[Regression] Well that was not the real cause of the performance loss. Because on kernel 2.6.27 I get really good performance (900-1200kpps) compared to 2.6.28 (git net-2.6). The cause of this problem (tracked down together with Robert Olsson) is that on 2.6.28 I have a lot less IRQs available. It seems max 34 IRQs. Due the reduced number of IRQs the NIU driver cannot get enough IRQs to the interfaces, and starts to use "IO-APIC" based IRQs. On kernel 2.6.28: My eth2 is ...

Please pull my for-linus branch: git pull git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci-2.6.git for-linus Another small fix for 2.6.28. If not fixed, this bug could end up causing IRQ 0 to get disabled when a PCI function is reset before init. Thanks, Jesse Sheng Yang (1): PCI: Fix disable IRQ 0 in pci_reset_function() drivers/pci/pci.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) --

Fix arch/x86/kernel/setup.c:592: warning: 'dmi_low_memory_corruption' defined but not used this is only used if CONFIG_X86_RESERVE_LOW_64K is defined. Signed-off-by: Richard A. Holden III <aciddeath@gmail.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ingo Molnar <mingo@redhat.com> Cc: H. Peter Anvin <hpa@zytor.com> --- arch/x86/kernel/setup.c | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-) diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index ...

Sorry for the shotgun mail, but in the end, it's not clear who can best answer my question(s). I'm currently trying to add documentation of all of the undocumented CLONE_* flags. One of these is CLONE_NEWNET, and I could use (quite a lot of) help. My questions: What is the current state of the network namespace implementation? Is it complete? What objects are considered part of the network namespace, and therefore distinct for a new network namespace? Is there any ...

Impact: Allow 64 bit PowerPC to trace modules with dynamic ftrace This adds code to handle the PPC64 module trampolines, and allows for PPC64 to use dynamic ftrace. Thanks to Paul Mackerras for these updates: - fix the mod and rec->arch.mod NULL checks. - fix to is_bl_op compare. Thanks to Milton Miller for: - finding the nasty race with using two nops, and recommending instead that I use a branch 8 forward. Signed-off-by: Steven Rostedt <srostedt@redhat.com> --- ...

Impact: Port 32 bit PowerPC dynamic ftrace This patch adds the necessary hooks to get PPC32 dynamic ftrace working. It does not handle modules. They are ignored by this patch. Signed-off-by: Steven Rostedt <srostedt@redhat.com> --- arch/powerpc/Kconfig | 2 +- scripts/recordmcount.pl | 7 ++++++- 2 files changed, 7 insertions(+), 2 deletions(-) diff --git a/arch/powerpc/Kconfig b/arch/powerpc/Kconfig index 9675e95..d64b629 100644 --- a/arch/powerpc/Kconfig +++ ...

Impact: add alignment option for recordmcount.pl script Align the __mcount_loc sections so that architectures with strict alignment requirements need not worry about performing unaligned accesses. This fixes an issue where I was seeing unaligned accesses, which are not supported on our architecture (the results of an unaligned access are undefined). Signed-off-by: Matt Fleming <matthew.fleming@imgtec.com> Signed-off-by: Steven Rostedt <srostedt@redhat.com> Signed-off-by: Ingo Molnar ...

Impact: use cleaner probe_kernel API over assembly Using probe_kernel_read/write interface is a much cleaner approach than the current assembly version. Signed-off-by: Steven Rostedt <srostedt@redhat.com> --- arch/powerpc/kernel/ftrace.c | 53 ++++++++++++++++------------------------- 1 files changed, 21 insertions(+), 32 deletions(-) diff --git a/arch/powerpc/kernel/ftrace.c b/arch/powerpc/kernel/ftrace.c index 24c023a..1adfbb2 100644 --- a/arch/powerpc/kernel/ftrace.c +++ ...

Impact: fix for irq off latency tracer When idle is called, interrupts are disabled, but the idle function will still wake up on an interrupt. The problem is that the interrupt disabled latency tracer will take this call to idle as a latency. This patch disables the latency tracing when going into idle. Signed-off-by: Steven Rostedt <srostedt@redhat.com> --- arch/powerpc/kernel/idle.c | 5 +++++ 1 files changed, 5 insertions(+), 0 deletions(-) diff --git a/arch/powerpc/kernel/idle.c ...

Hi Serge, What is the current status of CLONE_NEWUSER? I'm currently trying to test this flag in preparation for documenting it in the clone(2) man page, but am running into an ENOMEM error from the clone() call, which seems to occur after a failure in kobject_init_and_add() in the following call sequence: clone_user_ns() --> alloc_uid() --> uids_user_create() --> kobject_init_and_add() Are there already some test programs somewhere? Is there any documentation already available for this ...

applied to next. - k --

Serge, Eric, Below is a patch to document the CLONE_NEWUTS flag that was added in 2.6.19. Could you please review and let me know of improvements or inaccuracies? By the way, does anyone know where the UTS name in the uname() API comes from? My best guess is that it's from Unix Timesharing System, but I don't know this for sure. Cheers, Michael diff --git a/man2/clone.2 b/man2/clone.2 index 7212332..80f9caf 100644 --- a/man2/clone.2 +++ b/man2/clone.2 @@ -341,6 +340,33 @@ ...

Hi, as per Documentation/uml/*txt, I issued the following to get a direct hostfs: ./linux ubd0=/tftpboot/linux (/tftpboot/linux/sbin/init exists), but the UM kernel only gives: Failed to open '/home/tftpboot/linux', errno = 21 ubda: Can't open "/home/tftpboot/linux": errno = 21 List of all partitions: 6200 4 ubda driver: uml-blkdev So either UML lost some functionality sometime, or the documentation is incorrect. --

Does it survive if you boot with 'quiet'? Is this new behavior or have you successfully used netconsole before? -- Mathematics is the supreme nostalgia of our time. --

Hi It is x86_64 SMP The kernel freezes while transferring kernel log through netconsole, SysRq does not work at that moment. The kernel works fine without netconsole. Here is lspci, then dmesg, config at the end. 00:00.0 Host bridge [0600]: Intel Corporation 5000X Chipset Memory Controller Hub [8086:25c0] (rev 12) 00:02.0 PCI bridge [0604]: Intel Corporation 5000 Series Chipset PCI Express x4 Port 2 [8086:25e2] (rev 12) 00:03.0 PCI bridge [0604]: Intel Corporation 5000 Series Chipset PCI ...

proto_register could add another field somewhere and store the pointer to the name there? Then free the string on proto_unregister. --

Since the WM8350 driver was originally written the semantics for the identification registers of the chip have been clarified, allowing us to do an exact match on all the fields. This avoids mistakenly running on unsupported hardware. Also change to using the datasheet names more consistently for legibility and fix a printk() that should be dev_err(). Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com> --- drivers/mfd/wm8350-core.c | 54 ...

When booting the kernel in a Palm TX device, it's not possible to mount MMC cards because the driver doesn't seem to be automatically loaded. The card only works after sending an "add" command to the corresponding "uevent" file in "/sys". - Is there a way to configure the kernel to automatically load the MMC driver, so that the setup is not necessary ? (most drivers load while linux boots, it's not clear why it's not so with MMC) - Are there links/files where I can get more information about ...

