IB/uverbs: Release event file reference on ib_uverbs_create_cq() error
ib_uverbs_create_cq() should release the completion channel event file
if an error occurs after it looks it up. Also, if userspace asks for
a completion channel and we don't find it, an error should be returned
instead of silently creating a CQ without a completion channel.
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
If an operation fails after incrementing an object's reference count,
then it should decrement the reference count on the error path.
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il> Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
Dotan Barak [Fri, 6 Jan 2006 21:23:58 +0000 (13:23 -0800)]
IB/mthca: Add support for automatic path migration (APM)
Add code to modify QP operation to handle setting alternate paths for
connected QPs.
Signed-off-by: Dotan Barak <dotanb@mellanox.co.il> Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
IB/mthca: multiple fixes for multicast group handling
Multicast group management fixes:
. Fix leak of mailbox memory in error handling on multicast group operations.
. Free AMGM indices at detach and in attach error handling.
. Fix amount to shift for aligning next_gid_index in mailbox: it
starts at bit 6, not bit 5.
. Allocate AMGM index after end of MGM table, in the range num_mgms to
multicast table size - 1. Add some BUG_ON checks to catch cases
where the index falls in the MGM hash area.
. Initialize the list of QPs in a newly-allocated group from AMGM to 0
This is necessary since when a group is moved from AMGM to MGM (in the
case where the MGM entry has been emptied of QPs), the AMGM entry is
not reset to 0 (and we don't want an extra command to do that).
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il> Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
Fix a case where copying max_inline_data from a successful create_qp
capabilities output to create_qp input could cause EINVAL error:
mthca_set_qp_size must check max_inline_data directly against
max_desc_sz; checking qp->sq.max_gs is wrong since max_inline_data
depends on the qp type and does not involve max_sg.
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il> Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
Jack Morgenstein [Fri, 16 Dec 2005 03:59:01 +0000 (19:59 -0800)]
IB/mthca: Fix corner cases in max_rd_atomic value handling in modify QP
sae and sre bits should only be set when setting sra_max. Further, in
the old code, if the caller specifies max_rd_atomic = 0, the sre and
sae bits are still set, with the result that the QP ends up with
max_rd_atomic = 1 in effect.
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il> Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
Jack Morgenstein [Thu, 15 Dec 2005 22:36:24 +0000 (14:36 -0800)]
IB/mthca: Fix IB_QP_ACCESS_FLAGS handling.
This patch corrects some corner cases in managing the RAE/RRE bits in
the mthca qp context. These bits need to be zero if the user requests
max_dest_rd_atomic of zero. The bits need to be restored to the value
implied by the qp access flags attribute in a previous (or the
current) modify-qp command if the dest_rd_atomic variable is changed
to non-zero.
In the current implementation, the following scenario will not work:
RESET-to-INIT set QP access flags to all disabled (zeroes)
INIT-to-RTR set max_dest_rd_atomic=10, AND
set qp_access_flags = IB_ACCESS_REMOTE_READ | IB_ACCESS_REMOTE_ATOMIC
The current code will incorrectly take the access-flags value set in
the RESET-to-INIT transition.
We can simplify, and correct, this IB_QP_ACCESS_FLAGS handling: it is
always safe to set qp access flags in the firmware command if either
of IB_QP_MAX_DEST_RD_ATOMIC or IB_QP_ACCESS_FLAGS is set, so let's
just set it to the correct value, always.
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il> Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
Jack Morgenstein [Sat, 10 Dec 2005 00:40:14 +0000 (16:40 -0800)]
IB/mthca: don't change driver's copy of attributes if modify QP fails
Only change the driver's copy of the QP attributes in modify QP after
checking the modify QP command completed successfully.
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il> Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
Jack Morgenstein [Sat, 10 Dec 2005 00:38:04 +0000 (16:38 -0800)]
IB/mthca: correct log2 calculation
Fix thinko in rd_atomic calculation: ffs(x) - 1 does not find the next
power of 2 -- it should be fls(x - 1).
