The mouse button emulation calls input device methods from an input device. This causes funny lock nesting which is harmless as each device has its own locks. Give the nesting device its own lock classes so that lockdep will not consider them the same. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Dmitry Torokhov <dtor@mail.ru> --- drivers/macintosh/mac_hid.c | 6 ++++++ 1 files changed, 6 insertions(+), 0 deletions(-) diff --git a/drivers/macintosh/mac_hid.c ...

Since 43cc71eed1250755986da4c0f9898f9a635cb3bf, the platform modalias is prefixed with "platform:". Add MODULE_ALIAS() to the hotpluggable "input" platform drivers, to re-enable auto loading. [dbrownell@users.sourceforge.net: more drivers, registration fixes] Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Signed-off-by: Dmitry Torokhov <dtor@mail.ru> --- drivers/input/keyboard/aaed2000_kbd.c | 4 ++++ ...

Hi everyone. I would like to inform that kernel.org administration kindly provided us space for the following projects: . reiserfsprogs . reiser4 patches against vanilla kernels . reiser4progs The mentioned projects can be found here: http://www.kernel.org/pub/linux/kernel/people/edward/ All updates will be announced at reiserfs-devel mailing list. Thanks, Edward. --

On Fri, 18 Apr 2008 19:22:45 -0400 (EDT) #irq-remove doesn't seem to be included in the #ALL branch which I'm grabbing? --

From: Jeff Garzik <jeff@garzik.org> Acked-by: David S. Miller <davem@davemloft.net> --

Prefer 'void *dev_id' argument for passing data from request_irq() to each per-irq handler invocation. This change's main purpose is to prepare for the patchset in jgarzik/misc-2.6.git#irq-remove, that explores removal of the never-used 'irq' argument in each interrupt handler. Signed-off-by: Jeff Garzik <jgarzik@redhat.com> --- arch/x86/kernel/vm86_32.c | 7 ++++--- 1 files changed, 4 insertions(+), 3 deletions(-) diff --git a/arch/x86/kernel/vm86_32.c ...

What do you mean? I know at least one of two interrupt handlers in-tree that use their 'irq' arguments. --

Avoid confusion by /not/ passing an unused pointer to arm_rtc_interrupt() This change's main purpose is to prepare for the patchset in jgarzik/misc-2.6.git#irq-remove, that explores removal of the never-used 'irq' argument in each interrupt handler. Signed-off-by: Jeff Garzik <jgarzik@redhat.com> --- arch/arm/mach-integrator/time.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/arch/arm/mach-integrator/time.c b/arch/arm/mach-integrator/time.c index ...

- remove unused 'irq' argument from pfm_do_interrupt_handler() - remove pointless cast to void* - add KERN_xxx prefix to printk() - remove braces around singleton C statement - in tioce_provider.c, start tioce_dma_consistent() and tioce_error_intr_handler() function declarations in column 0 This change's main purpose is to prepare for the patchset in jgarzik/misc-2.6.git#irq-remove, that explores removal of the never-used 'irq' argument in each interrupt handler. Signed-off-by: ...

- remove pointless casts from void* - remove needless references to 'irq' function argument, when that information is already stored somewhere in a driver-private struct. - where the 'irq' function argument is known never to be used, rename it to 'dummy' to make this more obvious - remove always-false tests for dev_id==NULL - remove always-true tests for 'irq == host_struct->irq' - replace per-irq lookup functions and tables with a direct reference to data object obtained via ...

- remove always-true test - neaten request_irq() indentation This change's main purpose is to prepare for the patchset in jgarzik/misc-2.6.git#irq-remove, that explores removal of the never-used 'irq' argument in each interrupt handler. Signed-off-by: Jeff Garzik <jgarzik@redhat.com> --- arch/mips/pmc-sierra/msp71xx/msp_hwbutton.c | 5 +---- 1 files changed, 1 insertions(+), 4 deletions(-) diff --git a/arch/mips/pmc-sierra/msp71xx/msp_hwbutton.c ...

* prefer passing data to irq handler using 'void *dev_id' argument * remove references to 'irq' function arg that either duplicate a member of our private struct, or are always [true|false]. * add linux/interrupt.h include where needed This change's main purpose is to prepare for the patchset in jgarzik/misc-2.6.git#irq-remove, that explores removal of the never-used 'irq' argument in each interrupt handler. Signed-off-by: Jeff Garzik <jgarzik@redhat.com> --- drivers/atm/ambassador.c ...

On Fri, 18 Apr 2008 18:53:55 -0400 oops. It seems that I disabled it quite a long time ago. I'll resurrect That merges cleanly against the small number of isdn patches which are pending in -mm, so we're OK from that POV. --

On Fri, 18 Apr 2008 17:42:54 GMT Oh. Well that rather invalidates my earlier comments. It looks like this change could have been folded, but I understand that this sometimes gets wearisome and isn't terribly important if a) the fix doesn't repair build breakage and b) the fix doesn't fix runtime breakage and c) the fix fixes code which the git-bisect user won't have enabled in config anyway. Still. Do we need the set_fs() in there? __copy_from_user_inatomic() is a "__" ...

On Fri, 18 Apr 2008 17:05:06 GMT What is the reason for all the noinlines in this file? Something to do with getting proper wchan output? I guess it doesn't hurt from a documentation POV, but did you find that it was actually necessary? IOW: is gcc now capable of secretly inlining functions which are defined further ahead in the compilation unit? (did you actually "test" the wchan stuff, btw?) --

On Fri, 18 Apr 2008 19:00:57 GMT I expect that CONFIG_KGDB_TESTS_BOOT_STRING is always defined here. Or if not, that it is possible to arrange for that with sufficient Kconfig I dunno. Do we *really* need this feature? I mean, if you want to run the tests, you provide the boot parameter and that's it. Adding this Kconfig-time stuff on top seems excessive. --

On Fri, 18 Apr 2008 17:41:29 GMT Please just nuke all the interface comments in the header files. They duplicate the kernedoc comments at the definition site and we don't want to Well, these are trickier because there is an implementation of this function within each architecture. So I think that in this case it _does_ make sense to document the function in a common place, and the only common place is this header file. So please a) make this a kerneldoc comment and b) remove the ...

Hi, Well, I've tried to push for "generic" kmemcheck hooks so that they can be used by SLAB/SLOB too but Vegard has always talked me out of it. Of course, I have long forgotten the rationale. But anyway, slub_kmemcheck.h sounds okay and I wonder if we should put it in mm/ instead of include/linux/ as it's not supposed to be included by anyone except SLUB? And merged by Linus now. Pekka --

Thank you for your comments. We set it to zero to allow the compiler to optimize out code if CONFIG_KMEMCHECK=n. For instance if (kmemcheck_page_is_tracked(page) && !(s->flags & SLAB_NOTRACK)) { Will be turned into if(0 && !(s->flags & 0)), which can be completely optimized away. On the other hand, if SLAB_NOTRACK is non-zero, this The implementations of these prototypes are in mm/slub_kmemcheck.c. They are only ever called from slub.c since they represent the interface between ...

From e508eccb7a1dd8f4b0fd6056cdeff9ce93dad47c Mon Sep 17 00:00:00 2001 From: Vegard Nossum <vegard.nossum@gmail.com> Date: Fri, 4 Apr 2008 00:53:23 +0200 Subject: [PATCH] x86: add hooks for kmemcheck The hooks that we modify are: - Page fault handler (to handle kmemcheck faults) - Debug exception handler (to hide pages after single-stepping the instruction that caused the page fault) Also redefine memset() to use the optimized version if kmemcheck is enabled. Signed-off-by: Vegard ...

It's possible that the values used in and returned from jiffies_to_usecs() are incorrect because of truncation when variables of type u64 are involved. So a function specific to that type is used instead. This version implements a correction to jiffies_64_to_usecs() based on feedback from Randy Dunlap and Roman Zippel. Diff'd against: linux/kernel/git/torvalds/linux-2.6.git Signed-off-by: Jonathan Lim <jlim@sgi.com> --- a/include/linux/jiffies.h Thu Feb 14 18:04:14 PST 2008 +++ ...

On Fri, 18 Apr 2008 15:07:46 -0700 Well if there was a single CONFIG_MEMTEST then it would probably be sufficent to document it all within the Kconfig help for that option. But there isn't a CONFIG_MEMTEST. It appears that the presence or absence of the memtest feature is controlled by CONFIG_MEMTEST_BOOTPARAM which So... such users shouldn't add the memtest= boot parameter? I'd suggest the following: - Add a CONFIG_MEMTEST. No other config variable needed. - Put all the memtest ...

WARNING: at arch/x86/kernel/genapic_64.c:86 read_apic_id+0x78/0x80() Modules linked in: Pid: 1, comm: swapper Not tainted 2.6.25-2cca775baecbfede2fec20c99add709232311fe7 #3 Call Trace: [<ffffffff8023150f>] warn_on_slowpath+0x5f/0x80 [<ffffffff80299e9e>] ? open_namei+0x4e/0x6b0 [<ffffffff804594e5>] ? _spin_unlock_irqrestore+0x65/0x80 [<ffffffff80254105>] ? trace_hardirqs_on+0xd5/0x160 [<ffffffff804594c2>] ? _spin_unlock_irqrestore+0x42/0x80 [<ffffffff8021ed28>] read_apic_id+0x78/0x80 ...