I'm looking for a way to turn a Linux computer into a router as a front end for a small web hosting / spam filtering operation. Part of this is just educational for me. I want it to be web based controls and easy to use and install is the most important feature. I want to run it under Fedora so RPM based install would be great. Suggestions? --

On Wed, 19 Nov 2008 17:16:14 +0100 Your compiler appears to be faulty. You should report this to Ubuntu --

On Tue, 18 Nov 2008 Actually I need it for CONFIG_IA64_GENERIC and CONFIG_IA64_SGI_SN2. But, even so, unfortunately your suggestion doesn't build. I see numerous errors such as these: kernel/fork.o: In function `__crc_dma_get_required_mask': fork.c:(*ABS*+0xb0339303): multiple definition of `__crc_dma_get_required_mask' kernel/exit.o: In function `__crc_dma_get_required_mask': exit.c:(*ABS*+0x4a7c4115): multiple definiti CC fs/sysfs/group.o --

Someone is feeding the driver an sg list which has an odd number of bytes in an sg list entry which isnt the final one. Now I don't offhand know if that is valid or not but if it occurs then the libata code can't cope. Alan --

Dropping one parameter makes the function call a little faster, doesn't it? The results are the same, so if the string is constant, why not use it directly as the fmt string? -- Cheng Renquan, Shenzhen, China Lily Tomlin - "The trouble with the rat race is that even if you win, you're still a rat." --

Not that we should be able to measure it, but it should save some cycles parsing the format character and retrieving the string pointer It should be fine for strings, where you can be sure they can never contain a '%', sure. Thanks, Kay --

From: Cheng Renquan <crquan@gmail.com> Signed-off-by: Cheng Renquan <crquan@gmail.com> --- block/blk-sysfs.c | 2 +- block/elevator.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c index 21e275d..0b189e9 100644 --- a/block/blk-sysfs.c +++ b/block/blk-sysfs.c @@ -342,7 +342,7 @@ int blk_register_queue(struct gendisk *disk) return 0; ret = kobject_add(&q->kobj, kobject_get(&disk_to_dev(disk)->kobj), - "%s", ...

Hi Build fix for kerneldoc Jes

Hi It is 4way X86_64 The kernel does not boot. ... scsi0 : LSI SAS based MegaRAID driver Driver 'sd' needs updating - please use bus_type methods scsi 0:0:0:0: Direct-Access ATA SAMSUNG HE160HJ 0-24 PQ: 0 ANSI: 5 general protection fault: 0000 [#1] PREEMPT SMP DEBUG_PAGEALLOC last sysfs file: CPU 3 Modules linked in: Pid: 0, comm: swapper Not tainted 2.6.28-rc5-next-20081119 #6 RIP: 0010:[<ffffffff80240061>] [<ffffffff80240061>] get_next_timer_interrupt+0x11b/0x1f0 RSP: ...

The best solution is probably to either provide a "doesn't do highmem" in the scsi host template, or provide an appropriate DMA mask for the pci device to indicate it through that setting instead. -- Jens Axboe --

Thanks Alan, This seems to fix the problem. --

usb-storage doesn't allocate the memory. The memory is allocated by Jens, is there any way to tell the kernel that a device uses PIO and therefore its buffers shouldn't be allocated in high memory? For example, shouldn't a NULL dma_mask do this? If not, are there standard routines to set up bounce buffers for such devices? Alan Stern --

The DMA mask is currently set to NULL. Is that not appropriate for a device that can't do DMA? If not, then what would be appropriate? Also, is the patch above not correct? Alan Stern --

As soon as we have command line, so even before early_param. They just set vars, so it makes sense to do them as early as possible. This allows them to replace early_param, and fixes a bug in the new cpu_alloc implementation patches which was a complete PITA to find. Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> --- include/asm-generic/vmlinux.lds.h | 3 +++ include/linux/moduleparam.h | 26 ++++++++++++++++---------- init/main.c | 25 ...

Indeed, thanks! Fixed, Rusty. --

[RESENT, because LKML bounced some HTML that accidentally got put in the mail.] [CC+=John McCutchan, this time with hopefully a live email address; John, some context here: http://marc.info/?t=122633022400003&r=1&w=2 ] Evgeniy, I suspect that Christoph wants the same thing as I do: some thinking towards a future-proof design, rather than a quick hack to address the Again, appropriate flags in inotify_init1() could fix this -- e.g., only fill the field (and give an error if no perms) if a ...

(Linux-next again, but simple conversion...) Most cpumask functions started with cpus_: these have been replaced by cpumask_ ones which take struct cpumask pointers as expected. These four functions don't have good replacement names; fortunately they're rarely used, so we just change them over. Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Mike Travis <travis@sgi.com> Cc: paulus@samba.org Cc: mingo@redhat.com Cc: tony.luck@intel.com Cc: ralf@linux-mips.org Cc: Greg ...

(I'll be rinsing this through linux-next; testing feedback and even Acked-by's appreciated). Each SMP arch defines these themselves. Move them to a central location. Twists: 1) Some archs (m32, parisc, s390) set possible_map to all 1, so we add a CONFIG_INIT_ALL_POSSIBLE for this rather than break them. 2) mips and sparc32 '#define cpu_possible_map phys_cpu_present_map'. Those archs simply have phys_cpu_present_map replaced everywhere. 3) Alpha defined cpu_possible_map to ...

Remove CONFIG_APM_REAL_MODE_POWER_OFF like CONFIG_APM_POWER_OFF which has been done for linux-2.2.14pre8 (http://lkml.org/lkml/1999/11/23/3). Re-introducing CONFIG_APM_POWER_OFF got nack-ed. Stephen didn't bother to remove CONFIG_APM_REAL_MODE_POWER_OFF, let's get rid of it now. Reference: http://lkml.org/lkml/2008/5/7/97 Signed-off-by: Niels de Vos <niels.devos@wincor-nixdorf.com> --- arch/x86/Kconfig | 7 ------- arch/x86/kernel/apm_32.c | 4 ---- 2 files changed, 0 ...

Signed-off-by: Qinghuang Feng <qhfeng.kernel@gmail.com> --- diff --git a/drivers/mtd/ftl.c b/drivers/mtd/ftl.c index 9bf581c..c3bbf38 100644 --- a/drivers/mtd/ftl.c +++ b/drivers/mtd/ftl.c @@ -1099,7 +1099,7 @@ static struct mtd_blktrans_ops ftl_tr = { .owner = THIS_MODULE, }; -static int init_ftl(void) +static int __init init_ftl(void) { return register_mtd_blktrans(&ftl_tr); } --

Signed-off-by: Qinghuang Feng <qhfeng.kernel@gmail.com> --- diff --git a/drivers/mtd/nand/cmx270_nand.c b/drivers/mtd/nand/cmx270_nand.c index fa129c0..c429ea2 100644 --- a/drivers/mtd/nand/cmx270_nand.c +++ b/drivers/mtd/nand/cmx270_nand.c @@ -148,7 +148,7 @@ static int cmx270_device_ready(struct mtd_info *mtd) /* * Main initialization routine */ -static int cmx270_init(void) +static int __init cmx270_init(void) { struct nand_chip *this; const char *part_type; @@ -262,7 +262,7 ...

None of these (init|exit) functions is called from other functions which is outside the kernel module mechanism or kernel itself, so mark them as {__init|__exit}. --

Sorry for inconvenience but I'm out of office until Friday. I met Nishimura and disccussed to post rework patch. (means unififed patch of original+fix1,2,3,4 + consider about lock again) Thank you for handling. Regards, --

Acked-by: Christoph Lameter <cl@linux-foundation.org> --

In __sock_recv_timestamp() the additional SCM_TIMESTAMP[NS] is used. This has the same value as SO_TIMESTAMP[NS], so this is a purely cosmetic change. --- net/compat.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/net/compat.c b/net/compat.c index 67fb6a3..6ce1a1c 100644 --- a/net/compat.c +++ b/net/compat.c @@ -226,14 +226,14 @@ int put_cmsg_compat(struct msghdr *kmsg, int level, int type, int len, void *dat return 0; /* XXX: return error? check spec. */ ...

Instructions for time stamping outgoing packets are take from the socket layer and later copied into the new skb. --- include/net/ip.h | 1 + net/can/raw.c | 6 ++++++ net/ipv4/icmp.c | 2 ++ net/ipv4/ip_output.c | 2 ++ net/ipv4/raw.c | 1 + net/ipv4/udp.c | 4 ++++ 6 files changed, 16 insertions(+), 0 deletions(-) diff --git a/include/net/ip.h b/include/net/ip.h index bc026ec..9bc2b65 100644 --- a/include/net/ip.h +++ b/include/net/ip.h @@ ...

The new sk_buff->hwtstamp is used to transport time stamping instructions to the device driver (outgoing packets) and to return raw hardware time stamps back to user space (incoming or sent packets). Implements TX time stamping in software if the device driver doesn't support hardware time stamping. The new semantic for hardware/software time stamping around net_device->hard_start_xmit() is based on two assumptions about existing network device drivers which don't support hardware time ...

