David Fries [Tue, 22 Jan 2008 11:31:39 +0000 (03:31 -0800)]
W1: w1_therm.c is flagging 0C etc as invalid
The extra rom[0] check is flagging valid temperatures as invalid when
there is already a CRC data transmission check.
w1_therm_read_bin()
if (rom[8] == crc && rom[0])
verdict = 1;
Requiring rom[0] to be non-zero will flag as invalid temperature
conversions when the low byte is zero, specifically the temperatures 0C,
16C, 32C, 48C, -16C, -32C, and -48C.
The CRC check is produced on the device for the previous 8 bytes and is
required to ensure the data integrity in transmission. I don't see why the
extra check for rom[0] being non-zero is in there. Evgeniy Polyakov didn't
know either. Just for a check I unplugged the sensor, executed a
temperature conversion, and read the results. The read was all ff's, which
also failed the CRC, so it doesn't need to protect against a disconnected
sensor.
I have more extensive patches in the work, but these two trivial ones will
do for today. I would like to hear from people who use the ds2490 USB to
one wire dongle. 1 if you would be willing to test the patches as I
currently only have the one sensor on a short parisite powered wire, 2 if
there is any cheap sources for the ds2490.
Correct the decoding of negative C temperatures. The code did a binary OR
of two bytes to make a 16 bit value, but assignd it to an integer. This
caused the value to not be sign extended and to loose that it was a
negative number in the assignment.
Before the patch (in my freezer),
w1_slave
ed fe 4b 46 7f ff 03 10 e4 : crc=e4 YES
ed fe 4b 46 7f ff 03 10 e4 t=4078
With the patch,
e3 fe 4b 46 7f ff 0d 10 81 : crc=81 YES
e3 fe 4b 46 7f ff 0d 10 81 t=-17
It tried to fix long standing bugzilla entries, but the solution was
reported to break other systems. The reporter of
http://bugzilla.kernel.org/show_bug.cgi?id=9791
tracked it down to this commit and confirmed that reverting the patch
restores the correct behaviour. It's too late in the release cycle to
find a better solution than reverting the commit to avoid regressions.
Linus Torvalds [Tue, 22 Jan 2008 03:40:05 +0000 (19:40 -0800)]
Merge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
tc35815: Use irq number for tc35815-mac platform device id
[MIPS] Malta: Fix reading the PCI clock frequency on big-endian
[MIPS] SMTC: Fix build error.
Andrew G. Morgan [Tue, 22 Jan 2008 01:18:30 +0000 (17:18 -0800)]
Fix filesystem capability support
In linux-2.6.24-rc1, security/commoncap.c:cap_inh_is_capped() was
introduced. It has the exact reverse of its intended behavior. This
led to an unintended privilege esculation involving a process'
inheritable capability set.
To be exposed to this bug, you need to have Filesystem Capabilities
enabled and in use. That is:
- CONFIG_SECURITY_FILE_CAPABILITIES must be defined for the buggy code
to be compiled in.
- You also need to have files on your system marked with fI bits raised.
Randy Dunlap [Tue, 22 Jan 2008 01:18:25 +0000 (17:18 -0800)]
timer: fix section mismatch
The caller is __cpuinit.
Also, this code block and its caller are inside #ifdef CONFIG_HOTPLUG_CPU
blocks, so this code should reflect that config symbol's usage.
WARNING: vmlinux.o(.text+0x4252f): Section mismatch: reference to .init.text: (between 'timer_cpu_notify' and 'msleep')
Alan Cox [Tue, 22 Jan 2008 01:18:24 +0000 (17:18 -0800)]
keyspan: fix oops
If we get a data URB back from the hardware after we have put the tty to
bed we go kaboom. Fortunately all we need to do is process the URB without
trying to ram its contents down the throat of an ex-tty.
Atsushi Nemoto [Fri, 18 Jan 2008 16:15:52 +0000 (01:15 +0900)]
tc35815: Use irq number for tc35815-mac platform device id
The tc35815-mac platform device used a pci bus number and a devfn to
identify its target device, but the pci bus number may vary if some
bus-bridges are found. Use irq number which is be unique for embedded
controllers.
