--- ~Randy --

This patch adds a new recommended attribute named label into the NFSv4 file attribute structure. It also adds several new flags to allow the NFS client and server to determine if this attribute is supported and if it is being sent over the wire. Signed-off-by: David P. Quigley <dpquigl@tycho.nsa.gov> --- fs/nfs/nfs4proc.c | 3 +++ include/linux/nfs4.h | 2 ++ include/linux/nfs_fs_sb.h | 1 + include/linux/nfs_xdr.h | 4 ++++ include/linux/nfsd/export.h | ...

Two fields have been added to the nfs_fattr structure to carry the security label and its length. This has raised the need to provide lifecycle management for these values. This patch introduces two macros nfs_fattr_alloc and nfs_fattr_fini which are used to allocate and destroy these fields inside the nfs_fattr structure. These macros do not modify any other components of the structure so nfs_fattr_init still has to be used on these structures. In the event that CONFIG_SECURITY is not set these ...

This patch adds two new fields to the iattr structure. The first field holds a security label while the second contains the length of this label. In addition the patch adds a new helper function inode_setsecurity which calls the LSM to set the security label on the inode. Finally the patch modifies the necessary functions such that fsnotify_change can handle notification requests for dnotify and inotify. Signed-off-by: David P. Quigley <dpquigl@tycho.nsa.gov> --- fs/attr.c | ...

This patch adds two entries into the fs/KConfig file. The first entry NFS_V4_SECURITY_LABEL enables security label support for the NFSv4 client while the second entry NFSD_V4_SECURITY_LABEL enables security labeling support on the server side. These entries are currently marked as EXPERIMENTAL to indicate that at the moment they are only suitable for testing purposes and that the functionality they provide may change before it is removed from experimental status. Signed-off-by: David P. Quigley ...

Before the inode_xattr_getsecurity call was removed the caller would concatenate the security namespace prefix and the suffix provided by the lsm to obtain the security xattr. This hook provides the functionality to obtain the full LSM xattr name. The patch also provides implementations for the dummy security module and SELinux. This method is used instead of restoring the old method since it only requires an offset into the returned pointer to obtain the suffix. This approach is more efficient ...

The existing nfs4 xattr handlers do not accept xattr calls to the security namespace. This patch extends these handlers to accept xattrs from the security namespace in addition to the default nfsv4 acl namespace. Signed-off-by: David P. Quigley <dpquigl@tycho.nsa.gov> --- fs/nfs/nfs4proc.c | 54 +++++++++++++++++++++++++++++++++++++++++----------- 1 files changed, 42 insertions(+), 12 deletions(-) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index a1a4051..d7193df 100644 --- ...

This patch adds two entries into the fs/KConfig file. The first entry NFS_V4_SECURITY_LABEL enables security label support for the NFSv4 client while the second entry NFSD_V4_SECURITY_LABEL enables security labeling support on the server side. These entries are currently marked as EXPERIMENTAL to indicate that at the moment they are only suitable for testing purposes and that the functionality they provide may change before it is removed from experimental status. Signed-off-by: David P. Quigley ...

The new method for pulling argument for NFS from mount is through a text parsing system. This patch adds two new entries to the argument parsing code "securlty_label" and "nosecurity_label". Even though we use text across the user/kernel boundary internally we still pack a binary structure for mount info to be passed around. We add a flag for use in the nfs{4,}_mount_data struct to indicate that are using security labels. Finally we add the SELinux support to mark the labeling method as ...

There are several places where recommended attributes are implemented in the NFSv4 client code. This patch adds two functions to encode and decode the secid recommended attribute which makes use of the LSM hooks added earlier. It also adds code to grab the label from the file attribute structures and encode the label to be sent back to the server. Even though the code is there to encode a label to be sent back to the server there does not appear to be an interface to use it yet. Signed-off-by: ...

There currently doesn't exist a labeling type that is adequate for use with labeled nfs. Since NFS doesn't really support xattrs we can't use the use xattr labeling behavior. For this we developed a new labeling type. The native labeling type is used solely by NFS to ensure nfs inodes are labeled at runtime by the NFS code instead of relying on the SELinux security server on the client end. Signed-off-by: David P. Quigley <dpquigl@tycho.nsa.gov> --- security/selinux/hooks.c | 20 ...