Currently only mapping from clock source to system time is implemented. The interface could have been made more versatile by not depending on a clock source, but this wasn't done to avoid writing glue code elsewhere. The method implemented here is the one used and analyzed under the name "assisted PTP" in the LCI PTP paper: http://www.linuxclustersinstitute.org/conferences/archive/2008/PDF/Ohly_92221.pdf --- include/linux/clocksync.h | 139 +++++++++++++++++++++++++++++++++++++++++++++ ...

--- drivers/net/igb/igb_main.c | 30 ++++++++++++++++++++++++++++++ 1 files changed, 30 insertions(+), 0 deletions(-) diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c index 89ffc07..be8e2b8 100644 --- a/drivers/net/igb/igb_main.c +++ b/drivers/net/igb/igb_main.c @@ -34,6 +34,7 @@ #include <linux/ipv6.h> #include <net/checksum.h> #include <net/ip6_checksum.h> +#include <net/timestamping.h> #include <linux/mii.h> #include <linux/ethtool.h> #include ...

Adds register definitions and a clocksource accessing the NIC time. --- drivers/net/igb/e1000_regs.h | 28 +++++++++++ drivers/net/igb/igb.h | 3 + drivers/net/igb/igb_main.c | 105 ++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 136 insertions(+), 0 deletions(-) diff --git a/drivers/net/igb/e1000_regs.h b/drivers/net/igb/e1000_regs.h index 95523af..37f9d55 100644 --- a/drivers/net/igb/e1000_regs.h +++ b/drivers/net/igb/e1000_regs.h @@ -75,6 +75,34 @@ #define ...

Both TX and RX hardware time stamping are implemented. Due to hardware limitations it is not possible to verify reliably which packet was time stamped when multiple were pending for sending; this could be solved by only allowing one packet marked for hardware time stamping into the queue (not implemented yet). RX time stamping relies on the flag in the packet descriptor which marks packets that were time stamped. In "all packet" mode this flag is not set. TODO: also support that mode (even ...

At Wed, 19 Nov 2008 23:42:48 +0800, Well, the access in the kernel space is fairly similar as in the user space. It opens, issues ioctls, reads and writes. The difference is that you access via dedicated function calls instead of syscalls. There is no way to poke the driver internal from other drivers. Yes. But for the kernel space buffer, you'd need to fake the user-space pointer by snd_enter_user() and snd_leave_user(). See Not way to peek/poke the driver internals from the ...

No, my plan is not a sound card driver. It is an USB gadget audio driver. When an embedded system for example Blackfin board connects to a USB host (PC), PC will recognize this USB device as a USB Audio Class device. Generally, there should be a sound card on the embedded system. Our Blackfin board has an AD1980 ALSA sound card. The USB gadget audio driver will open this sound card and export this device to USB host PC by some USB audio class specific descriptors. Then the PC can playback ...

Hi, What switch is responsible for this logging: [ 94.025172] kobject: 'vcsa1': free name [ 94.025231] kobject: 'vcs1' (d2172ad8): kobject_add_internal: parent: 'vc', set: 'devices' [ 94.025308] kobject: 'vcs1' (d2172ad8): kobject_uevent_env [ 94.025315] kobject: 'vcs1' (d2172ad8): fill_kobj_path: path = '/class/vc/vcs1' [ 94.025352] kobject: 'vcsa1' (d23f00d8): kobject_add_internal: parent: 'vc', set: 'devices' [ 94.025398] kobject: 'vcsa1' (d23f00d8): kobject_uevent_env I can't ...

I've seen once the oops reported (the BUG() now @ line 1247), but I've never been able to reproduce it afterwards. I think that there still is a window open for a race here: 1314 struct cfq_data *cfqd = cic->key; 1315 =====> here cfq_exit_queue() can free cfqd and assign cic->key = NULL, and accessing cfqd->queue is not safe. [ If I'm not wrong :) ] 1316 if (cfqd) { 1317 struct request_queue *q = cfqd->queue; --

Hello, I am working on some project where I need to boot winxp from linux, and I am using kexec for it, but in some computers it doesn't work. When kexec jumps to my kernel, in kernel I am switching to real mode and try to call int13h to read mbr sector. But I have one computer(with AMI BIOS and this computer is newer than others) where this BIOS call doesn't work. I think it is the problem with data corruption of BIOS memory, and I also have compiled 2.6.27.6 kernel(where is applied ...

this is a very ugly patch, not really acceptable. printk methods should not be doing softirq processing - period. Ingo --

From: Ingo Molnar <mingo@elte.hu> Right, it's totally legit. --

This report: http://marc.info/?l=linux-netdev&m=122599341430090&w=2 shows local_bh_enable() is used in the wrong context (irqs disabled). It happens when a usual network path is called by netconsole, which simply turns off hardirqs around this all. This patch additionally disables softirqs to avoid possibility of enabling bh and calling do_softirq() with hardirqs disabled. Reported-by: Ferenc Wagner <wferi@niif.hu> Signed-off-by: Jarek Poplawski <jarkao2@gmail.com> --- ...

From: Ingo Molnar <mingo@elte.hu> Well, it sort of "has to". It calls the NAPI ->poll() method of the driver to try and make forward progress with TX reclaim so it can send new messages. It is very careful not to recursively invoke into ->poll() and other nasty situations. Didn't you write some of this code Ingo a very long time ago? :-))) Anyways, I'll look more closely at this and the original report, this never was a problem before. --

From: Ingo Molnar <mingo@elte.hu> See my other reply to Jarek's patch posting, we're supposed to ignore all RX packets in this situation, and all the logic appears to be there, but it's not happening on that person's system for some reason. --

All of the byteshifting versions were cribbed from the ARM implementation. I'm not sure if there was a particular reason for doing it in this order, but a lot of work seems to have gone in to minimize register usage. See include/asm-arm/unaligned.h circa 2.6.25. Harvey --

Try this, and you'll get oops immediately: # cd Documentation/accounting/ # gcc -o getdelays getdelays.c # mount -t cgroup -o debug xxx /mnt # ./getdelays -C /mnt/tasks Because a normal file's dentry->d_fsdata is a pointer to struct cftype, not struct cgroup. After the patch, it returns EINVAL if we try to get cgroupstats from a normal file. CC: Balbir Singh <balbir@linux.vnet.ibm.com> Signed-off-by: Li Zefan <lizf@cn.fujitsu.com> --- kernel/cgroup.c | 7 +++++-- 1 files ...

Forgot to mention this: The bug exists from 2.6.24 (when cgroupstats is introduced) to now, though I didn't test in all those versions. I guess not many people used cgroupstats, so it never gets triggered. I think it can be applied to 2.6.24 too. --

Good catch, thanks. --

I guess that Stephen carried this patch for a few weeks, but now he has stopped carrying linux-next build patches. Still broken. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ From: Randy Dunlap <randy.dunlap@oracle.com> Needs a header file for credentials struct: linux-next-20081023/fs/coda/file.c:177: error: dereferencing pointer to incomplete type Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com> cc: Jan Harkes <jaharkes@cs.cmu.edu> --- fs/coda/file.c | 1 + 1 file changed, 1 ...

