Tony Lindgren [Fri, 29 Sep 2006 07:45:10 +0000 (10:45 +0300)]
ARM: OMAP: Avoid updating system time for sub-jiffy interrupts
Updating system time and reprogramming timer can cause latency
issues on busy systems with lots of interrupts with constant
updating of time and reprogramming the system timer.
If a non-timer dyntick interrupt happens within a jiffy from
the last interrupt, updating time and reprogramming the timer
is unnecessary as we will get a timer interrupt soon anyways.
David Brownell [Thu, 28 Sep 2006 14:48:43 +0000 (17:48 +0300)]
musb_hdrc: musb procfs updates
Tighten up error handling in the procfs dump, and avoid returning
bogus values from the read() system call. Report single vs double
buffering for FIFO setup.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
David Brownell [Thu, 28 Sep 2006 14:47:44 +0000 (17:47 +0300)]
musb_hdrc: fix some bugs in the updated host side code
- Remove an oops in the queue fault cleanup path, caused by using the
wrong giveback routine (heavy weight vs lighter-weight). Move a few
key bits of code into that lighter weight version.
- Trigger that lighter-weight cleanup during URB completion, when the
URB in question can't yet been handed off to the hardware even if
it's at the queue head.
- Reduce syslog noise when those fault cleanup paths are triggered from
scatterlist calls by usb-storage.
- Paranoia: when scheduling, insist that the endpoint fifo isn't smaller
than required.
Oops observed courtesy of a slightly flakey usb-storage device, which
once again proves itself handy for testing.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
David Brownell [Thu, 28 Sep 2006 14:46:58 +0000 (17:46 +0300)]
musb_hdrc: cleanup, mostly related to CPPI
- Remove #ifdefs in favor of new is_cppi_enabled() macro.
- Trim some tx-start logic, inlining it and removing MGC_{Read,Write}Csr16()
- Remove remnant of bogus/non-existent host-side ISO bitfield.
For upstream merging, more such #ifdeffery still needs to be removed.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
David Brownell [Thu, 28 Sep 2006 14:46:12 +0000 (17:46 +0300)]
musb_hdrc: DaVinci-specific updates and cleanups
- Add header needed for DaVinci build.
- Remove some partially obsolete comments.
- Cleaner and more reproducible host side VBUS handling, including
waiting for it to fall after powerdown on rmmod.
- Tighten up various state machine transitions and diagnostics
Plus removing some generic #ifdeffery. The vbus updates make it work more like
the TUSB code, and the cleanups etc make upstream merging more practical.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
David Brownell [Thu, 28 Sep 2006 14:45:20 +0000 (17:45 +0300)]
musb_hdrc: add heuristic to make ethernet peripherals use RNDIS RX mode
When you're willing to trust the USB host not to commit certain errors, the
CPPI "rndis" RX mode can be used with the peripheral side g_ether driver.
(In other cases, rndis rx mode tends to misbehave _very_ rudely ...)
This patch adds a heuristic to detect when the safe scenario is likely, and
a module parameter to disable the heuristic in contexts when it's likely
to fail; for example, users of gadgetfs will likely need to disable it.
This gave about a 60% improvement in throughput with CDC Ethernet (!!) and a
slight reduction in CPU load. That's measured with TTCP, which pushes a lot
of full size Ethernet frames and thus can see three IRQs per packet reduced
to just one, and also get I/O overlap when double buffering packets. That is,
net 10 MByte/sec sustained, using double buffering but no DMA transfer queues;
it's unclear why TX is still just 6 MByte/sec.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
General update of the board file for Palm Tungsten E. Registers the
platform devices contained in the PDA (ROM chip, keypad, infra-red)
and updates the configuration for USB and MMC, whose config values
were previously guessed in most cases due to lack of documentation
(and now are confirmed by a number of users). Macros for GPIO pins are
moved to a file in include/asm-arm/arch-omap.