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il> Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
Jack Morgenstein [Sat, 10 Dec 2005 00:32:21 +0000 (16:32 -0800)]
IB/mthca: check RDMA limits
Add limit checking on rd_atomic and dest_rd_atomic attributes:
especially for max_dest_rd_atomic, a value that is larger than HCA
capability can cause RDB overflow and corruption of another QP.
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il> Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
Free the memory allocated in mthca_init_user_db_tab() when releasing
the db_tab in mthca_cleanup_user_db_tab().
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il> Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
Don't leak packet if it had a timeout, and don't leak timeout struct
if queue_packet() fails.
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il> Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
Sean Hefty [Wed, 30 Nov 2005 18:01:13 +0000 (10:01 -0800)]
IB/cm: avoid reusing local ID
Use an increasing local ID to avoid re-using identifiers while
messages may still be outstanding on the old ID. Without this, a
quick connect-disconnect-connect sequence can fail by matching
messages for the new connection with the old connection.
Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
Jack Morgenstein [Wed, 30 Nov 2005 17:55:22 +0000 (09:55 -0800)]
IB/mthca: fix QP size limits for mem-free HCAs
Unlike tavor, the max work queue size is an exact power of 2 for arbel
mode, despite what the documentation (of the QUERY_DEV_LIM firmware
command) says. Without this patch, on Arbel, we can start with a QP
of a valid size and get above the reported limit after rounding to the
next power of two.
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il> Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
Jack Morgenstein [Wed, 30 Nov 2005 00:57:01 +0000 (16:57 -0800)]
IB/uverbs: track multicast group membership for userspace QPs
uverbs needs to track which multicast groups is each qp
attached to, in order to properly detach when cleanup
is performed on device file close.
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il> Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
Roland Dreier [Tue, 29 Nov 2005 05:20:34 +0000 (21:20 -0800)]
IPoIB: reinitialize path struct's completion for every query
It's possible that IPoIB will issue multiple SA queries for the same
path struct. Therefore the struct's completion needs to be
initialized for each query rather than only once when the struct is
allocated, or else we might not wait long enough for later queries to
finish and free the path struct too soon.
ib_umad_write in user_mad.c is looking at rmpp_hdr field in MAD before
checking that the MAD actually has the RMPP header. So for a MAD
without RMPP header it looks like we are actually checking a bit
inside M_Key, or something.
Signed-off-by: Jack Morgenstein <jackm@mellanox.co.il> Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
Trond Myklebust [Fri, 25 Nov 2005 22:10:11 +0000 (17:10 -0500)]
SUNRPC: Funny looking code in __rpc_purge_upcall
In __rpc_purge_upcall (net/sunrpc/rpc_pipe.c), the newer code to clean up
the in_upcall list has a typo.
Thanks to Vince Busam <vbusam@google.com> for spotting this!
Trond Myklebust [Fri, 25 Nov 2005 22:10:06 +0000 (17:10 -0500)]
NFS: Fix a spinlock recursion inside nfs_update_inode()
In cases where the server has gone insane, nfs_update_inode() may end
up calling nfs_invalidate_inode(), which again calls stuff that takes
the inode->i_lock that we're already holding.
In addition, given the sort of things we have in NFS these days that
need to be cleaned up on inode release, I'm not sure we should ever
be calling make_bad_inode().
Fix up spinlock recursion, and limit nfs_invalidate_inode() to clearing
the caches, and marking the inode as being stale.
Thanks to Steve Dickson <SteveD@redhat.com> for spotting this.
David Gibson [Thu, 24 Nov 2005 02:34:56 +0000 (13:34 +1100)]
[PATCH] powerpc: More hugepage boundary case fixes
Blah. The patch [0] I recently sent fixing errors with
in_hugepage_area() and prepare_hugepage_range() for powerpc itself has
an off-by-one bug. Furthermore, the related functions
touches_hugepage_*_range() and within_hugepage_*_range() are also
buggy. Some of the bugs, like those addressed in [0] originated with
commit 7d24f0b8a53261709938ffabe3e00f88f6498df9 where we tweaked the
semantics of where hugepages are allowed. Other bugs have been there
essentially forever, and are due to the undefined behaviour of '<<'
with shift counts greater than the type width (LOW_ESID_MASK could
return non-zero for high ranges with the right congruences).
