It has been thought that the per-user file descriptors limit would also
limit the resources that a normal user can request via the epoll
interface. Vegard Nossum reported a very simple program (a modified
version attached) that can make a normal user to request a pretty large
amount of kernel memory, well within the its maximum number of fds. To
solve such problem, default limits are now imposed, and /proc based
configuration has been introduced. A new directory has been created,
named /proc/sys/fs/epoll/ and inside there, there are two configuration
points:
max_user_instances = Maximum number of devices - per user
max_user_watches = Maximum number of "watched" fds - per user
The current default for "max_user_watches" limits the memory used by epoll
to store "watches", to 1/32 of the amount of the low RAM. As example, a
256MB 32bit machine, will have "max_user_watches" set to roughly 90000.
That should be enough to not break existing heavy epoll users. The
default value for "max_user_instances" is set to 128, that should be
enough too.
This also changes the userspace, because a new error code can now come out
from EPOLL_CTL_ADD (-ENOSPC). The EMFILE from epoll_create() was already
listed, so that should be ok.
Any user on existing parisc 32- and 64bit-kernels can easily crash
the kernel and as such enforce a DSO.
A simple testcase is available here:
http://gsyprf10.external.hp.com/~deller/crash.tgz
The problem is introduced by the fact, that the handle_interruption()
crash handler calls the show_regs() function, which in turn tries to
unwind the stack by calling parisc_show_stack(). Since the stack contains
userspace addresses, a try to unwind the stack is dangerous and useless
and leads to the crash.
The fix is trivial: For userspace processes
a) avoid to unwind the stack, and
b) avoid to resolve userspace addresses to kernel symbol names.
While touching this code, I converted print_symbol() to %pS
printk formats and made parisc_show_stack() static.
An initial patch for this was written by Kyle McMartin back in August:
http://marc.info/?l=linux-parisc&m=121805168830283&w=2
Compile and run-tested with a 64bit parisc kernel.
Signed-off-by: Helge Deller <deller@gmx.de> Cc: Grant Grundler <grundler@parisc-linux.org> Cc: Matthew Wilcox <matthew@wil.cx> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Kyle McMartin <kyle@mcmartin.ca> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
A problem was found while reviewing the code after Bugzilla bug
http://bugzilla.kernel.org/show_bug.cgi?id=11796.
In ipc_addid(), the newly allocated ipc structure is inserted into the
ipcs tree (i.e made visible to readers) without locking it. This is not
correct since its initialization continues after it has been inserted in
the tree.
This patch moves the ipc structure lock initialization + locking before
the actual insertion.
kunmap() takes as argument the struct page that orginally got kmap()'d,
however the sg_miter_stop() function passed it the kernel virtual address
instead, resulting in weird stuff.
Somehow I ended up fixing this bug by accident while looking for a bug in
the same area.
Reported-by: kerneloops.org Acked-by: Tejun Heo <tj@kernel.org> Signed-off-by: Arjan van de Ven <arjan@linux.intel.com> Cc: Hugh Dickins <hugh@veritas.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
There are already various drivers having bigger label than 12 bytes. Most
of them fit well under 20 bytes but make column width exact so that
oversized labels don't mess up output alignment.
Signed-off-by: Jarkko Nikula <jarkko.nikula@nokia.com> Acked-by: David Brownell <david-b@pacbell.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When booting in a direct color mode, the penguin has dirty feet, i.e.,
some pixels have the wrong color. This is caused by
fb_set_logo_directpalette() which does not initialize the last 32 palette
entries.
Signed-off-by: Clemens Ladisch <clemens@ladisch.de> Acked-by: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Krzysztof Helt <krzysztof.h1@poczta.fm> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fixes a data corruption bug in pxa2xx_spi.c when operating in full duplex
mode with DMA and using buffers that overlap.
SPI transmit and receive buffers are allowed to be the same or to overlap.
However, this driver fails if such overlap is attempted in DMA mode
because it maps the rx and tx buffers in the wrong order. By mapping
DMA_FROM_DEVICE (read) before DMA_TO_DEVICE (write), it invalidates the
cache before flushing it, thus discarding data which should have been
transmitted.
The patch corrects the order of mapping. This bug exists in all versions
of pxa2xx_spi.c; similar bugs are in the drivers for two other SPI
controllers (au1500, imx).