On Wed, 19 Nov 2008 17:59:17 +1100 Stephen Rothwell wrote: drivers/media/video/v4l2-ioctl.c:1819: warning: format '%d' expects type 'int', but argument 3 has type 'struct v4l2_fract' drivers/media/video/v4l2-ioctl.c:1819: warning: format '%d' expects type 'int', but argument 4 has type 'struct v4l2_fract' drivers/media/video/v4l2-ioctl.c:1819: warning: format '%d' expects type 'int', but argument 5 has type 'struct v4l2_fract' config attached. --- ~Randy

Please use the address linux-net-drivers@solarflare.com, as listed in MAINTAINERS. There are several of us working on the driver, and they no longer include Michael Brown. So does the dependency need to be something like "depends on SFC && MTD && !(SFC=y && MTD=m)"? Ben. -- Ben Hutchings, Senior Software Engineer, Solarflare Communications Not speaking for my employer; that's the marketing department's job. They asked us to note that Solarflare product names are trademarked. --

eh? drivers/staging/go7007/go7007-v4l2.c:1338: error: too many arguments to function 'video_usercopy' config attached. --- ~Randy

That's due to the portions of the staging tree that is in Linus's tree, interacting with the v4l tree here. It's fixed in my staging tree that normally gets into -next :) thanks, greg k-h --

Hi all, Changes since 20081118: New tree: m68k-current Changed tree: m86k is now a git tree Undropped trees: firmware lblnet Dropped trees (temporarily): driver-core (build problem) usb (depends on driver-core) userns (it depends on creds) semaphore-removal (due to unfixed conflicts against Linus' tree) staging (depends on usb) The sched tree gained a conflict against Linus' tree. The net tree picked up a conflict from the cifs tree. The rr tree lost its ...

On Wed, 19 Nov 2008 17:59:17 +1100 Stephen Rothwell wrote: This can happen when these drivers are built as loadable modules or built into the kernel image. Loadable modules: ERROR: "print_ssid" [drivers/net/wireless/ipw2200.ko] undefined! ERROR: "print_ssid" [drivers/net/wireless/ipw2100.ko] undefined! (modular config is attached) In kernel image: ipw2200.c:(.text+0x1d6f37): undefined reference to `print_ssid' ipw2200.c:(.text+0x1d7466): undefined reference to ...

If it's a separate driver, perhaps it's sufficient just to make it a tristate and make it depend on SFC && MTD? -- dwmw2 --

hm, why trylock? Ingo --

I wholly agree on the scalability. The bigger the system, the more it needs to protect the integrity of cfs_rq[]->shares that the sum still adds up to tg->shares. Otherwise, the share distributed on each CPU's cfs_rq might go wildly and indirectly leads to fluctuation of effective total tg->shares. However, I have the same doubt that this will scale on large CPU system. Does CFS really have to iterate the whole task_group tree? - Ken --

I see why you want to do this, but introducing a global lock makes me sad :/ --

Yes, sadly. The weight of a per-cpu super-task representation depends on the group's task distribution over all tasks :/ (Dhaval, could you send Ken a copy of the paper we did on this?) The idea was that we balance the stuff usng the sched-domain tree and update it incrementally, and on the top level sched domain fix it all up. Will it scale, half-way, I'd say. It races a little, but should converge. The biggest issue is that we're running with 10 bit fixed point math, and on large cpu ...

When taking recursive faults in do_exit, if the io_context is not null, exit_io_context() is being called. But it might decrement the refcount more than once. It is better to leave this task alone. Signed-off-by: Nikanth Karthikesan <knikanth@suse.de> --- diff --git a/kernel/exit.c b/kernel/exit.c index 80137a5..22dedf5 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -1024,8 +1024,6 @@ NORET_TYPE void do_exit(long code) * task into the wait for ever nirwana as well. */ ...

In the update_shares() path leading to tg_shares_up(), the calculation of per-cpu cfs_rq shares is rather erratic even under moderate task wake up rate. The problem is that the per-cpu tg->cfs_rq load weight used in the sd_rq_weight aggregation and actual redistribution of the cfs_rq->shares are collected at different time. Under moderate system load, we've seen quite a bit of variation on the cfs_rq->shares and ultimately wildly affects sched_entity's load weight. This patch caches the ...

applied to tip/sched/core (for v2.6.29), thanks guys! Ingo --

Impact: clean up and fix for dyn ftrace filter selection The previous logic of the dynamic ftrace selection of enabling or disabling functions was complex and incorrect. This patch simplifies the code and corrects the usage. This simplification also makes the code more robust. Here is the correct logic: Given a function that can be traced by dynamic ftrace: If the function is not to be traced, disable it if it was enabled. (this is if the function is in the set_ftrace_notrace ...

Impact: fix of output of set_ftrace_filter The commit "ftrace: do not show freed records in available_filter_functions" Removed a bit too much from the set_ftrace_filter code, where we now see all functions in the set_ftrace_filter file even when we set a filter. Signed-off-by: Steven Rostedt <srostedt@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu> --- kernel/trace/ftrace.c | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-) diff --git ...

pulled into tip/tracing/urgent, thanks Steve! Ingo --

山鉧科技網頁設計 ＊我們的宗旨：客戶的每ㄧ件小事情，都是山鉧的大事情 ◆我們在推出企業形象網站包含前台網頁美工+後台管理程式 ＝＝限時限量專案價 只要$29,900 ↖(在送ㄧ年100MB不限流量網站空間) ↖我們做的不只是網站，而是您企業的入口 ㊣ㄧ個好的企業網站資料即時更新的速度是很重要的 ↖企業ｅ化的高品質團隊，打造您的網路門面 ∴選擇山鉧成就您的夢想 ～～～～～～～～～～～～～～～～～～～～～～～～～～ PS: 線上購物網站我們還可提供刷卡機制， 　　與線上列印帳單全省超商+郵局繳費......等金流服務機制 ～～～～～～～～～～～～～～～～～～～～～～～～～～ 歡迎來電洽詢黃專員（Sam）：0980119812 / 0938764395 ～～～～～～～～～～～～～～～～～～～～～～～～～～ 本公司另外提供關鍵字SEO排序服務 保證將您的網站在Yx / Gx ...排在第一頁 　　　　　　　　　　　　　　　　　　 歡迎來電詢問!!! ～～～～～～～～～～～～～～～～～～～～～～～～～～

applied to tip/core/urgent, thanks! Note, i extended the changelog, see the commit attached below. I think in 2.6.29 we want to convert struct dmar_drhd_unit's ->reg_base_addr to the phys_addr_t type, and use the %pR extension to print it out. There's other files affected as well by this type problem: drivers/pci/intr_remapping.c for example - it's just not built on ia64 right now because it depends on CONFIG_INTR_REMAP which is only 64-bit x86. Some goes for other fields as ...

panic log (printk times removed for clarity) : BUG: unable to handle kernel NULL pointer at 0000000000000007 IP: [<ffffffff810a8d43>] __kmalloc+0x7a/0xc4 PGD 13d04a067 PUD 13b908067 PMD 0 Oops: 0000 [1] PRREMPT SMP CPU 0 Modules linked in: nvidia(P) uvcvideo ohci1394 compat_ioctl32 ieee1394 btusb iwlagn sdhci_pci videodev sdhci iwlcore mmc_core v4l1_compat r8169 pcspkr bluetooth rfkill Pid: 2562, comm: udevd Tainted: P 2.6.27-gentoo-r2 #10 RIP: 0010:[<ffffffff810a8d43>] ...

On Wed, Nov 19, 2008 at 4:39 PM, John W. Linville There is a lot of effort to solve this I'm not sure what mailing list are u reading. We even presented test patches with alignment issues as Johannes few month ago but This is embarrassing also for us that we cannot locate this specific problem but it not like we are not doing anything.Not sure there are I'm not against merging these patches to wireless-testing but pushing them upstream before testing them is just plainly wrong and all I ...

Well, thanks for your opinion. Here is mine: The iwlwifi drivers have been hitting these problems for months, and I have seen no effort by your team to address the problems. It is possible that your team is working on them behind closed doors and will eventually throw something over the wall to us. I'm tired of waiting for that, and I imagine that hordes of iwlagn users are tired of waiting as well. Johannes has presented us with plausible fixes, and people are reporting that the ...

I'll get it in the next run, soon. John -- John W. Linville Linux should be at the core linville@tuxdriver.com of your literate lifestyle. --

On Wed, Nov 19, 2008 at 2:07 AM, John W. Linville IMHO its premature pushing ttese 2 patches up, they came in yestrday and nobody here has run tthe code --

I'm seeing this sometimes when I boot. It looks like mempool_alloc is falling over calling pool->alloc() because it is null. Other times it boots fine, and seems solid once it has got past this point. This is a Xen dom0 kernel, so lots of the low-level interrupt and DMA stuff is new code, but this seems to be above all that, and I don't think its related to anything I've done. And as I say, it seems pretty solid once it gets past this. Thanks, J ioc0: LSI53C1030 C0: ...