The good news is that I now have a testsuite which should pick up
things like this if they creep in again.
Stephen Rothwell [Tue, 22 Nov 2005 01:05:26 +0000 (12:05 +1100)]
[PATCH] powerpc: remove arch/powerpc/include hack for 64 bit
With the removal of include/asm-powerpc, we no longer need
arch/powerpc/include/asm for the 64 bit build. We also do not need
-Iarch/powerpc for the 64 bit build either.
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: Paul Mackerras <paulus@samba.org>
Felix Blyakher [Fri, 25 Nov 2005 05:42:13 +0000 (16:42 +1100)]
[XFS] Tight loop in xfs_finish_reclaim_all prevented the xfslogd to run
its queue of IO completion callbacks, thus creating the deadlock between
umount and xfslogd. Breaking the loop solves the problem.
Jasper Spaans [Thu, 24 Nov 2005 15:53:36 +0000 (16:53 +0100)]
[PATCH] fbcon: fix obvious bug in fbcon logo rotation code
This code fixes a tiny problem with the recent fbcon rotation changes:
fb_prepare_logo doesn't check the return value of fb_find_logo and that
causes a crash for my while booting.
Dave Airlie [Thu, 24 Nov 2005 10:41:14 +0000 (21:41 +1100)]
drm: fix quiescent locking
A fix for a locking bug which is triggered when a client tries to lock with
flag DMA_QUIESCENT (typically the X server), but gets interrupted by a signal.
The locking IOCTL should then return an error, but if DMA_QUIESCENT succeeds
it returns 0, and the client falsely thinks it has the lock. In addition
The client waits for DMA_QUISCENT and possibly DMA_READY without having the lock.
From: Thomas Hellstrom Signed-off-by: Dave Airlie <airlied@linux.ie>
David Härdeman [Wed, 23 Nov 2005 23:45:49 +0000 (15:45 -0800)]
[PATCH] USB: fix USB key generates ioctl_internal_command errors issue
On Wed, Nov 16, 2005 at 06:34:24PM -0800, Pete Zaitcev wrote:
>On Wed, 16 Nov 2005 23:52:32 +0100, David Härdeman <david@2gen.com> wrote:
>> usb-storage: waiting for device to settle before scanning
>> Vendor: I0MEGA Model: UMni1GB*IOM2K4 Rev: 1.01
>> Type: Direct-Access ANSI SCSI revision: 02
>> SCSI device sda: 2048000 512-byte hdwr sectors (1049 MB)
>> sda: Write Protect is off
>> sda: Mode Sense: 00 00 00 00
>> sda: assuming drive cache: write through
>> ioctl_internal_command: <8 0 0 0> return code = 8000002
>> : Current: sense key=0x0
>> ASC=0x0 ASCQ=0x0
>> SCSI device sda: 2048000 512-byte hdwr sectors (1049 MB)
>
>I think it's harmless. I saw things like that, and initially I plugged
>them with workarounds like this:
Thanks for the pointer, and yes, it is harmless, but it floods the
console with the messages which hides other (potentially important)
messages...following your example I've made a patch which fixes the
problem.
This should fix a suspend/resume issues that appear with OHCI on some
PPC hardware. The PCI layer should doesn't have the hooks needed for
such ASIC-specific hooks (in this case, software clock gating), so
this moves the code to do that into hcd-pci.c ... where it can be
done after the relevant PCI PM state transition (to/from D3).
David Brownell [Wed, 23 Nov 2005 23:45:37 +0000 (15:45 -0800)]
[PATCH] USB: EHCI updates split init/reinit logic for resume
Moving the PCI-specific parts of the EHCI driver into their own file
created a few issues ... notably on resume paths which (like swsusp)
require re-initializing the controller. This patch:
- Splits the EHCI startup code into run-once HCD setup code and
separate "init the hardware" reinit code. (That reinit code is
a superset of the "early usb handoff" code.)
