This cleans up the formatting in the vDSO linker script, mostly just the use of whitespace. It's intended to approximate the kernel standard conventions for indenting C, treating elements of the linker script about like initialized variable definitions. Signed-off-by: Roland McGrath <roland@redhat.com> CC: Sam Ravnborg <sam@ravnborg.org> --- arch/powerpc/kernel/vdso32/vdso32.lds.S | 219 +++++++++++++++++-------------- 1 files changed, 118 insertions(+), 101 deletions(-) diff --git ...

This cleans up the formatting in the vDSO linker script, mostly just the use of whitespace. It's intended to approximate the kernel standard conventions for indenting C, treating elements of the linker script about like initialized variable definitions. Signed-off-by: Roland McGrath <roland@redhat.com> CC: Sam Ravnborg <sam@ravnborg.org> --- arch/sh/kernel/vsyscall/vsyscall.lds.S | 77 +++++++++++++++++-------------- 1 files changed, 42 insertions(+), 35 deletions(-) diff --git ...

I am not subscribed to LKML, so please CC me in replies. I am reporting a regression when CONFIG_DEBUG_LOCKDEP is enabled in 2.6.23-git8. The error occurs immediately before loading init. Complete dmesg and kernel config are attached. [ 28.528074] VFS: Mounted root (ext3 filesystem) readonly. [ 28.528090] Freeing unused kernel memory: 212k freed [ 28.537431] Write protecting the kernel read-only data: 1036k [ 28.622874] WARNING: at kernel/lockdep.c:2658 check_flags() [ 28.625132] [ ...

This patch is a simple cleanup which should probably be applied to -mm (assuming I haven't messed it up). The next patch is an experimental patch which will require userspace support and is just RFC at this point. From 9fc0782de6e1287aaeebe8ad653b008f09b22c11 Mon Sep 17 00:00:00 2001 From: Serge E. Hallyn <serue@us.ibm.com> Date: Mon, 15 Oct 2007 17:33:24 -0400 Subject: [PATCH 1/2] capabilities: clean up file capability reading Simplify the vfs_cap_data structure. Also fix get_file_caps ...

This patch accomplishes the following goals: * kill the 'pci_enable_device ret val not checked' warning * eliminate the incorrect mucking with pci_dev::current_state via the following changes: * [minor bug fix] eliminate pci_set_power_state() call in resume, pci_enable_device() does so for us. * [bug fix] do not touch dev->current_state, pci_set_power_state() and other PCI layer functions manage this for us. * [minor bug fix, warning fix] check pci_enable_device() ret val in ...

From: Randy Dunlap <randy.dunlap@oracle.com> Fix filesystems docbook warnings. Warning(linux-2.6.23-git8//fs/debugfs/file.c:241): No description found for parameter 'name' Warning(linux-2.6.23-git8//fs/debugfs/file.c:241): No description found for parameter 'mode' Warning(linux-2.6.23-git8//fs/debugfs/file.c:241): No description found for parameter 'parent' Warning(linux-2.6.23-git8//fs/debugfs/file.c:241): No description found for parameter ...

From: Randy Dunlap <randy.dunlap@oracle.com> Fix kernel-api docbook warnings. Warning(linux-2.6.23-git8//drivers/message/fusion/mptscsih.c:2618): No description found for parameter 'sc' Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com> --- drivers/message/fusion/mptscsih.c | 10 +++------- 1 file changed, 3 insertions(+), 7 deletions(-) --- linux-2.6.23-git8.orig/drivers/message/fusion/mptscsih.c +++ linux-2.6.23-git8/drivers/message/fusion/mptscsih.c @@ -2605,14 +2605,10 @@ ...

Dave's tests were done with an early large blocksize patchset that had issues with readahead. More recent versions have the fixes by Fengguang that address the issue. -

It isn't that hard to add simple kset attributes, so don't go through all the gyrations of creating your own object type and show and store functions. Just use the functions that are already present. This makes things much simpler. Note, the version_str string violates the "one value per file" rule for sysfs. I suggest changing this now (individual files per type supported is one suggested way.) Cc: Michael A. Halcrow <mahalcro@us.ibm.com> Cc: Michael C. Thompson ...

> There is a /proc/sys/net/ipv4/ip_dynaddr sysctl in 2.6.21. Actually, it does look promising, thanks. Stefan -

this is the point of disagreement. the devices you can trivially enumerate can be handled easily and trivially, the ones that you can't may require more complex things to handle them, but that depends on the situation. If you only have one USB drive on a system you don't need to worry about what order USB hotplug events come in if you can just say 'the first USB drive'. mixing the different types of devices into one namespace complicates things in a couple of ways. 1. devices that used ...

Hi. I've noticed that with recent kernels (starting somewhere between 2.6.20 and 2.6.22) I sometimes get *lots* of disk activity on resume from suspend to ram. About 2/3 of the time, the system resumes normally but in the remaining 1/3 of the time, the hard drive light stays on almost solid and the machine is very, very slow to respond. The only way I've found to reliably recover from this is if I can get to a command prompt fast enough and do "shutdown -r now". There's not much cpu activity ...

Yup, they slipped through. I have fixups in my pile already. Will send them tomorrow when I'm actually awake. tglx -

That means it will only bother asking you if you've set EMBEDDED; otherwise its always on. J -

For the /proc/<pid>/pagemap code[1], we need to able to query how much virtual address space a particular task has. The trick is that we do it through /proc and can't use TASK_SIZE since it references "current" on some arches. The process opening the /proc file might be a 32-bit process opening a 64-bit process's pagemap file. x86_64 already has a TASK_SIZE_OF() macro: #define TASK_SIZE_OF(child) ((test_tsk_thread_flag(child, TIF_IA32)) ? IA32_PAGE_OFFSET : TASK_SIZE64) I'd like to ...

From: Fengguang Wu <wfg@mail.ustc.edu.cn> The "proportional set size" (PSS) of a process is the count of pages it has in memory, where each page is divided by the number of processes sharing it. So if a process has 1000 pages all to itself, and 1000 shared with one other process, its PSS will be 1500. - lwn.net: "ELC: How much memory are applications really using?" The PSS proposed by Matt Mackall is a very nice metic for measuring an process's memory footprint. So collect ...

Yeah, that looks like a pretty sane scheme. Do we want to be any more abstract about it? Perhaps instead of USER_SLAB, it should be USER_KERNEL_INTERNAL, or USER_KERNEL_USE. The slab itself is going away We could also Yeah, that looks like a pretty sane scheme. Do we want to be any more abstract about it? Perhaps instead of USER_SLAB, it should be USER_KERNEL_INTERNAL, or USER_KERNEL_USE. The slab itself is going away as we speak. For the bits that we want to export, we could also add ...

We're not. I just grabbed a random line to show the non-atomic accessors. Any actual one we'd need to add would be: #define __PageBuddy(page) __test_bit(PG_buddy, &(page)->flags) It looks like we don't have any of these non-atomic ones for plain __PageFoo(). So, we'd have to add them for each one that we wanted. Still not much work, and still satisfies the "grep test". :) -- Dave -

David, All of your comments looked pretty valid to me. I've refreshed that patch. I haven't even compile-tested this so there may be some fat fingering somewhere. I'll run compile tests on it now. -- Dave For the /proc/<pid>/pagemap code[1], we need to able to query how much virtual address space a particular task has. The trick is that we do it through /proc and can't use TASK_SIZE since it references "current" on some arches. The process opening the /proc file might be a 32-bit ...

This patchset is version 3 of my /proc/pid/pagemaps code. Rather than submit about 30 incremental patches atop an existing 20 or so where many of the intermediate states are broken and get undone anyway, I've respun this as a much smaller set of 11 patches. Changes in this series: - headers gone again (as recommended by Dave Hansen and Alan Cox) - 64-bit entries (as per discussion with Andi Kleen) - swap pte information exported (from Dave Hansen) - page walker callback for holes (from ...

Yeah, there are a bunch of flags that aren't mutually exclusive and we Perhaps we need something like: flags = page->flags; userflags = FLAG_BIT(USER_REFERENCED, flags & PG_referenced) | ... etc. for the flags we want to export. This will let us change to FLAG_BIT(USER_SLAB, PageSlab(page)) | if we make a virtual slab bit. And it shows up in grep. Unfortunately, i386 test_bit is an asm inline and not a macro so we can't hope for the compiler to fold up a bunch of ...

From: Matt Mackall <mpm@selenic.com> This pulls the shared map display code out of show_map and puts it in show_smap where it belongs. Signed-off-by: Matt Mackall <mpm@selenic.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Index: l/fs/proc/task_mmu.c =================================================================== --- l.orig/fs/proc/task_mmu.c 2007-10-14 13:37:08.000000000 -0500 +++ ...

From: Matt Mackall <mpm@selenic.com> This makes physical page map counts available to userspace. Together with /proc/pid/pagemap and /proc/pid/clear_refs, this can be used to monitor memory usage on a per-page basis. [bunk@stusta.de: make struct proc_kpagemap static] Signed-off-by: Matt Mackall <mpm@selenic.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton ...

I think that's overkill for something that has exactly one use of each. -- Mathematics is the supreme nostalgia of our time. -

There's no overkill at all, the current uses are already accessed with these bitshifts so there's no overhead when using an inlined function instead. To correctly access the pss, these bitshifts are required because the decision was made to use the lower PSS_DIV_BITS for rounding. Thus, you need to include accessor functions so that they are always accessed correctly now and in the future. David -

Use the generic pagewalker for smaps and clear_refs Signed-off-by: Matt Mackall <mpm@selenic.com> Index: l/fs/proc/task_mmu.c =================================================================== --- l.orig/fs/proc/task_mmu.c 2007-10-14 13:36:56.000000000 -0500 +++ l/fs/proc/task_mmu.c 2007-10-14 13:37:08.000000000 -0500 @@ -116,6 +116,7 @@ static void pad_len_spaces(struct seq_fi struct mem_size_stats { + struct vm_area_struct *vma; unsigned long resident; unsigned long ...

Make /proc/ page monitoring configurable This puts the following files under an embedded config option: /proc/pid/clear_refs /proc/pid/smaps /proc/pid/pagemap /proc/kpagecount /proc/kpageflags Signed-off-by: Matt Mackall <mpm@selenic.com> Index: l/fs/proc/base.c =================================================================== --- l.orig/fs/proc/base.c 2007-10-15 17:18:09.000000000 -0500 +++ l/fs/proc/base.c 2007-10-15 17:18:16.000000000 -0500 @@ -2031,7 +2031,7 @@ static const ...

From: Matt Mackall <mpm@selenic.com> This puts all the clear_refs code where it belongs and probably lets things compile on MMU-less systems as well. Signed-off-by: Matt Mackall <mpm@selenic.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Index: l/fs/proc/base.c =================================================================== --- l.orig/fs/proc/base.c 2007-10-14 13:35:08.000000000 ...

From: Matt Mackall <mpm@selenic.com> Reorder source so that all the code and data for each interface is together. Signed-off-by: Matt Mackall <mpm@selenic.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Index: l/fs/proc/task_mmu.c =================================================================== --- l.orig/fs/proc/task_mmu.c 2007-10-14 13:42:11.000000000 -0500 +++ ...