A version of this patch has been tested on kernel 2.6.20 using
verification of loopback data with: random transfer length, random
bits-per-word, random positive offsets (both larger and smaller than
transfer length) between the start of the rx and tx buffers, and varying
clock rates.
Signed-off-by: Ned Forrester <nforrester@whoi.edu> Cc: Vernon Sauder <vernoninhand@gmail.com> Cc: J. Scott Merritt <merrij3@rpi.edu> Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
I have received some reports of out-of-memory errors on some older AMD
architectures. These errors are what I would expect to see if
crypt_stat->key were split between two separate pages. eCryptfs should
not assume that any of the memory sent through virt_to_scatterlist() is
all contained in a single page, and so this patch allocates two
scatterlist structs instead of one when processing keys. I have received
confirmation from one person affected by this bug that this patch resolves
the issue for him, and so I am submitting it for inclusion in a future
stable release.
Note that virt_to_scatterlist() runs sg_init_table() on the scatterlist
structs passed to it, so the calls to sg_init_table() in
decrypt_passphrase_encrypted_session_key() are redundant.
Signed-off-by: Michael Halcrow <mhalcrow@us.ibm.com> Reported-by: Paulo J. S. Silva <pjssilva@ime.usp.br> Cc: "Leon Woestenberg" <leon.woestenberg@gmail.com> Cc: Tim Gardner <tim.gardner@canonical.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Impact: properly rebuild sched-domains on kmalloc() failure
When cpuset failed to generate sched domains due to kmalloc()
failure, the scheduler should fallback to the single partition
'fallback_doms' and rebuild sched domains, but now it only
destroys but not rebuilds sched domains.
The regression was introduced by:
| commit dfb512ec4834116124da61d6c1ee10fd0aa32bd6
| Author: Max Krasnyansky <maxk@qualcomm.com>
| Date: Fri Aug 29 13:11:41 2008 -0700
|
| sched: arch_reinit_sched_domains() must destroy domains to force rebuild
After the above commit, partition_sched_domains(0, NULL, NULL) will
only destroy sched domains and partition_sched_domains(1, NULL, NULL)
will create the default sched domain.
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com> Cc: Max Krasnyansky <maxk@qualcomm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch is required for all AMD SB600 revisions to avoid USB subsystem hang
symptom. The USB subsystem hang symptom is observed when the system has
multiple USB devices connected to it. In some cases a USB hub may be required
to observe this symptom.
This patch is required for AMD SB700 south bridge revision A12 and A13 to avoid
USB subsystem hang symptom. The USB subsystem hang symptom is observed when the
system has multiple USB devices connected to it. In some cases a USB hub may be
required to observe this symptom.
This patch works around the problem by correcting the internal register setting
that will help by changing the behavior of the internal logic to avoid the
USB subsystem hang issue. The change in the behavior of the logic does not
impact the normal operation of the USB subsystem.
There's a bug in the usbmon binary reader: When using read() to fetch
the packets and a packet's data is partially read, the next read call
will once again return up to len_cap bytes of data. The b_read counter
is not regarded when determining the remaining chunk size.
So, when dumping USB data with "cat /dev/usbmon0 > usbmon.trace" while
reading from a USB storage device and analyzing the dump file
afterwards it will get out of sync after a couple of packets.
Signed-off-by: Ingo van Lil <inguin@gmx.de> Signed-off-by: Pete Zaitcev <zaitcev@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Somewhere in the conversion of the RNDIS gadget code to the new
framework, the descriptor of its data interface seems to have
been copied from the CDC Ethernet driver. Unfortunately that
means it got a nonzero altsetting ... which is incorrect. Issue
uncovered by Richard Röjfors <richard.rojfors@endian.se>.
This patch fixes that problem, and resolves at least some cases
of Windows XP bluescreening itself.
Tested-by: Richard Röjfors <richard.rojfors@endian.se>. Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
It turns out that atomic_inc_return() returns the *new* value
not the original one, so the logic in rndis_response_available()
kept the first RNDIS response notification from getting out.
This prevented interoperation with MS-Windows (but not Linux).
Fix this to make RNDIS behave again.