Another crash, different output [ 2128.370238] monitor: Corrupted page table at address 82a4468 [ 2128.370257] PGD 10869067 PUD 23232323 BAD [ 2128.370271] Bad pagetable: 000d [1] SMP [ 2128.370276] CPU 2 [ 2128.370281] Modules linked in: i915 drm ipv6 dm_snapshot dm_mirror dm_log dm_mod coretemp w83627ehf hwmon_vid bttv ir_common compat_ioctl32 videodev v4l1_compat i2c_algo_bit v4l2_common videobuf_dma_sg videobuf_core btcx_risc tveeprom i2c_i801 shpchp pci_hotplug rng_core i2c_core iTCO_wdt ...

Hi Stephen, What's the purpose of your m68k-current tree? There should not be anything Thanks! Gr{oetje,eeting}s, Geert -- Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that. -- Linus Torvalds --

It's CVS that must die, not the poor server ;-) If you're actually going to pull the old CVS stuff into the new git repository and you happen to have a newer snapshot than the one which I've converted I can re-run my converter. Ralf --

Something I've run into while working on the locking stuff: what happens if the above ide_lock_hwgroup(hwgroup) sleeps for long enough to trigger the request timer? I'll think about the ramifications of your patch in the context of what I tested WRT unlocking whenever hwgroup->busy is cleared, and get back to you. Michael --

On Wed, Nov 19, 2008 at 2:10 PM, Michael Kerrisk I hesitate because I haven't looked at glibc in ages to see what it and other userland consumers are doing with inotify. But my original analysis still seems correct, and whatever we do we currently have inconsistencies, so this looks good to me. Acked-by: Robert Love <rlove@rlove.org> Robert --

On Wed, Nov 19, 2008 at 11:10 AM, Michael Kerrisk Seems sane enough. Acked-by: John McCutchan <john@johnmccutchan.com> -- John McCutchan <john@johnmccutchan.com> --

Hi Wang, This patch looks wrong to me; you can remove modules without /proc. Does it not compile with CONFIG_MODULE_UNLOAD=y and CONFIG_PROC_FS=n? Thanks, Rusty. --

The error is there, all right. However, proposed patch only hides the real problem. Building uml/i386 with CONFIG_HIGHMEM will result in a badly broken kernel with that patch, with no visible hints at the cause of problems. You've got it to link, but that's not enough. The problems you are seeing come from arch/x86/mm/highmem_32.c. It uses several variables from arch/x86/mm/init_32.c and you have copied them to arch/um. However, getting these variable defined is not going to make it work ...

Because %gs of 0 means "base too large, go to MSR". If you have a 32-bit base, then loading it into the gdt and loading %gs with the right selector is faster. wrmsr/rdmsr are slow instructions. J --

Ok, thanks, so I suppose now that only doing : asm volatile("movl %0,%%gs" :: "r" (0)); could corrupt the address of the PDA that resides actually in the MSR_GS_BASE. And that's why load_gs_index is used as it contains "swapgs" before and after the "mov to gs". Is that correct? Regards, Eric --

Yeah, it's fine to kill the arg. For consistency, change tracehook_consider_fatal_signal to match. They are specified as called with the siglock held, so it will indeed be easy and safe to check ->sighand if tracing code wants to distinguish the cases in the future. Thanks, Roland --

--

Well, who knows ? If you read my email carefully enough you should see that that is not the question or issue here. This was my first question to this list ever and I'm impressed with the good and constructive answers I've gotten, but I really do not understand the purpose of your response. /Fredrik On Wed, Nov 19, 2008 at 7:38 PM, Lennart Sorensen --

This won't work around the bug, it will just make its outcome less I doubt it, why use an isolation-capable IOMMU at all if not for the increased reliability? The majority of modern devices---those that you are likely to find on machines with an IOMMU---don't have DMA Calgary has a per-bus protection domain, both on x86 and PPC. Cheers, Muli -- The First Workshop on I/O Virtualization (WIOV '08) Dec 2008, San Diego, CA, http://www.usenix.org/wiov08/ <-> SYSTOR ...

a WARN_ON() can be acted upon much easier than silent/spurious data corruption. So printing a WARN_ON() will result in drivers being fixed a lot faster (and with a lot less debugging needed) than if we were intentionally letting DMA corruption happen. The WARN_ON() will be That would be _really_ nice to have. Ingo --

Hi, It would be interesting to know exactly what problem you are seeing. ARM I think is supposed to handle aliasing problems by flushing caches at appropriate points. It would be nice to know what's going wrong and whether we can cover those holes. We've traditionally avoided cache colouring in the page allocator for one reason or another. It should be a workable approach, but I think the way to get it into mainline is to first fix the problem using the existing cache flushing ...

dropping the BKL was a good idea, but the code flow was not investigated thoroughly enough, which caused this crash to trigger in -tip testing: calling utsname_sysctl_init+0x0/0x11 @ 1 initcall utsname_sysctl_init+0x0/0x11 returned 0 after 0 usecs calling init_sched_switch_trace+0x0/0xf @ 1 Testing tracer sched_switch: PASSED initcall init_sched_switch_trace+0x0/0xf returned 0 after 101562 usecs calling init_stack_trace+0x0/0xf @ 1 Testing tracer sysprof: PASSED initcall ...

On Tue, 18 Nov 2008 12:44:21 +0100 Yeah, as I wrote in a different thread, I think that it makes sense to enable SWIOTLB at all times: http://marc.info/?l=linux-kernel&m=122593287501949&w=2 An option asking an user if he has more than 4GB RAM or not, doesn't make sense much. --

I of course meant folding cpu and isspinlock into a combined bitfield (sorry for not being more clear), thereby saving space, this still takes 2*sizeof(int). We can safely take some bits from the cpu number as there currently are This is a bit icky still, I don't think TICKET_SHIFT is nessecarily the best macro to check on (other ticket lock implementations might not define it). A possible solution is to introduce a Kconfig variable HAVE_TICKET_LOCK and select that from ...

ach. So that's wrong anyway. There are arches which can't DMA stack Okay. A packed struct with a u8 followed by u16 which is required by the spec can't be fixed. unaligned helper is the only solution. I agree here. Allocating memory on the stack for a dma transfer is wrong. On PowerPC and X86 get_unaligned() does not behave any different than a Having a fixup in the exception handler like sparc does is probably little slower than the fixup here. On the other hand you would not have to fix ...

--

On Wed, Nov 19, 2008 at 2:56 PM, Andrew Morton Yes, I also prefer inline functions rather than macros here. Oh, I suggested Graf to run checkpatch.pl to find some issues before I sent out this patch. We can remove these 2 line spin_lock+spin_unlock and it also works. But maybe we will add some operation between spin_lock and spin_unlock here in the future, we'd like to keep them. P.S. also forward this patch to linux-arch Thanks --

On Wed, Nov 19, 2008 at 2:56 PM, Andrew Morton I guess Graf started this wiki recently although the patch exists for a long time. And Graf gave a presentation about this SMP on BF561 in AKA 2008 Linux kernel developer conference. If I found the link of this presentation, I will Exactly, SMP means hardware cache coherency. But BF561 dual core processor was designed almost 8 years ago. we have to do some workaround in software side. Fortunately, BF561 provides a L2 memory shared by both CoreA ...

that isnt a changelog though, it's documentation. perhaps we should put together a reduced version of the page for Documentation/blackfin/smp-like.txt ... -mike --

like this might attract some helpful comments. --

Cc, linux-arch -Bryan >

On Wed, Nov 19, 2008 at 2:56 PM, Andrew Morton As Nick said, include/asm-generic/bitops/non-atomic.h is the generic code. Thanks a lot for the review, I will forward this patchset to linux-arch. -Bryan --

Cc, linux-arch -Bryan --

--

Not in the absense of syscall in question. Patch applied, will push the tree tomorrow. --

i dont think ports should be forced to define syscalls they dont support. no-mmu cant do swap, so Blackfin doesnt define it ... -mike --

if it doesnt work on a platform, then it shouldnt be available. the arch list kind of implies that this is like x86 specific anyways ... -mike --

Thanks. There's no point in declaring this as inline, since it is always called indirectly, via on_each_cpu(). There would be a teeny point in inlining an on_each_cpu() function in uniprocessor code, because in that case it _would_ be inlined. But a) that's a bit dopey and b) this code you're patching is SMP-only. So... --- a/kernel/profile.c~kernel-profilec-fix-section-mismatch-warning-fix +++ a/kernel/profile.c @@ -544,7 +544,7 @@ static const struct file_operations proc }; ...