This one makes me worry a little bit. Are we sure that this won't expose a wee bit too much to userspace? I can see it making sense to clear the page refs, then inspect whether the page has been referenced again. But, I worry that people are going to start doing things like read NUMA, SPARSEMEM, or other internal information out of these. I've seen quite a few patches lately that do creative things with these *cough*clameter*cough*, and I worry that they're too fluid to get exposed to ...

Multicalls are expected to never fail, and the normal response to a failed multicall is very terse. In the interests of better debuggability, add some more verbose output. It may be worth turning this off once it all seems more tested. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> --- arch/x86/xen/multicalls.c | 25 +++++++++++++++++++++++++ 1 file changed, 25 insertions(+) =================================================================== --- ...

This adds a mechanism to register a callback function to be called once a batch of hypercalls has been issued. This is typically used to unlock things which must remain locked until the hypercall has taken place. [ Stable folks: pre-req for 2.6.23 bugfix "xen: deal with stale cr3 values when unpinning pagetables" ] Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Cc: Stable Kernel <stable@kernel.org> --- arch/x86/xen/multicalls.c | 29 ++++++++++++++++++++++++++--- ...

Crap, sorry about that. I'd just intended to send the overview and the four patches, which have explicit CC: stable lines in the patch descriptions. They are: Subject: [PATCH 06/12] xen: add batch completion callbacks Subject: [PATCH 07/12] xen: deal with stale cr3 values when unpinning pagetables Subject: [PATCH 10/12] xen: fix incorrect vcpu_register_vcpu_info hypercall argument Subject: [PATCH 12/12] xfs: eagerly remove vmap mappings to avoid upsetting Xen J -

Should be done unconditionally. One could remap the underlying physical space to include an MMIO region, and speculative reads from the cacheable virtual mapping of that region could move the robot arm, destroying the world. Zach -

When a pagetable is created, it is made globally visible in the rmap prio tree before it is pinned via arch_dup_mmap(), and remains in the rmap tree while it is unpinned with arch_exit_mmap(). This means that other CPUs may race with the pinning/unpinning process, and see a pte between when it gets marked RO and actually pinned, causing any pte updates to fail with write-protect faults. As a result, all pte pages must be properly locked, and only unlocked once the pinning/unpinning process ...

The conversion from using a slab cache to quicklist left some residual dead code. I note that in the conversion it now always allocates a whole page for the pgd, rather than the 32 bytes needed for a PAE pgd. Was this intended? Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Cc: Christoph Lameter <clameter@sgi.com> Cc: Andi Kleen <ak@suse.de> Cc: William Lee Irwin III <wli@holomorphy.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Thomas Gleixner <tglx@linutronix.de> --- ...

XFS leaves stray mappings around when it vmaps memory to make it virtually contigious. This upsets Xen if one of those pages is being recycled into a pagetable, since it finds an extra writable mapping of the page. This patch solves the problem in a brute force way, by making XFS always eagerly unmap its mappings. David Chinner says this shouldn't have any performance impact on filesystems with default block sizes; it will only affect filesystems with large block sizes. Signed-off-by: ...

This patch cleans up duplicate includes in arch/i386/xen/ Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com> Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> --- arch/x86/xen/enlighten.c | 1 - arch/x86/xen/mmu.c | 2 -- 2 files changed, 3 deletions(-) =================================================================== --- a/arch/x86/xen/enlighten.c +++ b/arch/x86/xen/enlighten.c @@ -25,7 +25,6 @@ #include <linux/mm.h> #include <linux/page-flags.h> #include ...

When sending a call-function IPI to a vcpu, yield if the vcpu isn't running. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> --- arch/x86/xen/smp.c | 14 ++++++++++---- arch/x86/xen/time.c | 6 ++++++ arch/x86/xen/xen-ops.h | 2 ++ 3 files changed, 18 insertions(+), 4 deletions(-) =================================================================== --- a/arch/x86/xen/smp.c +++ b/arch/x86/xen/smp.c @@ -360,7 +360,8 @@ int xen_smp_call_function_mask(cpumask_t ...

This patch refactors the paravirt_ops structure into groups of functionally related ops: pv_info - random info, rather than function entrypoints pv_init_ops - functions used at boot time (some for module_init too) pv_misc_ops - lazy mode, which didn't fit well anywhere else pv_time_ops - time-related functions pv_cpu_ops - various privileged instruction ops pv_irq_ops - operations for managing interrupt state pv_apic_ops - APIC operations pv_mmu_ops - operations for managing ...

PREFETCHNTA. -hpa -

I know about unordered stores (movnti & similar) --- they basically use write-combining method on memory that is normally write-back --- and they need sfence. But which one instruction does unordered load and needs lefence? Mikulas -

Also, for non-wb memory. I don't think the Intel document referenced says anything about this, but the AMD document says that loads can pass loads (page 8, rule b). This is why our rmb() is still an lfence. -

udev - persistent network rules in /etc/udev/rules.d/ -- Ondrej Zary -

Actually, I'm pretty happy sometimes with the current behavior: if the interface goes down and back up with the same AP within a short enough time, it typically gets the same IP and the router's NAT table still has the TCP connection live and things "just work". So I'd want to kill the connections not when the interface goes down, but in comes back up with a different IP. Stefan -

[ I suppose this is not the best place to ask this, but comp.os.linux.networking couldn't come up with a good answer and I can't think of any intermediate step between these two groups ;-( ] I'd like (as root, obviously) to kill some of the TCP connections visible in netstat. I've found `tcpkill' and `cutter' but `cutter' only kills TCP connections that go *though* the machine (in my case, the machine is not a router, so there aren't any such thu connections anyway) and `tcpkill' can only ...

It seemed to be doing the right thing here on an X31 (had to give it back to its owner; will retrieve and test again tomorrow to make sure I don't know that having another mixer is really the right answer. You want to have the buttons/hardware volume matching (and showing) on-screen changes to the system volume. Otherwise, things are just Having the volume keys reported makes it so that we get the on-screen display of the volume changing and /proc/acpi/ibm/volume matches what the ...

NAK. It is the completely wrong thing to do for IBM thinkpads which process volume and brightness completely in firmware. And the input subsystem maintainer has made it extremely clear in various threads that the input devices are *not* to be used as a notification service for on-screen-display or other such stuff. If you send volume and brightness *key* events to userspace, it is supposed to act on them and raise/lower brightness/volume, which is the wrong thing to do on thinkpads. Never ...

No, on Lenovo (and in general actually) the firmware should *not* touch the backlight. Otherwise if another driver touches it the driver and firmware will be out of sync, causing unexpected and undesirable behavior. We intend to fix this for the Intel driver at least (requiring both ACPI video driver and gfx driver updates), others will probably follow eventually. Jesse -

This is fixed in any release based on 2.6.16.31 or later. The relevant mainline patch is 1a047060a99f274a7c52cfea8159e4142a14b8a7 as below. So update your kernel package. NeilBrown commit 1a047060a99f274a7c52cfea8159e4142a14b8a7 Author: NeilBrown <neilb@suse.de> Date: Thu Oct 19 23:29:13 2006 -0700 [PATCH] knfsd: fix race that can disable NFS server This patch is suitable for just about any 2.6 kernel. It should go in 2.6.19 and 2.6.18.2 and possible even ...

The MIPS stuff was ACK'd by Ralf. Please pull from 'upstream-linus' branch of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6.git upstream-linus to receive the following updates: Documentation/networking/bonding.txt | 33 + arch/mips/au1000/common/prom.c | 61 +- arch/mips/au1000/common/setup.c | 5 +- arch/mips/au1000/db1x00/init.c | 10 +- arch/mips/au1000/mtx-1/init.c | 6 +- ...

This legacy define from the old buffer code is now only used in a single power pc driver than doesn't compile anyway. Signed-off-by: Alan Cox <alan@redhat.com> diff -u --exclude-from /usr/src/exclude --new-file --recursive linux.vanilla-2.6.23-mm1/include/linux/tty.h linux-2.6.23-mm1/include/linux/tty.h --- linux.vanilla-2.6.23-mm1/include/linux/tty.h 2007-10-15 15:03:38.000000000 +0100 +++ linux-2.6.23-mm1/include/linux/tty.h 2007-10-15 15:28:46.000000000 +0100 @@ -52,13 +52,6 @@ */ ...

Atari keyboard: incorporate additional review comments: o Kill reference to source file name o Return error value from input_register_device() instead of -ENOMEM Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com> Cc: Michael Schmitz <schmitz@biophys.uni-duesseldorf.de> --- arch/m68k/atari/atakeyb.c | 2 -- drivers/input/keyboard/atakbd.c | 7 ++++--- 2 files changed, 4 insertions(+), 5 deletions(-) --- ...

Use git log -p --follow arch/x86/kernel/io_apic_64.c where the "--follow" tells git to follow renames. And, of course, "git blame -C" will follow renames and copying of code across file boundaries too. NOTE! In both cases you may actually have to tell git to not limit its rename detection when it sees lots of files. You can do that once-and-for-all with git config --global diff.renamelimit 0 which should take care of it (although it seems that due to unlucky timing, the ...

Hi all, I'm trying to debug a weird problem with nfsd on a 2.6.16.27-0.6-smp kernel. 1 server: SuSE SLES 10 x86_64, config attached 256 clients: RHEL4 Update 4 2.6.9-42.ELsmp x86_64 Using nfs v3. The clients have been happily talking to the server for several days without incident. The weird thing is that at a certain point the socket opened on port 2049 on the NFS server is being closed for unknown reasons (or better for unknown reasons for me!). With unknown reasons I mean that I ...

Thomas, thanks for the quick reply. Here is the /etc/exports (all clients are accessing the same mountpoint): node0101:~ # cat /etc/exports # See the exports(5) manpage for a description of the syntax of this file. # This file contains a list of all directories that are to be exported to # other computers via NFS (Network File System). # This file used by rpc.nfsd and rpc.mountd. See their manpages for details # on how make changes in this file effective. /eni01 ...

On Mon, 15 Oct 2007 20:52:50 +0200 You may find the following useful http://en.wikipedia.org/Wiki/Copyright Please don't ask on this list for people to abuse copyright law. Alan -

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>

Looks fine to me. Acked-by: Matt Mackall <mpm@selenic.com> -- Mathematics is the supreme nostalgia of our time. -

Looks great. Acked-by: Matt Mackall <mpm@selenic.com> -- Mathematics is the supreme nostalgia of our time. -

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>

Re-Ping? This patch works (it has been working for months), is ACK'd by me, addresses some of the feedback given in August, and has not received any comments despite the SCSI maintainers being To'd or CC'd for weeks (months, if you include the older thread). Further, this patch is required to round out interrupt-driven media change notification, that which eliminates the CD-ROM polling that userspace has done since the dawn of time. It was not in the merge window SCSI push, while ...

On Mon, 15 Oct 2007 14:05:50 -0400 (EDT) Hi, Your patch looks quite clean in general; however it does make me wonder if it should either leverage or expand the arch/x86/pci/pcbios.c infrastructure for doing BIOS32 calls and share that.... it's kinda unpleasant as a general thought to have drivers poke this deep into various guts of the system/bios.... esp if the common code has to do something very similar already. Greetings, Arjan van de Ven -

Hp is providing a Hardware WatchDog Timer driver that will only work with the specific HW Timer located in the HP ProLiant iLO 2 ASIC. The iLO 2 HW Timer will generate a Non-maskable Interrupt (NMI) 9 seconds before physically resetting the server, by removing power, so that the event can be logged to the HP Integrated Management Log (IML), a Non-Volatile Random Access Memory (NVRAM). The logging of the event is performed using the HP ProLiant ROM via an Industry Standard access known as a BIOS ...

