Instead of always assigning an increasing inode number in new_inode
move the call to assign it into those callers that actually need it.
For now callers that need it is estimated conservatively, that is
the call is added to all filesystems that do not assign an i_ino
by themselves. For a few more filesystems we can avoid assigning
any inode number given that they aren't user visible, and for others
it could be done lazily when an inode number is actually needed,
but that's left for later patches.
Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Dave Chinner <dchinner@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Eric Dumazet [Sat, 23 Oct 2010 15:18:01 +0000 (11:18 -0400)]
fs: introduce a per-cpu last_ino allocator
new_inode() dirties a contended cache line to get increasing
inode numbers. This limits performance on workloads that cause
significant parallel inode allocation.
Solve this problem by using a per_cpu variable fed by the shared
last_ino in batches of 1024 allocations. This reduces contention on
the shared last_ino, and give same spreading ino numbers than before
(i.e. same wraparound after 2^32 allocations).
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: Nick Piggin <npiggin@suse.de> Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Split up inode_add_to_list/__inode_add_to_list. Locking for the two
lists will be split soon so these helpers really don't buy us much
anymore.
The __ prefixes for the sb list helpers will go away soon, but until
inode_lock is gone we'll need them to distinguish between the locked
and unlocked variants.
Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Stop abusing find_inode_fast for iunique and opencode the inode hash walk.
Introduce a new iunique_lock to protect the iunique counters once inode_lock
is removed.
Based on a patch originally from Nick Piggin.
Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Dave Chinner [Sat, 23 Oct 2010 10:58:09 +0000 (06:58 -0400)]
fs: Factor inode hash operations into functions
Before replacing the inode hash locking with a more scalable
mechanism, factor the removal of the inode from the hashes rather
than open coding it in several places.
Based on a patch originally from Nick Piggin.
Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Nick Piggin [Sat, 23 Oct 2010 10:55:17 +0000 (06:55 -0400)]
fs: Implement lazy LRU updates for inodes
Convert the inode LRU to use lazy updates to reduce lock and
cacheline traffic. We avoid moving inodes around in the LRU list
during iget/iput operations so these frequent operations don't need
to access the LRUs. Instead, we defer the refcount checks to
reclaim-time and use a per-inode state flag, I_REFERENCED, to tell
reclaim that iget has touched the inode in the past. This means that
only reclaim should be touching the LRU with any frequency, hence
significantly reducing lock acquisitions and the amount contention
on LRU updates.
This also removes the inode_in_use list, which means we now only
have one list for tracking the inode LRU status. This makes it much
simpler to split out the LRU list operations under it's own lock.
Signed-off-by: Nick Piggin <npiggin@suse.de> Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Dave Chinner [Sat, 23 Oct 2010 09:03:02 +0000 (05:03 -0400)]
fs: Convert nr_inodes and nr_unused to per-cpu counters
The number of inodes allocated does not need to be tied to the
addition or removal of an inode to/from a list. If we are not tied
to a list lock, we could update the counters when inodes are
initialised or destroyed, but to do that we need to convert the
counters to be per-cpu (i.e. independent of a lock). This means that
we have the freedom to change the list/locking implementation
without needing to care about the counters.
Based on a patch originally from Eric Dumazet.
[AV: cleaned up a bit, fixed build breakage on weird configs
Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Miklos Szeredi [Tue, 5 Oct 2010 10:31:09 +0000 (12:31 +0200)]
vfs: fix infinite loop caused by clone_mnt race
If clone_mnt() happens while mnt_make_readonly() is running, the
cloned mount might have MNT_WRITE_HOLD flag set, which results in
mnt_want_write() spinning forever on this mount.
Needs CAP_SYS_ADMIN to trigger deliberately and unlikely to happen
accidentally. But if it does happen it can hang the machine.
Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Al Viro [Sun, 24 Oct 2010 13:46:50 +0000 (09:46 -0400)]
gfs2: invalidate_inodes() is no-op there
In fill_super() we hadn't MS_ACTIVE set yet, so there won't
be any inodes with zero i_count sitting around.
In put_super() we already have MS_ACTIVE removed *and* we
had called invalidate_inodes() since then. So again there
won't be any inodes with zero i_count...