- Then it makes the PCI init code run both, and the resume code only
run the reinit code.
- It also removes needless pci wrappers around EHCI start/stop methods.
- Removes a byteswap issue that would be seen on big-endian hardware.
The HCD glue still doesn't actually provide a good way to do all this
run-one init stuff in one place though.
David Brownell [Wed, 23 Nov 2005 23:45:28 +0000 (15:45 -0800)]
[PATCH] USB: EHCI updates
This fixes some bugs in EHCI suspend/resume that joined us over the past
few releases (as usbcore, PCI, pmcore, and other components evolved):
- Removes suspend and resume recursion from the EHCI driver, getting
rid of the USB_SUSPEND special casing.
- Updates the wakeup mechanism to work again; there's a newish usbcore
call it needs to use.
- Provide simpler tests for "do we need to restart from scratch", to
address another case where PCI Vaux was lost. (In this case it was
restoring a swsusp snapshot, but there could be others.)
Un-exports a symbol that was temporarily exported.
A notable change from previous version is that this doesn't move
the spinlock init, so there's still a resume/reinit path bug.
Ian Abbott [Wed, 23 Nov 2005 23:45:23 +0000 (15:45 -0800)]
[PATCH] USB: ftdi_sio: new IDs for KOBIL devices
This patch adds two new devices to the ftdi_sio driver's device ID
table. The device IDs were supplied by Stefan Nies of KOBIL Systems for
two of their devices using the FTDI chip.
Damian Wrobel [Wed, 23 Nov 2005 23:45:17 +0000 (15:45 -0800)]
[PATCH] USB: SN9C10x driver - bad page state fix
This patch solves the following problem I've already discovered on the
latest 2.6.15-rc1-git1 kernel:
Nov 13 07:37:28 wrobel kernel: Bad page state at free_hot_cold_page (in process 'motion', page c164e020)
Nov 13 07:37:28 wrobel kernel: flags:0x40000400 mapping:00000000 mapcount:0 count:0
Nov 13 07:37:28 wrobel kernel: Backtrace:
Nov 13 07:37:28 wrobel kernel: [<c0146d86>] bad_page+0x85/0xbe
Nov 13 07:37:28 wrobel kernel: [<c0147629>] free_hot_cold_page+0x54/0x129
Nov 13 07:37:28 wrobel kernel: [<c01598c6>] __vunmap+0xa9/0xfe
Nov 13 07:37:28 wrobel kernel: [<c0154114>] vmalloc_to_page+0x34/0x55
Nov 13 07:37:28 wrobel kernel: [<c0159942>] vfree+0x27/0x35
Nov 13 07:37:28 wrobel kernel: [<f8a20292>] sn9c102_release_buffers+0x30/0x3f [sn9c102]
Nov 13 07:37:28 wrobel kernel: [<f8a231c2>] sn9c102_release+0x37/0xeb [sn9c102]
Nov 13 07:37:28 wrobel kernel: [<c0163e74>] __fput+0xa9/0x1aa
Nov 13 07:37:28 wrobel kernel: [<c01624f7>] filp_close+0x49/0x6d
Nov 13 07:37:30 wrobel kernel: [<c016258f>] sys_close+0x74/0x95
Nov 13 07:37:30 wrobel kernel: [<c0102ef9>] syscall_call+0x7/0xb
Nov 13 07:37:31 wrobel kernel: Trying to fix it up, but a reboot is needed
When attempting to hotadd a PCI card with a bridge on it, I saw
the kernel reporting resource collision errors even when there were
really no collisions. The problem is that the code doesn't skip
over "invalid" resources with their resource type flag not set.
Others have reported similar problems at boot time and for
non-bridge PCI card hotplug too, where the code flags a
resource collision for disabled ROMs. This patch fixes both
problems.