And it's called as kvm_free_physmem_slot(&old, &new); If a memory slot exists, the current code always deletes it and creates But the problem still exists even with this code. I checked. So if you have something working without modifying the kernel, can you post it? Regards, Anthony Liguori --

In this case, npages > 0 but !new.rmap is already allocated. But this is a new slot? The problem is that when we delete the slot, the rmap never gets freed. This means that if we delete a slot, then create a new slot which happens to be a different size, we use the old rmap and potentially overrun that buffer. So I think we need a fix that properly frees the rmap when the slot is destroyed. Regards, Anthony Liguori --

Oh yeah, it does get freed. The delete path ends up in a kvm_free_physmem_slot, which will effectively vfree() the rmap structure. In fact, my userspace use case worked totally properly when I deleted the slot prior to re-registering it. The problem here is when there is an already existant slot, and we are trying to change some information about it. The problem you are concerned basically does not exist, because it would raise only if we are changing the slot size. The code says: ...

Hello, On Wed, Nov 19, 2008 at 8:58 AM, FUJITA Tomonori This is the point I then want to make: we have pci_map_sg() users in other system than the block layer, the network and v4l2 subsystems, why cannot they benefit from coalescing? Should they copy the block layer coalescing implementation, or should that implementation be made more generic and live outside the block sub system? My intended use case is as follows. See the sg_write() call for the code ...

On Mon, 17 Nov 2008 19:15:32 -0800 Hm, what document did you read? We might need to fix it. pci_map_sg() is not a typical place to coalesce the entries of the sg list are physically adjacent. The block layer is the typical place. The dma operations are free to coalesce the entries that physically and virtually adjacent but there are not many that does. For example, by default, on x86, only AMD GART (x86_64) dma operation What's kinda of your driver? If it's a SCSI (or block) driver, ...

Oops, sorry about typos. On Wed, 19 Nov 2008 15:58:12 +0900 --

From: Yi Zou <yi.zou@intel.com> Why would FibreChannel over Ethernet need IPV4/IPV6 segmentation offloading? --

Thank you very much. Andrew, would you please consider merging this, it's short and working well. Thank you. Tomas M --

On Mon, 17 Nov 2008 08:44:28 +0100 Various people have tried this before, if I recall correctly. One was Akinobu Mita (cc'ed here). --

On Mon, 17 Nov 2008 15:38:16 +0800 Well. That's a bugfix, really. It fixes a bug which has been there for a very long time, in a way which is non-backward-compatible. Are there existing applications out there which will disastrously fail if we add a newline to the /proc/sys/fs/binfmt_misc/status contents? I somehow doubt it. But we shall find out :) --

Hi Christoph. So effectively you propose to have second generation of the inotify which will have additional pid field, which will be unused by all but the same uid events? If you want to return -EPERM, than it will be _always_ returned for non sysadmin capable user, which effectively makes it unusable. -- Evgeniy Polyakov --

Both -rc5 patches applied. Thanks Liam --

My preference would be for the patches to go through the powerpc tree unless there is a good reason for them to go via another tree. The style we use for the headline is "powerpc: Add xyz feature" or "powerpc/subsystem: Fix foo bug". As for the acked-by, I feel I first need to go through the whole series again with the changes you have made recently. Have you reworked the earlier patches to avoid introducing any bugs, rather than just fixing the bugs with later patches? If you haven't, ...

note that this inserts a lot of (unnecessary) serialization and a window of non-testing - by all likelyhood this will delay ppc ftrace to v2.6.30 or later kernels. Ingo --

I just pushed all the PowerPC patches on top of this port it works. I still need to rework the patches for Paul. -- Steve --

this is the diffstat: arch/powerpc/Kconfig | 2 + arch/powerpc/include/asm/ftrace.h | 14 +- arch/powerpc/include/asm/module.h | 16 ++- arch/powerpc/kernel/ftrace.c | 460 +++++++++++++++++++++++++++++++++--- arch/powerpc/kernel/idle.c | 5 + arch/powerpc/kernel/module_32.c | 10 + arch/powerpc/kernel/module_64.c | 13 + scripts/recordmcount.pl | 18 ++- 8 files changed, 495 insertions(+), 43 deletions(-) 90% of it creates new ...

Ah, ok. That makes sence (although such DVD's seem to be really broken). I'll merge the patch. Honza -- Jan Kara <jack@suse.cz> SUSE Labs, CR --

On Sun, 16 Nov 2008 20:52:19 +0100 Can we give these the same names and usage as their fatfs equivalents? --

Ok will rerun the tests tomorrow. Just got back from SC08 need some time to catch up. Looks like a lot of work was done on this issue. Thanks! --

Probably true. OTOH, I've seen indirect branches get compiled to direct branches or the common-case special cased into a direct branch if (object->fn == default_object_fn) default_object_fn(); That might be an easy way to test suspicions about CPU scheduler slowdowns... (adding a likely() there, and using likely profiling would help ensure you got the defualt case right). --

Definitely not in the second case - that one is called from inside readdir, lookup and symlink resolution code so that could lead to deadlocks IMHO. Regarding the first case in process_sequence, that is called only from udf_fill_super(). So there it might be safe to use GFP_KERNEL but I'm not quite sure either... So I'd leave GFP_NOFS there. Honza -- Jan Kara <jack@suse.cz> SUSE Labs, CR --

Hmm, but I see for example: static int shrink_icache_memory(int nr, gfp_t gfp_mask) { if (nr) { /* * Nasty deadlock avoidance. We may hold various FS locks, * and we don't want to recurse into the FS that called us * in clear_inode() and friends.. */ if (!(gfp_mask & __GFP_FS)) return -1; prune_icache(nr); } return ...

Both patches look fine. Thanks Martin. I've merged them to my UDF tree. -- Jan Kara <jack@suse.cz> SUSE Labs, CR --

Something like this would be a lot cleaner equivalent replacement: PUSHQ %rax /* rsp */ PUSHQ $(1<<9) /* eflags - interrupts on */ PUSHQ $__KERNEL_CS /* cs */ PUSHQ \child_rip /* rip */ cfi_map rip, 0 PUSHQ %rax /* orig rax */ as most of the really annoying CFI annotations in entry_64.S that obscruct code reading are just plain CFA offset modifications related to stack shuffling. [ Sidenote: trying to ...

i dont buy that argument at all. Yes, of course full "no changes to the current code at all" automation is hard. But _at minimum_ GAS should automate a large part of this and help us out syntactically: make it really easy to human-annotate instructions in a _minimal way_. Also, automate the easy bits while at it. Plus warn about missing annotations - nesting errors, etc. etc. Yeah. This is the second-best option - but has some limitations. Still it is much better than what we ...

Hmm, but if the assembler cannot auto generate it how should the assembler writer know if he should use the macro or the direct instruction without understanding CFI? Also what will the assembler reader do? Do they first have to understand CFI to understand everything? I personally would probably just resort to objdump -S in this situation. I think you're saying that for the user the macros would be just equivalent, but if that's true they could be just auto generated by the assembler. ...

Hi all, Here is a combined patch that moves "save_args" out-of-line for the interrupt macro and moves "error_entry" mostly out-of-line for the zeroentry and errorentry macros. The save_args function becomes really straightforward and easy to understand, with the possible exception of the stack switch code, which now needs to copy the return address of to the calling function. Normal interrupts arrive with ((~vector)-0x80) on the stack, which gets adjusted in ...