Signed-off-by: Richard Röjfors <richard.rojfors@endian.se> Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Alexey Starikovskiy <astarikovskiy@suse.de> Signed-off-by: Len Brown <len.brown@intel.com> Cc: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Alexey Starikovskiy <astarikovskiy@suse.de> Signed-off-by: Len Brown <len.brown@intel.com> Cc: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Restart current transaction if we recieved unexpected GPEs instead
of needed ones.
http://bugzilla.kernel.org/show_bug.cgi?id=11896
Signed-off-by: Alexey Starikovskiy <astarikovskiy@suse.de> Signed-off-by: Len Brown <len.brown@intel.com> Cc: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Alexey Starikovskiy <astarikovskiy@suse.de> Signed-off-by: Len Brown <len.brown@intel.com> Cc: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1155) fixes a bug in usbcore. When interfaces are
deleted, either because the device was disconnected or because of a
configuration change, the extra attribute files and child endpoint
devices may get left behind. This is because the core removes them
before calling device_del(). But during device_del(), after the
driver is unbound the core will reinstall altsetting 0 and recreate
those extra attributes and children.
The patch prevents this by adding a flag to record when the interface
is in the midst of being unregistered. When the flag is set, the
attribute files and child devices will not be created.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1165) makes a few small changes in the logic used by
ehci-hcd when it encounters a controller error:
Instead of printing out the masked status, it prints the
original status as read directly from the hardware.
It doesn't check for the STS_HALT status bit before taking
action. The mere fact that the STS_FATAL bit is set means
that something bad has happened and the controller needs to
be reset. With the old code this test could never succeed
because the STS_HALT bit was masked out from the status.
I anticipate that this will prevent the occasional "irq X: nobody cared"
problem people encounter when their EHCI controllers die.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Cc: David Brownell <david-b@pacbell.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1164) fixes a bug in the EHCI scheduler. The interval
value it uses is already in linear format, not logarithmically coded.
The existing code can sometimes crash the system by trying to divide
by zero.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Cc: David Brownell <david-b@pacbell.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fixes an obvious bug in cdc-acm by avoiding a recursive lock on
acm_start_wb()'s error path. Should apply towards 2.6.27 stable and
2.6.28.
=============================================
[ INFO: possible recursive locking detected ]
2.6.27-2-pae #109
---------------------------------------------
python/31449 is trying to acquire lock:
(&acm->write_lock){++..}, at: [<f89a0348>] acm_start_wb+0x5c/0x7b [cdc_acm]
but task is already holding lock:
(&acm->write_lock){++..}, at: [<f89a04fb>] acm_tty_write+0xe1/0x167 [cdc_acm]
other info that might help us debug this:
2 locks held by python/31449:
#0: (&tty->atomic_write_lock){--..}, at: [<c0260fae>] tty_write_lock+0x14/0x3b
#1: (&acm->write_lock){++..}, at: [<f89a04fb>] acm_tty_write+0xe1/0x167 [cdc_acm]
Add ehci_shutdown() or ohci_shutdown() calls to the USB
PS3 bus glue. ehci_shutdown() and ohci_shutdown() do some
controller specific cleanups not done by usb_remove_hcd().
Fixes errors on shutdown or reboot similar to these:
ps3-ehci-driver sb_07: HC died; cleaning up
irq 51: nobody cared (try booting with the "irqpoll" option)
This fixes a deadlock appearing with some USB peripheral drivers
when running CDC ACM gadget code.
The newish (2.6.27) CDC ACM event notification mechanism sends
messages (IN to the host) which are short enough to fit in most
FIFOs. That means that with some peripheral controller drivers
(evidently not the ones used to verify the notification code!!)
the completion callback can be issued before queue() returns.
The deadlock would come because the completion callback and the
event-issuing code shared a spinlock. Fix is trivial: drop
that lock while queueing the message.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The reason, is that ret is initialized with ENODEV instead of 0 _or_
the kmem cache is not freed in error case with no bus binding.
The difference between OF+PCI and OF only is
| 15148 804 32 15984 3e70 isp1760-of-pci.o
| 13748 676 8 14432 3860 isp1760-of.o
about 1.5 KiB.
Until there is a checkbox where the user *must* select atleast one item,
and may select multiple entries I don't make it selectable anymore.
Having a driver which can't be used under any circumstances is broken
anyway and I've seen distros shipping it that way.