On Mon, 15 Oct 2007 10:46:47 -0700 Seems a pretty fundamental change which could do with some careful benchmarking, methinks. See, your patch amounts to "do more seeks to improve one test case". Might be worth a code comment? -

On Mon, 15 Oct 2007 16:13:15 -0400 Gee. Expect the unexpected ;) One problem might be when kjournald is doing its ordered-mode data writeback at the start of commit. That writeout will now be higher-priority and might impact other tasks which are doing synchronous file overwrites (ie: no commits) or O_DIRECT reads or writes or just plain old reads. If the aggregate number of seeks over the long term is the same as before then of course the overall throughput should be the same, in which ...

There is a test in tgafb_mono_imageblit() for a colour image with a fall-back to cfb_imageblit(). The test is not necessary as the only caller, which is tgafb_imageblit(), checks it too and only invokes this function for monochrome images. It looks like a left-over from before some changes to tgafb_imageblit(). Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> --- Checked with checkpatch.pl and at the run time with an 8-bit TC frame buffer on a MIPS host. Please apply. ...

The pixel mask calculation in tgafb_mono_imageblit() uses a variable left-shift on a 32-bit data type by up to 32. Shifting by the width of a data type or more produces an unpredictable result according to the C standard. Rather than widening the data type this fix makes sure the count is between 0 and 31. The reason is not to penalise 32-bit platforms with operation on a "long long" type for a marginal case that is meant not to happen (blitting an image of a zero ...

Linus, please pull the lockdep tree from: git://git.kernel.org/pub/scm/linux/kernel/git/peterz/linux-2.6-lockdep.git v2.6.24-lockdep Ingo suggested to add me to the MAINTAINERS file for lockdep/lockstat. The rest of the pull contains a few fixes, a few annotations and a new 'feature'. The new thing is ensuring we don't hold any locks when returning to userspace. This will give a much earlier warning about imbalanced lock acquisitions. Peter --- Gregory Haskins (1): lockdep: ...

ad7877_read_adc, ad7877_read and ad7877_write are just used by the sysfs hooks. Touchscreen samples are read by the kthread using a different message struct. So far each sysfs invocation got its own storage for the spi message, which then is handed over to the SPI bus driver. The SPI bus driver serializes transfers in a kthread. Two different processes could access the drivers sysfs hooks. Using one ser_req per touch screen could require additional locking? Sure - I guess this was ...

oProfile Driver Module Patch for AMD Family10h Kernel 2.6.23 This patch is for controlling the upper 32bits of the event ctrl msrs. This includes the upper 4 bits of the event select and the Guest Only and Host Only bits This patch is necessary to make Event Based Profiling work reliably on a Family10h processor --- op_model_athlon.c | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-) --- Signed-off-by: Barry Kasindorf ...

heh. Incidentally i was thinking about using KVM for automated testing. Important pieces of hardware should get an in-KVM simulator/emulator, that way developers who do not own that hardware can do functionality testing too. So basically the highest-quality drivers would have an "inverse driver" in KVM, which simulates the hardware. (that model is evidently useful to the hardware maker even for new hardware: it can then also be used to test the Linux compatibility and Linux performance ...

Heh. I do agree that you likely find bugs, even if quite often it's exactly because the behaviour is something that will never happen on real hardware. But failure testing is very useful - I forget who it was who debugged some driver by taking a CD and just scrathing it mercilessly to induce read errors ;) Having a really *bad* HW emulator can certainly work that way too, even if it also would probably end up hitting just a few of the potential error paths.. Linus -

Using emulators to test device drivers is almost certain to be pointless. The problem with device drivers tends to be timing issues, odd hardware interactions, and lots of strange (and sometimes undocumented) behaviour and dependencies (eg things like "you have to wait 50us after setting the reset bit until the hardware has actually reset"). These are all things that you'd generally not catch in emulation - because the emulation by necessity is only going to be a very weak picture of ...

something like that wont enable 100% coverage (or even reasonable coverage for most hardware), so it's no replacement for actual hard testing, but it could push out the domain of minimally tested code quite a bit and increase the quality of the kernel. Races are always tough and so are bugs on the side of the hardware, but it's the silly boot-time crash showstoppers and "device does not work anymore" mistakes that causes us to lose most of the testers and early adopters. I'm not ...

the patch below fixes a bootup-crash bug merged via today's SCSI git merge: commit df3d80f5a5c74168be42788364d13cf6c83c7b9c Merge: 3d06f7a... c8e91b0... Author: Linus Torvalds <torvalds@woody.linux-foundation.org> Date: Mon Oct 15 08:19:33 2007 -0700 Merge master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6 ---------------------> Subject: scsi: fix crash in gdth_timeout() From: Ingo Molnar <mingo@elte.hu> fix the following bootup crash in gdth_timeout(): [ ...

Indeed. Maybe this is a better fix? That driver is pretty messy, and this should have been found ealier. James? Boaz? Linus --- drivers/scsi/gdth.c | 4 ++++ 1 files changed, 4 insertions(+), 0 deletions(-) diff --git a/drivers/scsi/gdth.c b/drivers/scsi/gdth.c index e8010a7..3ac080e 100644 --- a/drivers/scsi/gdth.c +++ b/drivers/scsi/gdth.c @@ -5213,6 +5213,10 @@ static int __init gdth_init(void) #endif /* CONFIG_PCI */ TRACE2(("gdth_detect() %d controller ...

phantom, synchronize_irq() on suspend Wait after disabling device's interrupt until the handler finishes its work if still in progress. Signed-off-by: Jiri Slaby <jirislaby@gmail.com> --- commit 7e792ef384190b517f2fb27cd0237fa30dbe0775 tree 17b15e5ab7c90eef0e7ae57e532839e81b831d58 parent 5c008a5651ee92ebe020dd5108a66a7db74fe41d author Jiri Slaby <jirislaby@gmail.com> Mon, 15 Oct 2007 15:52:21 +0200 committer Jiri Slaby <jirislaby@gmail.com> Mon, 15 Oct 2007 15:52:21 +0200 ...

phantom, improved data passing this new version guarantees amb_bit switch in small enough intervals, so that the device won't stop working in the middle of a movement anymore. However it preserves old (openhaptics) functionality. Signed-off-by: Jiri Slaby <jirislaby@gmail.com> --- commit e820ffda8d5f6c8a1d463955f6df7cf18544ca92 tree bba23481215cfcb6f3b7d6fec1d57d6902dc1098 parent 931e139b87d4fdb9fcae6a39cc0a157ff2fc07e1 author Jiri Slaby <jirislaby@gmail.com> Mon, 15 Oct 2007 16:03:38 ...

rocket, fix dynamic_dev tty - register_device unconditionally (non-pci dependent) to have also isa devices in /dev - unregister devices on module removal - don't set TTY_DRIVER_DYNAMIC_DEV twice (removed the one dependent on some macro) Signed-off-by: Jiri Slaby <jirislaby@gmail.com> Cc: Ferenc Wagner <wferi@niif.hu> --- commit afb1b8bbdc2cc13a8c2c7d9fff49a3098a370971 tree 49f12cbadea0b74968a7626301028fb0ccb31209 parent 4b741d759e5b898bf1bf19631d5e5b14a221ce52 author Jiri Slaby ...

rocket, don't re-set statics to 0 Signed-off-by: Jiri Slaby <jirislaby@gmail.com> --- commit ad37fddef12ce908078883fdc27297216c6d122e tree bf07d0f19abdc4220b1888e5385023c936fe3028 parent afb1b8bbdc2cc13a8c2c7d9fff49a3098a370971 author Jiri Slaby <jirislaby@gmail.com> Mon, 15 Oct 2007 13:53:01 +0200 committer Jiri Slaby <jirislaby@gmail.com> Mon, 15 Oct 2007 13:53:01 +0200 drivers/char/rocket.c | 12 ------------ 1 files changed, 0 insertions(+), 12 deletions(-) diff --git ...

rocket, remove potential leak in module_init if (controller && !request_region) then we leaked a tty driver struct, fix it by adding function deinit tail with goto-ing into it (and from other fail paths too) Signed-off-by: Jiri Slaby <jirislaby@gmail.com> --- commit 8aa70f167a6ae0f80af0d73a5365c81f2e146709 tree 3f44eb6baabbf501738c857b6114010ac408108e parent 2b2c0385d705ce1f4207595f8fda12c25ee39de3 author Jiri Slaby <jirislaby@gmail.com> Mon, 15 Oct 2007 15:13:01 +0200 committer Jiri ...

Ingo - this should go via netdev@vger.kernel.org.nz or dccp@vger.kernel.org.nz so it gets picked up by a maintainer. Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> -

Coverity (CID 1929) spotted the following: if a transfer buffer allocation fails, the last allocated urb is leaked (it hasn't been stored in dev->urb[] yet so em28xx_uninit_isoc misses it). The patch also includes a small typo fix. Signed-off-by: Florin Malita <fmalita@gmail.com> --- drivers/media/video/em28xx/em28xx-core.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/media/video/em28xx/em28xx-core.c b/drivers/media/video/em28xx/em28xx-core.c index ...

Subject: [PATCH try #3] Input/Joystick Driver: add support AD7142 joystick driver [try #2] Changelog: - Coding style issues fixed, passed checkpatch.pl - Kill uselss "ad7142_used" - Move request_irq to probe - Move i2c_check_functionality to probe - Error handling added [try #3] Changelog: - Use kzalloc - Use strlcpy - Use completion instead of wait queue - Move input_allocate_device and input_register_device to ad7142_probe() - Fix some issues pointed by LKML Cc: Andrey ...

Yes, I agree with you now, although I have a little concern about the Yes, no need input_unregister_device() here. Thanks a lot for you kindly review. I will resend update patch later. Best Regards, -Bryan Wu -

Hi Bryan, No, this is not going to work well: - you at least need to reinitialize the completion before enabling IRQ, otherwise you will spin in a very tight loop - if noone would touch the joystick ad7142_clsoe would() block infinitely because noone would signal the completion and ad7142_thread() would never stop. Completion is just not a good abstraction here... Please use work Don't you need to write something over i2c to tell the device to shut down? As it is now I expect the ...

On Mon, 15 Oct 2007 16:17:23 +0200 Did Paul Jackson's crash get fixed? -

so i dropped them and re-pushed. New shortlog below. Ingo ------------------> Alexey Dobriyan (1): sched: uninline scheduler Andi Kleen (4): sched: cleanup: remove unnecessary gotos sched: cleanup: refactor common code of sleep_on / wait_for_completion sched: cleanup: refactor normalize_rt_tasks sched: remove stale comment from sched_group_set_shares() Arjan van de Ven (1): Make scheduler debug file operations const Dhaval Giani (1): ...

