Bit 0 set: this is a chain entry, the next real entry is at ->page_link
with the two low bits masked off.
- Bit 1 set: this is the final entry in the sg entry. sg_next() will return
NULL when passed such an entry.
It's thus important that sg table users use the proper accessors to get
and set the page member.
Update documentation to the current state of affairs. Remove duplicated
method descruptions in exportfs.h and point to Documentation/filesystems/
Exporting instead. Add a little file header comment in expfs.c describing
what's going on and mentioning Neils and my copyright [1].
This one is a lot more complicated than the previous ones. XFS already had a
very clever scheme for supporting 64bit inode numbers in filehandles, and I've
reworked this to be some kind of a prototype for the generic 64bit inode
filehandle support.
Add the guts for the new filesystem API to exportfs.
There's now a fh_to_dentry method that returns a dentry for the object looked
for given a filehandle fragment, and a fh_to_parent operation that returns the
dentry for the encoded parent directory in case the file handle contains it.
There are default implementations for these methods that only take a callback
for an nfs-enhanced iget variant and implement the rest of the semantics.
This patchset is a medium scale rewrite of the export operations interface.
The goal is to make the interface less complex, and easier to understand from
the filesystem side, aswell as preparing generic support for exporting of
64bit inode numbers.
This touches all nfs exporting filesystems, and I've done testing on all of
the filesystems I have here locally (xfs, ext2, ext3, reiserfs, jfs)
This patch:
Add a structured fid type so that we don't have to pass an array of u32 values
around everywhere. It's a union of possible layouts.
As a start there's only the u32 array and the traditional 32bit inode format,
but there will be more in one of my next patchset when I start to document the
various filehandle formats we have in lowlevel filesystems better.
Also add an enum that gives the various filehandle types human- readable
names.
Note: Some people might think the struct containing an anonymous union is
ugly, but I didn't want to pass around a raw union type.
Bernhard Walle [Sun, 21 Oct 2007 23:42:01 +0000 (16:42 -0700)]
kexec: add BSS to resource tree
Add the BSS to the resource tree just as kernel text and kernel data are in
the resource tree. The main reason behind this is to avoid crashkernel
reservation in that area.
While it's not strictly necessary to have the BSS in the resource tree (the
actual collision detection is done in the reserve_bootmem() function before),
the usage of the BSS resource should be presented to the user in /proc/iomem
just as Kernel data and Kernel code.
Note: The patch currently is only implemented for x86 and ia64 (because
efi_initialize_iomem_resources() has the same signature on i386 and ia64).
This patch adds PageSelectiveInvalidation support replacing existing
DomainSelectiveInvalidation for intel_{map/unmap}_sg() calls and also
enables to mapping one big contiguous DMA virtual address which is mapped
to discontiguous physical address for SG map/unmap calls.
"Doamin selective invalidations" wipes out the IOMMU address translation
cache based on domain ID where as "Page selective invalidations" wipes out
the IOMMU address translation cache for that address mask range which is
more cache friendly when compared to Domain selective invalidations.
Here is how it is done.
1) changes to iova.c
alloc_iova() now takes a bool size_aligned argument, which
when when set, returns the io virtual address that is
naturally aligned to 2 ^ x, where x is the order
of the size requested.
Returning this io vitual address which is naturally
aligned helps iommu to do the "page selective
invalidations" which is IOMMU cache friendly
over "domain selective invalidations".
2) Changes to driver/pci/intel-iommu.c
Clean up intel_{map/unmap}_{single/sg} () calls so that
s/g map/unamp calls is no more dependent on
intel_{map/unmap}_single()
intel_map_sg() now computes the total DMA virtual address
required and allocates the size aligned total DMA virtual address
and maps the discontiguous physical address to the allocated
contiguous DMA virtual address.
In the intel_unmap_sg() case since the DMA virtual address
is contiguous and size_aligned, PageSelectiveInvalidation
is used replacing earlier DomainSelectiveInvalidations.
This config option (DMAR_FLPY_WA) sets up 1:1 mapping for the floppy device so
that the floppy device which does not use DMA api's will continue to work.
Once the floppy driver starts using DMA api's this config option can be turn
off or this patch can be yanked out of kernel at that time.
MSI interrupt handler registrations and fault handling support for Intel-IOMMU
hadrware.
This patch enables the MSI interrupts for the DMA remapping units and in the
interrupt handler read the fault cause and outputs the same on to the console.
Introduce intel_iommu=forcedac commandline option. This option is helpful to
verify the pci device capability of handling physical dma'able address greater
than 4G.