Reported-by: Roland Kletzing <devzero@web.de> Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
dpt_i2o.c::adpt_i2o_to_scsi() reads the value at (reply+5) which
should contain the length in bytes of the transferred data. This
would be correct if reply was a u32 *. However it is a void * here,
so we need to read the value at (reply+20) instead.
The value at (reply+5) is usually 0xff0000, which is apparently
'large enough' and didn't cause any trouble until 2.6.27 where
Mike Reed noted
(https://bugzilla.novell.com/show_bug.cgi?id=421330) that the
driver was incorrectly returning a SUCCESS status if the driver's
request to the firmware to abort a command failed. By doing so,
the mid-layer believed, incorrectly, that the command has
completed and has been returned (ultimately clearing
scsi_cmnd.request_buffer) yet the driver still has the command.
What should correctly happen is a mid-layer escalation
(device-reset, etc.) of recovery during which the driver will
eventually return the outstanding commands to the mid-layer.
Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit 69961c375288bdab7604e0bb1c8d22999bb8a347 ("[PATCH] m68k/Atari:
Interrupt updates") added a BUG_ON() with an incorrect upper bound
comparison, which causes an early crash on VME boards, where IRQ_USER is
8, cnt is 192 and NR_IRQS is 200.
Reported-by: Stephen N Chivers <schivers@csc.com.au> Tested-by: Kars de Jong <jongk@linux-m68k.org> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
According to ACPI spec when the status of some device is not present
but functional, the device is valid and the children of this device
should be enumerated. It means that the device should be added to
linux acpi device tree. But the device driver for this device should not
be loaded.
The detailed info can be found in the section 6.3.7 of ACPI 3.0b spec.
_STA may return bit 0 clear (not present) with bit 3 set (device is
functional). This case is used to indicate a valid device for which no
device driver should be loaded (for example, a bridge device.).
Children of this device may be present and valid. OS should continue
enumeration below a device whose _STA returns this bit combination
http://bugzilla.kernel.org/show_bug.cgi?id=3358
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com> Signed-off-by: Li Shaohua <shaohua.li@intel.com> Signed-off-by: Zhang Rui <rui.zhang@intel.com> Signed-off-by: Andi Kleen <ak@linux.intel.com> Signed-off-by: Len Brown <len.brown@intel.com> Cc: Holger Macht <hmacht@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
cpu_coregroup_map used to grab a mutex on s390 since it was only
called from process context.
Since c7c22e4d5c1fdebfac4dba76de7d0338c2b0d832 "block: add support
for IO CPU affinity" this is not true anymore.
It now also gets called from softirq context.
To prevent possible deadlocks change this in architecture code and
use a spinlock instead of a mutex.
Not all tvaudio chips allow controlling bass/treble. So, the driver
has a table with a flag to indicate if the chip does support it.
Unfortunately, the handling of this logic were broken for a very long
time (probably since the first module version). Due to that, an OOPS
were generated for devices that don't support bass/treble.
This were the resulting OOPS message before the patch, with debug messages
enabled:
icmpmsg_put() can happily corrupt kernel memory, using a static
table and forgetting to reset an array index in a loop.
Remove the static array since its not safe without proper locking.
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: Eric Dumazet <dada1@cosmosbay.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
At least the Vaio VGN-Z540N doesn't have this method, so let's not fail
to suspend just because it doesn't exist.
Signed-off-by: Adam Jackson <ajax@redhat.com> Acked-by: Mattia Dongili <malattia@linux.it> Cc: Len Brown <lenb@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Cc: Steve Conklin <sconklin@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
My last bugfix here (adding zone->lock) introduced a new problem: Using
page_zone(pfn_to_page(pfn)) to get the zone after the for() loop is wrong.
pfn will then be >= end_pfn, which may be in a different zone or not
present at all. This may lead to an addressing exception in page_zone()
or spin_lock_irqsave().
Now I use __first_valid_page() again after the loop to find a valid page
for page_zone().
The Freescale implementation of MPIC only allows a single CPU destination
for non-IPI interrupts. We add a flag to the mpic_init to distinquish
these variants of MPIC. We pull in the irq_choose_cpu from sparc64 to
select a single CPU as the destination of the interrupt.