Nice work... However it's a pity all the balancing stuff got wildly changed in 2.6.23 and then somewhat changed back again now. Despite appearances, a lot of those things weren't actually *completely* arbitrary values. I fear that it will make finding performance regressions harder than it should have... Anyway. -

YOU 'VE WON. You have won $655,000.00 Dollars in DE EUROMILLIONES ONLINE INT.LOTTERY SPAIN.For further development for Clearification and procedure please Contact,Dr.Pauly Anderson.E-mail:(sabadelltrust01@yahoo.es) TEL: +34 692 486 797. (1) Tic Nr: 6460DGH (2) Sr Nr: 0909AOB09 (3) LU Nr: 726726XZJHN (4) BT Nr: 2GH267XZZ1-5-42 (5) RF Nr 9527BCV-33-7-7-7 Regards. Mss.Sodic Lasy -

This patchset includes two patches. 1. Checks for the MCA fetures as early as possible and signedness fixup. 2. Minor coding style cleanup. -

remove one indent level Signed-off-by: Christoph Egger <Christoph.Egger@amd.com> Signed-off-by: Joerg Roedel <joerg.roedel@amd.com> --- arch/x86/kernel/cpu/mcheck/mce.c | 40 +++++++++++++++++++------------------- 1 files changed, 20 insertions(+), 20 deletions(-) diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c index c7246cc..e418c2f 100644 --- a/arch/x86/kernel/cpu/mcheck/mce.c +++ b/arch/x86/kernel/cpu/mcheck/mce.c @@ -45,26 +45,26 @@ void ...

Yes. But when we change legacy_init_iomem_resources(), then we should also change efi_initialize_iomem_resources(). But that's declared in <linux/efi.h> and so a change in ia64 code is required which I wanted to avoid. But that patch is for review of the idea. If nobody has objections, then I'll implement the IA64 change anyway -- and then the 3rd parameter can be added. Thanks, Bernhard -

This patch adds the BSS to the resource tree just as kernel text and kernel data are in the resource tree. The main reason behind this is to avoid crashkernel reservation in that area. Signed-off-by: Bernhard Walle <bwalle@suse.de> --- arch/x86/kernel/e820_32.c | 8 ++++++++ arch/x86/kernel/e820_64.c | 3 ++- arch/x86/kernel/efi_32.c | 3 +++ arch/x86/kernel/setup_32.c | 4 ++++ arch/x86/kernel/setup_64.c | 9 +++++++++ 5 files changed, 26 insertions(+), 1 ...

From: Ingo Molnar <mingo@elte.hu> I touched upon this in another reply. For forcedeth it should never be zero, it gets set in netif_napi_add() to RX_WORK_PER_LOOP which is unconditionally defined to 64. Afterwards napi->weight should never be modified and that is what is passed into n->poll(). Perhaps some memory scribble is causing it to be zero'd. -

Though it wasn't clear, that's an ACK I'll push the patch if nobody else beats me to it. -

From: Ingo Molnar <mingo@elte.hu> No, not in this case. The driver must only netif_rx_complete() That might help. at least once, perhaps budget is being passed erroneously in as zero? -

From: Ingo Molnar <mingo@elte.hu> Excellent catch: Acked-by: David S. Miller <davem@davemloft.net> -

From: Ingo Molnar <mingo@elte.hu> This is a driver bug, the work "budget" passed into a driver's ->poll() handler should never be exceeded. That's what this warning assertion is checking. What ethernet card is in your system and what driver is being used to drive it? -

ok, i've added such a patch. looking at the dev.c code - can napi_struct->weight be zero legitimately? If yes then the 0 gets passed to the driver and the driver would return 1 - violating the assertion. Ingo -

Two comments: 1) we have a vague definition of "RX work processed." Due to error conditions and goto's in that function, rx_processed_cnt may or may not equal the number of packets actually processed. 2) man I dislike these inline C statement combinations (ranting at original code style, not you). I would much rather waste a few extra lines of source code and make the conditions obvious: while (... && (rx_processed_cnt < limit)) { rx_processed_cnt++; ... } or ...

This is where we disagree. The existence of devices you cannot stably enumerate does not eliminate the existence of devices you trivially can. Pulling out the "IBM numa cluster with multiple SAS enclosures _and_ firewire" infrastructure to find the root partition on my hard drive may be good for the IBM numa clusters, but only at the expense of complicating this part of my laptop's infrastructure by an order of magnitude, and making embedded systems nearly impossible to put together. ...

And we are telling you that, in a modern hotplug world -- yes even on a laptop -- you are clinging too much to assumptions that were never 100% true in the first place, and much less so on today's laptops. When you can unplug a SATA drive from a laptop, and plug it back in via USB, you can see how unwise it is to hardcode device names into your fstab. We invented udev, sysfs, mount-by-label, mount-by-uuid, LVM and all sorts of other gadgets to make this problem go away. If you ...

randconfig testing of latest -git found this build bug: drivers/media/radio/radio-gemtek.c:557: error: 'VID_HARDWARE_GEMTEK' undeclared here (not in a function) config attached. Ingo

Problem: # CONFIG_VIDEO_V4L1_COMPAT is not set The VID_HARDWARE_* defines are only created if that is turned on. gemtek should depend on VIDEO_V4L1_COMPAT. *or* As far as I can tell, these VID_HARDWARE defines are entirely useless. I couldn't find any instance of them being used for anything other than setting the .hardware field of struct video_device, and I was unable to find any instance of that field being used. It looks like the whole thing is just unused cruft. Mauro, would ...

Remove obsolete V4L v1 reference. Signed-off-by: Pekka Seppänen <pexu@kapsi.fi> --- .hardware = VID_HARDWARE_GEMTEK was from pre-V4L2 era module. My bad, sorry folks. diff -up v4l-dvb/linux/drivers/media/radio/radio-gemtek.c{.orig,} --- v4l-dvb/linux/drivers/media/radio/radio-gemtek.c.orig 2007-10-15 15:52:38.000000000 +0300 +++ v4l-dvb/linux/drivers/media/radio/radio-gemtek.c 2007-10-15 15:52:53.000000000 +0300 @@ -555,7 +555,6 @@ static struct video_device gemtek_radio ...

Using the port of 2.4 code from Vitaly Bordug <vitb@kernel.crashing.org> and the actual algorithm used by the i2c driver of the DBox code on cvs.tuxboc.org from Tmbinc, Gillem (htoa@gmx.net). Renamed i2c-algo-8xx.c to i2c-algo-cpm.c and i2c-rpx.c to i2c-8xx.c. Added original i2c-rpx.c as i2c-8xx-ppc.c for pre-OF (arch ppc) devices. Signed-off-by: Jochen Friedrich <jochen@scram.de> --- This can be pulled from git://git.bocc.de/dbox2.git i2c arch/powerpc/boot/dts/mpc885ads.dts | ...

And so am I. I wouldn't change it, maybe they will contact you some later ;). Anyway the driver faces another problem with pci refcounting (it doesn't increment the counter) and the device might be removed from it any time the pci bus code decides. Are you willing to test to-pci-probing patches (i.e. patches which converts the driver to the model introduced in linux 2.4)? If not, I'll only increment the counter on modprobe and decrement it on rmmod, since it would be a safe (in the meaning ...

You are the man, Jiri! That indeed fixes the problem. Thank you very much. What a turnaround... Shouldn't Documentation/rocket.txt also be changed? I'm not sure whether that support email is still valid. -- Regards, Feri. -

From: Rob Landley <rob@landley.net> Add Documentation/block/00-INDEX Signed-off-by: Rob Landley <rob@landley.net> --- By the way, shouldn't request.txt be in htmldocs somewhere? Documentation/block/00-INDEX | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+) --- /dev/null 2007-04-23 10:59:00.000000000 -0500 +++ hg/Documentation/block/00-INDEX 2007-10-15 04:32:49.000000000 -0500 @@ -0,0 +1,20 @@ +00-INDEX + - This file +as-iosched.txt + - Anticipatory IO ...

I have not observed this case but it is possible to get a dirty page cache with clean buffer heads if we get a clean ramdisk page with buffer heads generated by a filesystem calling __getblk and then write to that page from user space through the block device. Then we just need to hit the proper window and try_to_free_buffers() will mark that page clean and eventually drop it. Ouch! To fix this use the generic __set_page_dirty_buffers in the ramdisk code so that when we mark a page dirty we ...

make_page_uptodate() is most hideous part I have run into. It has to know details about other layers to now what not to stomp. I think my incorrect simplification of this is what messed Cool. Triple buffering :) Although I guess that would only apply to metadata these days. Having a separate store would solve some of the problems, and probably remove the need for carefully specifying the ramdisk block size. We would still need the magic restictions on page allocations though and it we ...

The problem: When we are trying to free buffers try_to_free_buffers() will look at ramdisk pages with clean buffer heads and remove the dirty bit from the page. Resulting in ramdisk pages with data that get removed from the page cache. Ouch! Buffer heads appear on ramdisk pages when a filesystem calls __getblk() which through a series of function calls eventually calls init_page_buffers(). So to fix the mismatch between buffer head and page state this patch modifies init_page_buffers() ...

A minor one. It still leaves us with buffer heads out of sync with We actually allow that currently for clean pages which is part The core of my original fix was to modify init_page_buffers so that when we added buffers to a dirty page the buffers became dirty. Modifying the generic code is a bit spooky because it requires us to audit the kernel to make certain nothing else depends on the current behavior in odd ways. Although since init_page_buffers is only called when we are adding ...

Why do you say that? I guess it is _different_, by necessity(?) Is there anything that is really bad? I guess it's not nice for operating on the pagecache from its request_fn, but the alternative is to duplicate pages for backing store and buffer cache (actually that might not be a bad alternative really). -

I obviously agree ;-) Yes, that would solve the problem as well. As long as we fix the problem, I am happy. On the other hand, do you see any obvious problem with this "bandaid"? Christian -

How about the following - even more code simplification gained by this approach :-) I originally had the patch for removing the 'size' argument after a patch (series) that made bi_size unchanged. It seemed that patch would face a harder path upstream so I re-ordered them and missed this I must have missed something .... I've seen: A fix for a bi_end_io in jfs that I missed. A correction for that fix ("return 0" was remove instead of just the '0' removed) ...

s/before it is allocated/when it is never allocated/ -

I don't see the alternative -- I just see my patch, with the added blurbage: #12 - /usr/local/google/home/menage/kernel9/linux/kernel/cpuset.c ==== # action=edit type=text Should I be increasing my caffeine intake? -- I won't rest till it's the best ... Programmer, Linux Scalability Paul Jackson <pj@sgi.com> 1.925.600.0401 -

By making this cpus_equal() and not cpus_intersects(), you're trying to make sure that t->cpus_allowed is always equal to *cpus for each task in Yet by not doing any locking here to prevent a cpu from being hot-unplugged, you can race and allow the hot-unplug event to happen before calling set_cpus_allowed(). That makes this entire function a no-op with set_cpus_allowed() returning -EINVAL for every call, which isn't caught, and no error is reported to userspace. Now all the tasks in ...

> currently against an older kernel ah .. which older kernel? I tried it against the broken out 2.6.23-rc8-mm2 patch set, inserting it before the task-containersv11-* patches, but that blew up on me - three rejected hunks. Any chance of getting this against a current cgroup (aka container) kernel? Could you use the diff --show-c-function option when composing patches - they're easier to read that way - thanks. + if (!retval) { + cpus_allowed = cpuset_cpus_allowed(p); + if ...

Hi, Looking at the tlb_flush code path and its co-relation with ARCH_FREE_PTE_NR, on x86-64 architecture. I think we still don't use the ARCH_FREE_PTE_NR of 5350 as the caching value for the mmu_gathers structure, instead fallback to using 506 due to some typo errors in the code. Found this link in the archives. http://www.ussg.iu.edu/hypermail/linux/kernel/0509.1/1821.html I don't think anything has been done on this yet (looked at 2.6.23). Do let me know if its only a typo that needs ...