On Fri, 18 Apr 2008 18:47:21 GMT What was the reason for making this change? --

Subject typo. Please fix so people don't think this is a networking patch :-) --

Lots of stuff in that file is pretty damn wrong, and at it should probably just be deleted. Care to put this new documentation into: If you do: struct class block_class = { .name = "block", .dev_kobj = block_kobj, } in: block/genhd.c and put: if (!class->dev_kobj) class->dev_kobj = char_kobj; in: class_register() we can get rid of all this logic and the BLOCK #ifdef, right? Also: class_create(THIS_MODULE, "usb_device"); in: ...

Rather than stepping through all IO resources, then stepping through all MMIO resources, etc. (and traversing the list every time), we can just iterate over the resource list once directly. This can change the order in /sys, e.g., # cat /sys/devices/pnp0/00:07/resources # OLD state = active io 0x3f8-0x3ff irq 4 # cat /sys/devices/pnp0/00:07/resources # NEW state = active irq 4 io 0x3f8-0x3ff The old code artificially sorted resources by type; ...

The contents of include/linux/pnpbios.h are used only inside the PNPBIOS backend, so this file doesn't need to be visible outside PNP. This patch moves the contents into an existing PNPBIOS-specific file, drivers/pnp/pnpbios/pnpbios.h. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> --- drivers/pnp/pnpbios/bioscalls.c | 1 drivers/pnp/pnpbios/core.c | 1 drivers/pnp/pnpbios/pnpbios.h | 136 ++++++++++++++++++++++++++++++++++++ drivers/pnp/pnpbios/proc.c | 2 ...

The interfaces for registering protocols, devices, cards, and resource options should only be used inside the PNP core. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Acked-By: Rene Herman <rene.herman@gmail.com> --- drivers/pnp/base.h | 24 +++++++++++++++++++++++- include/linux/pnp.h | 34 ---------------------------------- 2 files changed, 23 insertions(+), 35 deletions(-) Index: ...

Move warnings about _CRS and _PRS problems to the place where we actually make the ACPI calls. Then we don't have to pass around acpi_status values any more than necessary. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Index: work8/drivers/pnp/pnpacpi/pnpacpi.h =================================================================== --- work8.orig/drivers/pnp/pnpacpi/pnpacpi.h 2008-04-17 15:58:17.000000000 -0600 +++ work8/drivers/pnp/pnpacpi/pnpacpi.h 2008-04-17 15:58:22.000000000 -0600 @@ ...

PNP used to have a fixed-size pnp_resource_table for tracking the resources used by a device. This table often overflowed, so we've had to increase the table size, which wastes memory because most devices have very few resources. This patch replaces the table with a linked list of resources where the entries are allocated on demand. This removes messages like these: pnpacpi: exceeded the max number of IO resources 00:01: too many PORTs (max 40) References: ...

Given a struct resource, this returns the type (IO, MEM, IRQ, DMA). Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Index: work8/drivers/pnp/base.h =================================================================== --- work8.orig/drivers/pnp/base.h 2008-04-17 14:07:25.000000000 -0600 +++ work8/drivers/pnp/base.h 2008-04-17 14:07:37.000000000 -0600 @@ -18,6 +18,7 @@ int pnp_check_dma(struct pnp_dev * dev, int idx); void pnp_init_resource(struct resource *res); +int ...

Add a pnp_add_mem_resource() that can be used by all the PNP backends. This consolidates a little more pnp_resource_table knowledge into one place. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> --- drivers/pnp/base.h | 3 +++ drivers/pnp/resource.c | 26 ++++++++++++++++++++++++++ drivers/pnp/interface.c | 17 +++++++---------- drivers/pnp/isapnp/core.c | 9 +++------ drivers/pnp/pnpacpi/rsparser.c | 32 ++++++++------------------------ ...

This encapsulates the code to locate a new pnp_resource of the desired type. Currently this uses the pnp_resource_table, but it will soon change to find a resource in a linked list. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Index: work8/drivers/pnp/resource.c =================================================================== --- work8.orig/drivers/pnp/resource.c 2008-04-17 15:04:41.000000000 -0600 +++ work8/drivers/pnp/resource.c 2008-04-18 11:08:33.000000000 -0600 @@ -516,6 ...

This removes more direct references to pnp_resource_table. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Index: work8/drivers/pnp/manager.c =================================================================== --- work8.orig/drivers/pnp/manager.c 2008-04-10 16:50:03.000000000 -0600 +++ work8/drivers/pnp/manager.c 2008-04-10 16:59:08.000000000 -0600 @@ -232,17 +232,17 @@ */ void pnp_init_resources(struct pnp_dev *dev) { - struct pnp_resource_table *table = &dev->res; - int ...

Remove some PNP_MAX_* uses. The pnp_resource_table isn't dynamic yet, but with pnp_get_resource(), we can start moving away from the table size constants. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Acked-By: Rene Herman <rene.herman@gmail.com> --- drivers/pnp/interface.c | 43 ++++++++++++++++++------------------ drivers/pnp/quirks.c | 13 ++++++----- drivers/pnp/resource.c | 56 +++++++++++++++++++++++++++--------------------- drivers/pnp/system.c | 21 ...

This removes a few more references to the pnp_resource_table. No functional change. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> --- drivers/pnp/isapnp/core.c | 87 ++++++++++++++++++++-------------------- drivers/pnp/pnpacpi/rsparser.c | 88 ++++++++++++++++++++++------------------- drivers/pnp/pnpbios/rsparser.c | 82 +++++++++++++++++++++----------------- 3 files changed, 138 insertions(+), 119 deletions(-) Index: ...

There are no remaining references to the PNP_MAX_* constants or the pnp_resource_table structure outside of the PNP core. Make them private to the PNP core. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> --- include/linux/pnp.h | 14 ++------------ drivers/pnp/base.h | 12 ++++++++++++ drivers/pnp/core.c | 8 ++++++++ drivers/pnp/isapnp/core.c | 4 ++-- drivers/pnp/pnpacpi/rsparser.c | 10 ++++++---- drivers/pnp/pnpbios/rsparser.c ...

This removes more direct references to pnp_resource_table. Note that this does not allow adding new resources beyond the preallocated ones, so even though it does not reference pnp_resource_table directly, it is still limited to the fixed size of that table. When the preallocation is removed, we'll have to change this so it can add new resources. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Index: ...

Use dev_printk() when possible for more informative error messages. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Acked-By: Rene Herman <rene.herman@gmail.com> --- drivers/pnp/isapnp/core.c | 38 ++++++++++++++++++-------------------- drivers/pnp/pnpacpi/rsparser.c | 20 ++++++++++++-------- drivers/pnp/pnpbios/rsparser.c | 36 ++++++++++++++---------------------- 3 files changed, 44 insertions(+), 50 deletions(-) Index: ...

... it's not being set. Rene. --

Rene Herman <rene.herman@gmail.com> recently removed the only in-tree driver uses of: pnp_init_resource_table() pnp_manual_config_dev() pnp_resource_change() in this change: http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=109c53... These are no longer used in the PNP core either, so we can just remove them completely. It's possible that there are out-of-tree drivers that use these interfaces. They should ...

Add pnp_init_resources(struct pnp_dev *) to replace pnp_init_resource_table(), which takes a pointer to the pnp_resource_table itself. Passing only the pnp_dev * reduces the possibility for error in the caller and removes the pnp_resource_table implementation detail from the interface. Even though pnp_init_resource_table() is exported, I did not export pnp_init_resources() because it is used only by the PNP core. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Acked-By: Rene Herman ...

This changes pnp_clean_resource_table() to take a pnp_dev pointer rather than a pnp_resource_table pointer. This reduces the visibility of pnp_resource_table and removes an opportunity for error in the caller. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Acked-By: Rene Herman <rene.herman@gmail.com> Index: work7/drivers/pnp/manager.c =================================================================== --- work7.orig/drivers/pnp/manager.c 2008-03-25 11:03:59.000000000 -0600 +++ ...

A future change will change pnp_mem_flags() from a "#define that simplifies to an lvalue" to "an inline function that returns the flags value." Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Index: work7/drivers/pnp/quirks.c =================================================================== --- work7.orig/drivers/pnp/quirks.c 2008-04-02 13:08:09.000000000 -0600 +++ work7/drivers/pnp/quirks.c 2008-04-02 13:10:34.000000000 -0600 @@ -114,6 +114,7 @@ static void ...

Add pnp_alloc_card() to allocate a struct pnp_card and fill in the protocol, instance number, and initial PNP ID. Now it is always valid to use dev_printk() on any pnp_card pointer. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Acked-By: Rene Herman <rene.herman@gmail.com> --- drivers/pnp/base.h | 1 + drivers/pnp/card.c | 28 +++++++++++++++++++++++++--- drivers/pnp/isapnp/core.c | 11 ++++------- 3 files changed, 30 insertions(+), 10 deletions(-) Index: ...

pnpacpi_encode_ext_irq() should set resource->data.extended_irq, not resource->data.irq. This has been wrong since at least 2.6.12. I haven't seen any bug reports, but it's clearly incorrect. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Index: work8/drivers/pnp/pnpacpi/rsparser.c =================================================================== --- work8.orig/drivers/pnp/pnpacpi/rsparser.c 2008-04-10 12:26:08.000000000 -0600 +++ work8/drivers/pnp/pnpacpi/rsparser.c 2008-04-10 ...