Signed-off-by: Andrzej Zaborowski <balrog@zabor.org> Signed-off-by: Tony Lindgren <tony@atomide.com>
[PATCH] ARM: OMAP: correct misc 15xx and non-15xx platform code
Disable accesses to SOFT_REQ_REG2 and ULPD_SOFT_DISABLE_REQ_REG
registers for 15xx processors that don't have these registers. Enable
level 2 interrupt handler for processors that identify as OMAP 15xx
(e.g 310) and not 1510 specifically. Also fix the following compiler
warning (only visible with CONFIG_OMAP_RESET_CLOCKS):
arch/arm/mach-omap1/clock.c: In function 'omap1_clk_disable_unused':
arch/arm/mach-omap1/clock.c:634: warning: 'return' with a value, in
function returning void
Signed-off-by: Andrzej Zaborowski <balrog@zabor.org> Signed-off-by: Tony Lindgren <tony@atomide.com>
DSP HW-related initialization is deferred until the first DSP device
node is opened. This allows us to keep DSP powered down before the
DSP SW is loaded.
Signed-off-by: Hiroshi DOYU <Hiroshi.DOYU@nokia.com> Signed-off-by: Juha Yrjola <juha.yrjola@solidboot.com>
Here's the patch that updates DSP Gateway driver to 3.3.1, that
supports omap2 DSP. It also contains generic mailbox interface
considering IVA use on 2420, but this patch itself doesn't contain any
IVA driver.
Regarding omap2, this code is tested only on our 2420 custom board, so
feedbacks from testing on other chips / boards will be greatly
appreciated.
Signed-off-by: Hiroshi DOYU <Hiroshi.DOYU@nokia.com> Signed-off-by: Juha Yrjola <juha.yrjola@solidboot.com>
Juha Yrjola [Thu, 14 Sep 2006 11:40:09 +0000 (14:40 +0300)]
ARM: OMAP: Add support for forcing osc_ck on
Some boards might use the oscillator clock for powering
external peripherals. Add support for making sure osc_ck
stays active even when trying to go to sleep.
Signed-off-by: Juha Yrjola <juha.yrjola@solidboot.com>
Merge branch 'fixes' of git://git.linux-nfs.org/pub/linux/nfs-2.6
* 'fixes' of git://git.linux-nfs.org/pub/linux/nfs-2.6:
NFS: Fix nfs_page use after free issues in fs/nfs/write.c
NFSv4: Fix incorrect semaphore release in _nfs4_do_open()
NFS: Fix Oopsable condition in nfs_readpage_sync()
Those changes were needed for EFI-environment Intel macs, but broke some
newer Intel 965 boards, so for now it's better to revert to our old
2.6.17 behaviour and at least avoid introducing any new breakage.
Andi Kleen has a set of patches that work with both EFI and the broken
Intel 965 boards, which will be applied once they get wider testing.
Cc: Arjan van de Ven <arjan@infradead.org> Cc: Edgar Hucek <hostmaster@ed-soft.at> Cc: Andi Kleen <ak@suse.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
* git://git.infradead.org/mtd-2.6:
[MTD] Use SEEK_{SET,CUR,END} instead of hardcoded values in mtdchar lseek()
MTD: Fix bug in fixup_convert_atmel_pri
[JFFS2][SUMMARY] Fix a summary collecting bug.
[PATCH] [MTD] DEVICES: Fill more device IDs in the structure of m25p80
MTD: Add lock/unlock operations for Atmel AT49BV6416
MTD: Convert Atmel PRI information to AMD format
fs/jffs2/xattr.c: remove dead code
[PATCH] [MTD] Maps: Add dependency on alternate probe methods to physmap
[PATCH] MTD: Add Macronix MX29F040 to JEDEC
[MTD] Fixes of performance and stability issues in CFI driver.
