Jason Uhlenkott [Thu, 19 Jul 2007 08:50:16 +0000 (01:50 -0700)]
drivers/edac: mod race fix i82875p
If ERRSTS indicates that there's no error then we don't need to bother reading
the other registers.
In addition to making the common case faster, this actually fixes a small race
where we don't see an error but we clear the error bits anyway, potentially
wiping away info on an error that happened in the interim (or where a CE
arrives between the first and second read of ERRSTS, causing us to falsely
claim "UE overwrote CE").
Douglas Thompson [Thu, 19 Jul 2007 08:50:15 +0000 (01:50 -0700)]
drivers/edac: add mips and ppc visibility
1) Remove an old CVS ID string
2) change EDAC from a tristate option to a simple bool option
3) In addition to the X86 arch, PPC and MIPS also have drivers in the
submission queue. This patch turns on the EDAC flag for those archs. Each
driver will have its respective 'depends on ARCH' set.
Douglas Thompson [Thu, 19 Jul 2007 08:50:13 +0000 (01:50 -0700)]
drivers/edac: cleanup spaces-gotos after Lindent messup
This patch fixes some remnant spaces inserted by the use of Lindent.
Seems Lindent adds some spaces when it shoulded. These have been fixed.
In addition, goto targets have issues, these have been fixed
in this patch.
Douglas Thompson [Thu, 19 Jul 2007 08:50:10 +0000 (01:50 -0700)]
drivers/edac: add device sysfs attributes
Added new controls for the edac_device and edac_mc sysfs folder.
These can be initialized by the low level driver to provide misc
controls into the low level driver for its use
Douglas Thompson [Thu, 19 Jul 2007 08:50:01 +0000 (01:50 -0700)]
drivers/edac: cleanup workq ifdefs
The origin of this code comes from patches at sourceforge, that
allow EDAC to be updated to various kernels. With kernel version 2.6.20 a
new workq system was installed, thus the patches needed to be modified
based on the kernel version. For submitting to the latest kernel.org
those #ifdefs are removed
Jason Uhlenkott [Thu, 19 Jul 2007 08:49:48 +0000 (01:49 -0700)]
drivers/edac: new inte 30x0 MC driver
Here's a driver for the Intel 3000 and 3010 memory controllers,
relative to today's Sourceforge code drop. This has only had light
testing (I've yet to actually see it handle a memory error) but it
detects my hardware correctly.
Douglas Thompson [Thu, 19 Jul 2007 08:49:47 +0000 (01:49 -0700)]
drivers/edac: mod use edac_core.h
In the refactoring of edac_mc.c into several subsystem files,
the header file edac_mc.h became meaningless. A new header file
edac_core.h was created. All the files that previously included
"edac_mc.h" are changed to include "edac_core.h".
Dave Jiang [Thu, 19 Jul 2007 08:49:46 +0000 (01:49 -0700)]
drivers/edac: add new nmi rescan
Provides a way for NMI reported errors on x86 to notify the EDAC
subsystem pending ECC errors by writing to a software state variable.
Here's the reworked patch. I added an EDAC stub to the kernel so we can
have variables that are in the kernel even if EDAC is a module. I also
implemented the idea of using the chip driver to select error detection
mode via module parameter and eliminate the kernel compile option.
Please review/test. Thx!
Also, I only made changes to some of the chipset drivers since I am
unfamiliar with the other ones. We can add similar changes as we go.
Eric Wollesen [Thu, 19 Jul 2007 08:49:39 +0000 (01:49 -0700)]
drivers/edac: new intel 5000 MC driver
Eric Wollesen ported the Bluesmoke Memory Controller driver (written by Doug
Thompson) for the Intel 5000X/V/P (Blackford/Greencreek) chipset to the in
kernel EDAC model.
This patch incorporates the module for the 5000X/V/P chipset family
Douglas Thompson [Thu, 19 Jul 2007 08:49:36 +0000 (01:49 -0700)]
drivers/edac: add edac_device class
This patch adds the new 'class' of object to be managed, named: 'edac_device'.