This has been seen with several Hitachi drives. There's speculation that it's due to a bug in the drive firmware: http://groups.google.it/group/linux.kernel/browse_thread/thread/a4bd3c19565a2009/38981... https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.22/+bug/137470 A workaround: Index: /etc/rc.local =================================================================== --- .orig/etc/rc.local 2007-10-13 ...

If memchr argument is longer than strlen(kp->name), there will be some weird result. It will casuse duplicate filenames in sysfs for the "nousb". kernel warning messages are as bellow: sysfs: duplicate filename 'usbcore' can not be created WARNING: at fs/sysfs/dir.c:416 sysfs_add_one() [<c01c4750>] sysfs_add_one+0xa0/0xe0 [<c01c4ab8>] create_dir+0x48/0xb0 [<c01c4b69>] sysfs_create_dir+0x29/0x50 [<c024e0fb>] create_dir+0x1b/0x50 [<c024e3b6>] kobject_add+0x46/0x150 [<c024e2da>] ...

On Mon, 15 Oct 2007 00:42:23 -0500 Acked-by: Arjan van de Ven <arjan@linux.intel.com> -

On Mon, 15 Oct 2007 00:34:28 -0500 Acked-by: Arjan van de Ven <arjan@linux.intel.com> -

On Mon, 15 Oct 2007 00:38:01 -0500 Acked-by: Arjan van de Ven <arjan@linux.intel.com> -

When rounding a relative timeout we need to use round_jiffies_relative(). Signed-off-by: Anton Blanchard <anton@samba.org> --- diff --git a/drivers/net/wireless/b43/main.c b/drivers/net/wireless/b43/main.c index c141a26..41049a4 100644 --- a/drivers/net/wireless/b43/main.c +++ b/drivers/net/wireless/b43/main.c @@ -2392,7 +2392,7 @@ out_requeue: if (b43_debug(dev, B43_DBG_PWORK_FAST)) delay = msecs_to_jiffies(50); else - delay = round_jiffies(HZ * 15); + delay = ...

05 - use superio-locks in rest of drivers/hwmon/*.c this patch is compile-tested only, please review for sanity before you try running them. Things to look for - missing superio_release(), opportunities to use superio_devid(), superio_inw(), etc. Signed-off-by: Jim Cromie <jim.cromie@gmail.com> --- hwmon-superio-others Kconfig | 6 +++ f71805f.c | 86 +++++++++++++++------------------------------- it87.c | 80 +++++++++++++++--------------------------- ...

03 - use superio-locks in drivers/hwmon/pc87360 this driver keeps the slot for only during __init, since it only needs the sio-port to read the ISA addresses of the Logical Devices in the chip, which are then used exclusively. Signed-off-by: Jim Cromie <jim.cromie@gmail.com> --- hwmon-superio-pc87360 Kconfig | 1 pc87360.c | 73 +++++++++++++++++++++++++------------------------------------- 2 files changed, 31 insertions(+), 43 deletions(-) diff -ruNp -X dontdiff -X ...

04 - use superio-locks in drivers/char/pc8736x_gpio this driver keeps the slot for the lifetime of the driver ( __init til __exit ), since the driver needs the sio-port to change pin configurations. patches 03,04 were tested on a soekris 4801 a year ago, the box is currently busy. Together they sanity-test the sharing of a reservation with 2 different life-cycles. Signed-off-by: Jim Cromie <jim.cromie@gmail.com> --- hwmon-superio-pc8736x-gpio Kconfig | 1 pc8736x_gpio.c ...

this patchset (on hwmon-git) re-introduces superio_locks module, previously RFC'd here, where I 'borrowed' another thread.. http://marc.info/?l=linux-kernel&m=115821759424601&w=2 The module shares out slots/shared-reservations containing a mutex, so that multiple modules can coordinate access to the sio-port. Im crossposting - LKML for more reviewers, lm-sensors for folks with the hardware and (perhaps) more interest. If its not too late, please consider for 2.6.24-rc0 01 - adds ...

Whenever a read error is found, we should attempt to overwrite with correct data to 'fix' it. However when do a 'check' pass (which compares data blocks that are successfully read, but doesn't normally overwrite) we don't do that. We should. Signed-off-by: Neil Brown <neilb@suse.de> ### Diffstat output ./drivers/md/raid1.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff .prev/drivers/md/raid1.c ./drivers/md/raid1.c --- .prev/drivers/md/raid1.c 2007-10-15 ...

When an array is started read-only, MD_RECOVERY_NEEDED can be set but no recovery will be running. This causes 'sync_action' to report the wrong value. We could remove the test for MD_RECOVERY_NEEDED, but doing so would leave a small gap after requesting a sync action, where 'sync_action' would still report the old value. So make sure that for a read-only array, 'sync_action' always returns 'idle'. Signed-off-by: Neil Brown <neilb@suse.de> ### Diffstat output ./drivers/md/md.c | ...

Following are 5 minor patches for md in current -mm. The first 4 are suitable to flow into 2.6.24. The last fixes a small bug in Dan Williams' patches currently in -mm, which are not scheduled for 2.6.24. Thanks, NeilBrown [PATCH 001 of 5] md: Fix a bug in some never-used code. [PATCH 002 of 5] md: 'sync_action' in sysfs returns wrong value for readonly arrays [PATCH 003 of 5] md: Expose the degraded status of an assembled array through sysfs [PATCH 004 of 5] md: Make sure read ...

Right you are. remap_vmalloc_range doesn't turn the passed vmalloc area into user memory (it creates a completely new mapping). Presumably it either wants to copy_to_user to that new mapping, or memcpy to ->vmalloc? Would the former be an attempt to avoid some I don't know why one would be finding remap_vmalloc_range to fail it mmap time but not later? Should just do it at mmap time and if that is failing, then work out why (or ask linux-mm for help). Actually there is probably a window ...

git stores to my best knowledge only a single author. Infrequently we need a list but then people have solved it putting relevant people in s-o-b and by give credit in changelog. With the current definition you need to supply BOTH a from: and a s-o-b. I usually request a s-o-b when it is missing no matter what other content I did not realise that "concensus" in this context refered to both the one that give the "Acked-by" and the author. Viewing it this way I agree with the intent and the ...

Maybe this: Allow removal of select from Kconfig files Difficulty: 4 Many config options depend on other options is unrelated submenus. As a result, people have complained about not being able to select the desired option because they finding all dependencies is too complicated. Select solves this problem and creates a near-identical new one. Now it is just as hard to turn some options _off_ as it was before to turn others _on_. The solution would be to have smarter tools that give ...

On Mon, 15 Oct 2007 16:23:52 +0200 Thank you Sam, I have added your project ideas to the page. -- "Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it." - Brian W. Kernighan -

Even now, make -j8 really pays off on bigiron AMD. -

Thanks for adding these. Sam -

another config thing that would be nice would be to take something like Rob Landley's miniconfig tool and make it work well enough to be integrated (it creates a version of .config that only contains the things that need to be set, not everything that's at a default that doesn't make any difference) David Lang -

make -j works fine with an unique Makefile, if that's the question. Xav -

Is there already a make config option that will do a good job at setting a default .config file based on what is already running on a system? I get tiered of trimming down my .config for my laptop build so it takes less than 30min to build a kernel. Bonus credit to additional "expert" options (like those powertop puts out) for target uses, laptop, HPC, home file share, embedded targets.... Oh, and lets make the expert configs easily extensible. -

could you explain better what you need? I think I've already such tools ;-) ciao cate -

On Mon, 15 Oct 2007 18:40:34 +0800 You would be the third student to take on that project simultaneously. That is not a problem for me (on the contrary, it increases the chances of one codebase being likeable to Linus), but it does decrease the chances of your patch being the one to make it upstream. Still, it should be a fun project to implement and benchmark, so go ahead if you want. -- "Debugging is twice as hard as writing the code in the first place. Therefore, if you write the ...

On Mon, 15 Oct 2007 11:10:32 -0600 I've written down the basic description: http://kernelnewbies.org/KernelProjects/Raid1ReadBalancing Could you add any ideas that you have to the page? It is a wiki, so anybody can edit the site (after creating an account). -- "Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it." - Brian W. Kernighan -

I'm also quite interested in what compsci students can do for the kernel project. I'm currently doing a little embedded development and research at school, but I and a few others would jump at the chance to work on the kernel (besides finding duplicate problems that the x86 merge is already taking care of, of course. ;) Also (as an aside), we're looking at redoing our operating systems curriculum out here at school...anyone aware of (relatively good) OS curricula? (time scope: ...

that would be cool. --mgross -

Am Sun, 14 Oct 2007 20:49:42 -0700 Here's a version against mainline 2.6.23: Thanks, Hans -- Add uio document to DocBook compilation target. `make *docs' doesn't generate "The Userspace I/O HOWTO", the user space I/O document written in DocBook. Signed-off-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com> Signed-off-by: Hans J. Koch <hjk@linutronix.de> Index: linux-2.6.23/Documentation/DocBook/Makefile =================================================================== --- ...

Thanks, tglx -

Hi Thomas, Thanks for the review, updates to follow. Anton -- Pull the copy_to_user out of hrtimer_nanosleep and into the callers (common_nsleep, sys_nanosleep) in preparation for converting compat_sys_nanosleep to use hrtimers. Signed-off-by: Anton Blanchard <anton@samba.org> --- diff --git a/include/linux/hrtimer.h b/include/linux/hrtimer.h index 540799b..7a9398e 100644 --- a/include/linux/hrtimer.h +++ b/include/linux/hrtimer.h @@ -300,7 +300,7 @@ hrtimer_forward(struct ...

I can get to this later in the week, if you feel the urge in the meantime go for it :) Anton -- Pull the copy_to_user out of hrtimer_nanosleep and into the callers (common_nsleep, sys_nanosleep) in preparation for converting compat_sys_nanosleep to use hrtimers. Signed-off-by: Anton Blanchard <anton@samba.org> Acked-by: Arnd Bergmann <arnd@arndb.de> --- diff --git a/include/linux/hrtimer.h b/include/linux/hrtimer.h index 540799b..7a9398e 100644 --- ...

applied to pata_pcmcia -

I only had one, in /app (lm-sensors-2.10.2) -- which has been removed, thanks.. Justin. -

Arg, you were correct-- my error, it had been a special version that I=20 had to install to get the coretemp several months ago, it is working now. $ sensors abituguru3-isa-00e0 Adapter: ISA adapter CPU Core: +1.35 V (min +0.00 V, max +1.60 V) DDR2: +2.00 V (min +1.60 V, max +2.40 V) DDR2 VTT: +1.00 V (min +0.80 V, max +1.20 V) CPU VTT 1.2V: +1.22 V (min +0.95 V, max +1.45 V) MCH & PCIE 1.5V: +1.50 V (min +1.20 ...

Output from ls -lH /sys/class/hwmon/hwmon*/device: /sys/class/hwmon/hwmon0/device: total 0 lrwxrwxrwx 1 root root 0 2007-10-15 03:59 driver -> ../../../bus/platfor= m/drivers/abituguru3 -r--r--r-- 1 root root 4096 2007-10-15 03:59 fan1_alarm -r--r--r-- 1 root root 4096 2007-10-15 04:06 fan1_alarm_enable -r--r--r-- 1 root root 4096 2007-10-15 04:06 fan1_beep -r--r--r-- 1 root root 4096 2007-10-15 03:59 fan1_input -r--r--r-- 1 root root 4096 2007-10-15 03:59 fan1_label -r--r--r-- 1 ...