This simplifies IRQ resource parsing slightly by computing all the IORESOURCE_IRQ_* flags at the same time. This also keeps track of shareability information when parsing options from _PRS. Previously we ignored shareability in _PRS. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Acked-By: Rene Herman <rene.herman@gmail.com> Index: work7/drivers/pnp/pnpacpi/rsparser.c =================================================================== --- ...

This moves some of the pnp_id knowledge out of the backends and into the PNP core. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Acked-By: Rene Herman <rene.herman@gmail.com> --- drivers/pnp/base.h | 2 +- drivers/pnp/driver.c | 28 +++++++++++++++++++++------- drivers/pnp/isapnp/core.c | 24 +++++++++++------------- drivers/pnp/pnpacpi/core.c | 27 +++------------------------ drivers/pnp/pnpbios/core.c | 10 +--------- ...

pnp_add_id() doesn't need to be exposed outside the PNP core, so move the declaration to an internal header file. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Acked-By: Rene Herman <rene.herman@gmail.com> Index: work7/include/linux/pnp.h =================================================================== --- work7.orig/include/linux/pnp.h 2008-03-21 13:31:06.000000000 -0600 +++ work7/include/linux/pnp.h 2008-03-21 13:31:08.000000000 -0600 @@ -403,7 +403,6 @@ /* protocol helpers */ ...

Add pnp_alloc_dev() to allocate a struct pnp_dev and fill in the protocol, instance number, and initial PNP ID. Now it is always valid to use dev_printk() on any pnp_dev pointer. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Acked-By: Rene Herman <rene.herman@gmail.com> --- drivers/pnp/base.h | 1 + drivers/pnp/core.c | 28 +++++++++++++++++++++++++--- drivers/pnp/isapnp/core.c | 11 +++++------ drivers/pnp/pnpacpi/core.c | 19 +++---------------- ...

pnp_add_card_id() doesn't need to be exposed outside the PNP core, so move the declaration to an internal header file. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Acked-By: Rene Herman <rene.herman@gmail.com> Index: work7/drivers/pnp/base.h =================================================================== --- work7.orig/drivers/pnp/base.h 2008-04-02 15:17:49.000000000 -0600 +++ work7/drivers/pnp/base.h 2008-04-02 15:19:07.000000000 -0600 @@ -4,6 +4,7 @@ void ...

These are used only in drivers/pnp/isapnp/core.c, so no need to expose them to the world. Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Acked-By: Rene Herman <rene.herman@gmail.com> Index: work7/drivers/pnp/isapnp/core.c =================================================================== --- work7.orig/drivers/pnp/isapnp/core.c 2008-03-21 11:52:12.000000000 -0600 +++ work7/drivers/pnp/isapnp/core.c 2008-03-21 11:53:24.000000000 -0600 @@ -88,6 +88,14 @@ #define _LTAG_MEM32RANGE 0x85 ...

https://bugzilla.redhat.com/show_bug.cgi?id=441622 ... on a Thinkpad X41T when docking a running laptop. ====================================================== [ INFO: hard-safe -> hard-unsafe lock order detected ] 2.6.25-0.195.rc8.git1.fc9.i686 #1 ------------------------------------------------------ kacpi_notify/64 [HC0[0]:SC0[0]:HE0:SE1] is trying to acquire: (sequence_lock){--..}, at: [kobject_uevent_env+379/867] kobject_uevent_env+0x17b/0x363 and this task is already holding: ...

On Fri, 18 Apr 2008 16:05:42 -0400 tpyo --

Currently ftrace allocates a trace buffer for every possible CPU. Work is being done to change it to only online CPUs and add hooks to hotplug CPUS. This patch lays out the infrastructer for such a change. Signed-off-by: Steven Rostedt <srostedt@redhat.com> --- kernel/trace/trace.c | 38 ++++++++++++++++++++++++-------------- 1 file changed, 24 insertions(+), 14 deletions(-) Index: ...

Andrew Morton suggested using strict_strtoul over simple_strtoul. This patch replaces them in ftrace. Signed-off-by: Steven Rostedt <srostedt@redhat.com> --- kernel/trace/trace.c | 28 ++++++++++++++++++++++------ 1 file changed, 22 insertions(+), 6 deletions(-) Index: linux-sched-devel.git/kernel/trace/trace.c =================================================================== --- linux-sched-devel.git.orig/kernel/trace/trace.c 2008-04-18 15:47:40.000000000 -0400 +++ ...

Andrew Morton mentioned some clean ups that should be done to ftrace. This patch does some of the simple clean ups. Signed-off-by: Steven Rostedt <srostedt@redhat.com> --- kernel/trace/trace.c | 25 ++++++++++++------------- 1 file changed, 12 insertions(+), 13 deletions(-) Index: linux-sched-devel.git/kernel/trace/trace.c =================================================================== --- linux-sched-devel.git.orig/kernel/trace/trace.c 2008-04-18 15:47:22.000000000 -0400 +++ ...

I don't think this was a good idea: commit 5b0e508415a83989fe704b4718a1a214bc333ca7 Author: Jan Beulich <jbeulich@novell.com> Date: Mon Mar 10 13:11:17 2008 +0000 x86: prevent unconditional writes to DebugCtl MSR It's already a bug if there is any unconditional use of the MSR. Silenting ignoring it is just wrong. There was such a bug before this fix: commit 4ba51fd75cc3789be83f0d6f878dabbb0cb19bca Author: Roland McGrath <roland@redhat.com> Date: Thu Apr 3 14:18:55 2008 ...

Shadowed variable and integer as NULL pointer fixes: drivers/misc/kgdbts.c:877:6: warning: symbol 'sys_open_test' shadows an earlier one drivers/misc/kgdbts.c:537:27: originally declared here drivers/misc/kgdbts.c:378:22: warning: Using plain integer as NULL pointer drivers/misc/kgdbts.c:386:22: warning: Using plain integer as NULL pointer drivers/misc/kgdbts.c:468:30: warning: Using plain integer as NULL pointer drivers/misc/kgdbts.c:472:15: warning: Using plain integer as NULL ...

From: Adrian Bunk <bunk@kernel.org> dlm_print_rsb() can now become static. Signed-off-by: Adrian Bunk <bunk@kernel.org> Signed-off-by: David Teigland <teigland@redhat.com> --- fs/dlm/lock.c | 2 +- fs/dlm/lock.h | 1 - 2 files changed, 1 insertions(+), 2 deletions(-) diff --git a/fs/dlm/lock.c b/fs/dlm/lock.c index 8f250ac..1e9e8eb 100644 --- a/fs/dlm/lock.c +++ b/fs/dlm/lock.c @@ -165,7 +165,7 @@ void dlm_print_lkb(struct dlm_lkb *lkb) lkb->lkb_grmode, ...

Move the code that handles cluster posix locks from gfs2 into the dlm so that it can be used by both gfs2 and ocfs2. Signed-off-by: David Teigland <teigland@redhat.com> --- fs/dlm/Makefile | 1 + fs/dlm/dlm_internal.h | 2 + fs/dlm/main.c | 7 + fs/dlm/plock.c | 439 ++++++++++++++++++++++++++++++++++++++++ fs/gfs2/locking/dlm/Makefile | 2 +- fs/gfs2/locking/dlm/lock_dlm.h | 11 +- fs/gfs2/locking/dlm/main.c | ...

When a NOQUEUE request fails, the rsb res_master field is unnecessarily reset to -1, instead of leaving the valid master setting in place. We want to save the looked-up master values while the rsb is on the "toss list" so that another lookup can be avoided if the rsb is soon reused. The fix is to simply leave res_master value alone. Signed-off-by: David Teigland <teigland@redhat.com> --- fs/dlm/lock.c | 3 +-- 1 files changed, 1 insertions(+), 2 deletions(-) diff --git a/fs/dlm/lock.c ...

If a node is removed from a lockspace, and then added back before the dlm is notified of the removal, the dlm will not detect the removal and won't clear the old state from the node. This is fixed by using a list of added nodes so the membership recovery can detect when a newly added node is already in the member list. Signed-off-by: David Teigland <teigland@redhat.com> --- fs/dlm/config.c | 48 +++++++++++++++++++++++++++++++++++++++--------- fs/dlm/config.h | 3 ++- ...

Linus, please pull from the for-linus branch at git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394-2.6.git for-linus to receive the following IEEE 1394/ FireWire subsystem updates. In the queue, as reported on April 2: - raw1394, video1394, dv1394: [improvement] These drivers are no longer bound to unit directory devices. This has always been unnecessary and got in the way of more specialized out-of-tree drivers. - firewire-core: [fix] Recognize changes ...

Hi, Please consider pulling the following patches. The tree is almost unchanged from the patch posting with the exception of the addition of Roel's patch to fix a calculation in the logging code. Once these patches go in, I'll look into creating a follow-up patch as suggested by Christoph. There are no changes outside of the GFS2 code in this patch set, Steve. The following changes since commit 4a72ef9f7ad36ea0fb3b42f31b2b0c5b9871969a: Linus Torvalds (1): Merge branch ...