As a peer of the 'edac_mc' class of object, it provides a non-memory centric
view of an ERROR DETECTING device in hardware. It provides a sysfs interface
and an abstraction for varioius EDAC type devices.
Multiple 'instances' within the class are possible, with each 'instance'
able to have multiple 'blocks', and each 'block' having 'attributes'.
At the 'block' level there are the 'ce_count' and 'ue_count' fields
which the device driver can update and/or call edac_device_handle_XX()
functions. At each higher level are additional 'total' count fields,
which are a summation of counts below that level.
This 'edac_device' has been used to capture and present ECC errors
which are found in a a L1 and L2 system on a per CORE/CPU basis.
Douglas Thompson [Thu, 19 Jul 2007 08:49:33 +0000 (01:49 -0700)]
drivers/edac: split out functions to unique files
This is a large patch to refactor the original EDAC module in the kernel
and to break it up into better file granularity, such that each source
file contains a given subsystem of the EDAC CORE.
Originally, the EDAC 'core' was contained in one source file: edac_mc.c
with it corresponding edac_mc.h file.
Now, there are the following files:
edac_module.c The main module init/exit function and other overhead
edac_mc.c Code handling the edac_mc class of object
edac_mc_sysfs.c Code handling for sysfs presentation
edac_pci_sysfs.c Code handling for PCI sysfs presentation
edac_core.h CORE .h include file for 'edac_mc' and 'edac_device' drivers
edac_module.h Internal CORE .h include file
This forms a foundation upon which a later patch can create the 'edac_device'
class of object code in a new file 'edac_device.c'.
Douglas Thompson [Thu, 19 Jul 2007 08:49:31 +0000 (01:49 -0700)]
drivers/edac: add edac_mc_find API
This simple patch adds an important CORE API for EDAC that EDAC drivers can
use to find their edac_mc control structure by passing a mem_ctl_info
'instance' value
Rusty Russell [Thu, 19 Jul 2007 08:49:22 +0000 (01:49 -0700)]
lguest: the guest code
lguest is a simple hypervisor for Linux on Linux. Unlike kvm it doesn't need
VT/SVM hardware. Unlike Xen it's simply "modprobe and go". Unlike both, it's
5000 lines and self-contained.
Performance is ok, but not great (-30% on kernel compile). But given its
hackability, I expect this to improve, along with the paravirt_ops code which
it supplies a complete example for. There's also a 64-bit version being
worked on and other craziness.
But most of all, lguest is awesome fun! Too much of the kernel is a big ball
of hair. lguest is simple enough to dive into and hack, plus has some warts
which scream "fork me!".
This patch:
This is the code and headers required to make an i386 kernel an lguest guest.
Rusty Russell [Thu, 19 Jul 2007 08:49:21 +0000 (01:49 -0700)]
lguest: export symbols for lguest as a module
lguest does some fairly lowlevel things to support a host, which
normal modules don't need:
math_state_restore:
When the guest triggers a Device Not Available fault, we need
to be able to restore the FPU
__put_task_struct:
We need to hold a reference to another task for inter-guest
I/O, and put_task_struct() is an inline function which calls
__put_task_struct.
access_process_vm:
We need to access another task for inter-guest I/O.
map_vm_area & __get_vm_area:
We need to map the switcher shim (ie. monitor) at 0xFFC01000.
J. Bruce Fields [Thu, 19 Jul 2007 08:49:18 +0000 (01:49 -0700)]
nfsd: return errors, not NULL, from export functions
I converted the various export-returning functions to return -ENOENT instead
of NULL, but missed a few cases.
This particular case could cause actual bugs in the case of a krb5 client that
doesn't match any ip-based client and that is trying to access a filesystem
not exported to krb5 clients.