block2mtd.c: Make kernel boot command line arguments work (try 4)
[MTD NAND] Fix lookup error in nand_get_flash_type()
remove #error on !PCI from pmc551.c
MTD: [NAND] Fix the sharpsl driver after breakage from a core conversion
[MTD] NAND: OOB buffer offset fixups
make fs/jffs2/nodelist.c:jffs2_obsolete_node_frag() static
[PATCH] [MTD] NAND: fix dead URL in Kconfig
Dave Kleikamp [Tue, 19 Sep 2006 03:12:33 +0000 (20:12 -0700)]
[PATCH] EXT2: Remove superblock lock contention in ext2_statfs
Fix a performance degradation introduced in 2.6.17. (30% degradation
running dbench with 16 threads)
Commit 21730eed11de42f22afcbd43f450a1872a0b5ea1, which claims to make
EXT2_DEBUG work again, moves the taking of the kernel lock out of
debug-only code in ext2_count_free_inodes and ext2_count_free_blocks and
into ext2_statfs.
Matthew Wilcox [Tue, 19 Sep 2006 03:12:27 +0000 (20:12 -0700)]
[PATCH] headers_check: Clean up asm-parisc/page.h for user headers
Remove definitions of PAGE_* from the user view
Delete unnecessary comments referring to the size of pages
Only include <asm-generic> if we're in __KERNEL__
Signed-off-by: David Woodhouse <dwmw2@infradead.org> Signed-off-by: Matthew Wilcox <matthew@wil.cx> Signed-off-by: Kyle McMartin <kyle@parisc-linux.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
as it turns out, the bug is caused by handle_level_irq(), which if it
races with another CPU already handling this IRQ, it _unmasks_ the IRQ
line on the way out. This is not how 2.6.17 works, and we introduced
this bug in one of the early genirq cleanups right before it went into
-mm. (the bug was not in the genirq patchset for a long time, and we
didnt notice the bug due to the lack of -rt rebase to the new genirq
code. -rt, and hardirq-preemption in particular opens up such races much
wider than anything else.)
Signed-off-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
It turns out that eflags is important to save and restore not just
because of iopl, but due to the magic bits like the NT bit, which we
don't want leaking between different threads.
Tested-by: Mike Galbraith <efault@gmx.de> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
[ATM] CLIP: Do not refer freed skbuff in clip_mkip().
[NET]: Drop tx lock in dev_watchdog_up
[PACKET]: Don't truncate non-linear skbs with mmaped IO
[NET]: Mark frame diverter for future removal.
[NETFILTER]: Add secmark headers to header-y
[ATM]: linux-atm-general mailing list is subscribers only
[ATM]: [he] when transmit fails, unmap the dma regions
[TCP] tcp-lp: update information to MAINTAINERS
[TCP] tcp-lp: bug fix for oops in 2.6.18-rc6
[BRIDGE]: random extra bytes on STP TCN packet
[IPV6]: Accept -1 for IPV6_TCLASS
[IPV6]: Fix tclass setting for raw sockets.
[IPVS]: remove the debug option go ip_vs_ftp
[IPVS]: Make sure ip_vs_ftp ports are valid
[IPVS]: auto-help for ip_vs_ftp
[IPVS]: Document the ports option to ip_vs_ftp in kernel-parameters.txt
[TCP]: Turn ABC off.
[NEIGH]: neigh_table_clear() doesn't free stats
* master.kernel.org:/home/rmk/linux-2.6-arm:
[ARM] 3815/1: headers_install support for ARM
[ARM] 3794/1: S3C24XX: do not defined set_irq_wake when no CONFIG_PM
[ARM] 3793/1: S3C2412: fix wrong serial info struct
[ARM] 3780/1: Fix iop321 cpuid
[ARM] 3786/1: pnx4008: update defconfig
[ARM] 3785/1: S3C2412: Fix idle code as default uses wrong clocks
[ARM] 3784/1: S3C2413: fix config for MACH_S3C2413/MACH_SMDK2413
Ben Dooks [Mon, 18 Sep 2006 12:30:17 +0000 (13:30 +0100)]
[ARM] 3793/1: S3C2412: fix wrong serial info struct
Patch from Ben Dooks
The S3C2440 serial info struct is being passed
through the S3C2412 serial info struct probe
routine.