Chris Mason [Mon, 23 Aug 2010 14:47:55 +0000 (10:47 -0400)]
aio: bump i_count instead of using igrab
The aio batching code is using igrab to get an extra reference on the
inode so it can safely batch. igrab will go ahead and take the global
inode spinlock, which can be a bottleneck on large machines doing lots
of AIO.
In this case, igrab isn't required because we already have a reference
on the file handle. It is safe to just bump the i_count directly
on the inode.
Benchmarking shows this patch brings IOP/s on tons of flash up by about
2.5X.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Randy Dunlap [Fri, 15 Oct 2010 23:32:27 +0000 (16:32 -0700)]
fs: move exportfs since it is not a networking filesystem
Move the EXPORTFS kconfig symbol out of the NETWORK_FILESYSTEMS block
since it provides a library function that can be (and is) used by other
(non-network) filesystems.
This also eliminates a kconfig dependency warning:
warning: (XFS_FS && BLOCK || NFSD && NETWORK_FILESYSTEMS && INET && FILE_LOCKING && BKL) selects EXPORTFS which has unmet direct dependencies (NETWORK_FILESYSTEMS)
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com> Cc: Dave Chinner <david@fromorbit.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Alex Elder <aelder@sgi.com> Cc: xfs-masters@oss.sgi.com Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
365b1818 ("add f_flags to struct statfs(64)") resized f_spare within
struct statfs which caused a UML crash. There is no need to copy f_spare.
Signed-off-by: Richard Weinberger <richard@nod.at> Reported-by: Toralf Förster <toralf.foerster@gmx.de> Tested-by: Toralf Förster <toralf.foerster@gmx.de> Cc: Christoph Hellwig <hch@lst.de> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Jeff Dike <jdike@addtoit.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Eric Dumazet [Tue, 5 Oct 2010 07:32:55 +0000 (09:32 +0200)]
fs: allow for more than 2^31 files
Andrew,
Could you please review this patch, you probably are the right guy to
take it, because it crosses fs and net trees.
Note : /proc/sys/fs/file-nr is a read-only file, so this patch doesnt
depend on previous patch (sysctl: fix min/max handling in
__do_proc_doulongvec_minmax())
Thanks !
[PATCH V4] fs: allow for more than 2^31 files
Robin Holt tried to boot a 16TB system and found af_unix was overflowing
a 32bit value :
<quote>
We were seeing a failure which prevented boot. The kernel was incapable
of creating either a named pipe or unix domain socket. This comes down
to a common kernel function called unix_create1() which does:
atomic_inc(&unix_nr_socks);
if (atomic_read(&unix_nr_socks) > 2 * get_max_files())
goto out;
The function get_max_files() is a simple return of files_stat.max_files.
files_stat.max_files is a signed integer and is computed in
fs/file_table.c's files_init().
In our case, mempages (total_ram_pages) is approx 3,758,096,384
(0xe0000000). That leaves max_files at approximately 1,503,238,553.
This causes 2 * get_max_files() to integer overflow.
</quote>
Fix is to let /proc/sys/fs/file-nr & /proc/sys/fs/file-max use long
integers, and change af_unix to use an atomic_long_t instead of
atomic_t.
get_max_files() is changed to return an unsigned long.
get_nr_files() is changed to return a long.
unix_nr_socks is changed from atomic_t to atomic_long_t, while not
strictly needed to address Robin problem.
Before patch (on a 64bit kernel) :
# echo 2147483648 >/proc/sys/fs/file-max
# cat /proc/sys/fs/file-max
-18446744071562067968
Reported-by: Robin Holt <holt@sgi.com> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Acked-by: David Miller <davem@davemloft.net> Reviewed-by: Robin Holt <holt@sgi.com> Tested-by: Robin Holt <holt@sgi.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Jan Kara [Mon, 4 Oct 2010 09:37:37 +0000 (11:37 +0200)]
isofs: Fix isofs_get_blocks for 8TB files
Currently isofs_get_blocks() was limited to handle only 4TB files on 32-bit
architectures because of unnecessary use of iblock variable which was signed
long. Just remove the variable. The error messages that were using this
variable should have rather used b_off anyway because that is the block we
are currently mapping.
Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
__block_write_begin and block_prepare_write are identical except for slightly
different calling conventions. Convert all callers to the __block_write_begin
calling conventions and drop block_prepare_write.
Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Add a new helper to write out the inode using the writeback code,
that is including the correct dirty bit and list manipulation. A few
of filesystems already opencode this, and a lot of others should be
using it instead of using write_inode_now which also writes out the
data.
Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Linus Torvalds [Mon, 25 Oct 2010 17:59:31 +0000 (10:59 -0700)]
Merge branch 'davinci-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/khilman/linux-davinci
* 'davinci-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/khilman/linux-davinci: (50 commits)
davinci: fix remaining board support after io_pgoffst removal
davinci: mityomapl138: make file local data static
arm/davinci: remove duplicated include
davinci: Initial support for Omapl138-Hawkboard
davinci: MityDSP-L138/MityARM-1808 read MAC address from I2C Prom
davinci: add tnetv107x touchscreen platform device
input: add driver for tnetv107x touchscreen controller
davinci: add keypad config for tnetv107x evm board
davinci: add tnetv107x keypad platform device
input: add driver for tnetv107x on-chip keypad controller
net: davinci_emac: cleanup unused cpdma code
net: davinci_emac: switch to new cpdma layer
net: davinci_emac: separate out cpdma code
net: davinci_emac: cleanup unused mdio emac code
omap: cleanup unused davinci mdio arch code
davinci: cleanup mdio arch code and switch to phy_id
net: davinci_emac: switch to new mdio
omap: add mdio platform devices
davinci: add mdio platform devices
net: davinci_emac: separate out davinci mdio
...
Fix up trivial conflict in drivers/input/keyboard/Kconfig (two entries
added next to each other - one from the davinci merge, one from the
input merge)
Linus Torvalds [Mon, 25 Oct 2010 17:08:21 +0000 (10:08 -0700)]
Merge branch 'for-linus' of git://git.open-osd.org/linux-open-osd
* 'for-linus' of git://git.open-osd.org/linux-open-osd:
exofs: Remove inode->i_count manipulation in exofs_new_inode
fs/exofs: typo fix of faild to failed
exofs: Set i_mapping->backing_dev_info anyway
exofs: Cleaup read path in regard with read_for_write
Borislav Petkov [Mon, 25 Oct 2010 16:15:22 +0000 (18:15 +0200)]
x86-32, mm: Remove duplicated include
Commit b40827fa7268 ("x86-32, mm: Add an initial page table for core
bootstrapping") added an include directive which is needless and is
taken care of by a previous one. Remove it.
Boaz Harrosh [Sat, 16 Oct 2010 08:14:01 +0000 (19:14 +1100)]
exofs: Remove inode->i_count manipulation in exofs_new_inode
exofs_new_inode() was incrementing the inode->i_count and
decrementing it in create_done(), in a bad attempt to make sure
the inode will still be there when the asynchronous create_done()
finally arrives. This was very stupid because iput() was not called,
and if it was actually needed, it would leak the inode.
However all this is not needed, because at exofs_evict_inode()
we already wait for create_done() by waiting for the
object_created event. Therefore remove the superfluous ref counting
and just Thicken the comment at exofs_evict_inode() a bit.
While at it change places that open coded wait_obj_created()
to call the already available wrapper.
CC: Dave Chinner <dchinner@redhat.com> CC: Christoph Hellwig <hch@lst.de> CC: Nick Piggin <npiggin@kernel.dk> Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Linus Torvalds [Mon, 25 Oct 2010 15:32:05 +0000 (08:32 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6: (365 commits)
ALSA: hda - Disable sticky PCM stream assignment for AD codecs
ALSA: usb - Creative USB X-Fi volume knob support
ALSA: ca0106: Use card specific dac id for mute controls.
ALSA: ca0106: Allow different sound cards to use different SPI channel mappings.
ALSA: ca0106: Create a nice spot for mapping channels to dacs.
ALSA: ca0106: Move enabling of front dac out of hardcoded setup sequence.
ALSA: ca0106: Pull out dac powering routine into separate function.
ALSA: ca0106 - add Sound Blaster 5.1vx info.
ASoC: tlv320dac33: Use usleep_range for delays
ALSA: usb-audio: add Novation Launchpad support
ALSA: hda - Add workarounds for CT-IBG controllers
ALSA: hda - Fix wrong TLV mute bit for STAC/IDT codecs
ASoC: tpa6130a2: Error handling for broken chip
ASoC: max98088: Staticise m98088_eq_band
ASoC: soc-core: Fix codec->name memory leak
ALSA: hda - Apply ideapad quirk to Acer laptops with Cxt5066
ALSA: hda - Add some workarounds for Creative IBG
ALSA: hda - Fix wrong SPDIF NID assignment for CA0110
ALSA: hda - Fix codec rename rules for ALC662-compatible codecs
ALSA: hda - Add alc_init_jacks() call to other codecs
...