TF_MASK is no longer defined, use X86_EFLAGS_TF. Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com> --- In current -linus tree. arch/x86/kernel/kgdb.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/x86/kernel/kgdb.c b/arch/x86/kernel/kgdb.c index 8c7e555..24362ec 100644 --- a/arch/x86/kernel/kgdb.c +++ b/arch/x86/kernel/kgdb.c @@ -375,12 +375,12 @@ int kgdb_arch_handle_exception(int e_vector, int signo, int err_code, newPC = linux_regs->ip; ...

Acked-by: Ingo Molnar <mingo@elte.hu> Ingo --

Add Timur TAbi as the maintainer for the Freescale QE library, the Freescale QE UART device driver, the Freescale SOC sound drivers, and the Crystal Semiconductor CS4270 device driver. Signed-off-by: Timur Tabi <timur@freescale.com> --- This patch is for 2.6.26. MAINTAINERS | 25 +++++++++++++++++++++++++ 1 files changed, 25 insertions(+), 0 deletions(-) diff --git a/MAINTAINERS b/MAINTAINERS index 90dcbbc..7a8bc0a 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -1104,6 +1104,12 @@ ...

Hey Rusty, [...] Please CC me on API changes... Cheers, Michael --

Andi, I wrote the following description of the core_pattern pipe feature. Does this seem okay? Piping core dumps to a program Since kernel 2.6.19, Linux supports an alternate syntax for the /proc/sys/kernel/core_pattern file. If the first character of this file is a pipe symbol (|), then the remainder of the line is interpreted as a program to be executed. Instead of being written to a disk file, the core dump is given as ...

Amen on that. And to add confusion here, my makeit script started dying when I do a remake after changing a config option, cuz it can't find a System.map file to delete, even if I copy it in from /boot. I had to delete the make -- Cheers, Gene "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) Check it out, send me comments, and dance joyously in the streets, -- Linus Torvalds announcing ...

[Cyrill Gorcunov - Fri, Apr 18, 2008 at 11:38:02PM +0400] | [Linus Torvalds - Fri, Apr 18, 2008 at 08:43:35AM -0700] | | | | | | On Fri, 18 Apr 2008, Matthew wrote: | | > | | > I can "reproduce" this on an Intel Core 2 Duo 6600 (Conroe) with a P5W | | > DH Deluxe mainboard (by Asus) since at least 2.6.25-rc8 (or possibly | | > also rc7) | | | | Can you bisect it? | | | | Linus | | it seems drivers/acpi/thermal.c has a small nit | | 881: /* sys I/F for generic thermal sysfs ...

Hi everyone, hi Linus, congratulations on this new great kernel-release :) I've another "regression" to report for 2.6.25: it's concerning much higher temperatures being read out by the "coretemp" kernel-module in comparison to 2.6.24* series e.g. where temperatures were around 40-47°C they are now constantly jumping around 55-70°C (even in idle !) several other users/testers have reported this issue too (both on zen-sources [heavy patched] & latest gentoo-sources) ...

And then, after killing the make clean statement, it still dies when making a -- Cheers, Gene "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) Check it out, send me comments, and dance joyously in the streets, -- Linus Torvalds announcing 2.0.27 --

The coretemp kernel module reports 25°C on my PC when idle, and 34°C after having performed some computations (lmbench2). This looks normal. This test has been performed with a vanilla 2.6.25 kernel and a Core 2 Duo E6750 CPU (Asus motherboard). The 2.6.22 and 2.6.24 kernels report an incorrect temperature on the same system however (10°C). So there is either an issue with the patches that have been applied to your kernel or the behavior of the coretemp module for the 6600 and E6750 CPU's is ...

Add modules.order to the list of files that shoud be ignored when using diff on a built kernel tree. Signed-off-by: Ben Dooks <ben-linux@fluff.org> diff -rup linux-2.6.25/Documentation/dontdiff linux-2.6.25-dontdiff/Documentation/dontdiff --- linux-2.6.25/Documentation/dontdiff 2008-04-17 09:48:05.000000000 +0100 +++ linux-2.6.25-dontdiff/Documentation/dontdiff 2008-04-18 16:16:45.000000000 +0100 @@ -142,6 +142,7 @@ mkprep mktables mktree modpost +modules.order modversions.h* ...

Fix build breakage introduced in commit "[POWERPC] i2c: OF helpers for the i2c API". If i2c-core is compiled as module, the helper needs to be compiled as module, as well. Rename i2c.c to of_i2c.c to avoid name space conflict. Signed-off-by: Jochen Friedrich <jochen@scram.de> --- drivers/of/Kconfig | 2 +- drivers/of/Makefile | 2 +- drivers/of/{i2c.c => of_i2c.c} | 0 3 files changed, 2 insertions(+), 2 deletions(-) rename drivers/of/{i2c.c => of_i2c.c} ...

Subject: [PATCH 2/4] jbd: ordered data integrity fix In ordered mode, if a buffer being dirtied exists in the committing transaction, we write the buffer to the disk, move it from the committing transaction to the running transaction, then dirty it. But we don't have to remove the buffer from the committing transaction when the buffer couldn't be written out, otherwise it breaks the ordered mode rule. Signed-off-by: Hidehiro Kawai <hidehiro.kawai.ez@hitachi.com> --- fs/jbd/transaction.c | ...

Subject: [PATCH 3/4] jbd: abort when failed to log metadata buffers If we failed to write metadata buffers to the journal space and succeeded to write the commit record, stale data can be written back to the filesystem as metadata in the recovery phase. To avoid this, when we failed to write out metadata buffers, abort the journal before writing the commit record. Signed-off-by: Hidehiro Kawai <hidehiro.kawai.ez@hitachi.com> --- fs/jbd/commit.c | 3 +++ 1 file changed, 3 ...

Subject: [PATCH 4/4] jbd/ext3: fix error handling for checkpoint io When a checkpointing IO fails, current JBD code doesn't check the error and continue journaling. This means latest metadata can be lost from both the journal and filesystem. This patch leaves the failed metadata blocks in the journal space and aborts journaling in the case of log_do_checkpoint(). To achieve this, we need to do: 1. don't remove the failed buffer from the checkpoint list where in the case of ...

Subject: [PATCH 1/4] jbd: strictly check for write errors on data buffers In ordered mode, we should abort journaling when an I/O error has occurred on a file data buffer in the committing transaction. But there can be data buffers which are not checked for error: (a) the buffer which has already been written out by pdflush (b) the buffer which has been unlocked before scanned in the t_locked_list loop This patch adds missing error checks and aborts ...

Hi, this was discussed before, but I thought the problem was solved. Apparently not... After upgrading to 2.6.25, jiffies on my 2xdual-core IBM x3650 boxes seem to jump back and forth as can be seen on the dmesg output. Did not happen in 2.6.24. As I did not follow the rc-cycles for 2.6.25, I cannot really say when it started. Config attached. Cheers Martin ------------------------------------------------------ Martin Knoblauch email: k n o b i AT knobisoft DOT de www: ...

This patch implements various helpers to support OF bindings for the i2c API. Signed-off-by: Jochen Friedrich <jochen@scram.de> Acked-by: David S. Miller <davem@davemloft.net> Acked-by: Stephen Rothwell <sfr@canb.auug.org.au> --- drivers/of/Kconfig | 6 +++ drivers/of/Makefile | 1 + drivers/of/of_i2c.c | 115 ++++++++++++++++++++++++++++++++++++++++++++++++ include/linux/of_i2c.h | 24 ++++++++++ 4 files changed, 146 insertions(+), 0 deletions(-) create mode 100644 ...

CONGRATULATIONS ...YOUR E-MAIL WON... SELECTED E-MAIL WON YOU CONTACT MR PAUL GREEN FOR MORE INFORMATION FOR THE CLAIM OF YOUR $950,000.00 AS SOON AS POSSIBLE VIA EMAIL (APEXTRUSTAGENCY@AIM.COM) OR CALL +34628652243 --

we have hex_asc() in include/linux/kernel.h already. --

Hello Adrian, I noticed that blk_get_queue and blk_put_queue were unexported in the 2.6.25 kernel. Are you aware that this breaks DRBD (an out-of-tree kernel module) ? Bart. --

Problem: I run car navigation software on linux, using a gps. Getting data from the gps unit in a timely fashion is important, as the software says thing like "turn left in 50m". That is useless however, if the message arrives 10s after you passed the intersection! The gps unit reports location once per second. I frequently see delays of up to 13 seconds. I drive down a street and make a turn or stop. Then I look at map display, and count 13 updates before the map reflects what I did. This ...

It boots up fine with that patch below and: tornado boot # grep STACKPROTECT /boot/config-2.6.25-mm1-wip CONFIG_CC_STACKPROTECTOR_ALL=y CONFIG_CC_STACKPROTECTOR=y In fact I'm running with it applied right now and it all seems good so far, so I guess that's confirmation that it is just the test itself which is problematic? Reuben --

Ah, more fun caused by Dan's /sys/dev/... patch. Dan, this is causing a lot of problems, I'm going to drop it for now until the build and run-time errors get resolved. thanks, greg k-h --

At least the following files aren't removed by "make mrproper": Module.markers arch/x86/kernel/acpi/realmode/wakeup.lds crypto/.tmp_aes_generic.ver fs/.tmp_buffer.ver kernel/time/.tmp_timekeeping.ver Noticed by "git-ls-files -o". *.ver , I think, remain after abruptdly terminated build. --

I may have reported this same one previously against an earlier -mm, or somebody did, or something... ;) x86_64 kernel, Core2 Duo T7200, Dell Latitude D820 laptop... [ 0.060388] ACPI: Core revision 20080321 [ 0.070079] ------------[ cut here ]------------ [ 0.070082] WARNING: at arch/x86/kernel/genapic_64.c:86 read_apic_id+0x41/0x7c() [ 0.070085] Modules linked in: [ 0.070089] Pid: 1, comm: swapper Not tainted 2.6.25-mm1 #1 [ 0.070091] [ 0.070092] Call Trace: [ ...