Intel IOMMU: Avoid memory allocation failures in dma map api calls
Intel IOMMU driver needs memory during DMA map calls to setup its internal
page tables and for other data structures. As we all know that these DMA map
calls are mostly called in the interrupt context or with the spinlock held by
the upper level drivers(network/storage drivers), so in order to avoid any
memory allocation failure due to low memory issues, this patch makes memory
allocation by temporarily setting PF_MEMALLOC flags for the current task
before making memory allocation calls.
We evaluated mempools as a backup when kmem_cache_alloc() fails
and found that mempools are really not useful here because
1) We don't know for sure how much to reserve in advance
2) And mempools are not useful for GFP_ATOMIC case (as we call
memory alloc functions with GFP_ATOMIC)
(akpm: point 2 is wrong...)
With PF_MEMALLOC flag set in the current->flags, the VM subsystem avoids any
watermark checks before allocating memory thus guarantee'ing the memory till
the last free page. Further, looking at the code in mm/page_alloc.c in
__alloc_pages() function, looks like this flag is useful only in the
non-interrupt context.
If we are in the interrupt context and memory allocation in IOMMU driver fails
for some reason, then the DMA map api's will return failure and it is up to
the higher level drivers to retry. Suppose, if upper level driver programs
the controller with the buggy DMA virtual address, the IOMMU will block that
DMA transaction when that happens thus preventing any corruption to main
memory.
So far in our test scenario, we were unable to create any memory allocation
failure inside dma map api calls.
Actual intel IOMMU driver. Hardware spec can be found at:
http://www.intel.com/technology/virtualization
This driver sets X86_64 'dma_ops', so hook into standard DMA APIs. In this
way, PCI driver will get virtual DMA address. This change is transparent to
PCI drivers.
Intel IOMMU: IOVA allocation and management routines
This code implements a generic IOVA allocation and management. As per Dave's
suggestion we are now allocating IO virtual address from Higher DMA limit
address rather than lower end address and this eliminated the need to preserve
the IO virtual address for multiple devices sharing the same domain virtual
address.
Also this code uses red black trees to store the allocated and reserved iova
nodes. This showed a good performance improvements over previous linear
linked list.
When devices are under a p2p bridge, upstream transactions get replaced by the
device id of the bridge as it owns the PCIE transaction. Hence its necessary
to setup translations on behalf of the bridge as well. Due to this limitation
all devices under a p2p share the same domain in a DMAR.
We just cache the type of device, if its a native PCIe device
or not for later use.
This patch supports the upcomming Intel IOMMU hardware a.k.a. Intel(R)
Virtualization Technology for Directed I/O Architecture and the hardware spec
for the same can be found here
http://www.intel.com/technology/virtualization/index.htm
FAQ! (questions from akpm, answers from ak)
> So... what's all this code for?
>
> I assume that the intent here is to speed things up under Xen, etc?
Yes in some cases, but not this code. That would be the Xen version of this
code that could potentially assign whole devices to guests. I expect this to
be only useful in some special cases though because most hardware is not
virtualizable and you typically want an own instance for each guest.
Ok at some point KVM might implement this too; i likely would use this code
for this.
> Do we
> have any benchmark results to help us to decide whether a merge would be
> justified?
The main advantage for doing it in the normal kernel is not performance, but
more safety. Broken devices won't be able to corrupt memory by doing random
DMA.
Unfortunately that doesn't work for graphics yet, for that need user space
interfaces for the X server are needed.
There are some potential performance benefits too:
- When you have a device that cannot address the complete address range an
IOMMU can remap its memory instead of bounce buffering. Remapping is likely
cheaper than copying.
- The IOMMU can merge sg lists into a single virtual block. This could
potentially speed up SG IO when the device is slow walking SG lists. [I
long ago benchmarked 5% on some block benchmark with an old MPT Fusion; but
it probably depends a lot on the HBA]
And you get better driver debugging because unexpected memory accesses from
the devices will cause a trappable event.
>
> Does it slow anything down?
It adds more overhead to each IO so yes.
This patch:
Add support for early detection and parsing of DMAR's (DMA Remapping) reported
to OS via ACPI tables.
DMA remapping(DMAR) devices support enables independent address translations
for Direct Memory Access(DMA) from Devices. These DMA remapping devices are
reported via ACPI tables and includes pci device scope covered by these DMA
remapping device.