Linus Torvalds [Mon, 25 Oct 2010 15:30:48 +0000 (08:30 -0700)]
Merge branch 'for-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/dvrabel/uwb
* 'for-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/dvrabel/uwb:
uwb: Orphan the UWB and WUSB subsystems
uwb: Remove the WLP subsystem and drivers
Linus Torvalds [Mon, 25 Oct 2010 15:28:13 +0000 (08:28 -0700)]
Merge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mjg59/platform-drivers-x86
* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mjg59/platform-drivers-x86: (44 commits)
eeepc-wmi: Add cpufv sysfs interface
eeepc-wmi: add additional hotkeys
panasonic-laptop: Simplify calls to acpi_pcc_retrieve_biosdata
panasonic-laptop: Handle errors properly if they happen
intel_pmic_gpio: fix off-by-one value range checking
IBM Real-Time "SMI Free" mode driver -v7
Add OLPC XO-1 rfkill driver
Move hdaps driver to platform/x86
ideapad-laptop: Fix Makefile
intel_pmic_gpio: swap the bits and mask args for intel_scu_ipc_update_register
ideapad: Add param: no_bt_rfkill
ideapad: Change the driver name to ideapad-laptop
ideapad: rewrite the sw rfkill set
ideapad: rewrite the hw rfkill notify
ideapad: use EC command to control camera
ideapad: use return value of _CFG to tell if device exist or not
ideapad: make sure we bind on the correct device
ideapad: check VPC bit before sync rfkill hw status
ideapad: add ACPI helpers
dell-laptop: Add debugfs support
...
Linus Torvalds [Mon, 25 Oct 2010 15:19:14 +0000 (08:19 -0700)]
Merge branch 'next-devicetree' of git://git.secretlab.ca/git/linux-2.6
* 'next-devicetree' of git://git.secretlab.ca/git/linux-2.6:
mtd/m25p80: add support to parse the partitions by OF node
of/irq: of_irq.c needs to include linux/irq.h
of/mips: Cleanup some include directives/files.
of/mips: Add device tree support to MIPS
of/flattree: Eliminate need to provide early_init_dt_scan_chosen_arch
of/device: Rework to use common platform_device_alloc() for allocating devices
of/xsysace: Fix OF probing on little-endian systems
of: use __be32 types for big-endian device tree data
of/irq: remove references to NO_IRQ in drivers/of/platform.c
of/promtree: add package-to-path support to pdt
of/promtree: add of_pdt namespace to pdt code
of/promtree: no longer call prom_ functions directly; use an ops structure
of/promtree: make drivers/of/pdt.c no longer sparc-only
sparc: break out some PROM device-tree building code out into drivers/of
of/sparc: convert various prom_* functions to use phandle
sparc: stop exporting openprom.h header
powerpc, of_serial: Endianness issues setting up the serial ports
of: MTD: Fix OF probing on little-endian systems
of: GPIO: Fix OF probing on little-endian systems
Ralf Baechle [Sun, 24 Oct 2010 21:23:50 +0000 (22:23 +0100)]
MIPS: MT: Fix build error iFPU affinity code
Commit b0ae19811375 ("security: remove unused parameter from
security_task_setscheduler()") broke the build of
arch/mips/kernel/mips-mt-fpaff.c. The function arguments were
unnecessary, not the semicolon ...
Linus Torvalds [Mon, 25 Oct 2010 15:05:29 +0000 (08:05 -0700)]
Merge branch 'ieee1394-removal' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394-2.6
* 'ieee1394-removal' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394-2.6:
ieee1394: remove the old IEEE 1394 driver stack
ieee1394: move init_ohci1394_dma to drivers/firewire/
Fix trivial change/delete conflict: drivers/ieee1394/eth1394.c is
getting removed, but was modified by the networking merge.
Yoshihisa Abe [Mon, 25 Oct 2010 06:03:45 +0000 (02:03 -0400)]
Coda: push BKL regions into coda_upcall()
Now that shared inode state is locked using the cii->c_lock, the BKL is
only used to protect the upcall queues used to communicate with the
userspace cache manager. The remaining state is all local and we can
push the lock further down into coda_upcall().