This is to deal with the fact that the default smp affinity was
changed by commit 18404756765c713a0be4eb1082920c04822ce588 ("genirq:
Expose default irq affinity mask (take 3)") to be all CPUs.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
But blk_recalc_rq_segments might expect that req->biotail and the
previous bio in the req are supposed be merged into one
segment. blk_recalc_rq_segments might also expect that next_req->bio
and the next bio in the next_req are supposed be merged into one
segment. In such case, we merge two requests that can't be merged
here. Later, blk_rq_map_sg gives more segments than it should.
We need to keep track of segment size in blk_recalc_rq_segments and
use it to see if two requests can be merged. This patch implements it
in the similar way that we used to do for hw merging (virtual
merging).
The below is a simplistic fix for "make deb-pkg"; it splits the
firmware out to a linux-firmware-image package and adds an
(unversioned) Suggests to the linux package for this firmware.
Signed-Off-By: Jonathan McDowell <noodles@earth.li> Acked-by: Frans Pop <elendil@planet.nl> Signed-off-by: Sam Ravnborg <sam@ravnborg.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Ivan Kuten <ivan.kuten@promwad.com> Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Oops. Part of the hugetlb private reservation code was not fully
converted to use hstates.
When a huge page must be unmapped from VMAs due to a failed COW,
HPAGE_SIZE is used in the call to unmap_hugepage_range() regardless of
the page size being used. This works if the VMA is using the default
huge page size. Otherwise we might unmap too much, too little, or
trigger a BUG_ON. Rare but serious -- fix it.
Signed-off-by: Adam Litke <agl@us.ibm.com> Cc: Jon Tollefson <kniht@linux.vnet.ibm.com> Cc: Mel Gorman <mel@csn.ul.ie> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Make sure we can tell if the GPE storm workaround gets activated,
and avoid flooding the logs afterwards.
http://bugzilla.kernel.org/show_bug.cgi?id=11841
"plenty of line "ACPI: EC: non-query interrupt received,
switching to interrupt mode" in dmesg"
Signed-off-by: Alan Jenkins <alan-jenkins@tuffmail.co.uk> Acked-by: Alexey Starikovskiy <astarikovskiy@suse.de> Signed-off-by: Len Brown <len.brown@intel.com> Cc: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Since commit bc45b1d39a925b56796bebf8a397a0491489d85c acpi tables are
allowed to have an empty signature and /sys/firmware/acpi/tables uses the
signature as filename. Applications using naive recursion through /sys
loop forever. A possible solution would be: (replacing the zero length
filename with the string "NULL")
http://bugzilla.kernel.org/show_bug.cgi?id=11539
Acked-by: Zhang Rui <rui.zhang@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Len Brown <len.brown@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Like mac80211 did, this driver makes 'clever' use of skb->cb to pass
information along with an skb as it is requeued from the virtual device
to the physical wireless device. Unfortunately, that trick no longer
works...
Unlike mac80211, code complexity and driver apathy makes this hack
the best option we have in the short run. Hopefully someone will
eventually be motivated to code a proper fix before all the effected
hardware dies.
(Above text by me. Johannes officially disavows all knowledge of this
hack. -- JWL)
Signed-off-by: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
It turns out that if one registers a struct platform_device, the
platform device code expects that platform_device.device->driver points
to a struct driver inside a struct platform_driver.
This is not the case with the ipmi-si, ipmi-msghandler and ibmaem
drivers, which causes the suspend/resume hook functions to jump off into
nowhere, causing a crash. Make this assumption hold true for these
three drivers.
Signed-off-by: Darrick J. Wong <djwong@us.ibm.com> Acked-by: Corey Minyard <cminyard@mvista.com> Cc: Jean Delvare <khali@linux-fr.org> Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Same fix as commit c7cf72dcadb: when 'start' and 'end' are less than a
cacheline apart and 'start' is unaligned we are done after cleaning and
invalidating the first cacheline.
Signed-off-by: Nicolas Pitre <nico@marvell.com> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
netif_carrier_off was called too early at the probe. In case of failure
or simply bad timing, this can cause a fatal error since linkwatch_event
might run too soon.
Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net> Cc: Alex Chiang <achiang@hp.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The current code read nothing but zeros on big-endian (wrong part of the
32bits). This caused poor performance on big-endian machines. Though this
issue did not cause the system to crash, the performance is significantly
better with the fix so I view it as critical bug fix.
Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net> Cc: Alex Chiang <achiang@hp.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When the PMF flag is set, the driver can access the HW freely. When the
driver is unloaded, it should not access the HW. The problem caused fatal
errors when "ethtool -i" was called after the calling instance was unloaded
and another instance was already loaded
Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net> Cc: Alex Chiang <achiang@hp.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Take care to handle register 0xa228 exactly as in the HAL released by
Atheros. This change is required to make ath5k work again on my system
since commit 2203d6be (ath5k: Misc hw_reset updates), thus fixing a
regression in 2.6.27 and therefore hopefully eligible for inclusion into
a stable release.
v2: Only overwrite initial register values on later revisions of AR5212
chips.
v3: Use standard macros to manipulate the register.
Signed-off-by: Elias Oltmanns <eo@nebensachen.de> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Based on a patch by Elias Oltmanns, we call ath5k_init in resume even
if we didn't previously open the device. Besides starting up the
device unnecessarily, this also causes an oops on rmmod because
mac80211 will not invoke ath5k_stop and softirqs are left running after
the module has been unloaded. Add a new state bit, ATH_STAT_STARTED,
to indicate that we have been started up.
After a s2ram / resume cycle, resetting the key cache does not work
unless it is deferred until after the hardware has been reinitialised by
a call to ath5k_hw_reset(). This fixes a regression introduced by
"ath5k: fix suspend-related oops on rmmod".
Reported-by: Toralf Förster <toralf.foerster@gmx.de> Signed-off-by: Elias Oltmanns <eo@nebensachen.de> Signed-off-by: Bob Copeland <me@bobcopeland.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Avoid the sleep by changing iwl_scan_cancel_timeout with
iwl_scan_cancel and simply returning on failure if the scan persists.
This will cause hardware decryption to fail and we'll handle a few more
frames with software decryption.
Signed-off-by: John W. Linville <linville@tuxdriver.com> Cc: Holger Macht <hmacht@novell.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Daemons that need to be launched while the rootfs is read-only can now
poll /proc/mounts to be notified when their O_RDWR requests may no
longer end in EROFS.
Cc: Kay Sievers <kay.sievers@vrfy.org> Cc: Neil Brown <neilb@suse.de> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix a stack corruption caused by a corrupted hfs filesystem. If the
catalog name length is corrupted the memcpy overwrites the catalog btree
structure. Since the field is limited to HFS_NAMELEN bytes in the
structure and the file format, we throw an error if it is too long.
Cc: Roman Zippel <zippel@linux-m68k.org> Signed-off-by: Eric Sesterhenn <snakebyte@gmx.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
It seems that some cards are slightly out of spec and occasionally
will not be able to complete a write in the alloted 250 ms [1].
Incease the timeout slightly to allow even these cards to function
properly.
[1] http://lkml.org/lkml/2008/9/23/390
Signed-off-by: Pierre Ossman <drzeus@drzeus.cx> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix the __pfn_to_page(pfn) macro so that it doesn't evaluate its
argument twice in the CONFIG_DISCONTIGMEM=y case, because 'pfn' may
be a result of a funtion call having side effects.
For example, the hibernation code applies pfn_to_page(pfn) to the
result of a function returning the pfn corresponding to the next set
bit in a bitmap and the current bit position is modified on each
call. This leads to "interesting" failures for CONFIG_DISCONTIGMEM=y
due to the current behavior of __pfn_to_page(pfn).
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Acked-by: Pavel Machek <pavel@suse.cz> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
ehc->last_reset is used to ensure that resets are not issued too
close to each other. It's initialized to jiffies minus one minute
on EH entry. However, when new links are initialized after PMP is
probed, new links have zero for this timestamp resulting in long wait
depending on the current jiffies.
This patch makes last_set considered iff ATA_EHI_DID_RESET is set, in
which case last_reset is always initialized. As an added precaution,
WARN_ON() is added so that warning is printed if last_reset is
in future.
This problem is spotted and debugged by Shane Huang.
Make request_key() instantiate the per-user keyrings so that it doesn't oops
if it needs to get hold of the user session keyring because there isn't a
session keyring in place.