For detailed info on the specification of "Intel(R) Virtualization Technology
for Directed I/O Architecture" please see
http://www.intel.com/technology/virtualization/index.htm
Jan Kara [Sun, 21 Oct 2007 23:41:40 +0000 (16:41 -0700)]
ext2: avoid rec_len overflow with 64KB block size
With 64KB blocksize, a directory entry can have size 64KB which does not
fit into 16 bits we have for entry length. So we store 0xffff instead and
convert the value when read from / written to disk.
J. Bruce Fields [Sun, 21 Oct 2007 23:41:38 +0000 (16:41 -0700)]
dcache: don't expose uninitialized memory in /proc/<pid>/fd/<fd>
Well, it's not especially important that target->d_iname get the contents
of dentry->d_iname, but it's important that it get initialized with
*something*, otherwise we're just exposing some random piece of memory to
anyone who reads the link at /proc/<pid>/fd/<fd> for the deleted file, when
it's still held open by someone.
I've run a test program that copies a short (<36 character) name ontop of a
long (>=36 character) name and see that the first time I run it, without
this patch, I get unpredicatable results out of /proc/<pid>/fd/<fd>.
Yasunori Goto [Sun, 21 Oct 2007 23:41:37 +0000 (16:41 -0700)]
memory hotplug: make kmem_cache_node for SLUB on memory online avoid panic
Fix a panic due to access NULL pointer of kmem_cache_node at discard_slab()
after memory online.
When memory online is called, kmem_cache_nodes are created for all SLUBs
for new node whose memory are available.
slab_mem_going_online_callback() is called to make kmem_cache_node() in
callback of memory online event. If it (or other callbacks) fails, then
slab_mem_offline_callback() is called for rollback.
In memory offline, slab_mem_going_offline_callback() is called to shrink
all slub cache, then slab_mem_offline_callback() is called later.
Yasunori Goto [Sun, 21 Oct 2007 23:41:36 +0000 (16:41 -0700)]
memory hotplug: rearrange memory hotplug notifier
Current memory notifier has some defects yet. (Fortunately, nothing uses
it.) This patch is to fix and rearrange for them.
- Add information of start_pfn, nr_pages, and node id if node status is
changes from/to memoryless node for callback functions.
Callbacks can't do anything without those information.
- Add notification going-online status.
It is necessary for creating per node structure before the node's
pages are available.
- Move GOING_OFFLINE status notification after page isolation.
It is good place for return memory like cache for callback,
because returned page is not used again.
- Make CANCEL events for rollingback when error occurs.
- Delete MEM_MAPPING_INVALID notification. It will be not used.
- Fix compile error of (un)register_memory_notifier().
Rusty Russell [Sun, 21 Oct 2007 23:41:35 +0000 (16:41 -0700)]
i386: paravirt boot sequence
This patch uses the updated boot protocol to do paravirtualized boot.
If the boot version is >= 2.07, then it will do two things:
1. Check the bootparams loadflags to see if we should reload the
segment registers and clear interrupts. This is appropriate
for normal native boot and some paravirtualized environments, but
inapproprate for others.
2. Check the hardware architecture, and dispatch to the appropriate
kernel entrypoint. If the bootloader doesn't set this, then we
simply do the normal boot sequence.
Rusty Russell [Sun, 21 Oct 2007 23:41:33 +0000 (16:41 -0700)]
update boot spec to 2.07
Updates for version 2.07 of the boot protocol. This includes:
load_flags.KEEP_SEGMENTS- flag to request/inhibit segment reloads
hardware_subarch - what subarchitecture we're booting under
hardware_subarch_data - per-architecture data
The intention of these changes is to make booting a paravirtualized
kernel work via the normal Linux boot protocol.