Dmitri Vorobiev [Mon, 14 Jan 2008 21:27:46 +0000 (00:27 +0300)]
[MIPS] Malta: Fix reading the PCI clock frequency on big-endian
The JMPRS register on Malta boards keeps a 32-bit CPU-endian
value. The readw() function assumes that the value it reads is a
little-endian 16-bit number. Therefore, using readw() to obtain
the value of the JMPRS register is a mistake. This error leads
to incorrect reading of the PCI clock frequency on big-endian
during board start-up.
Change readw() to __raw_readl().
This was tested by injecting a call to printk() and verifying
that the value of the jmpr variable was consistent with current
setting of the JP4 "PCI CLK" jumper.
In that patch, David L added a icmp_out_count() in
ip_push_pending_frames(), remove icmp_out_count() from
icmp_reply(). But he forgot to remove icmp_out_count() from
icmp_send() too. Since icmp_send and icmp_reply will call
icmp_push_reply, which will call ip_push_pending_frames, a duplicated
increment happened in icmp_send.
This patch remove the icmp_out_count from icmp_send too.
Patrick McHardy [Mon, 21 Jan 2008 01:25:14 +0000 (17:25 -0800)]
[NET]: rtnl_link: fix use-after-free
When unregistering the rtnl_link_ops, all existing devices using
the ops are destroyed. With nested devices this may lead to a
use-after-free despite the use of for_each_netdev_safe() in case
the upper device is next in the device list and is destroyed
by the NETDEV_UNREGISTER notifier.
The easy fix is to restart scanning the device list after removing
a device. Alternatively we could add new devices to the front of
the list to avoid having dependant devices follow the device they
depend on. A third option would be to only restart scanning if
dev->iflink of the next device matches dev->ifindex of the current
one. For now this seems like the safest solution.
With this patch, the veth rtnl_link_ops unregistration can use
rtnl_link_unregister() directly since it now also handles destruction
of multiple devices at once.
David S. Miller [Mon, 21 Jan 2008 00:39:03 +0000 (16:39 -0800)]
[NEIGH]: Revert 'Fix race between neigh_parms_release and neightbl_fill_parms'
Commit 9cd40029423701c376391da59d2c6469672b4bed (Fix race between
neigh_parms_release and neightbl_fill_parms) introduced device
reference counting regressions for several people, see:
When packets are flood-forwarded to multiple output devices, the
bridge-netfilter code reuses skb->nf_bridge for each clone to store
the bridge port. When queueing packets using NFQUEUE netfilter takes
a reference to skb->nf_bridge->physoutdev, which is overwritten
when the packet is forwarded to the second port. This causes
refcount unterflows for the first device and refcount leaks for all
others. Additionally this provides incorrect data to the iptables
physdev match.
Unshare skb->nf_bridge by copying it if it is shared before assigning
the physoutdev device.
Reported, tested and based on initial patch by
Jan Christoph Nordholz <[email protected]>.
[IPV6] ROUTE: Make sending algorithm more friendly with RFC 4861.
We omit (or delay) sending NSes for known-to-unreachable routers (in
NUD_FAILED state) according to RFC 4191 (Default Router Preferences
and More-Specific Routes). But this is not fully compatible with RFC
4861 (Neighbor Discovery Protocol for IPv6), which does not remember
unreachability of neighbors.
So, let's avoid mixing sending algorithm of RFC 4191 and that of RFC
4861, and make the algorithm more friendly with RFC 4861 if RFC 4191
is disabled.
Issue was found by IPv6 Ready Logo Core Self_Test 1.5.0b2 (by TAHI
Project), and has been tracked down by Mitsuru Chinen
<[email protected]>.
Eric Dumazet [Fri, 18 Jan 2008 12:30:21 +0000 (04:30 -0800)]
[IPV4] FIB_HASH : Avoid unecessary loop in fn_hash_dump_zone()
I noticed "ip route list" was slower than "cat /proc/net/route" on a
machine with a full Internet routing table (214392 entries : Special
thanks to Robert ;) )
David S. Miller [Fri, 18 Jan 2008 12:21:39 +0000 (04:21 -0800)]
[NET]: Fix interrupt semaphore corruption in Intel drivers.