Hmm.. Yes. I initially had a file called features that I intended to use for Thanks for the review! -- Balbir --

| | > +static inline int siginfo_from_ancestor_ns(siginfo_t *info) | > +{ | > + return SI_FROMUSER(info) && (info->si_pid == 0); | > +} | | Yes, this is problem... I doubt we can rely on !si_pid here. | More on this later. BTW, rather than clearing SIG_FROM_USER in send_signal(), can we keep it till we dequeue the signal ? Yes, collect_signal() would need to consider this flag. But when we dequeue, we can note that it was from user and use that in the siginfo_from_ancestor() ...

Acked-by: Eric Miao <eric.miao@marvell.com> Andrew, Could you please help merge this as well? Thanks. -----Original Message----- From: Mike Rapoport [mailto:mike@compulab.co.il] Sent: Wednesday, November 19, 2008 4:51 PM Cc: rpurdie@linux.intel.com; linux-kernel@vger.kernel.org; kernel-janitors@vger.kernel.org; Eric Miao; Mike Rapoport; rpurdie@linux.intel.com Subject: [PATCH] drivers/video/backlight/da903x.c: introduce one more missing kfree One more error handling code should have ...

it's in tip/tracing/urgent: agreed, it's messy. At minimum the ordering is wrong: it should not return the iterator but 'ret' - the _iterator_ value can then be a side-effect (dependent on the return value being fine). the usage site clearly shows the problem: static int tracing_open(struct inode *inode, struct file *file) { int ret; __tracing_open(inode, file, &ret); return ret; } that could then be a simple: static int tracing_open(struct inode ...

-----BEGIN PGP SIGNED MESSAGE----- Just to confirm then, its not necessary if I'm not checking the driver data for the quirk? -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkkkHQEACgkQwuRiAT9o60+WHACg8N419HjyNxVeX/ozhRH7Wdrx J80AoLXsSnLSAidbNYM6BID5j6OMh4ny =dk3Z -----END PGP SIGNATURE----- --

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Added quirks for the N-Trig digitizer. Signed-off-by: Rafi Rubin <rafi@seas.upenn.edu> - --- drivers/hid/Kconfig | 7 ++++ drivers/hid/Makefile | 1 + drivers/hid/hid-core.c | 1 + drivers/hid/hid-dummy.c | 4 +- drivers/hid/hid-ids.h | 3 ++ drivers/hid/hid-ntrig.c | 82 +++++++++++++++++++++++++++++++++++++++++++++++ 6 files changed, 96 insertions(+), 2 deletions(-) create mode 100644 ...

Please keep the driver as simple as possible. Whenever there are any new devices in the future, we could easily extend it so that quirk data in Probe function is necesary, you need to parse the report descriptor (hid_parse()), activate the hardware (hid_hw_start()), etc. Thanks, -- Jiri Kosina SUSE Labs --

(top-posting repaired. Please don't top-post!) I haven't seen any patch which alters hpwdt_pretimeout() and there is no such patch in linux-next. Perhaps it got lost? --

This was the patch (see below). It's in the linux-2.6-watchdog-next tree now, so it should go into the linux-next tree soon. The other patch is in the linux-2.6-watchdog-next tree also. Kind regards, Wim. ---------------------------------------------------------- commit ed22ea64f9efe4531be5130c0f77130c2ad74130 Author: Bernhard Walle <bwalle@suse.de> Date: Sun Oct 26 15:59:37 2008 +0100 [WATCHDOG] hpwdt: Fix kdump when using hpwdt When the "hpwdt" module is loaded ...

Wasn't completely clear to me, but it would indeed be much cleaner. Hi Tom, What did you test exactly? Thanks in advance, Wim. --

I tested changing the return value to NOTIFY_OK always. -----Original Message----- From: Wim Van Sebroeck [mailto:wim@iguana.be] Sent: Wednesday, November 19, 2008 5:11 PM To: Bernhard Walle; Mingarelli, Thomas Cc: Andrew Morton; linux-kernel@vger.kernel.org; stable@kernel.org Subject: Re: [PATCH] [WATCHDOG] [hpwdt] Set the mapped BIOS address space as executable Wasn't completely clear to me, but it would indeed be much cleaner. Hi Tom, What did you test exactly? Thanks in ...

Seems like a X problem to me.... can you try to run X with fbdev Cc yourself, I always do group replies... Pavel -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html --

On Mon, 17 Nov 2008 16:16:06 +0000 32bit is large enough for dma segment boundary mask, I think. Since the popular value of the mask is 0xffffffff. So the above code (mask + 1 ?) works wrongly if the size of mask is 32bit (well, I guess that you talk about the dma_mask (and coherent_dma_mask) in struct device. The dma segment boundary mask represents the different I think that 'unsigned long' is better for the dma segment boundary mask since it represents the hardware limitation. The ...

Already fixed in 2.6.27.7-rc2. thanks, greg k-h --

On Thu, 13 Nov 2008 15:43:34 +0100 I'd prefer if include/linux/mmc could stay clean and just have core stuff since the header files are basically the reference documentation for the subsystem. But if you can't find any other way... --=20 -- Pierre Ossman WARNING: This correspondence is being monitored by the Swedish government. Make sure your server uses encryption for SMTP traffic and consider using PGP for end-to-end encryption.

Markus, On Wed, Nov 19, 2008 at 8:20 PM, stephane eranian To follow-up on this, the other issue with this code is that you should not decrease those two mm fields if the buffer was not allocated by ds.c. So I think, the modification I sent you in my patch is actually valid (just for another reason). --

I'm having a deja vu. We had this discussion before. You reported those issues and I fixed them. Same for the PEBS size; and Andi Kleen asked to exclude ds.c from the build instead of guarding the .c file and to use the mm semaphore in ds_allocate_buffer(). That thread ended in the multiplexing discussion and my fixes never got in. I'll send a patch covering those problems next week. Regarding access to the interrupt threshold, we never completed our discussion. If we look towards ...

Markus, You are right about the reserved field, it was missing from my code but that was harmless. I had to hack ds.c some more to make forward progress with PEBS. First of all my PEBS code is in a kernel module, so all PEBS functions have to be exported. Furhtermore, I need a ds_get_pebs_thres() and ds_set_pebs_thres() calls. But the one key problem is ds_validate_access(). I had to disable this function. The problem is that with perfmon there can be several threads who need access to ...

You are right that there are 9 fields. However, the 9th field is double the size of the other fields and the entire structure is padded. You're right about the PEBS size. I will send a patch some time this week. thanks, markus. --------------------------------------------------------------------- Intel GmbH Dornacher Strasse 1 85622 Feldkirchen/Muenchen Germany Sitz der Gesellschaft: Feldkirchen bei Muenchen Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes ...

The second removal of invalid key corrupt the eeprom pointer in this line if (!test_and_clear_bit(priv->stations[sta_id].sta.key.key_offset, &priv->ucode_key_table)) as discovered by Yi so this patch also fix the immediate failure We are just not sure in what flow the key is removed second time and maybe there is an other issue behind it. The full log will be appreciated but you've already helped a lot Thanks for your time --

Actually we (or maybe just me) gave up on those for now. We went back an forth on the 'system set' and what it supposed to mean. Both Paul J. and Paul M. were against the concept and especially backward compatibility with existing user-space tools that use cpusets. Plus it's really really easy to setup the 'system' set from user-space and I just ended up writing 'syspart' thing that I mentioned before. Similar thing happened to "managing irqs via cpusets" idea. Peter and I wanted to represent ...

What you described is almost exactly what I did in my original cpu isolation patch, which did get NAKed :). Basically I used global cpu_isolated_map and exposed 'isolated' bit, etc. I do not see how 'partfs' that you described would be different from 'cpusets' that we have now. Just ignore 'tasks' files in the cpusets and you already have your 'partfs'. You do _not_ have to use cpuset for assigning tasks if you do not want to. Just use them to define sets of cpus and keep all the tasks ...

Already did. It's all in mainline. The part you quoted was just pointing out that the original approach was not correct. Max --

[Empty message]

I do not see any benefits in exposing a special 'isolated' bit and have it do the same thing that the cpu hotplug already does. As I explained in other threads cpu hotplug is a _perfect_ fit for the isolation purposes. In order to isolate a CPU dynamically (ie at runtime) we need to flush pending work, flush chaches, move tasks and timers, etc. Which is _exactly_ what cpu hotplug code does when it brings CPU down. There is no point in reimplementing it. btw It sounds like you misunderstood the ...