The GCC stackprotector option is a no-go for me, and causes 100% repeatable fatal oopses on boot with my x86_64 box. This is not new to 2.6.25-mm1 - but was also present in 2.6.24-rc8-mm2 (2.6.24-rc8-mm1 was good, but this option didn't exist then). It seems that enabling the stackprotector option: tornado boot # diff -u config-2.6.25-mm1 config-2.6.25-mm1.old --- config-2.6.25-mm1 2008-04-18 22:40:15.000000000 +1000 +++ config-2.6.25-mm1.old 2008-04-18 20:09:38.000000000 ...

I think so. Well, put it this way... I haven't made any changes to it, this is the standard/current gcc that has been in Gentoo Portage for the last while. > Can you do a (with the test enabled in the config/code) do a > make kernel/panic.s > and send me that file? (this allows me to see what code your gcc generated for the test) Done - posted up in the web directory along with the other files (saves the possible grief of MUA mangling). I'm about to reboot to try Ingo's test ...

hm, does it boot up fine with the attached patch and stackprotector enabled? It appears that your system got to the self-test so stackprotector is working mostly - it's just that the self-test went wrong. Ingo -----------------> Subject: x86: disable stackprotector selftest From: Ingo Molnar <mingo@elte.hu> Date: Fri Apr 18 15:21:49 CEST 2008 Signed-off-by: Ingo Molnar <mingo@elte.hu> --- kernel/panic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) Index: ...

While debugging something on parisc a few weeks ago, I noticed kallsyms was all busted because print_symbol was printing the descriptor address. Since we don't use addr[0] for the function address, and we don't want a mess of #ifdef ARCH1 #elseif ARCH2 garbage, move the definitions to <asm/processor.h> Signed-off-by: Kyle McMartin <kyle@mcmartin.ca> --- include/asm-ia64/processor.h | 6 ++++++ include/asm-parisc/processor.h | 8 ++++++++ include/asm-powerpc/processor.h | 8 ...

ack on all three. Please fold #3 into #1 and add both to git-powerpc. --

ah, I misread it. I thought this was changing the __attribute__((weak)) which patch #1 added. But it's changing a different one. I though that whatever ;) --

Well, I copied over from something else, looks like there's a -lot- of 0 users here... Shouldn't there be some kind of SLAB_DEFAULT that contain those things ? Ben. --

The powerpc kernel stacks need to be naturally aligned, as they contain the thread info at the bottom, which is obtained by clearing the low bits of the stack pointer. However, when using 64K pages (the stack is smaller than a page), we use kmalloc to allocate it, which doesn't provide that guarantee. It appeared to work so far... until one enables SLUB debugging which then returns unaligned pointers. Ooops... This patch fixes it by using a slab cache with enforced alignment for those. It ...

[PATCH 04/04] This patch makes use of the target ids specified by a previous write to /proc/self/task/<tid>/next_id as the ids to use to allocate the next upid nrs. Upper levels upid nrs that are not specified in next_pids file are left to the kernel choice. Signed-off-by: Nadia Derbey <Nadia.Derbey@bull.net> --- include/linux/pid.h | 2 kernel/fork.c | 3 - kernel/pid.c | 141 +++++++++++++++++++++++++++++++++++++++++++++------- 3 files changed, 126 insertions(+), ...

[PATCH 02/04] This patch proposes the procfs facilities needed to feed the id(s) for the next task to be forked. say n is the number of pids to be provided through procfs: if an echo "LONG<n> X0 X1 ... X<n-1>" > /proc/self/task/<tid>/next_id is issued, the next task to be forked will have its upid nrs set as follows (say it is forked in a pid ns of level L): level upid nr L ----------> X0 .. L - i ------> Xi .. L - n + 1 --> X<n-1> Then, for levels L-n down to level 0, ...

I still think, in its current form, this is pretty dangerous. It might be nice to have a central place to set ids, but there's no real safety here for when we get a mishmash of 50 of these things. We have a queue and we fix the order in which things are inserted, but how do we control the order in which the kernel extracts them? Can we 100% conform to that order in the future and be consistent? Won't that also be dictated by the way userspace behaves? If we're going to go this route, I ...

Please pull from the for-linus branch: git pull git://oss.sgi.com:8090/xfs/xfs-2.6.git for-linus This will update the following files: Documentation/filesystems/xfs.txt | 15 +- fs/xfs/Kconfig | 12 - fs/xfs/linux-2.6/kmem.c | 6 +- fs/xfs/linux-2.6/xfs_aops.c | 12 +- fs/xfs/linux-2.6/xfs_buf.c | 8 +- fs/xfs/linux-2.6/xfs_buf.h | 8 +- fs/xfs/linux-2.6/xfs_cred.h | 2 +- fs/xfs/linux-2.6/xfs_export.c | 14 ...

What is the maximum numbet of pages which an unpriviliged user can This loop is a fancy way of doing --

This patch modifies tun to allow a vringfd to specify the send buffer. The user does a write to push out packets from the buffer. Again we use the 'struct virtio_net_hdr' to allow userspace to send GSO packets. In this case, it can hint how much to copy, and the other pages will be made into skb fragments. Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> --- drivers/net/tun.c | 410 +++++++++++++++++++++++++++++++++++++++++-------- include/linux/if_tun.h | 1 2 files ...

It turns out the lguest (and possibly kvm) want the addresses in the ring buffer to only cover a certain part of memory, and be offset. It makes sense that this be an ioctl. Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> --- Documentation/ioctl-number.txt | 3 +-- drivers/char/vring.c | 37 +++++++++++++++++++++++++++++++++++-- include/linux/vring.h | 9 +++++++++ 3 files changed, 45 insertions(+), 4 deletions(-) diff -r caeeb48d478a ...

Well, it helps if he's CCed.... I'm happy to work *with someone* on the documentation (pointless to do it on my own -- how do I know what Rusty's *intended* behavior for the interface is), and review, and testing. --

Generally, the other end of the virtio ring doesn't need to see where you're up to in consuming the ring. However, in order for an external entity to understand it, it must be exposed. For example, if you want to save and restore a virtio_ring, but you're not the consumer because the kernel is using it directly. Fortunately, we have room to expand: the ring is always a whole number of pages and there's hundreds of bytes of padding after the avail ring and the used ring, whatever the number of ...

We generally try to avoid it - it allows users to dos the box. Although I suspect that direct-io presently permits users to transiently pin an amount of memory which is proportional to the number of disks upon which they can acked-by: me. --

virtio introduced a ring structure ABI for guest-host communications (currently used by lguest and kvm). Using this same ABI, we can create a nice fd version. This is useful for efficiently passing packets to and from the tun, for example. Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> --- drivers/char/Kconfig | 9 + drivers/char/Makefile | 2 drivers/char/vring.c | 400 ++++++++++++++++++++++++++++++++++++++++++++++++++ include/linux/vring.h | 58 +++++++ 4 files ...

This patch modifies tun to allow a vringfd to specify the receive buffer. Because we can't copy to userspace in bh context, we queue like normal then use the "pull" hook to actually do the copy. We use struct virtio_net_hdr prepended to packets in the ring to allow userspace to receive GSO packets in future (at the moment, the tun driver doesn't tell the stack it can handle them, so these cases are never taken). This will need to be something that userspace tells us it can ...

page->page_cgroup is unsigned long. Should be cleared by 0UL. Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Index: linux-2.6.25/include/linux/memcontrol.h =================================================================== --- linux-2.6.25.orig/include/linux/memcontrol.h +++ linux-2.6.25/include/linux/memcontrol.h @@ -30,7 +30,7 @@ struct mm_struct; extern void mm_init_cgroup(struct mm_struct *mm, struct task_struct *p); extern void mm_free_cgroup(struct mm_struct ...