Several of the Intel ethernet drivers keep an atomic counter used to
manage when to actually hit the hardware with a disable or an enable.
The way the net_rx_work() breakout logic works during a pending
napi_disable() is that it simply unschedules the poll even if it
still has work.
This can potentially leave interrupts disabled, but that is OK
because all of the drivers are about to disable interrupts
anyways in all such code paths that do a napi_disable().
Unfortunately, this trips up the semaphore used here in the Intel
drivers. If you hit this case, when you try to bring the interface
back up it won't enable interrupts. A reload of the driver module
fixes it of course.
So what we do is make sure all the sequences now go:
Russell King [Sun, 20 Jan 2008 10:18:12 +0000 (10:18 +0000)]
Merge branch 'omap-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap-2.6
* 'omap-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap-2.6:
ARM: OMAP1: Fix compile for board-nokia770
ARM: OMAP1: Keymap fix for f-sample and p2-sample
Dan Williams [Fri, 18 Jan 2008 22:26:10 +0000 (23:26 +0100)]
[ARM] 4748/1: dca: source drivers/dca/Kconfig in arch/arm/Kconfig to fix warning
'select' used by config symbol 'INTEL_IOATDMA' refers to undefined symbol 'DCA'
Although drivers/dma is currently the only user future drivers outside of
drivers/dma may select this option so it is better to add this to
arch/arm/Kconfig than move DCA to drivers/dma/Kconfig.
Russell King [Fri, 11 Jan 2008 12:23:58 +0000 (12:23 +0000)]
[ARM] pxa: don't rely on r2 being preserved over a function call
r2 is not guaranteed to be preserved over a function call, so relying
on it to store the link register over the call to sleep_phys_sp() is
unreliable. Store the link register on the stack instead.
Rusty Russell [Sun, 20 Jan 2008 00:54:18 +0000 (11:54 +1100)]
Remove bogus duplicate CONFIG_LGUEST_GUEST entry.
It was moved to arch/x86/lguest/Kconfig, but I lost the deletion part in a
patch suffle. My confused one-liner "fix" to turn it on is also reverted: 84f7466ee20cc094aa38617abfa2f3834871f054
Alan Cox [Sat, 19 Jan 2008 15:51:26 +0000 (15:51 +0000)]
pata_pdc202xx_old: Fix crashes with ATAPI
The PDC202xx older devices do not support ATAPI DMA via the usual
interfaces. What documentation I have isn't sufficient to support DMA and
it isn't clear if the Windows drivers do this or it is possible at all.
(Neither do the drivers/ide old drivers)
So turn it ATAPI DMA off, these are disk optimised controllers.
* git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog:
[WATCHDOG] clarify watchdog operation in documentation
[WATCHDOG] Revert "Stop looking for device as soon as one is found"
Arjan van de Ven [Fri, 18 Jan 2008 21:49:33 +0000 (22:49 +0100)]
x86: add support for the latest Intel processors to Oprofile
The latest Intel processors (the 45nm ones) have a model number of 23
(old ones had 15); they're otherwise compatible on the oprofile side.
This patch adds the new model number to the oprofile code.
Al Viro [Sun, 13 Jan 2008 14:17:35 +0000 (14:17 +0000)]
3c574, 3c515 bitfields abuse
wn3_config is shared by these cards; the way we deal with it is both bad C
(union abuse) and broken on big-endian. For 3c515 it's less serious (ISA
cards are quite rare outside of little-endian boxen), but 3c574 is a pcmcia
one and that'd better be endian-independent... Fix is the same in both
cases.
S2io: Fixed synchronization between scheduling of napi with card reset and close
- Fixed synchronization between scheduling of napi with card reset and close
by moving the enabling and disabling of napi to card up and card down
functions respectively instead of open and close.