Long time no see :-) Yes, there has been quite a bit of feedback on this patch, and using EV_SYN the way you suggest came up. Thanks - will redo. Henrik --

Hi Henrik, Sorry for the long silence. I don't think utilizing button events for this is a good idea. I'd rather just start reporting extended touch events for a finger and signal end of sub-packet with something like EV_SYN/SYN_MT_REPORT. -- Dmitry --

At Fri, 14 Nov 2008 10:57:10 -0800, AFAIK, tasklet_hi_schedule() uses another softirq vector with the highest priority (the first item to be scanned). Regarding the ALSA codes, there is no big reason to use hi_* version. Could you check the patch below and see whether you get any regressions? thanks, Takashi --- diff --git a/sound/core/rawmidi.c b/sound/core/rawmidi.c index 39672f6..002777b 100644 --- a/sound/core/rawmidi.c +++ b/sound/core/rawmidi.c @@ -151,7 +151,7 @@ static ...

If you look at bfq_update_hw_tag(), the logic introduced by the patch you mention is still there; BFQ starts with ->hw_tag = 1, and updates it every 32 valid samples. What changed WRT your patch, apart from the number of samples, is that the condition for a sample to be valid is: bfqd->rq_in_driver + bfqd->queued >= 5 while in your patch it is: cfqd->rq_queued > 5 || cfqd->rq_in_driver > 5 We preferred the first one because that sum better reflects the number of requests that ...

BFQ's tag detection logic is broken in the same way that CFQ's used to be. Explanation is in this patch: ============================x8============================ commit 45333d5a31296d0af886d94f1d08f128231cab8e Author: Aaron Carroll <aaronc@gelato.unsw.edu.au> Date: Tue Aug 26 15:52:36 2008 +0200 cfq-iosched: fix queue depth detection CFQ's detection of queueing devices assumes a non-queuing device and detects if the queue depth reaches a certain threshold. Under ...

Sorry, I forgot the patch... This seems to solve the problem with your workload here, does it work for you? [ The magic number would not appear in a definitive fix... ] --- diff --git a/block/bfq-iosched.c b/block/bfq-iosched.c index 83e90e9..e9b010f 100644 --- a/block/bfq-iosched.c +++ b/block/bfq-iosched.c @@ -1322,10 +1322,12 @@ static void bfq_update_io_seektime(struct bfq_data *bfqd, /* * Don't allow the seek distance to get too large from the - * odd fragment, pagein, ...

--

Thanks. Eric --

I see. I have already pushed the bits that ignore the mouse interface on the hardware in question in HID code. But that should be perfectly fine, Sure, no problem, I will keep an eye on incoming input patches before you are fully settled again, and if anything new urgent comes, I will merge it for .28. -- Jiri Kosina SUSE Labs --

These usb devices provide several different functions through the same interface. By default, it looks like a regular mouse interface, which No harm done. :-) Thanks, Henrik --

That I fully understand. But I though that the device is so much HID standard non-compliant, that even basic functionality is not possible with the generic HID driver, and therefore blacklisting it immediately can't do any harm. That's what I understood previously. But if this is wrong, and the HID code can get at least basic functionality from the device, I will revert the blacklist addition. Thanks, -- Jiri Kosina SUSE Labs --

Do you know if this was pre or post the root-domain code? Here is a reference to the commit: http://git.kernel.org/?p=3Dlinux/kernel/git/torvalds/linux-2.6.git;a=3Dco= mmit;h=3D57d885fea0da0e9541d7730a9e1dcf734981a173 A bisection that shows when this last worked for you would be very appreciated if you have the time, Dimitri. Regards, -Greg

I beleive that's the intended behaviour. We always put cpus that are not balanced into null sched domains. This was done since day one (ie when cpuisol= option was introduced) and cpusets just followed the same convention. I think the idea is that we want to make balancer a noop on those processors. We could change cpusets code to create a root sched domain for each cpu I guess. But can we maybe scale cpupri some other way ? Max --

I think a NULL sched domain, as it is now, is fine. --

Heh...well, as the guy that wrote root-domans, I can definitively say It sounds like the problem with my code is that "null sched domain" translates into "default root-domain" which is understandably unexpected by Dimitri (and myself). Really I intended root-domains to become associated with each exclusive/disjoint cpuset that is created. In a way, non-balanced/isolated cpus could be modeled as an exclusive cpuset with one member, but that is somewhat beyond the scope of the root-domain code ...

It was pre root-domain. That behavior was replaced by addition of the sched_load_balance flag with the following commit (though it was actually removed even earlier): --

On Sun, 16 Nov 2008 00:03:02 +0000 I don't see where knowledge about the card gets into the picture here. All of this sounds like information that is gathered from platform data and/or ios information. -- -- Pierre Ossman Linux kernel, MMC maintainer http://www.kernel.org rdesktop, core developer http://www.rdesktop.org WARNING: This correspondence is being monitored by the Swedish government. Make sure your server uses encryption for SMTP traffic and ...

On Mon, 17 Nov 2008 10:28:46 +0100 Any side-effects besides from dmesg noise? IOW should I push for .28? -- -- Pierre Ossman Linux kernel, MMC maintainer http://www.kernel.org rdesktop, core developer http://www.rdesktop.org WARNING: This correspondence is being monitored by the Swedish government. Make sure your server uses encryption for SMTP traffic and consider using PGP for end-to-end encryption. --

It needs task_lock(current). --

Sorry for delay in replying to this. And sorry for causing regression on some ppc platforms. I totally agree with what Dave said above. ALL_CPUS is a sane default, platform code has to sanity check masks passed via set_affinity() calls anyway. So I beleive it should be fixed in the platform code. Max --

This patch provides a driver for SGI RTC clocks and timers. This provides a high resolution clock and timer source using the SGI system-wide synchronized RTC clock/timer hardware. Signed-off-by: Dimitri Sivanich <sivanich@sgi.com> --- drivers/clocksource/Makefile | 1 drivers/clocksource/rtc_uv.c | 399 +++++++++++++++++++++++++++++++++++++++ drivers/misc/Kconfig | 9 kernel/time/clockevents.c | 1 kernel/time/clocksource.c | 1 5 files changed, 411 ...

Looking at the above, it seems that you're spending an inordinate amount of time in generic_match too. Could you see if the following patch helps? Cheers Trond ------------------------------------------------------------------------------- From: Trond Myklebust <Trond.Myklebust@netapp.com> SUNRPC: Fix up generic_match() Cut down on the number of similar copies of group_info. Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com> --- net/sunrpc/auth_generic.c | 20 ...

No, you don't want to :-). Pavel -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html --

Yeah, kexec is definitely a clue. My guess is that we got some sort of left over completion. Regardless of the status of this particular bug or not, I think it would be a good idea to add some checks for when a command is attempted removed from a queue it isn't currently on. -- Jens Axboe --

Sure. I have 2 patches. One is a fix for CCISS_DEBUG printk formats that I I don't know that the patch will tell us which call it is. The added code is inside removeQ() and addQ(), not near the calls to them. -- ~Randy

Done (separately). I need to back up a bit. Yesterday these BUGs happened consistenly, so I wondered why. Then I recalled that for debugging another bug/problem, I had changed the test system's normal boot kernel from 2.6.25 to 2.6.18-8. The test system is used to build and then boot the new kernel *via kexec*, so it's quite possible (or certain) that something in the kexec world has been fixed since 2.6.18. I don't recall seeing this problem lately when using 2.6.25 to kexec/boot the ...

I agree, I'll fix. -- mikem --

Hi, Andi Kleen wrote: > N

On Wed, Jul 25, 2007 at 12:03 AM, Duncan Sands Duncan, as you remember we discussed 1 year ago speed issues with Speedtouch driver. Please let me know if the driver has been fixed to support higher speeds? Or is it still limited to 3Mbaud top ? Regards, MK --