That's pretty cosmetic. In fact it could be argued that we shouldn't do this. "0" is zero regardless of the type of the LHS. So if we later change page_cgroup to u32 or u16 or `struct superblock *', the old page_reset_bad_cgroup() will happily continue to work. There's no need to make this code aware of page_cgroup's type, to this extent? --

It's CONFIG_DISCONTIGMEM_MANUAL. BTW, I attached the .config file. pls check it. Thanks, -Shi

On Fri, 18 Apr 2008 13:43:34 +0800 Hmmm....ok. I'll check DISCONTIGMEM's boot codes. Thanks, -Kame --

On Fri, 18 Apr 2008 12:32:56 +0900 I reproduced the hang with his config and confirmed his fix works well. But I can't find why...I'll dig a bit more. Thanks, -Kame --

hm, fishy. Perhaps the architecture isn't zeroing the memmap arrays? Or perhaps that page was used and then later freed before we got to memmap_init_zone() and was freed with a non-zero ->page_cgroup. Which is unlikely given that page.page_cgroup was only just added and is only present if CONFIG_CGROUP_MEM_RES_CTLR. --

.config is shared in previous mails. kexec/kdump has not been used. Thanks -Shi --

The comment from Hugh is correct, which implies that in this case page->cgroup The mem_map array should be cleared. I need to see the code to check where the Please share your .config? Is this a kexec/kdump reboot by any chance? -- Warm Regards, Balbir Singh Linux Technology Center IBM, ISTL --

But if ((pfn & (pageblock_nr_pages -1))) is not correct. for calling set_pageblock_migrationtype() once in a pageblock, ok, I'll test today if I have a chance. At least, I think I can test this My concern here is, memmap out-of-zone is not initialized and not marked as PG_reserved...sholdn't we initialize existing memmap even if they are out-ot-zone ? I think all existing mem_map for memory hole should be initialized properly. Thanks, -Kame --

Thanks! This resolves the warnings here. Now I'll take a shot at Regards, Dan --

The deprecated config option CONFIG_USB_DEVICE_CLASS causes class devices with duplicate major:minor numbers to be registered. In effect they represent a usb specific address space for major:minor numbers so add 'usb' as a directory along side 'block' and 'char'. Cc: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Dan Williams <dan.j.williams@intel.com> --- This boots up without warning on my dev system. We end up with the following situation: $ ls -l /sys/dev/usb/189:0 lrwxrwxrwx 1 ...

On Thu, Apr 17, 2008 at 8:08 PM, Andrew Morton I'll take a look. And Greg shares my suspicion that there might be other duplicate devices like these USB class devices, so need to look Thanks, Dan --

Hm, no they do not, they are not a new address space, we are just reusing them as the other user wasn't using them, and the code is deprecated and will be removed eventually. Neither of these char devices are really hooked up to anything within the kernel, so it doesn't matter yet. I wonder how many other duplicates we have floating around... thanks, greg k-h --

This breaks out the ptrace handling from get_signal_to_deliver into a new subroutine. The actual code there doesn't change, and it gets inlined into nearly identical compiled code. This makes the function substantially shorter and thus easier to read, and it nicely isolates the ptrace magic. Signed-off-by: Roland McGrath <roland@redhat.com> --- kernel/signal.c | 71 ++++++++++++++++++++++++++++++++---------------------- 1 files changed, 42 insertions(+), 29 deletions(-) diff --git ...

Implements an alternative iret with popf and return so trap and exception handlers can return to the NMI handler without issuing iret. iret would cause NMIs to be reenabled prematurely. x86_32 uses popf and far return. x86_64 has to copy the return instruction pointer to the top of the previous stack, issue a popf, loads the previous esp and issue a near return (ret). It allows placing immediate values (and therefore optimized trace_marks) in NMI code since returning from a breakpoint would be ...

echo "1" >/proc/sys/net/core/xfrm_larval_drop --

The patch is developed against 2.6.25-rc8 + some git patches, but I have forward-ported it to current git. I *guess* you are talking about commit aad8a278 (USB: support more Huawei data card product IDs), as I could not find . The header of this patch talks about three changes (declaration of unusual devices, disabling product ID matching in driver.c and finally declaring the product in option.c). I find only the first and the last point if this reflected in the patch, and there are *no* ...

What is this patch against? We have a change that went into 2.6.25 after 2.6.25-rc9 that should make this patch not be needed. Can you try that release and verify this? thanks, greg k-h --

Unless you have a qemu with the NMI patches, kgdb does not work on SMP with qemu. The very first test is going to fail because the IPI sent by the kernel is not handled in qemu's hardware emulation. Jason. --

On Thu, 17 Apr 2008 19:55:46 -0400 I dropped git-selinux and that crash seems to have gone away. It took about five minutes before, but would presumably have happened earlier if I'd reduced the cache size. btw, wouldn't this --- a/security/selinux/netnode.c~a +++ a/security/selinux/netnode.c @@ -190,7 +190,7 @@ static int sel_netnode_insert(struct sel if (sel_netnode_hash[idx].size == SEL_NETNODE_HASH_BKT_LIMIT) { struct sel_netnode *tail; tail = list_entry(node->list.prev, ...

cool! Jason: i think that particular self-test should be repeated 1000 times before reporting success ;-) Ingo --

Not really. In the selinux case we don't know that it'll break at compile Yeah, #error would work too. --

Oops, no, and that makes sense. I now picked up qemu 0.9.1 and applied the three NMI/SMI patches by Jan Kiszka. So in qemu it seems to run fine now, except that I need to prod it sometimes (it gets stuck in cpu_clock() and I have to break/continue from gdb to make it proceed). Oh, there it made it to 1056, and gdb can't interrupt anymore. Hmm. This is probably not a very good But booting with nosmp on real hardware gets easily above 100,000 iterations of the loop (before I reboot), so it ...

hm, there's sel_netnode_free() in the stackframe - that's from security/selinux/netnode.c. Andrew, any recent changes in that area? Ingo --

ok I found a way that works for me: From: Arjan van de Ven <arjan@linux.intel.com> Subject: [PATCH] stackprotector: turn not having the right gcc into an #error If the user selects the stack-protector config option, but does not have a gcc that has the right bits enabled (for example because it isn't build with a glibc that supports TLS, as is common for cross-compilers, but also because it may be too old), then the runtime test fails right now. Andrew rightfully points out that this is a ...

I assume this was SMP? While I had not tried it yet, my guess would have been this did not happen on a UP kernel. If it does occur on a UP kernel it means the problem is squarely between the task scheduling after the exception is handled and the kgdb state logic for re-entering the debug state after a single step exception occurs. It seems reasonable to go for 1000 iterations of this particular test to declare success as pointed out by Ingo. Previous versions of kgdb handled some of the ...

Try rmmod usb / insmod usb around suspend to see if it is usb-specific, or if something went seriously wrong in core. Or you might just bisect it ;-). Pavel -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html --

Looks good to me, although before I fix this let me try and figure out why this code is causing the machine to puke all over itself. Priorities you know :) -- paul moore linux @ hp --

that came in via the UV-APIC patchset but the warning is entirely harmless. At that point we've got a single CPU running only so preemption of that code to another CPU is not possible. native_smp_prepare_cpus() should probably just disable preemption, that way we could remove all those ugly preempt disable-enable calls from the called functions - per the patch below. (not boot tested yet - might provoke atomic-scheduling warnings if i forgot about some schedule point in this rather ...

On Thu, 17 Apr 2008 16:40:34 -0700 Disabling CPA_DEBUG didn't help. It's still hanging. The final initcall is init_kgdbts() and disabling KGDB prevents the hang. --

In this case you do not have to disable kgdb, but just disable the kgdb test suite. Certainly I would be interested to know where it is failing as it would indicate that there is a regression that is caused by a change that occurred somewhere else in the kernel or a latent defect in kgdb was triggered. The kgdb test suite exercises a number of kernel fault systems as well as arch specific single stepping when it runs and when it fails it is likely worth it to track down which test failed and ...

On Fri, Apr 18, 2008 at 2:03 AM, Andrew Morton Andrew, you don't seem to have slab debugging enabled: # CONFIG_DEBUG_SLAB is not set And quite frankly, the oops looks unlikely to be a slab bug but rather a plain old slab corruption cause by the callers... Pekka --

Keep in mind that slab might have been corrupted by someone else much earlier but we didn't notice due to the lack of CONFIG_SLAB_DEBUG. --

oop, there's more: sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00 sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA firewire_core: created device fw0: GUID 00016c2000174bad, S400 PM: Device usb4 failed to restore: error -113 eth0: Link is Up 100 Mbps Full Duplex, Flow Control: RX/TX eth0: 10/100 speed: disabling TSO PM: Device usb5 failed to restore: error -113 PM: Device usb7 failed to restore: error -113 sd 0:0:0:0: [sda] Starting disk PM: Image ...

I've reverted the -mm only change to that file in git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/selinux-2.6.git#for-akpm commit f777964ad75cf4a119d911d12e81948d2402677f Author: James Morris <jmorris@namei.org> Date: Fri Apr 18 20:27:24 2008 +1000 Revert "SELinux: Made netnode cache adds faster" This reverts commit 6bf8f41d4efdf9d4eeb4f7df9c591e281f7da93e. Possible cause of slab corruption in -mm. -- James Morris <jmorris@namei.org> --

that should be the patch below. Ingo ------------> Subject: x86: disable preemption in native_smp_prepare_cpus From: Ingo Molnar <mingo@elte.hu> Date: Fri Apr 18 11:07:10 CEST 2008 Signed-off-by: Ingo Molnar <mingo@elte.hu> --- arch/x86/kernel/smpboot.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) Index: linux-x86.q/arch/x86/kernel/smpboot.c =================================================================== --- linux-x86.q.orig/arch/x86/kernel/smpboot.c +++ ...

On Fri, 18 Apr 2008 00:28:58 -0700 I'm totally fine with that, but I think I need Sam's help on making that happen the right way; this is going to need makefile fu L( Sam: Basically what I need is that if the scripts/gcc-x86_64-has-stack-protector.sh script fails, the build aborts with a message/#error that says that the compiler is not capable of supporting this feature. Right now the script is used like this: stackp := $(CONFIG_SHELL) ...