Linus Torvalds [Sun, 21 Oct 2007 16:57:55 +0000 (09:57 -0700)]
Merge branch 'for-linus' of master.kernel.org:/pub/scm/linux/kernel/git/cooloney/blackfin-2.6
* 'for-linus' of master.kernel.org:/pub/scm/linux/kernel/git/cooloney/blackfin-2.6:
Blackfin arch: update boards files
Blackfin arch: dma add some API and cleanup bf54x DMA definition
Blackfin arch: cleanup and promote the general purpose timers api to a core blackfin component
Blackfin arch: add a cheesy install target
Blackfin arch: add functions for converting between sclks and usecs
Blackfin arch: add assembly function for doing 64bit unsigned division
Blackfin arch: -mno-fdpic works
Blackfin arch: use "char bfin_board_name[]" rather than "char *bfin_board_name" per discussion on lkml as the former uses less storage
Blackfin arch: Fixing Bug: balance calls to get_task_mm with corresponding mmput calls
Blackfin serial driver Kconfig: depend on DMA not being enabled rather than a specific DMA size
Blackfin arch: Fix bug: missing CHIPID register field definition of BF54x
Blackfin arch: Fix up /proc/cpuinfo so it is like everyone else
Blackfin arch: Optimization - no need to make additional math here
Blackfin arch: force irq_flags into the .data section
Blackfin arch BF548 defconfig: enable watchdog by default
Blackfin arch: add new processor ADSP-BF52x arch/mach support
Linus Torvalds [Sun, 21 Oct 2007 15:54:32 +0000 (08:54 -0700)]
Merge branch 'audit.b43' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/audit-current
* 'audit.b43' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/audit-current:
[PATCH] audit: watching subtrees
[PATCH] new helper - inotify_evict_watch()
[PATCH] new helper - inotify_clone_watch()
[PATCH] new helpers - collect_mounts() and release_collected_mounts()
[PATCH] pass dentry to audit_inode()/audit_inode_child()
Nick Piggin [Sun, 21 Oct 2007 04:57:41 +0000 (06:57 +0200)]
nobh: nobh_write_end fix
This path mustn't have been tested :( I did attempt to exercise it
by injecting failures here, but I suspect PageMappedToDisk may have
been getting in the way. Will need more of a look, although I think
nobh mode is OK for an -rc1 (it shouldn't eat anyone's data).
Bernd Schmidt [Sun, 21 Oct 2007 14:32:27 +0000 (22:32 +0800)]
Blackfin arch: Fixing Bug: balance calls to get_task_mm with corresponding mmput calls
We must balance calls to get_task_mm with corresponding mmput calls, otherwise
refcounting is screwed up and mms don't get freed when their task exits.
Robin Getz [Sun, 21 Oct 2007 09:03:31 +0000 (17:03 +0800)]
Blackfin arch: Fix up /proc/cpuinfo so it is like everyone else
Fix up /proc/cpuinfo so it is like everyone else, and gets
parsed by various applications properly. Still needs some tweaking on
parts without full L1 sram, like 532, 531, so it doesn't print out L1
bank info that doesn't exist.
Mike Frysinger [Sun, 21 Oct 2007 16:19:31 +0000 (00:19 +0800)]
Blackfin arch: force irq_flags into the .data section
force irq_flags into the .data section by initializing it to
the hardware masks that cannot be disabled. this way if we
use irq enable/disable functions before the .bss has been
zeroed out (as does our l1 relocate/dma functions), we dont
hit a problem where bss contains bogus crap.
Al Viro [Sun, 22 Jul 2007 12:04:18 +0000 (08:04 -0400)]
[PATCH] audit: watching subtrees
New kind of audit rule predicates: "object is visible in given subtree".
The part that can be sanely implemented, that is. Limitations:
* if you have hardlink from outside of tree, you'd better watch
it too (or just watch the object itself, obviously)
* if you mount something under a watched tree, tell audit
that new chunk should be added to watched subtrees
* if you umount something in a watched tree and it's still mounted
elsewhere, you will get matches on events happening there. New command
tells audit to recalculate the trees, trimming such sources of false
positives.
Note that it's _not_ about path - if something mounted in several places
(multiple mount, bindings, different namespaces, etc.), the match does
_not_ depend on which one we are using for access.
Linus Torvalds [Sun, 21 Oct 2007 03:19:15 +0000 (20:19 -0700)]
Merge branch 'master' of hera.kernel.org:/pub/scm/linux/kernel/git/kyle/parisc-2.6
* 'master' of hera.kernel.org:/pub/scm/linux/kernel/git/kyle/parisc-2.6: (29 commits)
[PARISC] fix uninitialized variable warning in asm/rtc.h
[PARISC] Port checkstack.pl to parisc
[PARISC] Make palo target work when $obj != $src
[PARISC] Zap unused variable warnings in pci.c
[PARISC] Fix tests in palo target
[PARISC] Fix palo target
[PARISC] Restore palo target
[PARISC] Attempt to clean up parisc/Makefile
[PARISC] Fix infinite loop in /proc/iomem
[PARISC] Quiet sysfs_create_link __must_check warnings in pdc_stable
[PARISC] Squelch pci_enable_device __must_check warning in superio
[PARISC] Kill off broken irqstack code
[PARISC] Remove hardcoded uses of PAGE_SIZE
[PARISC] Clean up pointless ASM_PAGE_SIZE_DIV use
[PARISC] Kill off the last vestiges of ASM_PAGE_SIZE
[PARISC] Kill off ASM_PAGE_SIZE use
[PARISC] Beautify parisc vmlinux.lds.S
[PARISC] Clean up a resource_size_t warning in sba_iommu
[PARISC] Kill incorrect cast warning in unwinder
[PARISC] Kill zone_to_nid printk warning
...