J. Bruce Fields [Thu, 19 Jul 2007 08:49:18 +0000 (01:49 -0700)]
nfsd: fix possible read-ahead cache and export table corruption
The value of nperbucket calculated here is too small--we should be rounding up
instead of down--with the result that the index j in the following loop can
overflow the raparm_hash array. At least in my case, the next thing in memory
turns out to be export_table, so the symptoms I see are crashes caused by the
appearance of four zeroed-out export entries in the first bucket of the hash
table of exports (which were actually entries in the readahead cache, a
pointer to which had been written to the export table in this initialization
code).
Andrew Morton [Thu, 19 Jul 2007 08:49:17 +0000 (01:49 -0700)]
move page writeback acounting out of macros
page-writeback accounting is presently performed in the page-flags macros.
This is inconsistent and a bit ugly and makes it awkward to implement
per-backing_dev under-writeback page accounting.
Remove is_in_rom() function. It doesn't actually serve the purpose it was
intended to. If you look at the use of it _access_ok() (which is the only use
of it) then it is obvious that most of memory is marked as access_ok. No
point having is_in_rom() then, so remove it.
In die_if_kernel() start the stack dump at the exception-time SP, not at the
SP with all the saved registers; the stack below exception-time sp contains
only exception-saved values and is already printed in details just before.
The fix to that race in alloc_fresh_huge_page() which could give an illegal
node ID did not need nid_lock at all: the fix was to replace static int nid
by static int prev_nid and do the work on local int nid. nid_lock did make
sure that racers strictly roundrobin the nodes, but that's not something we
need to enforce strictly. Kill nid_lock.
There is check_reset() -- global function in drivers/isdn/sc/
There is check_reset -- variable holding module param in aacraid driver.
On allyesconfig they clash with:
LD drivers/built-in.o
drivers/isdn/built-in.o: In function `check_reset':
: multiple definition of `check_reset'
drivers/scsi/built-in.o:(.data+0xe458): first defined here
ld: Warning: size of symbol `check_reset' changed from 4 in drivers/scsi/built-in.o to 219 in drivers/isdn/built-in.o
ld: Warning: type of symbol `check_reset' changed from 1 to 2 in drivers/isdn/built-in.o
David Howells [Thu, 19 Jul 2007 08:49:09 +0000 (01:49 -0700)]
FRV: work around a possible compiler bug
Work around a possible bug in the FRV compiler.
What appears to be happening is that gcc resolves the
__builtin_constant_p() in kmalloc() to true, but then fails to reduce the
therefore constant conditions in the if-statements it guards to constant
results.
When compiling with -O2 or -Os, one single spurious error crops up in
cpuup_callback() in mm/slab.c. This can be avoided by making the memsize
variable const.
Johannes Berg [Thu, 19 Jul 2007 08:49:02 +0000 (01:49 -0700)]
lockdep debugging: give stacktrace for init_error
When I started adding support for lockdep to 64-bit powerpc, I got a
lockdep_init_error and with this patch was able to pinpoint why and where
to put lockdep_init(). Let's support this generally for others adding
lockdep support to their architecture.
Peter Zijlstra [Thu, 19 Jul 2007 08:48:59 +0000 (01:48 -0700)]
lockdep: various fixes
- update the copyright notices
- use the default hash function
- fix a thinko in a BUILD_BUG_ON
- add a WARN_ON to spot inconsitent naming
- fix a termination issue in /proc/lock_stat
'contentions' and 'acquisitions' are the number of such events measured (since
the last reset). The waittime- and holdtime- (min, max, total) numbers are
presented in microseconds.
If there are any contention points, the lock class is presented in the block
format (as i_mutex and tree_lock above), otherwise a single line of output is
presented.
The output is sorted on absolute number of contentions (read + write), this
should get the worst offenders presented first, so that:
Peter Zijlstra [Thu, 19 Jul 2007 08:48:56 +0000 (01:48 -0700)]
lockstat: core infrastructure
Introduce the core lock statistics code.