Jay Cliburn [Tue, 15 Jan 2008 01:56:41 +0000 (19:56 -0600)]
atl1: fix frame length bug
The driver sets up the hardware to accept a frame with max length
equal to MTU + Ethernet header + FCS + VLAN tag, but we neglect to
add the VLAN tag size to the ingress buffer. When a VLAN-tagged
frame arrives, the hardware passes it, but bad things happen
because the buffer is too small. This patch fixes that.
Thanks to David Harris for reporting the bug and testing the fix.
Some BIOS's break if Wake On Lan is enabled, and the machine
can't boot. Better to have some user's have to call ethtool to
enable WOL than to break a single user's boot.
Jay Vosburgh [Fri, 18 Jan 2008 00:25:02 +0000 (16:25 -0800)]
bonding: fix lock ordering for rtnl and bonding_rwsem
Fix the handling of rtnl and the bonding_rwsem to always be acquired
in a consistent order (rtnl, then bonding_rwsem).
The existing code sometimes acquired them in this order, and sometimes
in the opposite order, which opens a window for deadlock between ifenslave
and sysfs.
Jay Vosburgh [Fri, 18 Jan 2008 00:25:01 +0000 (16:25 -0800)]
bonding: Fix up parameter parsing
A recent change to add an additional hash policy modified
bond_parse_parm, but it now does not correctly match parameters passed in
via sysfs.
Rewrote bond_parse_parm to handle (a) parameter matches that
are substrings of one another and (b) user input with whitespace (e.g.,
sysfs input often has a trailing newline).
Jay Vosburgh [Fri, 18 Jan 2008 00:25:00 +0000 (16:25 -0800)]
bonding: release slaves when master removed via sysfs
Add a call to bond_release_all in the bonding netdev event
handler for the master. This releases the slaves for the case of, e.g.,
"echo -bond0 > /sys/class/net/bonding_masters", which otherwise will spin
forever waiting for references to be released.
Jay Vosburgh [Fri, 18 Jan 2008 00:24:59 +0000 (16:24 -0800)]
bonding: fix locking during alb failover and slave removal
alb_fasten_mac_swap (actually rlb_teach_disabled_mac_on_primary)
requries RTNL and no other locks. This could cause dev_set_promiscuity
and/or dev_set_mac_address to be called with improper locking.
Changed callers to hold only RTNL during calls to alb_fasten_mac_swap
or functions calling it. Updated header comments in affected functions to
reflect proper reality of locking requirements.
Jay Vosburgh [Fri, 18 Jan 2008 00:24:58 +0000 (16:24 -0800)]
bonding: fix ASSERT_RTNL that produces spurious warnings
Move an ASSERT_RTNL down to where we should hold only RTNL;
the existing check produces spurious warnings because we hold additional
locks at _bh, tripping a debug warning in spin_lock_mutex().
Jay Vosburgh [Fri, 18 Jan 2008 00:24:57 +0000 (16:24 -0800)]
bonding: fix locking in sysfs primary/active selection
Fix the functions that store the primary and active slave
options via sysfs to hold the correct locks in the correct order.
The bond_change_active_slave and bond_select_active_slave
functions both require rtnl, bond->lock for read and curr_slave_lock for
write_bh, and no other locks. This is so that the lower level
mode-specific functions (notably for balance-alb mode) can release locks
down to just rtnl in order to call, e.g., dev_set_mac_address with the
locks it expects (rtnl only).
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6:
[NIU]: Fix 1G PHY link state handling.
[NET]: Fix TX timeout regression in Intel drivers.
Nigel Cunningham [Thu, 17 Jan 2008 23:21:21 +0000 (15:21 -0800)]
Fix unbalanced helper_lock in kernel/kmod.c
call_usermodehelper_exec() has an exit path that can leave the
helper_lock() call at the top of the routine unbalanced. The attached
patch fixes this issue.
Framebuffer is ok only with default parameters only (it is 1280x800-8@60). If
parameters are video=radeonfb:1280x800-32@60 then xres, yres and xres_virtual
are ok but yres_virtual is 1024. It can be corrected by fbset utility so I
think it can be corrected in the driver code also.