Yoshihisa Abe [Mon, 25 Oct 2010 06:03:44 +0000 (02:03 -0400)]
Coda: add spin lock to protect accesses to struct coda_inode_info.
We mostly need it to protect cached user permissions. The c_flags field
is advisory, reading the wrong value is harmless and in the worst case
we hit a slow path where we have to make an extra upcall to the
userspace cache manager when revalidating a dentry or inode.
* git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6: (110 commits)
sh: i2c-sh7760: Replase from ctrl_* to __raw_*
sh: clkfwk: Shuffle around to match the intc split up.
sh: clkfwk: modify for_each_frequency end condition
sh: fix clk_get() error handling
sh: clkfwk: Fix fault in frequency iterator.
sh: clkfwk: Add a helper for rate rounding by divisor ranges.
sh: clkfwk: Abstract rate rounding helper.
sh: clkfwk: support clock remapping.
sh: pci: Convert to upper/lower_32_bits() helpers.
sh: mach-sdk7786: Add support for the FPGA SRAM.
sh: Provide a generic SRAM pool for tiny memories.
sh: pci: Support secondary FPGA-driven PCIe clocks on SDK7786.
sh: pci: Support slot 4 routing on SDK7786.
sh: Fix up PMB locking.
sh: mach-sdk7786: Add support for fpga gpios.
sh: use pr_fmt for clock framework, too.
sh: remove name and id from struct clk
sh: free-without-alloc fix for sh_mobile_lcdcfb
sh: perf: Set up perf_max_events.
sh: perf: Support SH-X3 hardware counters.
...
Fix up trivial conflicts (perf_max_events got removed) in arch/sh/kernel/perf_event.c
Linus Torvalds [Mon, 25 Oct 2010 14:44:27 +0000 (07:44 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu: (21 commits)
m68knommu: convert to using tracehook_report_syscall_*
m68knommu: some boards use fixed phy for FEC ethernet
m68knommu: support the external GPIO based interrupts of the 5272
m68knommu: mask of vector bits in exception word properly
m68knommu: change to new flag variables
m68knommu: Fix MCFUART_TXFIFOSIZE for m548x.
m68knommu: add basic mmu-less m548x support
m68knommu: .gitignore vmlinux.lds
m68knommu: stop using __do_IRQ
m68knommu: rename PT_OFF_VECTOR to PT_OFF_FORMATVEC.
m68knommu: add support for Coldfire 547x/548x interrupt controller
m68k{nommu}: Remove unused DEFINE's from asm-offsets.c
m68knommu: whitespace cleanup in 68328/entry.S
m68knommu: Document supported chips in intc-2.c and intc-simr.c.
m68knommu: fix strace support for 68328/68360
m68knommu: fix default starting date
arch/m68knommu: Removing dead 68328_SERIAL_UART2 config option
arch/m68knommu: Removing dead RAM_{16,32}_MB config option
arch/m68knommu: Removing dead M68KFPU_EMU config option
arch/m68knommu: Removing dead RELOCATE config option
...
Heiko Carstens [Mon, 25 Oct 2010 14:10:54 +0000 (16:10 +0200)]
[S390] topology: export cpu topology via proc/sysinfo
Export the cpu configuration topology via sysinfo. Two new lines are
introduced:
CPU Topology HW: 0 0 0 4 6 4
CPU Topology SW: 0 0 0 0 4 24
The HW line describes the cpu topology nesting levels when the maximum
nesting level is used to get the corresponding SYSIB.
The SW line describes what Linux is actually using. In this case it
supports only two levels (CONFIG_SCHED_BOOK off) and therefore the
hardware folded the two lower levels in the SYSIB response block.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Stefan Haberland [Mon, 25 Oct 2010 14:10:49 +0000 (16:10 +0200)]
[S390] dasd: ignore unsolicited interrupts for DIAG
For the DASD DIAG discipline IO is started through special diagnose
calls. Unsolicited interrupts may contain information about the device
itself. But this information is not needed because the device is not
used directly.
Fix the case that an unimplemented dicipline function may be called
by ignoring unsolicited interrupts for the DIAG disciplin.