Signed-off-by: Ben Dooks <ben-linux@fluff.org> Signed-off-by: Thomas Glexiner <tglx@linutronix.de> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
David S. Miller [Mon, 18 Sep 2006 08:47:13 +0000 (01:47 -0700)]
[OPENPROMIO]: Handle current_node being NULL correctly.
If the user tries to traverse to the next node of the
last node, we get NULL in current_node and a zero phandle
returned. That's fine, but if the user tries to obtain
properties in that state, we try to dereference a NULL
pointer in the downcall to the of_*() routines.
So protect against that.
Signed-off-by: David S. Miller <davem@davemloft.net>
Herbert Xu [Mon, 18 Sep 2006 07:22:30 +0000 (00:22 -0700)]
[NET]: Drop tx lock in dev_watchdog_up
Fix lockdep warning with GRE, iptables and Speedtouch ADSL, PPP over ATM.
On Sat, Sep 02, 2006 at 08:39:28PM +0000, Krzysztof Halasa wrote:
>
> =======================================================
> [ INFO: possible circular locking dependency detected ]
> -------------------------------------------------------
> swapper/0 is trying to acquire lock:
> (&dev->queue_lock){-+..}, at: [<c02c8c46>] dev_queue_xmit+0x56/0x290
>
> but task is already holding lock:
> (&dev->_xmit_lock){-+..}, at: [<c02c8e14>] dev_queue_xmit+0x224/0x290
>
> which lock already depends on the new lock.
This turns out to be a genuine bug. The queue lock and xmit lock are
intentionally taken out of order. Two things are supposed to prevent
dead-locks from occuring:
1) When we hold the queue_lock we're supposed to only do try_lock on the
tx_lock.
2) We always drop the queue_lock after taking the tx_lock and before doing
anything else.
>
> the existing dependency chain (in reverse order) is:
>
> -> #1 (&dev->_xmit_lock){-+..}:
> [<c012e7b6>] lock_acquire+0x76/0xa0
> [<c0336241>] _spin_lock_bh+0x31/0x40
> [<c02d25a9>] dev_activate+0x69/0x120
This path obviously breaks assumption 1) and therefore can lead to ABBA
dead-locks.
I've looked at the history and there seems to be no reason for the lock
to be held at all in dev_watchdog_up. The lock appeared in day one and
even there it was unnecessary. In fact, people added __dev_watchdog_up
precisely in order to get around the tx lock there.
The function dev_watchdog_up is already serialised by rtnl_lock since
its only caller dev_activate is always called under it.
So here is a simple patch to remove the tx lock from dev_watchdog_up.
In 2.6.19 we can eliminate the unnecessary __dev_watchdog_up and
replace it with dev_watchdog_up.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
Patrick McHardy [Mon, 18 Sep 2006 06:59:57 +0000 (23:59 -0700)]
[PACKET]: Don't truncate non-linear skbs with mmaped IO
Non-linear skbs are truncated to their linear part with mmaped IO.
Fix by using skb_copy_bits instead of memcpy.
Signed-off-by: Patrick McHardy <kaber@trash.net> Acked-by: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru> Signed-off-by: David S. Miller <davem@davemloft.net>
The code for frame diverter is unmaintained and has bitrotted.
The number of users is very small and the code has lots of problems.
If anyone is using it, they maybe exposing themselves to bad packet attacks.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org> Signed-off-by: David S. Miller <davem@davemloft.net>
James Morris [Thu, 14 Sep 2006 04:04:55 +0000 (21:04 -0700)]
[NETFILTER]: Add secmark headers to header-y
This patch includes xt_SECMARK.h and xt_CONNSECMARK.h to the kernel
headers which are exported via 'make headers_install'. This is needed to
allow userland code to be built correctly with these features.
Please apply, and consider for inclusion with 2.6.18 as a bugfix.