Steps to reproduce: video=radeonfb:1280x800-32@60 or
video=radeonfb:1280x800-16@60
Carsten Otte [Thu, 17 Jan 2008 23:21:17 +0000 (15:21 -0800)]
#ifdef very expensive debug check in page fault path
This patch puts #ifdef CONFIG_DEBUG_VM around a check in vm_normal_page
that verifies that a pfn is valid. This patch increases performance of the
page fault microbenchmark in lmbench by 13% and overall dbench performance
by 7% on s390x. pfn_valid() is an expensive operation on s390 that needs a
high double digit amount of CPU cycles. Nick Piggin suggested that
pfn_valid() involves an array lookup on systems with sparsemem, and
therefore is an expensive operation there too.
The check looks like a clear debug thing to me, it should never trigger on
regular kernels. And if a pte is created for an invalid pfn, we'll find
out once the memory gets accessed later on anyway. Please consider
inclusion of this patch into mm.
Sam Ravnborg [Thu, 17 Jan 2008 23:21:12 +0000 (15:21 -0800)]
mm: fix section mismatch warning in page_alloc.c
With CONFIG_HOTPLUG=n and CONFIG_HOTPLUG_CPU=y we saw
following warning:
WARNING: mm/built-in.o(.text+0x6864): Section mismatch: reference to .init.text: (between 'process_zones' and 'pageset_cpuup_callback')
The culprit was zone_batchsize() which were annotated __devinit but used
from process_zones() which is annotated __cpuinit. zone_batchsize() are
used from another function annotated __meminit so the only valid option is
to drop the annotation of zone_batchsize() so we know it is always valid to
use it.
Jesper Nilsson [Thu, 17 Jan 2008 23:21:11 +0000 (15:21 -0800)]
CRIS v10: vmlinux.lds.S: ix kernel oops on boot and use common defines
- Move alignment to page size of init data outside ifdef for BLK_DEV_INITRD.
The reservation up to page size of memory after init data was previously
not done if BLK_DEV_INITRD was undefined.
This caused a kernel oops when init memory pages were freed after startup,
data placed in the same page as the last init memory would also be freed
and reused, with disastrous results.
- Use macros for initcalls and .text sections.
- Replace hardcoded page size constant with PAGE_SIZE define.
- Change include/asm-cris/page.h to use the _AC macro to instead
of testing __ASSEMBLY__.
Johannes Weiner [Thu, 17 Jan 2008 23:21:08 +0000 (15:21 -0800)]
cpufreq: Initialise default governor before use
When the cpufreq driver starts up at boot time, it calls into the default
governor which might not be initialised yet. This hurts when the
governor's worker function relies on memory that is not yet set up by its
init function.
This migrates all governors from module_init() to fs_initcall() when being
the default, as was already done in cpufreq_performance when it was the
only possible choice. The performance governor is always initialized early
because it might be used as fallback even when not being the default.
Fixes at least one actual oops where ondemand is the default governor and
cpufreq_governor_dbs() uses the uninitialised kondemand_wq work-queue
during boot-time.
Francois Romieu [Thu, 10 Jan 2008 22:53:15 +0000 (23:53 +0100)]
ipg: fix Tx completion irq request
The current logic will only request an ack for the first pending
packet. No irq is triggered as soon as the CPU submits a few
packets a bit quickly. Let's request an irq for every packet
instead.
Francois Romieu [Thu, 10 Jan 2008 22:40:59 +0000 (23:40 +0100)]
ipg: plug Tx completion leak
The Tx skb release could not free more than one skb per call.
Add it to the fact that the xmit handler does not check for
a queue full condition and you have a recipe to leak quickly.
Let's release every pending Tx descriptor which has been given
back to the host CPU by the network controller. The xmit handler
suggests that it is done through the IPG_TFC_TFDDONE bit.
Remove the former "curr" computing: it does not produce anything
usable in its current form.
Andres Salomon [Thu, 17 Jan 2008 17:01:30 +0000 (12:01 -0500)]
Input: psmouse - fix potential memory leak in psmouse_connect()
If we successfully call input_register_device() in psmouse_connect()
but sysfs_create_group() fails, we'll enter the error path without
ever having called input_unregister_device() potentially leaking
memory.
projects / linux.git / log