Rajesh Shah [Wed, 23 Nov 2005 23:44:54 +0000 (15:44 -0800)]
[PATCH] PCI Express Hotplug: clear sticky power-fault bit
Per the PCI Express spec, the power-fault-detected bit in the
slot status register can be set anytime hardware detects a power
fault, regardless of whether the slot has a device populated in
it or not. This bit is sticky and must be explicitly cleared.
This patch is needed to allow hot-add after such a power fault
has been detected.
Jean Delvare [Wed, 23 Nov 2005 23:44:31 +0000 (15:44 -0800)]
[PATCH] hwmon: Fix missing it87 fan div init
Fix a bug where setting the low fan speed limits will not work if no
data was ever read through the sysfs interface and the fan clock
dividers have not been explicitely set yet either. The reason is that
data->fan_div[nr] may currently be used before it is initialized from
the chip register values. The fix is to explicitely initialize
data->fan_div[nr] before using it.
Jean Delvare [Wed, 23 Nov 2005 23:44:26 +0000 (15:44 -0800)]
[PATCH] hwmon: Fix lm78 VID conversion
Fix the lm78 VID reading, which I accidentally broke while making
this driver use the common vid_from_reg function rather than
reimplementing its own in 2.6.14-rc1.
Jody McIntyre [Wed, 23 Nov 2005 23:44:03 +0000 (15:44 -0800)]
[PATCH] Clarify T: field in MAINTAINERS
Pavel Machek points out that for git repos, what we include is not
actually a URL. It is undesirable to use a URL since git repos can be
accessed in many different ways.
Olaf Rempel [Thu, 24 Nov 2005 03:04:08 +0000 (19:04 -0800)]
[BRIDGE]: recompute features when adding a new device
We must recompute bridge features everytime the list of underlying
devices changes, or we might end up with features that are not
supported by all devices (eg. NETIF_F_TSO)
This patch adds the missing recompute when adding a device to the bridge.
Signed-off-by: Olaf Rempel <razzor@kopf-tisch.de> Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv4/netfilter/ip_conntrack_netlink.c: In function 'ctnetlink_dump_table':
net/ipv4/netfilter/ip_conntrack_netlink.c:409: warning: implicit declaration of function 'local_bh_disable'
net/ipv4/netfilter/ip_conntrack_netlink.c:427: warning: implicit declaration of function 'local_bh_enable'
Signed-off-by: Benoit Boissinot <benoit.boissinot@ens-lyon.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: David S. Miller <davem@davemloft.net>
David Gibson [Wed, 23 Nov 2005 21:37:45 +0000 (13:37 -0800)]
[PATCH] powerpc: fix for hugepage areas straddling 4GB boundary
Commit 7d24f0b8a53261709938ffabe3e00f88f6498df9 fixed bugs in the ppc64 SLB
miss handler with respect to hugepage handling, and in the process tweaked
the semantics of the hugepage address masks in mm_context_t.
Unfortunately, it left out a couple of necessary changes to go with that
change. First, the in_hugepage_area() macro was not updated to match,
second prepare_hugepage_range() was not updated to correctly handle
hugepages regions which straddled the 4GB point.
The latter appears only to cause process-hangs when attempting to map such
a region, but the former can cause oopses if a get_user_pages() is
triggered at the wrong point. This patch addresses both bugs.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
If unregister_console() is inadvertently called while no consoles are
registered, it will crash trying to dereference NULL pointer. It is
necessary to fix that because register_console() provides no indication
that it actually registered the console passed in. In fact, it may well
decide not to register it based on various things...
(akpm: It'd be better to make register_console() return something and fix the
callers. All 106 of them...)
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Jim Keniston [Wed, 23 Nov 2005 21:37:42 +0000 (13:37 -0800)]
[PATCH] kprobes: Fix return probes on sys_execve
Fix a bug in kprobes that can cause an Oops or even a crash when a return
probe is installed on one of the following functions: sys_execve,
do_execve, load_*_binary, flush_old_exec, or flush_thread. The fix is to
remove the call to kprobe_flush_task() in flush_thread(). This fix has
been tested on all architectures for which the return-probes feature has
been implemented (i386, x86_64, ppc64, ia64). Please apply.