Acked-by: David Howells <dhowells@redhat.com> -

At least in case of RHEL, such backports never were automatic. In any case, RHEL 2.1 and 3 do not receive new drivers anymore. We only do bugfixes if something comes up. Realistically speaking, 2.4 kernels are just too old for anyone to use. So, I think it would be best for Did you verify if this works? We use pre-swapped descriptors in 2.4. The above very clearly is a use-after-free, in case the device was open across a disconnect. Solution: Use minor_table_mutex to lock dev->open_count ...

I sent the equivalent patch to John Linville yesterday. Larry -

Initialize "err" in drivers/net/wireless/b43legacy/main.c::b43legacy_start() in case of returning an uninitialized value. From: Adrian Bunk <bunk@kernel.org> Signed-off-by: WANG Cong <xiyou.wangcong@gmail.com> --- drivers/net/wireless/b43/main.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) Index: linux-2.6.23-mm1/drivers/net/wireless/b43legacy/main.c =================================================================== --- ...

From: Noriaki TAKAMIYA <takamiya@po.ntts.co.jp> Patch applied, thanks everyone! -

I'll forward this to Jeff, thanks. Auke -

The how/why of the packet rejection probably isn't all that important, but the most likely scenario based on the ICMP error code is that the router simply does not know about the CIPSO IP option type and is dropping the packet as a result. I'd be very surprised to see a standard router in general use which has policy to perform packet level access control using CIPSO options/labels. -- paul moore linux security @ hp -

Hi, I have encountered the same problem which was reported in http://lkml.org/lkml/2007/9/29/45 For the first one "usbcore duplicated sysfs filename" , I have submit a patch to fix it. For the "bInterfaceNumber" one, I have no idea, the same problem still -

I'd appreciate that. Do you, by any chance, use aic94xx driver too (drivers/scsi/aic94xx/*)? I am a desktop Linux user, so far I don't use suspend at all. Sorry. -- vda -

Hmm, I would have waited for this to happen before committing the change depending on it not to keep things known-broken then... That's what I do with my bits, e.g. I have an update to platform code waiting till sb1250-mac.c changes come over here. Maciej -

I've added linux-scsi which should be cc'd on all SCSI issues. I don't quite believe this, though. The requirement for struct scatterlist is the same before and after the gid scsi-misc patch. If the compile fails with git-scsi-misc because of a missing scatterlist include, it should also fail with vanilla 2.6.23 without the git patch ... could you see if you can find out why it doesn't? thanks, James -

I suspect you're correct, yes. -- I won't rest till it's the best ... Programmer, Linux Scalability Paul Jackson <pj@sgi.com> 1.925.600.0401 -

Ah, right, sorry ... on the ball now. I thought you were saying that the scsi_error.c compilation was failing. In that case, the correct fix is actually to move the scatterlist include from scsi_error.c (where the scatterlist was originally used locally) into scsi_eh.h, like this. James diff --git a/drivers/scsi/scsi_error.c b/drivers/scsi/scsi_error.c index d29f846..ebaca4c 100644 --- a/drivers/scsi/scsi_error.c +++ b/drivers/scsi/scsi_error.c @@ -24,7 +24,6 @@ #include ...

there's a new build failure with latest -git: drivers/built-in.o: In function `ssb_ohci_probe': ohci-hcd.c:(.text+0xbff39): undefined reference to `ssb_device_enable' ohci-hcd.c:(.text+0xbff6f): undefined reference to `ssb_admatch_base' ohci-hcd.c:(.text+0xbff8b): undefined reference to `ssb_admatch_size' ohci-hcd.c:(.text+0xbffe5): undefined reference to `ssb_device_disable' [...] config attached. Ingo

hm, it builds fine now. Weird - i'll double-check i got the right config. have triggered a new SSB build failure meanwhile: drivers/built-in.o:: more undefined references to `pcmcia_access_configuration_register' follow config attached. Ingo

the patch below fixes it for me. PCI seems to have a similar bug as well. Ingo -----------------> Subject: ssb: fix build failure From: Ingo Molnar <mingo@elte.hu> fix build failure if PCMCIA=m but SSB=y: (fix symmetric bug for PCI too.) Signed-off-by: Ingo Molnar <mingo@elte.hu> --- drivers/ssb/Kconfig | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) Index: linux/drivers/ssb/Kconfig =================================================================== --- ...

Thanks, I'll add it to the patch list. I guess people often look in the documentation for email addresses, so it's nice to keep it current. -- Jens Axboe -

Thanks, pulled and pushed out. Can you check that applesmc looks sane after the merge conflict? The conflict itself looked pretty trivial, and it compiles for me, but I'd still like somebody who knows those things to double-check.. Linus -

Yep, it looks good from here. Thanks. -- Dmitry -

Umm, not quite, from my experiences with pre-production wireless drivers, (another story, another time) fancy stuff is being done in udev to make sure that your gigabit card is always assigned to eth0. -- Julian Calaby Email: julian.calaby@gmail.com -

Well, triggered by. (This documentation stuff makes me poke into corners of the kernel I ordinarily otherwise avoid, for various reasons. I don't currently have the luxury of saying "beats me how this bit works, not my Sorry about that. My social skills are finite, I tend to exhaust them when I do too much at once. :( This discussion has clarified for me that my objection isn't the scsi layer itself, it's the /dev/sd? namespace combining devices that would otherwise be /dev/hda, ...

I think a close analogy would be that after a partition is mounted you don't need to know the path to the hard drive, and that is already true today. when you mount a drive (or assign and IP address to a network I also have a 3TB raid I built at home, it uses 3ware cards and a dozen 300G IDE drives. since the 3ware driver is classified as SCSI if a drive fails all the other drives get renumbered on the next boot and it's painful to figure out which drive has a problem. I have to reboot ...

Ah, so it was about your documentation work. I already forgot the context of your previous inquiries. Alas the tone of them already did some damage, leading to responses like these. ... The Linux SCSI subsystems don't consume, they provide services; nowadays not only for SCSI hardware and SCSI protocols but also for a number of subsystems whose tasks are similar enough to SCSI subsystems to make the SCSI core and upper SCSI layer useful to them too. BTW: | Now that IDE disks have been ...

I was just trying to use the strangeness in a large distributor's first attempt at this functionality as an evidence that it's apparently not trivial to get even the common cases right under the new model, while the common cases used to be trivial to get right under the old model. (Or at least it seemed so to me.) I think I've exhausted this line of argument, though, and will stop now. Rob -- "One of my most productive days was throwing away 1000 lines of code." - Ken Thompson. -

Low-level networking drivers suggest a default interface name (per interface or as a template like eth%d into which the networking core inserts a lowest spare number). Userspace can rename interfaces, but nevertheless it's nice to have different default kernel names for ethernet, wlan etc.. Could low-level SCSI drivers provide similar name templates which give a hint on the transport involved? It's a bit more difficult as with networking interfaces though because - SCSI devices can have ...

No, it doesn't. James Bottomley has been exceedingly polite and helpful, as were several other people on the linux-scsi list when I asked them about this stuff back in August. Religion, politics, and anything remotely related to hotplug appear to be topics to avoid in polite company if you want it to remain polite. (My gripes with scsi mostly have to do with device enumeration. My attempts to use sysfs also have to do with device enumeration. I've spotted a trend here.) Rob -- ...

How do you define real SCSI ? The definition of SCSI in the kernel is "a device that accept the SCSI command set" (more precisely "a suitably large subset a the SCSI command set". It looks as if you definition of SCSI is "a device that is sold with written SCSI on the box and that attaches to a card with SCSI written on the box"; is it correct ? The host is the expansion card that connects the device to the motherboard. If it is emulated this means that it is not a ...

On Mon, 15 Oct 2007 03:36:15 -0500 that's a choice Ubuntu made in their udev scripts... if you don't like it, complain to them. I'm surprised you would even need to care about what device name things are though.... with mount-by-label (deployed for a bunch of years now in most distros), and various helpful links like /dev/cdrom .... anyway.. if you don't like your distros udev configuration, lkml is the wrong forum. -

Ok, I'll bite. If it's all "real" scsi, why does ioctl(SG_EMULATED_HOST) They're the same thing? Do you mean that on a system with both, going: ifconfig eth1 66.92.53.140 ifconfig ppp 192.168.0.42 Would be functionally equivalent to: ifconfig eth1 192.168.0.42 ifconfig ppp 66.92.53.140 So if on one boot the addresses are assigned the first way, and upon reboot they're assigned in the second way by exact the same set of commands... well that's not IMPORTANT, is it? (Or ...

About 6 weeks ago, on a 2.6.23-rc kernel, I accidentally typed "make -j", and left off the 4 before I hit the return key. About 2-3 minutes later, the box locked pretty tight. I managed to switch to a VT console before I lost total control of X (took many, many minutes to do the switch), but after many minutes, managed to get logged into the console, but I wasn't able to get a ps command to complete so I could start killing processes. (I probably should have just done a "killall make" right ...

I hate to go completely offtopic here, but disks are so incredibly slow when compared to RAM that there is really nothing the kernel can do about this. Presumably the job will finish, given infinite time. How much swap do you have configured? You really shouldn't configure so much unless you do want the kernel to actually use it all, right? Because if we're not really conservative about OOM killing, then the user who actually really did want to use all the swap they configured gets angry ...

Sorry about that. Not my intent. I was aiming more at "I'm trying to document this and I don't understand how it works at all, or why it does things this way. It seems backwards from what I would expect." Rob -- "One of my most productive days was throwing away 1000 lines of code." - Ken Thompson. -

I gave it about half an hour, then it locked solid and stopped writing to the disk at all. (I gave it another 5 minutes at that point, then held down the power button.) Two words: "Software suspend". I've actually been thinking of increasing it I tend to lower "swappiness" and when that happens all sorts of stuff goes weird. Software suspend used to say says it can't free enough memory if I put swappiness at 0 (dunno if it still does). This time the OOM killer never ...

OK, so could we get back to the original discussion? The question I think you meant to ask is "does SCSI use the block layer, and if so; how?" The answer is yes (just do an ls /sys/block on any scsi machine). The how is that it bascially uses the block layer as a service library (i.e. most SCSI services are built on top of those already provided by block). The email you cited was basically from our one area of confusion: SCSI and block both provide services to decode the SG_IO ioctl. This ...

Ah, but it could. If you had more than one IDE controller (which is even possible on laptops; the Fujitsu P7120 is one that I'm familiar with that has more than one), the initialisation order *of the It's not something anyone particularly set out to do, it's just how it worked out. It was justified by saying "ok, this goes from a 99% solution to a 96% solution, but there's 100% solution called uuids". I don't particularly agree with this line of argumentation, but it did hold sway. -- ...

They *are* SCSI devices. USB storage is a SCSI over USB transport. ATAPI is a SCSI over ATA transport. SAS is much the same thing, as is FC, and it continues. With the exception of ATA disk for historical reasons SCSI essentially For the emedded CF using world we could do with a truely dumb ATA only CF driver, possibly even with pure polled support that used neither the IDE or the ATA layer. Alan -

If you hate USB storage devices using scsi, please use the ub driver, When did usb-storage devices ever show up as /dev/usb0? USB flash disks are really SCSI devices, look at the USB storage spec for proof of that. thanks, greg k-h -

/somewhat/ true I'm afraid: libata uses the SCSI layer for ATAPI devices because they are essentially bridges to SCSI devices. It uses the SCSI layer for ATA devices because the SCSI layer provided a huge amount of infrastructure that would need to have been otherwise duplicated, /then/ massaged into coordinating between <jgarzik's ATA layer> and <SCSI layer> when dealing with ATAPI. There is also a detail that was of /huge/ value when introducing a new device class: distro ...