Signed-off-by: David Howells <dhowells@redhat.com> Tested-by: Steve French <smfrench@gmail.com> Tested-by: Rutger Nijlunsing <rutger.nijlunsing@gmail.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Some machines don't have the pullup/down on their reset
pin, so configuring the reset generating pin as input makes
them reset immediately. Fix that by making reset pin direction
configurable.
This fixes the boot problem on Sharp Zaurus c3000
Signed-off-by: Dmitry Baryshkov <dbaryshkov@gmail.com> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Pavel Machek <pavel@suse.cz> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
In some BIOSes, every _STA method call will send a notification again,
this cause freeze. And in some BIOSes, it appears _STA should be called
after _DCK. This tries to avoid calls _STA, and still keep the device
present check.
http://bugzilla.kernel.org/show_bug.cgi?id=10431
Signed-off-by: Shaohua Li <shaohua.li@intel.com> Signed-off-by: Len Brown <len.brown@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This avoids a Microcode error as reported in:
http://www.intellinuxwireless.org/bugzilla/show_bug.cgi?id=1650
http://bugzilla.kernel.org/show_bug.cgi?id=11806
http://marc.info/?l=linux-wireless&m=122437145211886&w=2
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch fixes suspend to RAM after by moving
notify_mac out of iwlwifi mutex
http://bugzilla.kernel.org/show_bug.cgi?id=11845
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com> Signed-off-by: Tomas Winkler <tomas.winkler@intel.com> Tested-by: Carlos R. Mafra <crmafra2@gmail.com> Tested-by: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Use correct DMA_MASK: 4964 and 5000 support 36 bit addresses for
pci express memory access.
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com> Signed-off-by: Zhu Yi <yi.zhu@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
This patch returns success and empty scan on scans requests that were
rejected because issued too early. The cached bss list from previous
scanning will be returned by mac80211.
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com> Signed-off-by: Zhu Yi <yi.zhu@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch removes the HT flags from RXON when moving from HT to legacy.
This avoids keeping those flags set and possibly miss configuring firmware.
If we are configured in HT, fat channel: channel 1 above, and move later
to legacy channel 11, we need to clear the FAT channel control flags in
RXON. If we don't, the firmware will understand this as channel 11 above
which is not possible due to regulatory constraints, leading to firmware
crash.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com> Reviewed-by: Tomas Winkler <tomas.winkler@intel.com> Signed-off-by: Zhu Yi <yi.zhu@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch disables power save upon association and enables it back
after association. This allows to associate to AP on a radar channel
if power save is enabled.
Radar and passive channels are not allowed for TX (required for association)
unless RX is received but PS may close the radio and no RX will be received
effectively failing association.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com> Signed-off-by: Mohamed Abbas <mohamed.abbas@intel.com> Signed-off-by: Tomas Winkler <tomas.winkler@intel.com> Signed-off-by: Zhu Yi <yi.zhu@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch allows consecutive scans requests when driver is in
unassociated state.
Signed-off-by: Ron Rindjunsky <ron.rindjunsky@intel.com> Signed-off-by: Esti Kummer <ester.kummer@intel.com> Signed-off-by: Tomas Winkler <tomas.winkler@intel.com> Signed-off-by: Zhu Yi <yi.zhu@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Mark dca_init as a subsys_initcall since it needs to be ready to go
before dependent drivers start registering themselves.
Reported-and-tested-by: Mark Rustad <mark_rustad@Xiotech.com> Acked-by: Maciej Sosnowski <maciej.sosnowski@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
async_tx.callback should be checked for the first
not the last descriptor in the chain.
Signed-off-by: Maciej Sosnowski <maciej.sosnowski@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Error handling needs to be modified in dma_pin_iovec_pages().
It should return NULL instead of ERR_PTR
(pinned_list is checked for NULL in tcp_recvmsg() to determine
if iovec pages have been successfully pinned down).
In case of error for the first iovec,
local_list->nr_iovecs needs to be initialized.
Signed-off-by: Maciej Sosnowski <maciej.sosnowski@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
If the ioatdma driver is loaded but not used it does not allocate descriptors.
Before it frees channel resources it should first be sure
that they have been previously allocated.