Fixed trivial conflict in include/asm-parisc/tlbflush.h manually
* ssh://master.kernel.org/pub/scm/linux/kernel/git/sam/kbuild:
kbuild: restore arch/{ppc/xtensa}/boot cflags
kconfig: set title bar in xconfig
kbuild: fix toplevel Makefile/depmod
Kyle McMartin [Sat, 20 Oct 2007 20:23:00 +0000 (13:23 -0700)]
[PARISC] fix uninitialized variable warning in asm/rtc.h
get_rtc_time, in the case that PDC returns that the battery is bad, returns
an unmodified rtc_time arg to the caller, which then uses uninitialized
values. Fix this by memset-ing the arg with zeroes, so it will at least be
cleared if we return failure.
Milton Miller [Sat, 20 Oct 2007 08:58:03 +0000 (03:58 -0500)]
kbuild: restore arch/{ppc/xtensa}/boot cflags
Commit 9a39e273d4df0560c724c5fe71f6314a0583ca2b removed the boot directory
addition to CFLAGS that was being used by the subdirectory builds. For the
other files, that patch set EXTRA_CFLAGS, but Makefile.build explicitly
sets that to empty as it is explicitly for a single directory only.
Append to KBUILD_CFLAGS instead.
David Brownell notes that this causes a regression visible in the
drivers/usb/gadget Kconfig file:
"That Kconfig hasn't changed (other than adding new drivers), and it's
worked that way for several years now ... so the issue seems to be
changes in menuconfig/kconfig/etc semantics.
The issue is that when USB_GADGET=m, it's no longer possible to
configure peripheral controller drivers as modules ... the
controller drivers can now only be configured for static linkage.
It should be making a choice of one of the controller drivers which
could work on the target system, and allow that driver to be linked
either as a module (ok iff USB_GADGET=m) or statically."
Reverting this commit resolves the problem, and also fixes a second
problem that David noticed: various dependent options couldn't be enabled.
* git://git.kernel.org/pub/scm/linux/kernel/git/bunk/trivial: (74 commits)
fix do_sys_open() prototype
sysfs: trivial: fix sysfs_create_file kerneldoc spelling mistake
Documentation: Fix typo in SubmitChecklist.
Typo: depricated -> deprecated
Add missing profile=kvm option to Documentation/kernel-parameters.txt
fix typo about TBI in e1000 comment
proc.txt: Add /proc/stat field
small documentation fixes
Fix compiler warning in smount example program from sharedsubtree.txt
docs/sysfs: add missing word to sysfs attribute explanation
documentation/ext3: grammar fixes
Documentation/java.txt: typo and grammar fixes
Documentation/filesystems/vfs.txt: typo fix
include/asm-*/system.h: remove unused set_rmb(), set_wmb() macros
trivial copy_data_pages() tidy up
Fix typo in arch/x86/kernel/tsc_32.c
file link fix for Pegasus USB net driver help
remove unused return within void return function
Typo fixes retrun -> return
x86 hpet.h: remove broken links
...
Olof Johansson [Sat, 20 Oct 2007 02:04:20 +0000 (21:04 -0500)]
Fix build break in tsi108.c
Fix build break:
drivers/net/tsi108_eth.c: In function 'tsi108_init_one':
drivers/net/tsi108_eth.c:1633: error: expected ')' before 'dev'
drivers/net/tsi108_eth.c:1633: warning: too few arguments for format
make[2]: *** [drivers/net/tsi108_eth.o] Error 1
Drivers that use lro functions should depend on INET, otherwise they
may not link correctly. Let's not select INET. Select should be used
only for library-like code, not to enable subsystems.
Randy Dunlap [Fri, 19 Oct 2007 00:16:20 +0000 (17:16 -0700)]
NAPI: kconfig prompt and deleted doc file
- make the kconfig NAPI option prompt consistent across all net drivers
(other than EXPERIMENTAL; can it now be removed also, or is the new
napi_struct implementation now EXPERIMENTAL ?)
- remove comment about the now-deleted NAPI_HOWTO.txt file
- clean up typos in Tulip NAPI & Interrupt Mitigation
projects / linux.git / log