Acked-by: Shannon Nelson <shannon.nelson@intel.com> FYI - this code gets heavily revised by ioat patches in -mm that Andrew is planning on merging soon. -

Thanks. Sorry about missing your CC on the libata-wide ata_link changes. We should have poked the maintainers on that, for the drivers we cannot build ourselves. -

Will test tomorrow/Wednesday. -- Russell King Linux kernel 2.6 ARM Linux - http://www.arm.linux.org.uk/ maintainer of: -

Hi Geert, I'd be more happy if we returned real error reported by input_register_device() instead of substituting it with -ENOMEM. -- Dmitry -

On Fri, 12 Oct 2007, Gregory Haskins wrote: Again no description. -

-- Acked-by: Steven Rostedt <rostedt@goodmis.org> This patch seems reasonable to me. I'll pull it into the queue unless there's any NACKs. -

-- Well if this fails to terminate, then we have a major bug. Perhaps we can add something like this: { int count = NR_CPUS * 2; while (push_rt_tasks(this_rq) && count--) ; BUG_ON(!count); } Since we should do it really at most CPU times, and I added a CPU * 2 just to be safe that we don't have a unrealistic point of moving tasks onto other CPUS and have them finish before we finish this loop. But really, I don't think we need to worry too much about that while ...

After thinking about this over the weekend, the task_cpu optimization doesn't make sense. It made sense in my original design before I integrated with Steve's series, but not here. I think the second optimization (this_cpu) still makes sense though, so I will update this -

Indeed! Having this in the kernel would allow us to enable RT scheduling for PulseAudio by default without bad effects. I was thinking about adding some kind of babysitting process to userspace -- but doing this as an RLIMIT in the kernel strikes me a much better idea! I think it would make a lot of sense to make the API very similar to RLIMIT_CPU, i.e. also send out SIGXCPU and SIGKILL, with the single difference that RLIMIT_CPU sends out a signal depending on the total CPU time used for the ...

Why only SHCED_FIFO and not SCHED_RR? I guess, put_prev_task() / set_curr_task() should be called (for the case of task_running(p)) to make it group-scheduler-friendly (as it's done e.g. in sched_setscheduler()). (normilize_task() should probably do the same) -- Best regards, Dmitry Adamushko -

Wouldn't it be better to WARN_ON, and simply not entering lazy mode? I am concerned that this is 32-bit specific. But hey: We could wrap it here, but the best solution may be just to define this macro for 64-bit, and make it everyone benefits. So yeah, this is a concern here, but I don't think anything should be changed in this patch to address it so... so... ok ;-) -- Glauber de Oliveira Costa. "Free as in Freedom" http://glommer.net "The less confident you are, the more serious you ...

Can be used by any controller that supports SNotification (or is able to capture and parse the raw FIS's). And FWIW, Tejun's PMP work also uses a few bits of same async-notify stuff. I agree with your review items, and will look at those if others don't beat me to it. Jeff -

Hi, The Open Firmware exception handler is begin invoked, on power5 lpar. boot: 2.6.23.1 Please wait, loading kernel... Allocated 03500000 bytes for kernel @ 01c00000 Elf64 kernel loaded... Loading ramdisk... ramdisk loaded 0020d000 @ 06b00000 DEFAULT CATCH!, exception-handler=fff00300 at %SRR0: c00000000202705c %SRR1: 8000000000003002 Open Firmware exception handler entered from non-OF code Client's Fix Pt Regs: 00 0223100003080000 00000000018bfac0 0000000004bbeaf8 ...

Sorry - I should have been more precise - vmap should never be used in performance critical paths on default configs. Log recovery will trigger vmap/vunmap usage, so this is probably what you are seeing. Cheers, Dave. -- Dave Chinner Principal Engineer SGI Australian Software Group -

Hm, well I saw the problem with a filesystem made with mkfs.xfs with no options, so there must be at least *some* vmapping going on there. J -

Yeah, it would be possible. The easiest way would just be to shoot down all lazy vmaps (because you're doing the global IPIs anyway, which are the expensive thing, at which point you may as well purge the rest of your lazy mappings). OK, I see. Because even though it is technically safe where we are using it (because nothing writes through the mappings after the page is freed), a corrupted guest could use the same window to do bad vmap is slightly harder than kmap in some respects. ...

Yes, as Dave said, vmap (more specifically: vunmap) is very expensive because it generally has to invalidate TLBs on all CPUs. I'm looking at some more general solutions to this (already have some batching / lazy unmapping that replaces the XFS specific one), however they are still likely going to leave vmap mappings around after freeing the page. We _could_ hold on to the pages as well, but that's pretty inefficient. The memory cost of keeping the mappings around tends to be well under 1% ...

With defaults - little effect as vmap should never be used. It's only when you start using larger block sizes for metadata that this becomes an issue. The CONFIG_XEN workaround should be fine until we get a proper vmap cache.... Cheers, Dave. -- Dave Chinner Principal Engineer SGI Australian Software Group -

You're keeping vmaps around for already freed pages? That will be a big problem for proper PAT support, which needs to track all mappings to memory. It's not just a problem for Xen. In fact I suspect it is already broken with DRM or AGP for example which can use UC and WC mappings -- if you keep the mapping around and DRM or AGP turns the page in another mapping uncacheable you're creating an illegal cache attribute alias. These are known to occasionally create cache corruptions on several ...

Probably. David -

You're probably right since struct lsap_cb get inserted into the hashbin. Regards, Hinko -- ČETRTA POT, d.o.o., Kranj Planina 3 4000 Kranj Slovenia, Europe Tel. +386 (0) 4 280 66 03 E-mail: hinko.kocevar@cetrtapot.si Http: www.cetrtapot.si -

hrm, OK. But it could have been controlled by a more-user-friendly runtime knob, I guess. Not that it's worth changing that now. -

thanks, great fixes! They should show up in the next sched-devel git tree. Ingo -

I enabled every option I thought might be useful. I booted the system, fired a serial console, bumped the priority of the serial port IRQ handler to 80 and the priority of the shell in the serial console to 70, set the console log level to 9 using SysRq. I then connected via ssh, loaded the driver module (the output showed up in the serial console), started 4 processes, and had a complete system lock-up within 10 seconds. Nothing on the serial console :-( The system didn't even ...

Hello. Oh, nice. Thank you. I recompiled the kernel using the same kernel today. However, I couldn't reproduce slab's leak today. -

OK, thanks. Greetings, Rafael -

I get funny SIGBUS' like so: fault if (->page_mkwrite() < 0) nfs_vm_page_mkwrite() nfs_write_begin() nfs_flush_incompatible() nfs_wb_page() nfs_wb_page_priority() nfs_sync_mapping_wait() nfs_wait_on_request_locked() nfs_wait_on_request() nfs_wait_bit_interruptible() return -ERESTARTSYS SIGBUS trying to figure out what to do about this... -

Hmmm... It sounds like the fault handler should deliver the appropriate signal, should ->page_mkwrite() return ERESTARTSYS, and then retry the access instruction that caused the fault when the signal handler has finished running. David -

Why? If someone is interrupting the write, then a SIGBUS is pretty much expected. Trond -

Does work fine, thanks. -

Hi Andrew, While running regular cpu-offline tests on 2.6.23-mm1, I hit the following lockdep warning. It was triggered because some of the per-cpu counters and thus their locks are accessed from IRQ context. This can cause a deadlock if it interrupts a cpu-offline thread which is transferring a dead-cpu's counts to the global counter. Please find the patch for the same below. Tested on i386. Thanks and Regards gautham. =====================Warning! ...

Hmmm... Good point. Yuck. David -

I don't think the fault handler is currently in any position to do that ATM. It is possible to make it interruptible in some contexts, but faults from kernel code may not be able to cope. -

Do you know any more about when this happened? Was it during a reboot, or after you unmounted some device or volume? Have you seen it again? Which filesystem(s) do you use? -- Dave -

I am experiencing some highly annoying but intermitent freezing on a pentium4 2.80G HT/SMT box, when doing normal desktop work with 2.6.23-rt1. The same crippling behavior does not occur on a Core 2 Due T7200 2.0G SMP, so I suspect it's something due specific to the SMT scheduling support (Hyper-Threading). But can't tell for sure, obviously :) The symptoms are noticeable primarily as some X/GUI intermitent freezing, sometimes only one application, then several and ultimately the whole ...

this was full of spaces, and should be converted to tabs (I did this manually) In the future, make sure to run your patch through scripts/checkpatch.pl. Some of that script's complaints are a bit changed this to [ich9_2port_sata] = since we have started using the (C99? gcc?) method of initializing piix_port_info[] array entries based on the given index, rather than simply noting the index in a comment and depending on the order of other entries not to change. Jeff -

You were asked to rebase your patches against the latest -rt kernel. And you bluntly refused to. I knew exactly why I ignored the patches when I was moving -rt forward. Simply because I had not enough time to review them in detail. Ingo acked them in principle, but we had subtle wreckage from such changes before and I had no intention to put more risk into a non trivial merge. Sorry, you have provided sloppy patches more than once and requested us to pick random bugfixes from a broken ftp ...

-- You're right I missed that. (I even read it). But for fixes like this, (or any patches that are not in the tree), you really need to resend the series. When Ingo posts patches to LKML on CFS, if there's a little fix like this, There wouldn't be if we didn't have to go looking for patches on patches that are out of the series. If you see that a series is broken, don't patch against it. Resend the series! Patches must be against upstream, unless they are more RFC (like what Gregory ...

In hindsight I should have resent, but I didn't .. Really it's a matter of frustration on my part. Where I submit patches that should be included , but they are ignored for no clear reason .. Then I submit again, and the patches are ignored again for no clear reason.. Comparing mainline and -rt isn't very interesting since , -rt has few patches sent in against it .. Andrew is bound to loose a patch here and there on accident, but -rt has so few patches sent in that we shouldn't In all ...

That's a bit hard to parse. On Thu, 11 Oct 2007 18:23:40 +0800 There's a typo there. Also the text is pretty meaningless - I'd fix it up Probably this shouldn't be here at all - DEBUG is passed in from the gcc It is not a good idea to create files in /proc which have spaces in their names. Yes, userspace _should_ be able to cope with that in all cases, but all software sucks, even including userspace ;) That's a bit ugly. Perhaps using container_of() would be clearer. ...

On Mon, 15 Oct 2007 18:03:35 -0400 err, umm, yeah. But the same argument applies: it is imprudent to have space-containing records in /proc/interrupts. However it seems that we've already done that in several places so I guess any /proc/interrupts-parsing programs are already coping with it OK. -

Blackfin's have a synchronous Serial Peripheral pORT (SPORT). Unlike SPI, UART, I2C, or CAN interfaces which are designed for specific industry standard compatible communication only, the SPORT support a variety (software programmable) serial data communication protocols: - A-law or µ-law companding according to G.711 specification - Multichannel or Time-Division-Multiplexed (TDM) modes - Stereo Audio I2S Mode - TDM Modes for Multi-Channel audio codecs - H.100 Telephony standard ...