Signed-off-by: Maciej Sosnowski <maciej.sosnowski@intel.com> Tested-by: Tom Picard <tom.s.picard@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
- the register is defined for the 8169 chipset only and there is
no 8169 beyond RTL_GIGA_MAC_VER_06.
- only the lower 3 bytes of the register are valid
Previously I assumed that the receive queues of candidates don't
change during the GC. This is only half true, nothing can be received
from the queues (see comment in unix_gc()), but buffers could be added
through the other half of the socket pair, which may still have file
descriptors referring to it.
This can result in inc_inflight_move_tail() erronously increasing the
"inflight" counter for a unix socket for which dec_inflight() wasn't
previously called. This in turn can trigger the "BUG_ON(total_refs <
inflight_refs)" in a later garbage collection run.
Fix this by only manipulating the "inflight" counter for sockets which
are candidates themselves. Duplicating the file references in
unix_attach_fds() is also needed to prevent a socket becoming a
candidate for GC while the skb that contains it is not yet queued.
For "unlock" cycles to 16bit devices in 8bit compatibility mode we need
to use the byte addresses 0xaaa and 0x555. These effectively match
the word address 0x555 and 0x2aa, except the latter has its low bit set.
Most chips don't care about the value of the 'A-1' pin in x8 mode,
but some -- like the ST M29W320D -- do. So we need to be careful to
set it where appropriate.
cfi_send_gen_cmd is only ever passed addresses where the low byte
is 0x00, 0x55 or 0xaa. Of those, only addresses ending 0xaa are
affected by this patch, by masking in the extra low bit when the device
is known to be in compatibility mode.
[dwmw2: Do it only when (cmd_ofs & 0xff) == 0xaa]
v4: Fix stupid typo in cfi_build_cmd_addr that failed to compile
I'm writing this patch way to late at night.
v3: Bring all of the work back into cfi_build_cmd_addr
including calling of map_bankwidth(map) and cfi_interleave(cfi)
So every caller doesn't need to.
v2: Only modified the address if we our device_type is larger than our
bus width.
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: David Woodhouse <David.Woodhouse@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When 'start' and 'end' are less than a cacheline apart and 'start' is
unaligned we are done after cleaning and invalidating the first
cacheline. So check for (start < end) which will not walk off into
invalid address ranges when (start > end).
This issue was caught by drivers/dma/dmatest.
2.6.27 is susceptible.
Cc: <stable@kernel.org> Cc: Haavard Skinnemoen <haavard.skinnemoen@atmel.com> Cc: Lothar Wafmann <LW@KARO-electronics.de> Cc: Lennert Buytenhek <buytenh@marvell.com> Cc: Eric Miao <eric.miao@marvell.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The thread_should_wake() function trawls through the list of 'very
dirty' eraseblocks, determining whether the background GC thread should
wake. Doing this without holding the appropriate locks is a bad idea.
OLPC Trac #8615
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
deflate_mutex protects the globals lzo_mem and lzo_compress_buf. However,
jffs2_lzo_compress() unlocks deflate_mutex _before_ it has copied out the
compressed data from lzo_compress_buf. Correct this by moving the mutex
unlock after the copy.
In addition, document what deflate_mutex actually protects.
Signed-off-by: Geert Uytterhoeven <Geert.Uytterhoeven@sonycom.com> Acked-by: Richard Purdie <rpurdie@openedhand.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: David Woodhouse <David.Woodhouse@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Adding a spare to a raid10 doesn't cause recovery to start.
This is due to an silly type in
commit 6c2fce2ef6b4821c21b5c42c7207cb9cf8c87eda
and so is a bug in 2.6.27 and .28-rc.
Thanks to Thomas Backlund for bisecting to find this.
Cc: Thomas Backlund <tmb@mandriva.org> Cc: George Spelvin <linux@horizon.com> Signed-off-by: NeilBrown <neilb@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Date: Thu, 6 Nov 2008 19:41:24 +1100
Subject: md: linear: Fix a division by zero bug for very small arrays.
We currently oops with a divide error on starting a linear software
raid array consisting of at least two very small (< 500K) devices.
The bug is caused by the calculation of the hash table size which
tries to compute sector_div(sz, base) with "base" being zero due to
the small size of the component devices of the array.
Fix this by requiring the hash spacing to be at least one which
implies that also "base" is non-zero.