Signed-off-by: James Morris <jmorris@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Sorry that the patch submited yesterday still contain a small bug.
This version have already been test for hours with BT connections. The
oops is now difficult to reproduce.
Signed-off-by: Wong Hoi Sing Edison <hswong3i@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
This patch should add support for -1 as "default" IPv6 traffic class,
as specified in IETF RFC3542 ยง6.5. Within the kernel, it seems tclass
< 0 is already handled, but setsockopt, getsockopt and recvmsg calls
won't accept it from userland.
Signed-off-by: Remi Denis-Courmont <rdenis@simphalempin.com> Acked-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Turn Appropriate Byte Count off by default because it unfairly
penalizes applications that do small writes. Add better documentation
to describe what it is so users will understand why they might want to
turn it on.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org> Signed-off-by: David S. Miller <davem@davemloft.net>
neigh_table_clear() doesn't free tbl->stats.
Found by Alexey Kuznetsov. Though Alexey considers this
leak minor for mainstream, I still believe that cleanup
code should not forget to free some of the resources :)
At least, this is critical for OpenVZ with virtualized
neighbour tables.
Signed-Off-By: Kirill Korotaev <dev@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Frank Pavlic [Fri, 15 Sep 2006 14:27:02 +0000 (16:27 +0200)]
[PATCH] s390: qeth driver fixes [6/6]
[PATCH 9/9] s390: qeth driver fixes [6/6]
From: Frank Pavlic <fpavlic@de.ibm.com>
- Hipersockets has no IPV6 support, thus prevent issueing
SETRTG_IPV6 control commands on Hipersockets devices.
- fixed error handling in qeth_sysfs_(un)register
Signed-off-by: Frank Pavlic <fpavlic@de.ibm.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
Frank Pavlic [Fri, 15 Sep 2006 14:26:52 +0000 (16:26 +0200)]
[PATCH] s390: qeth driver fixes [5/6]
[PATCH 8/9] s390: qeth driver fixes [5/6]
From: Frank Pavlic <fpavlic@de.ibm.com>
fix kernel panic in qdio queue handling.
qeth_qdio_clear_card() could be invoked by 2 CPUs
simultaneously (for example reboot event and recovery).
Signed-off-by: Frank Pavlic <fpavlic@de.ibm.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
Frank Pavlic [Fri, 15 Sep 2006 14:26:34 +0000 (16:26 +0200)]
[PATCH] s390: qeth driver fixes [4/6]
[PATCH 7/9] s390: qeth driver fixes [4/6]
From: Frank Pavlic <fpavlic@de.ibm.com>
- fix kernel crash due to race,
set card->state to SOFTSETUP after
card and card->dev are initialized properly.
- remove CONFIG_QETH_PERF_STATS, use sysfs attribute instead,
as we want to have the ability to turn on/off the
statistics at runtime.
Signed-off-by: Frank Pavlic <fpavlic@de.ibm.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
Frank Pavlic [Fri, 15 Sep 2006 14:26:19 +0000 (16:26 +0200)]
[PATCH] s390: qeth driver fixes [3/6]
[PATCH 6/9] s390: qeth driver fixes [3/6]
From: Frank Pavlic <fpavlic@de.ibm.com>
fixed kernel panic caused by qeth driver:
Using a bonding device qeth driver will realloc
headroom for every skb coming from the bond device.
Once this happens qeth frees the original skb and
set the skb pointer to the new realloced skb.
Under heavy transmit workload (e.g.UDP streams) through bond
network device the qdio output queue might get full.
In this case we return with EBUSY from qeth_send_packet.
Returning to qeth_hard_start_xmit routine
the skb address on the stack still points to the old address,
which has been freed before.
Returning from qeth_hard_start_xmit with EBUSY results in
requeuing the skb. In this case it corrupts the qdisc queue
and results in kernel panic.
Signed-off-by: Frank Pavlic <fpavlic@de.ibm.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
projects / linux-2.6-omap-h63xx.git / log