It gets stuck in kgdb_roundup_cpus(), verified by putting a printk() before and after this call (in kgdb_handle_exception()). Simple, but effective :-) Vegard -- "The animistic metaphor of the bug that maliciously sneaked in while the programmer was not looking is intellectually dishonest as it disguises that the error is the programmer's own creation." -- E. W. Dijkstra, EWD1036 --

It uses RCU-style updates and has been designed to be lockless from the ground up. Mathieu -- Mathieu Desnoyers Computer Engineering Ph.D. Student, Ecole Polytechnique de Montreal OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F BA06 3F25 A8FE 3BAE 9A68 --

At Thu, 17 Apr 2008 20:38:57 +0200, I applied it to my git tree. Thanks. --

Note that my little patch purely resolves the situation we were seeing on the client side. We were using Netapps for servers. We're actually using that small patch, against 2.6.22, in production right now and it has still solved our issue. -Aaron --

Yes, I think xtime is updated by the tick timer. system_timer->offset() may be updated by different mechanisms, depending No. As I said, the function usually has "gettimeoffset" in the name and it's specific to arm so it would be under arch/arm. For instance, on Xscale boards it's ixp2000_gettimeoffset() in arch/arm/mach-ixp2000/core.c. For more information you might want to look at the code in arch/arm or else talk to the arm developers on the linux-arm-kernel mailing ...

Won't this work as well? --- In ext4_mb_init_backend() 'i' is of type ext4_group_t. Since unsigned, i >= 0 is always true, so fix hot spins after err_freebuddy: and -meta: and prevent decrements when zero. Signed-off-by: Roel Kluin <12o3l@tiscali.nl> --- diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c index ef97f19..054cd33 100644 --- a/fs/ext4/mballoc.c +++ b/fs/ext4/mballoc.c @@ -2572,13 +2572,13 @@ static int ext4_mb_init_backend(struct super_block *sb) meta_group_info[j] = ...

Updated one. diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c index 28b5ada..572d809 100644 --- a/fs/ext4/mballoc.c +++ b/fs/ext4/mballoc.c @@ -2223,7 +2223,7 @@ static noinline void ext4_mb_store_history(struct ext4_allocation_context *ac) static int ext4_mb_init_backend(struct super_block *sb) { - ext4_group_t i; + long i; /* should be able to store group number */ int j, len, metalen; struct ext4_sb_info *sbi = EXT4_SB(sb); int num_meta_group_infos = @@ -2257,6 +2257,7 ...

Do I interpert this correct when I envision a call-chain like this: register_devise(A, some_parent) lock_device(A, NESTING_PARENT) D->probe() register_device(B, A) lock_device(B, NESTING_PARENT) That would work iff register_device() sets a tree-level class on B that Yes, associate a class with each level like this: static struct lockdep_class_key device_tree_class[MAX_DEVICE_TREE_DEPTH]; register_device(child, parent) { ... child->depth = parent->depth + ...

Lockdep only cares about class order; if you always lock NESTING_PARENT -> NESTING_CHILD it won't complain; but it will not require you to hold the parent lock (although I can provide you with an annotation to that effect if you want; lockdep_assert_held(dev->parent->lock) - or something like that). As long as you're sure there aren't any actual deadlocks; and ensure you use your annotated classes in the same order: tree_level_n -> tree_level_n+1 and tree_leve_n_sub_m -> ...

I have done basic testing on this patchset and have not encountered any problems from the Smack side. I have not Casey Schaufler casey@schaufler-ca.com --

Unfortunately, I must say, Smartmedia is not supported on linux for any reasonable definition of the word. First, with all the respect to Jorn, alauda driver can only be considered "proof of concept". It does not try to abstract any smartmedia functionality. Second, ssfdc is hopelessly obsolete and requires rewrite. Third, there's no attempt made by mtd to support advanced functionality present in many smartmedia and all memorystick readers: adapter side copy and multi page programming (this ...

Oh, aufs was not listed in the bug report. that's an out of tree module, right? Is it completely up to date and safe with 2.6.25? -Eric --

And it's a completely crappy one. I personally would through away every bug report with that piece of junk involved. --

Well, yes, and we treat shutdown bugs as such. A filesystem shutdown is effectively a filesystem panic and is indicative of either a corruption or a bug. The reality is that it takes time to triage such a problem that only occurs on one workload on one set of identical machines once every day or two. This does not make the problem a release blocker, though. The other side of it is that problems like this in Linux are often the result of a bug in a lower layer and not XFS itself. Given ...

From: Patrick McHardy <kaber@trash.net> Even though there have been discussions to do other things, this patch moves things in the forward direction so I have applied it. Thanks. --

I guessed it was required for 2.6.24 based on the patch dates (and a fuzzy recollection of which release the "bad" commit went in). The date on the "bad" commit was before the release of 2.6.24, so I assumed the fix should be applied in 2.6.24 as well. Mea culpa. -- Chuck Lever chuck[dot]lever[at]oracle[dot]com --

I don't object to the point of the patch (changing KERN_ERR to KERN_WARNING), but there is a more serious bug lurking here that corrupts memory, and if we're doing a -stable patch, we should fix it. The test is: if (i >= PNP_MAX_IRQ && !warned) { printk(...); warned = 1; return; } But it should be: if (i >= PNP_MAX_IRQ) { if (!warned) { printk(...); warned = 1; } return; --

Looks good to me. Thanks! --

Dnia 2008-04-17, o godz. 14:33:15 Len wanted only entry in MAINTAINERS for compal driver. -- Cezary --

we do an automatic build test (of both vmlinux and of modules) of over 80 non-x86 architecture configs, amongst them are sparc64 and sparc configs: http://www.tglx.de/autoqa-cgi/index?run=86&tree=1 for the latest version of our trees all the 80 non-x86 configs (and all 96 configs in total) built successfully. if in the sparc64 row you click on the green "OK" button (which signals that the build was successful) you can get the build log and see all 7 warnings that get triggered ...

ok, good point, i zapped this change [it was in the x86/testing section anyway, i.e. not to be pushed upstream as an x86 change] - and if it's resubmitted it should be sent to -mm anyway. Arjan, what do you think? Ingo --

Does it meany Linux give up implementing DTrace way of tracing/instrumntation ? In last time I observe more and more signs inroducing parallel ways of tracing/instrumentations infrasctructures in Linux kernel where all this can be rolled into only one .. common. Strange but some of this tracing/instrumentations does not uses "zero cost probes" but "near zero cost probes" (like this) and this will result only more and more bloated kernel code with statically injected ...

quite so. Your change was promising from the get go and the latest iteration was definitely a good one and is very much mergable. That it also helps improve other architectures is the icing on the cake. Andi will have to prove his points by coming up with competing benchmark results - you certainly did your fair share to back up your change with numbers. (and Andi, if/when you do so, please Cc: Alexander too in the future. Dont you want him to be able to reply to your complaints ...

btw., several percent of TPC-C is bad... so i take back the 'SLAB is out' observation for now :-/ Ingo --

Why is is deprecated anyway? I think we all agreed that killing the export is good, but I don't see any way how we could kill the core useage. --

are you sure? This seems to assume machine is thinkpad if kstrdup fails... which is very wrong. -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html --

[Pavel Machek - Fri, Apr 18, 2008 at 02:41:12PM +0200] | Hi! | | > Fix potential NULL pointer dereference if kstrdup failed | > | > Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com> | > | > --- | > | > Index: linux-2.6.git/drivers/misc/thinkpad_acpi.c | > =================================================================== | > --- linux-2.6.git.orig/drivers/misc/thinkpad_acpi.c 2008-04-16 20:35:34.000000000 +0400 | > +++ linux-2.6.git/drivers/misc/thinkpad_acpi.c 2008-04-16 ...

how about without acpi=off? can you make sure acpi=off works with previous kernel in that box? YH --

according to patch from Suresh in x86.git, one node still only have one bdata. YH --

Applied the lot, thank you Mark. -- Dmitry --

Hi - At least two answers apply. The markers being put in should be chosen with the concurrence of the subsystem maintainer, who should help identify those key quantities that are likely to be both long-lived and good diagnostic value. So that's the discussion we're having right now: which values should be passed. If they're long-lived, then they can be given a long-lived name - and it doesn't have to be a low-level C variable name. (There's no reason why we can't also have a slew of ...

*blink* so all this is so you don't have to put a declarion in a header file? How about we put these premanent markers in a header - Mathieu says there are <200. Surely that's not too much trouble. Then you can keep this trace_mark() (perhaps trace_printf() is a better So instead of writing normal C code and placing a declarion in a header, you've come up with a scheme that needs to duplicate a text string to check integrity. Sounds like a real good way to confuse people. --

static struct tracer no_tracer __read_mostly = { We should generate an error in this case, rather than just copying in a wrong value. The necessity to keep the 63 and 64 in sync is fragile. sizeof() would I guess this really should be advanced by the number of bytes which This can be grossly inefficient. A machine can (I believe) have 1024 possible CPUs with only four present. It should size this storage by the number of online CPUs and implement the So I stared at the data ...

Andrew, thanks for taking you time in looking through this patch. I saw that in my checkpatch run. But the API was strange to me (my first time I've seen strict_strtoul). I'll look more at how to use FYI, I'm currently working on a tutorial for OLS. Yes, I'm giving a tutorial for ftrace. I'm now writing a LaTeX document of all the features. One thing that will come out of this is a Documentation/ftrace.txt Last night, while writing the above mentioned document, I thought ...