Lock statistics provides lock wait-time and hold-time (as well as the count
of corresponding contention and acquisitions events). Also, the first few
call-sites that encounter contention are tracked.
Lock wait-time is the time spent waiting on the lock. This provides insight
into the locking scheme, that is, a heavily contended lock is indicative of
a too coarse locking scheme.
Lock hold-time is the duration the lock was held, this provides a reference for
the wait-time numbers, so they can be put into perspective.
1)
lock
2)
... do stuff ..
unlock
3)
The time between 1 and 2 is the wait-time. The time between 2 and 3 is the
hold-time.
The lockdep held-lock tracking code is reused, because it already collects locks
into meaningful groups (classes), and because it is an existing infrastructure
for lock instrumentation.
Currently lockdep tracks lock acquisition with two hooks:
lock()
lock_acquire()
_lock()
... code protected by lock ...
unlock()
lock_release()
_unlock()
We need to extend this with two more hooks, in order to measure contention.
lock_contended() - used to measure contention events
lock_acquired() - completion of the contention
These are then placed the following way:
lock()
lock_acquire()
if (!_try_lock())
lock_contended()
_lock()
lock_acquired()
... do locked stuff ...
unlock()
lock_release()
_unlock()
(Note: the try_lock() 'trick' is used to avoid instrumenting all platform
dependent lock primitive implementations.)
It is also possible to toggle the two lockdep features at runtime using:
Peter Zijlstra [Thu, 19 Jul 2007 08:48:53 +0000 (01:48 -0700)]
lockdep: sanitise CONFIG_PROVE_LOCKING
Ensure that all of the lock dependency tracking code is under
CONFIG_PROVE_LOCKING. This allows us to use the held lock tracking code for
other purposes.
Peter Zijlstra [Thu, 19 Jul 2007 08:48:53 +0000 (01:48 -0700)]
fix raw_spinlock_t vs lockdep
Use the lockdep infrastructure to track lock contention and other lock
statistics.
It tracks lock contention events, and the first four unique call-sites that
encountered contention.
It also measures lock wait-time and hold-time in nanoseconds. The minimum and
maximum times are tracked, as well as a total (which together with the number
of event can give the avg).
All statistics are done per lock class, per write (exclusive state) and per read
(shared state).
The statistics are collected per-cpu, so that the collection overhead is
minimized via having no global cachemisses.
This new lock statistics feature is independent of the lock dependency checking
traditionally done by lockdep; it just shares the lock tracking code. It is
also possible to enable both and runtime disabled either component - thereby
avoiding the O(n^2) lock chain walks for instance.
This patch:
raw_spinlock_t should not use lockdep (and doesn't) since lockdep itself
relies on it.
Jan Harkes [Thu, 19 Jul 2007 08:48:46 +0000 (01:48 -0700)]
coda: block signals during upcall processing
We ignore signals for about 30 seconds to give userspace a chance to see the
upcall. As we did not block signals we ended up in a busy loop for the
remainder of the period when a signal is received.
Jan Harkes [Thu, 19 Jul 2007 08:48:46 +0000 (01:48 -0700)]
coda: cleanup for upcall handling path
Make the code that processes upcall responses more straightforward, uncovered
at least one bad assumption. We trusted that vc_inuse would be 0 when upcalls
are aborted, however the device may have been reopened.
Jan Harkes [Thu, 19 Jul 2007 08:48:43 +0000 (01:48 -0700)]
coda: allow removal of busy directories
A directory without children may still be busy when it is the cwd for some
process. We can safely remove such a directory because the VFS prevents
further operations. Also we don't need to call d_delete as it is already
called in vfs_rmdir.
Jan Harkes [Thu, 19 Jul 2007 08:48:43 +0000 (01:48 -0700)]
coda: fix nlink updates for directories
The Coda client sets the directory link count to 1 when it isn't sure how many
subdirectories we have. In this case we shouldn't change the link count in
the kernel when a subdirectory is created or removed.
projects / linux.git / log