Since this is for some special aim (not default for most classes, at least not for sched_fair) I'd suggest to change names: default_timeslice_fair() and .default_timeslice to something like eg.: rr_timeslice_fair() and .rr_timeslice or rr_interval_fair() and .rr_interval (maybe with this "default" before_"rr_" if necessary). On the other hand man (2) sched_rr_get_interval mentions that: "The identified process should be running under the SCHED_RR scheduling policy". Also this place seems ...

Another thing to take into account is that kconfig was recently changed to save ARCH and CROSS_COMPILE in the .config file: http://lwn.net/Articles/253889/ Presumably that means you'd only have to specify your arch and cross compiler during config, and then if you re-used that config it would re-use those settings. But the existing makefile discards anything that isn't explicitly overridden on the make command line at each stage of the build. It seems to me any fix should only reset ...

ACK patches 1-3, but they still do not apply. Since net-2.6.24 has gone upstream, you may send diffs against torvalds/linux-2.6.git if you like -- there are no outstanding dependencies at this time. -

applied -

Trapping DMA and ignoring will make sense. I don't know how it can be done. What does ignoring DMA mean here? Acutual DMA does not take place but device thinks DMA is complete? Trapping DMA and ignoring in kdump kernel (if possible), seems to be better than letting DMA happen. This will reduce chances of corruption of second kernel. Can you please give little more details in terms of how to trap DMA. I think Right now ppc64 just uses the TCE tables from first kernel and allows the ...

On Thu, Oct 11, 2007 at 08:59:58AM +0100, Andy Whitcroft wrote: > > /* get fid did for hardware pstate transition */ > > pstate = index & HW_PSTATE_MASK; > > - if (pstate > MAX_HW_PSTATE) > > + if (pstate > data->max_hw_pstate); > > return 0; > > checkpatch picked up this dodgy if. I suspect that the ';' is wrong > from the context. Indeed. Mark/Andreas, want to send me a new patch against latest Linus HEAD ? I didn't apply the old one yet, as it conflicted with the ...

On Mon, Oct 15, 2007 at 11:23:45AM -0500, Mark Langsdorf wrote: > On Monday 15 October 2007 10:40, Dave Jones wrote: > > On Thu, Oct 11, 2007 at 08:59:58AM +0100, Andy Whitcroft wrote: > > > > > > /* get fid did for hardware pstate transition */ > > > > pstate = index & HW_PSTATE_MASK; > > > > - if (pstate > MAX_HW_PSTATE) > > > > + if (pstate > data->max_hw_pstate); > > > > return 0; > > > > > > checkpatch picked up this dodgy if. I suspect that the ';' is wrong ...

Yeah, OK, I can live with that. I'll fold it into the base patch, and update the lazymode patch accordingly. J -

On Mon, 15 Oct 2007 13:06:14 -0700 (PDT) <looks> OK, in ipv4_local_port_range() we have if (range[1] <= range[0]) ret = -EINVAL; which will prevent the crashes. But is it good to disallow high=low? This disallows a port range of one single port. Unless "high" is exclusive. But ie: inclusive. also inclusive. -

From: Andrew Morton <akpm@linux-foundation.org> I'm pretty sure we took care of this, but maybe not :-) -

Writing 15361 (i.e. 0x3C01) to ACPI_REGISTER_PM1A_CONTROL appears to hang my system in acpi_os_write_port(). What can I do about that? Is it a BIOS issue? a kernel issue? a hardware issue? (All my results are attached to the bug report.) Regards. -

I have the same problem in 2.6.23.1 (cf. my bug report in the database) I'll ask the manufacturer whether they could get poweroff to work. Regards. -

We shouldn't need that. What happens if you add an msleep(5) before ehci-hcd::ehci_run() drops ehci_cf_port_reset_rwsem? The theory there being that the switch triggered by setting CF doesn't take effect instantaneously, contrary to the effective assumption of that code. A delay of 5 msec seems like it should be more than enough, but that's kind of a guess ... it's good to keep that low, since unfortunately that's in the critical path for OLPC "resume from idle". - Dave -

From: David Miller <davem@davemloft.net> Bad news, even with the rwsem after a lot more testing I can still trigger the hang in ohci_hub_control() :-( I think we need to go back to considering the total serialization approach to this problem. -

On Mon, 15 Oct 2007 21:59:49 +0900 Yes that value means that TCP stays in slow start until the first loss. This makes TCP behave as expected in the RFC's. -- Stephen Hemminger <shemminger@linux-foundation.org> -

Looks reasonable to me. (sorry for the belated reply... I also saw this in gregkh's patch series. Since the email was addressed to me, I figure I should at least say "yes I read it". Dunno if "Yes-I-read-it-by:" is the same as "Acked by:", if it is, then acked-by me.. ) --linas -

"Trusting people or their opinions" is only about use of the english language, and not that intersting to bring up here. Surely you know that lots of people here have english as a secondary language only. Intersting for me to know, but I never claimed that linux will work on your laptop, so no: You can't take my word for that, because I never gave it! It is well known that some laptops don't work with linux, I have no idea if yours will work, I don't even know what kind it is. I told you the ...

It's more expensive than nothing, sure. However in real code, algorithmic complexity, cache misses and cacheline bouncing tend to be much bigger issues. I can't think of a place in the kernel where smp_rmb matters _that_ much. seqlocks maybe (timers, dcache lookup), vmscan... Obviously removing the lfence is not going to hurt. Maybe we even gain 0.01% performance in someone's workload. Also, remember: if loads are already in-order, then lfence is a noop, right? (in practice it seems to ...

Yes, I still can't believe this, but after some more reading I start to admit such things can happen in computer "science" too... I've mentioned a lost performance, but as a matter of fact I've been more concerned with the problem of truth: From: Intel(R) 64 and IA-32 Architectures Software Developer's Manual Volume 3A: "7.2.2 Memory Ordering in P6 and More Recent Processor Families ... 1. Reads can be carried out speculatively and in any order. ..." So, it looks to me ...

I'd say that's exactly what Intel wanted. It's pretty common (we do it all the time in the kernel too) to create an API which places a stronger requirement on the caller than is actually required. It can make changes much less painful. Has performance really been much problem for you? (even before the lfence instruction, when you theoretically had to use a locked op)? I mean, I'd struggle to find a place in the Linux kernel where there is actually a measurable difference anywhere... and we're ...

Hello. OK. I posted new one that uses existing API at http://lkml.org/lkml/2007/10/11/140 . By the way, what do you think of new primitives shown below? I'd like to use list_for_each_cookie() and list_add_tail_mb() if acceptable. ----- Start of code ----- /********** Existing API **********/ #include <stdio.h> #include <stdlib.h> static inline void mb(void) {;} static inline void prefetch(void *p) {;} #define offsetof(TYPE, MEMBER) ((size_t) &((TYPE *)0)->MEMBER) #define ...

With the new code Linus already merged, signalfd does not attach to the sighand anymore, so the "orphaned sighand" behaviour is no more there. An exec() will carry the fd over, and you will be able to use the fd in the same way you did before the exec(). If sigpending()/sigwaitinfo() will It'll return the signals that would be normally returned to the thread with the standard signal delivery methods. That is, calling thread private signals, and calling thread-group shared ...

Well, this is actually the bit that made cancel_work_sync() be written in the first place. The short story is the netlink lock is most probably held at this point (depending on the usage of phy_disconnect()) and there is also an event waiting in the queue that requires the lock, so if flush_scheduled_work() is called here a deadlock will happen. Let me find a reference for a longer ...

Hi - Our team is farther along adapting to this change against 2.6.23-mm1, and we have run into a complication. It's more of a distribution issue. We would prefer to retain systemtap's capability to build instrumentation for a kernel other than the currently running one. Such instrumentation can be then copied and run on a distinct machine. This has meant relying on development data: make install_headers + Makefiles (as packaged by Fedora/RHEL), and to a lesser extent separated debugging ...

My suggestion applies as well to modules as anything else. What "like Module.symvers" means is something like: name1 vmlinux %s name2 fs/nfs/nfs %d All the modules built by the same kernel build go into this one file. Modules packaged separately for the same kernel could provide additional That's clearly straightforward to do given the basic markers data structures. It does not address the need for an offline list of markers available in a particular kernel build or set of modules ...

No, It must not be cleared here because we can't enter in the accounting = code between kvm_guest_enter(void) and kvm_guest_exit(void). This is why the accounting code clears it. I put a kvm_guest_exit() only for symmetry. Laurent --=20 ---------------- Laurent.Vivier@bull.net ----------------- "Given enough eyeballs, all bugs are shallow" E. S. Raymond

I think we can merge your patches, as the userspace interface seems fine. To make the accounting work for virtual cpu accounting found on ppc and s390 we can later add an additional patch that also deals with interruptible guest contexts. So something like this should work: Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com> --- drivers/kvm/kvm.h | 8 ++++++++ kernel/sched.c | 2 ++ 2 files changed, 10 insertions(+) Index: ...

thx - zapped. Ingo -

It's fine for me. Laurent --=20 ---------------- Laurent.Vivier@bull.net ----------------- "Given enough eyeballs, all bugs are shallow" E. S. Raymond

IMHO, I think it is better to let kvm_guest_exit() empty (you can remove = it, if you want): 1st case: - unset PF_VCPU in kvm_guest_exit(), all the tick is always for system ti= me. Guest time is always 0. 1st case and half: - like 1st case but we move kvm_guest_exit() as you propose and the reaso= n of the interrupt is the tick interrupt. The tick is for guest time only. I t= hink the probability is very low. 2nd case: - don't unset PF_VCPU in kvm_guest_exit(), all the tick is ...

If the guest is executing for 10% of the time, the probability is But then even execution in ->handle_exit() is accounted as guest time, which is wrong. -- error compiling committee.c: too many arguments to function -

FYI, KVM abstracted its guest-mode entry code recently so this did not apply - find below the reworked patch. Ingo -------------------> Subject: sched: guest CPU accounting: maintain guest state in KVM From: Laurent Vivier <Laurent.Vivier@bull.net> Modify KVM to update guest time accounting. [ mingo@elte.hu: ported to 2.6.24 KVM. ] Signed-off-by: Laurent Vivier <Laurent.Vivier@bull.net> Acked-by: Avi Kivity <avi@qumranet.com> Signed-off-by: Ingo Molnar <mingo@elte.hu> --- ...

I think you know that better than me. But is there homogeneity in probability ? I mean, if the guest has a lot I/O, it is interrupted by them and the probability to be interrupted by a tick is lower than the time passed in = System time and User time are wrong too as the tick is accounted to the s= ide where it appears, even if CPU has executed code from the other side in a sub-part of the tick. It's not a good argument. Laurent --=20 ---------------- Laurent.Vivier@bull.net ...

I would move this line        if (p->flags & PF_VCPU) {                account_guest_time(p, cputime); ------>               p->flags &= ~PF_VCPU; <---------                return;        } into kvm_guest_exit. Otherwise a guest that is running very long in guest context would only get the first tick accounted as guest time, no? Besides that, this looks good and should work for kvm on s390 as well. Thanks Laurent. Christian -

Sorry for ambiguity. With current adapter firmware the driver is working but future firmware updates may return sense data larger than 96 bytes, causing overflow on scp->sense_buffer and a kernel crash. This fix should be backported to earlier kernels. HighPoint Linux Team -

This one hasn't been merged yet. -- Frank -