BACKGROUND
Up to now, we have called kprobe_flush_task() under two situations: when a
task exits, and when it execs. Flushing kretprobe_instances on exit is
correct because (a) do_exit() doesn't return, and (b) one or more
return-probed functions may be active when a task calls do_exit(). Neither
is the case for sys_execve() and its callees.
Initially, the mistaken call to kprobe_flush_task() on exec was harmless
because we put the "real" return address of each active probed function
back in the stack, just to be safe, when we recycled its
kretprobe_instance. When support for ppc64 and ia64 was added, this safety
measure couldn't be employed, and was eventually dropped even for i386 and
x86_64. sys_execve() and its callees were informally blacklisted for
return probes until this fix was developed.
Acked-by: Prasanna S Panchamukhi <prasanna@in.ibm.com> Signed-off-by: Jim Keniston <jkenisto@us.ibm.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Hugh Dickins [Wed, 23 Nov 2005 21:37:40 +0000 (13:37 -0800)]
[PATCH] mm: fill arch atomic64 gaps
alpha, sparc64, x86_64 are each missing some primitives from their atomic64
support: fill in the gaps I've noticed by extrapolating asm, follow the
groupings in each file. But powerpc and parisc still lack atomic64.
Signed-off-by: Hugh Dickins <hugh@veritas.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: "David S. Miller" <davem@davemloft.net> Cc: Andi Kleen <ak@muc.de> Cc: Nick Piggin <nickpiggin@yahoo.com.au> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Hugh Dickins [Wed, 23 Nov 2005 21:37:39 +0000 (13:37 -0800)]
[PATCH] mm: powerpc ptlock comments
Update comments (only) on page_table_lock and mmap_sem in arch/powerpc.
Removed the comment on page_table_lock from hash_huge_page: since it's no
longer taking page_table_lock itself, it's irrelevant whether others are; but
how it is safe (even against huge file truncation?) I can't say.
Signed-off-by: Hugh Dickins <hugh@veritas.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Hugh Dickins [Wed, 23 Nov 2005 21:37:38 +0000 (13:37 -0800)]
[PATCH] mm: unbloat get_futex_key
The follow_page changes in get_futex_key have left it with two almost
identical blocks, when handling the rare case of a futex in a nonlinear vma.
get_user_pages will itself do that follow_page, and its additional
find_extend_vma is hardly any overhead since the vma is already cached. Let's
just delete the follow_page block and let get_user_pages do it.
Hugh Dickins [Wed, 23 Nov 2005 21:37:37 +0000 (13:37 -0800)]
[PATCH] mm: update split ptlock Kconfig
Closer attention to the arithmetic shows that neither ppc64 nor sparc really
uses one page for multiple page tables: how on earth could they, while
pte_alloc_one returns just a struct page pointer, with no offset?
Well, arm26 manages it by returning a pte_t pointer cast to a struct page
pointer, harumph, then compensating in its pmd_populate. But arm26 is never
SMP, so it's not a problem for split ptlock either.
And the PA-RISC situation has been recently improved: CONFIG_PA20 works
without the 16-byte alignment which inflated its spinlock_t. But the current
union of spinlock_t with private does make the 7xxx struct page significantly
larger, even without debug, so disable its split ptlock.
Changing the #define to an inline function breaks on non-SMP builds,
since wuite a few places in the kernel do not implement the ipi handler
when compiling for UP.
Dave Airlie [Wed, 23 Nov 2005 11:12:59 +0000 (22:12 +1100)]
drm: move is_pci to the end of the structure
We memset the structure across opens except for the flags. The correct
fix is more intrusive but this should fix a problem with bad iounmaps
seen on AGP radeons acting like PCI ones.
projects / linux-2.6 / log