Signed-off-by: Stefan Haberland <stefan.haberland@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
The dasd interrupt handler needs to distinguish solicited from
unsolicited interrupts, as unsolicited interrupts may require special
handling (e.g. summary unit checks) and solicited interrupts require
proper error recovery for the failed I/O request.
The interrupt handler needs to check several bit fields in the
interrupt response block (irb) to make this distinction.
So far our check of the status control bits has not been specific
enough, which may lead to a failed request getting just retried
instead of the necessary error recovery.
Signed-off-by: Stefan Weinhuber <wein@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Sebastian Ott [Mon, 25 Oct 2010 14:10:46 +0000 (16:10 +0200)]
[S390] dasd: fix use after free in dbf
Writing to /proc/dasd/statistics while the debug level of the
generic dasd debug entry is set to DBF_DEBUG will lead to an
use after free when accessing the debug entry later.
Since for the format string "%s" in the s390 dbf only a pointer
to the string is stored in the debug feature and the buffer used
here is freed afterwards.
To fix this just remove the debug message.
Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
[S390] kvm: Fix badness at include/asm/mmu_context.h:83
commit 050eef364ad700590a605a0749f825cab4834b1e
[S390] fix tlb flushing vs. concurrent /proc accesses
broke KVM on s390x. On every schedule a
Badness at include/asm/mmu_context.h:83 appears. s390_enable_sie
replaces the mm on the __running__ task, therefore, we have to
increase the attach count of the new mm.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Function ccw_device_cancel_halt_clear may cause an unexpected kernel
panic if a clear function is currently active at the subchannel for
which it is called. In that case, the iretry counter used to determine
the number of retries is never initialized, leading to an immediate
failure of the function which results in a kernel panic.
Fix this by initializing the iretry counter when the function is
first called. Also replace the kernel panic with a return code: a
single malfunctioning I/O device should not automatically cause a
system-wide kernel panic.
Signed-off-by: Peter Oberparleiter <peter.oberparleiter@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Heiko Carstens [Mon, 25 Oct 2010 14:10:43 +0000 (16:10 +0200)]
[S390] topology: change default
Switch default value of the kernel parameter 'topology' from off to on.
Various performance measurements have finally shown that there are no
(known) regressions anywhere.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Heiko Carstens [Mon, 25 Oct 2010 14:10:42 +0000 (16:10 +0200)]
[S390] smp: use correct cpu address in print_cpu_info()
Up to now print_cpu_info() uses the cpu address stored in it's local
lowcore to print a message to the console. The cpu address in the
lowcore is (in this case) however not the physical cpu address of the
local cpu. It's the address of the cpu that issued the sigp restart
which started the local cpu.
Fix this by using the store cpu address instruction instead.
It's not that anybody really cares since this is broken since more than
ten years...
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
[S390] remove ieee_instruction_pointer from thread_struct
The ieee_instruction_pointer can not be read from user space anymore
since git commit 613e1def6b52c399a8b72a5e11bc2e57d2546fb8, the ptrace
interface always returns zero. Remove it from the thread_struct. It
is still present in the user_regs_struct for compatability reasons.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Do the setup of the stack overflow argument for the sixth system
call parameter right before the branch to the system call function.
That simplifies the system call parameter access code.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Read all required fields for program checks from the lowcore in the
first level interrupt handler in entry[64].S. If the context that
caused the fault was enabled for interrupts we can now re-enable the
irqs in entry[64].S.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Heiko Carstens [Mon, 25 Oct 2010 14:10:36 +0000 (16:10 +0200)]
[S390] pgtable: move pte_mkhuge() from hugetlb.h to pgtable.h
All architectures besides s390 have pte_mkhuge() defined in pgtable.h.
So move the function to pgtable.h on s390 as well.
Fixes a compile error introduced with "hugetlb: hugepage migration core"
in linux-next which only happens on s390.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Sebastian Ott [Mon, 25 Oct 2010 14:10:30 +0000 (16:10 +0200)]
[S390] chsc: use the global page to determine the chp desriptor
chsc_determine_channel_path_desc is called by a wrapper
who allocates a response struct. The response data
is then memcpy'ed to this response struct by
chsc_determine_channel_path_desc.
Change chsc_determine_base_channel_path_desc to use the
global chsc_page and deliver it to the function doing
the actual chsc call. The channel path desriptor is
then directly read from the response data.