Hm, I'd need to think about it more. There's more to NMI's than just the popf. J --

I wonder if we could simply paravirtualize the popf instruction, which seems to be the only one requiring to run in ring 0. Mathieu -- Mathieu Desnoyers Computer Engineering Ph.D. Student, Ecole Polytechnique de Montreal OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F BA06 3F25 A8FE 3BAE 9A68 --

On Mon, 14 Apr 2008 22:39:15 -0400 Hm, no answer on that. Might as well just remove the const. Can you please add the patch below? A couple things were wrong here: - a struct in lxfb should've been static. - 'const' and '__initdata' don't mix, so the consts have been dropped - get_modedb() in both drivers should've been marked w/ __init Signed-off-by: Andres Salomon <dilinger@debian.org> --- drivers/video/geode/gxfb_core.c | 8 ++++---- ...

http://developer.petalogix.com/wiki/UserGuide/AdvancedTopics/EnablingUnc It probably makes sense to grab: arch/microblaze/mm/dma-coherent.c from git.xilinx.com and drop consistent.c. I believe this is the more 'modern' way to implement this functionality. I think it's easiest for I'd recommend dropping anything having to do with PCI: I don't know that there is much interest in it on microblaze. Other comments: memset/memmove/memcpy are broken under gcc4. There are fixed versions of ...

Swap on loopback could deadlock with whatever file system, not only FUSE. The correct and safe way to use swap with FUSE is using the fuseblk fstype and a swap file on the FUSE file system (so it's exactly the same as one would do with an in-kernel file system). This completely avoids the usage of user space during swapping. This is in use for over a year with ntfs-3g, e.g. by Knoppix and Mr Klaus Knopper wrote only positive things to me so far. Szaka -- NTFS-3G: ...

Presumably if "user space isn't involved at all", it must require that user space has granted caching rights to the kernel over a FUSE cache coherency protocol? Otherwise I don't see how the kernel could coherently cache file pages for some kinds of FUSE filesystems. (E.g. sshfs, for example: every operation must surely invoke a user space request or involve granting a caching right to the kernel, to keep accesses coherent with other users of the same remote files). Ergo, either its not ...

Correct me if I'm wrong, but one place where caches seem necessary is for lookup. My file system already has an inode number; my understanding is that the kernel inode cache and dcache are caching the FUSE inode by filename and its hashed inode number. In FUSE, on open, I'm passed a filename which I then have to resolve into an inode # via my own lookup. The VFS does the path_lookup as part of sys_open, and since I get to put private data into the struct inode, I'll generally already have ...

Thankfully you need none, it's already there by FUSE and the kernel. The trick is exactly that you can have the kernel performance and the left is moved to user space with typically negligible performance overhead which is usually well compensated with faster delivered new features and bug fixes. The completely unoptimized ntfs-3g read/write saturates my USB disks with 25-30 MB/sec using 8% (read), 35% (write) CPU time on a 2.5 GHz Core 2 Duo with cold caches. If you have the free ...

Yes, that makes sense. I expect this could be done by extending the existing requests with a flag saying the result is already cached. And then the filesystem can either reply with a special "cached data Oh well, you can turn off caching if it bothers you :) OTOH it would be rather hard (and probably against the point) to try to extend the sftp protocol to handle cache coherency. Sshfs is not meant to be a normal filesystem (although some people are trying to use it for home directories ...

Sure. I'll go on record saying that omfs_fuse was written over a weekend and can not be considered a good example of anything. I already had a proto-libomfs that I had made for omfsck and mkomfs, and wanted to see how easy it would be to wire that stuff up into a FUSE fs. Pretty easily, it turned out. -- Bob Copeland %% www.bobcopeland.com --

Ah, I'll have to take a closer look at these cards and see if I can figure out what drives the link layer... But barring the discovery of another TI chip, I'm assuming an attempt to reproduce the generation issue will be fruitless. Well, based on the positive results we've seen thus far with TI controllers and this patch in Fedora kernels: Signed-off-by: Jarod Wilson <jwilson@redhat.com> -- Jarod Wilson jwilson@redhat.com --

I wasnt sure how it was evaluated, i dont really know alsa alot, so i Yes, sorry for delay, for some reason my mail client decided not to update the view.. weird.. You should know that this is WITH my modification.. About problems.. well.. now that you do mention it, when i play something without a .asoundrc, i get: ALSA lib dlmisc.c:118:(snd_dlsym_verify) unable to verify version for symbol _snd_pcm_empty_open ALSA lib pcm.c:2108:(snd_pcm_open_conf) symbol _snd_pcm_empty_open is not ...

OK, but that AFAICS is a lock manager implementation issue, not an API No gfs2, only ext3. Thanks, Miklos --

This is definitely the cleanest way to do this from my pov. Although I'm slightly concerned about the proliferation of weak symbols, I'm much more concerned with adding more include-order-dependent arch overrides. :) speaking with the parisc dunce-cap on, Kyle --

Well, looking there, I saw we already used weak symbols for that so what about the patch below ? If you're ok, I'll re-send with appropriate sob & adapted powerpc part. Cheers, Ben. Index: linux-work/init/main.c =================================================================== --- linux-work.orig/init/main.c 2008-03-26 10:39:25.000000000 +1100 +++ linux-work/init/main.c 2008-04-18 13:10:35.000000000 +1000 @@ -504,6 +504,10 @@ void __init __attribute__((weak)) smp_se { } +void ...

Aye. Though I prefer the term virtual geometry. But thats cosmetics. Sorry for beeing unclear, and many thanks for untangling my post. If anyones interested in even more Details about C/H/S adressing and so on, there's a very good document about that to be found here: http://www.mossywell.com/boot-sequence/ --

OK, race problem between threads is fixed. Thanks for finding the issue! The new patch series is coming next. Regards, Nadia --

Sorry to be a bit though, but _what_ I'm supposed to patch? It's in s2ram sources? Thanks! Romano -- Sorry for the disclaimer --- ¡I cannot stop it! -- La presente comunicación tiene carácter confidencial y es para el exclusivo uso del destinatario indicado en la misma. Si Ud. no es el destinatario indicado, le informamos que cualquier forma de distribución, reproducción o uso de esta comunicación y/o de la información contenida en la misma están estrictamente prohibidos por la ...

Thanks. Committed to linux1394-2.6.git/master. I will submit it to upstream in a week or so. -- Stefan Richter -=====-==--- -=-- =--=- http://arcgraph.de/sr/ --

Here are some microbenchmark results. Test setup is a 2-way IA64 with a single 15k RPM 73GiB SCSI disk with TCQ depth set to 1. Workloads are generated with FIO: 128 processes issuing synchronous, O_DIRECT, 16KiB block size requests. Figures are quoted as average (stdev). CFQ (i=0) means idle window disabled. All other tunables are default. ==================================x8======================================= Random ...

technically you can compute x4 with the original value prior to isolating the least-significant one-bit -- the compiler probably can't figure this i'm never sure if it's better to use | or + here... i bet it depends on what native operations the processor has... and depends on how ?: are implemented. -dean --

<snip> Damn, x86 is also using functions instead of macros. Sorry about that; I'll have another try. Ben. -- Ben Hutchings, Senior Software Engineer, Solarflare Communications Not speaking for my employer; that's the marketing department's job. --

ooh, I just discovered kstack=. That should be hoisted up to core kernel then pushed out to other architectures, not hidden in x86_64 code. But first I guess it should be make to work - it fails miserably, because only one of our seventy fix stack-dumping functions actually looks at it. This: --- a/arch/x86/kernel/traps_64.c~a +++ a/arch/x86/kernel/traps_64.c @@ -426,7 +426,7 @@ _show_stack(struct task_struct *tsk, str printk(" %016lx", *stack++); touch_nmi_watchdog(); ...

The mount time compared to JFFS2 on a 700MiB partition went down from We did some initial stress tests and so far it looks pretty robust. Great job ! Thanks, tglx --

Thanks, this seems has solved the problem. Now, I can fetch from stable branch over http without any problems. Richard --

From: Krzysztof Halasa <khc@pm.waw.pl> I'll fix it properly because I know you won't do the work. You don't need to keep making excuses. --

I'd suggest to clean up the source of this mess instead; this is quite horrible. --

Samuel, Why did you write sa_family_t here? Dave M already said sizeof(short), which is the same thing, and I see that in net/unix/af_unix.c::unix_getname() there is: u = unix_sk(sk); unix_state_lock(sk); if (!u->addr) { sunaddr->sun_family = AF_UNIX; sunaddr->sun_path[0] = 0; *uaddr_len = sizeof(short); Cheers, Michael -- Michael Kerrisk Linux man-pages maintainer; ...

Oh sorry my bad! I investigated again and found 2 problem in your test program. 1. text segment isn't locked. if strong memory pressure happned, kernel may drop program text region. then your test program suddenly slow down. please use mlockall(MCL_CURRENT) before large buffer allocation. 2. repeat open/close to /proc/meminfo. in the fact, open(2) system call use a bit memory. if call open(2) in strong memory pressure, doesn't return until memory freed enough. ...

Hi, When capturing some traces with dynamic tick we were noticing the interrupt latency seems to go up a good amount. If you look at the trace the gpio IRQ is now offset a good amount. Good news I guess is its pretty predictable. * If we couple this with progressively higher latency C-States we see that IO speed can fall by a good amount, especially for PIO mixes. Now if QOS is maintained you may or may-not care. I was wondering what thoughts of optimizing this might be. One thought ...