As a result we get rid of the additional allocation
for the response struct.
Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Sebastian Ott [Mon, 25 Oct 2010 14:10:29 +0000 (16:10 +0200)]
[S390] chsc: consolidate memory allocations
Most wrappers around the channel subsystem call have their own logic
to allocate memory (with proper alignment) or use preallocated or
static memory. This patch converts most users of the channel
subsystem call to use the same preallocated page (proteced by a
spinlock).
Note: The sei_page which is used in our crw handler to call
"store event information" has to coexist, since
a) in crw context, while accessing the sei_page, sleeping is allowed
(which will conflict with the spinlock protection of the chsc_page)
b) in crw context, while accessing the sei_page, channel subsystem
calls are allowed (which itself would require the page).
Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Sebastian Ott [Mon, 25 Oct 2010 14:10:28 +0000 (16:10 +0200)]
[S390] chsc: initialization fixes
This patch fixes:
* kfree vs. free_page usage
* structure definition for determine_css_characteristics
* naming convention for the chsc init function
* deregistration of crw handlers in the cleanup path
Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Sebastian Ott [Mon, 25 Oct 2010 14:10:26 +0000 (16:10 +0200)]
[S390] cio: fix memleak in resume path
If a ccwdevice is lost during hibernation and a different
ccwdevice is attached to the same subchannel, we will
deregister the old ccw device and register the new one.
Since deregistration is not allowed in this context, we
handle this action later. However, some parts of the
registration process for the new device were started anyway,
so that the old device structure is no longer accessible.
Fix this by deferring both actions to the afterwards
scheduled subchannel event.
Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Stefan Haberland [Mon, 25 Oct 2010 14:10:24 +0000 (16:10 +0200)]
[S390] dasd fix dump_sense_dbf
The dasd_eckd_dump_sense_dbf function uses a macro for s390 debug
feature that can handle up to 8 parameters (for the DASD device
driver).
Fix the function to use only the maximum number of parameters.
Signed-off-by: Stefan Haberland <stefan.haberland@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Remove the call to iucv_unregister() and remove the goto label
as unregistering is the last step in the hvc_iucv initialization.
If iucv_register() fails, simply clean up hvc terminals and free
resources.
Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Heiko Carstens [Mon, 25 Oct 2010 14:10:21 +0000 (16:10 +0200)]
[S390] cmm: fix crash on case conversion
When the cmm module is compiled into the kernel it will crash when
writing to the R/O data section.
Reason is the lower to upper case conversion of the "sender" module
parameter which ignored the fact that the pointer is preinitialized.
Introduced with 41b42876 "cmm, smsgiucv_app: convert sender to
uppercase"
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Heiko Carstens [Mon, 25 Oct 2010 14:10:18 +0000 (16:10 +0200)]
[S390] switch_to: dont restore/save access & fpu regs for kernel threads
If the previous task was a kernel thread there is no need to save the
contents of the fpu and access registers since they aren't used in
kernel mode.
For the same reason it is not necessary to restore these registers if
the next task is a kernel thread.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Improve performance of the sske operation by using the nonquiescing
variant if the affected page has no mappings established. On machines
with no support for the new sske variant the mask bit will be ignored.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Use the store indication bit in the translation exception code on
page faults to avoid the protection faults that immediatly follow
the page fault if the access has been a write.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Heiko Carstens [Mon, 25 Oct 2010 14:10:12 +0000 (16:10 +0200)]
[S390] cpu hotplug/idle: move cpu_die call to enabled context
There is no difference if cpu_die is called from enabled or disabled
context. Except that the fast_gup code might be called via
cpu_die -> idle_task_exit -> __mm_drop -> crst_table_free. Which in
turn grabs and releases a spinlock using the _bh ops, which is not
allowed in irq disabled context, since spin_unlock_bh will
unconditionally enable interrupts again.
To get rid of the warning emitted by the softirq code just move the
code to enabled context.
In this case this doesn't fix a bug, we just get rid of a warning.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Christof Schmitt [Mon, 25 Oct 2010 14:10:10 +0000 (16:10 +0200)]
[S390] set ARCH_HAS_SG_CHAIN for s390
Set ARCH_HAS_SG_CHAIN for scatter lists on s390. Without this flag the
SCSI code limits the maximum number of segments, so set it to make
proper use of the FCP channel hardware.
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
projects / linux-beck.git / log