The existing nfs4 xattr handlers do not accept xattr calls to the security namespace. This patch extends these handlers to accept xattrs from the security namespace in addition to the default nfsv4 acl namespace. Signed-off-by: David P. Quigley <dpquigl@tycho.nsa.gov> --- fs/nfs/nfs4proc.c | 54 +++++++++++++++++++++++++++++++++++++++++----------- 1 files changed, 42 insertions(+), 12 deletions(-) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index a1a4051..d7193df 100644 --- ...

This patch implements the encoding of a MAC label on the server side to be sent across the wire to the NFSv4 client. At this time there is no method of receiving a label from the client to be set on the server. Signed-off-by: David P. Quigley <dpquigl@tycho.nsa.gov> --- fs/nfsd/nfs4xdr.c | 90 ++++++++++++++++++++++++++++++++++++++++++++++++++- fs/nfsd/vfs.c | 7 ++++ security/security.c | 1 + 3 files changed, 97 insertions(+), 1 deletions(-) diff --git a/fs/nfsd/nfs4xdr.c ...

This patch set is the first submission to fs-devel and lkml for the purpose of code review. To test the patch set you need patches to nfs-utils as well. Since this is just a code review I haven't posted the patch to nfs-utils however if you want to test the code feel free to e-mail me and I will send you the necessary patch. Out of all of the functionality we have prototyped I have narrowed it down to these items which I believe is the solid base for initial kernel inclusion. These patches ...

* Make idecs_register() return pointer to hwif structure instead of hwif's index. * Store pointer to hwif structure instead of hwif's index in ide_info_t (fixes ide-cs for IDE[6-9]_MAJOR major numbers). * Remove no longer needed ide_major[]. * Tell IDE layer to not manage resources by setting hwif->mmio flag. * Use {request,release}_region() for resources management. * Use driver name for resources management. Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com> --- ...

All host drivers using ide_unregister()/module_exit() have been fixed to manage resources themselves so this function can be removed now. There should be no functional changes caused by this patch. Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com> --- drivers/ide/ide.c | 25 ------------------------- include/linux/ide.h | 1 - 2 files changed, 26 deletions(-) Index: b/drivers/ide/ide.c =================================================================== --- ...

My vision on the direction we should take wrt cpu isolation. Next on the list would be figuring out a nice solution to the workqueue flush issue. --

Keep the affinity of unbound IRQs within the system set. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> --- arch/alpha/kernel/irq.c | 2 - include/linux/irq.h | 7 ----- kernel/irq/manage.c | 62 ++++++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 63 insertions(+), 8 deletions(-) Index: linux-2.6/arch/alpha/kernel/irq.c =================================================================== --- linux-2.6.orig/arch/alpha/kernel/irq.c +++ ...

cpu isolation doesn't offer anything over cpusets, hence remove it. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> --- kernel/sched.c | 24 +++--------------------- 1 file changed, 3 insertions(+), 21 deletions(-) Index: linux-2.6/kernel/sched.c =================================================================== --- linux-2.6.orig/kernel/sched.c +++ linux-2.6/kernel/sched.c @@ -6217,24 +6217,6 @@ cpu_attach_domain(struct sched_domain *s rcu_assign_pointer(rq->sd, sd); } ...

"one that caters the general purpose OS" ... a tad terse on the documentation ;). I guess what you have is a new cpumask_t cpu_system_map, which is the union of the CPUs of all the cpusets marked 'system', where to a rough approximation the CPUs -not- in that cpumask are what we would have called the isolated CPUs by the old code? In any case, if this patch survives its birth, it will need an added change for some file in the Documentation directory. Could we get the term 'cpu' in the ...

Adds macros similar to min/max/min_t/max_t. Also, change the variable names used in the min/max macros to avoid shadowed variable warnings when min/max min_t/max_t are nested. Small formatting changes to make all the macros have a similar form. Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com> --- Andrew, here's a cleanup of the max/min macros and the introduction of a clamp/clamp_t macro. There are users for this in libata (the FIX macro you've already seen) and in ...

There's a known suspend problem with 2.6.25-rc3 that has been fixed already in the Linus' tree. Can you test the current head of the Linus' tree, please? Rafael --

This does not look like known problem, actually... s2ram segfaults somewhere in emulator...? Pavel -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html --

Add ide_legacy_device_add() helper for use by legacy VLB host drivers (+ convert them to use it). There should be no functional changes caused by this patch. Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com> --- drivers/ide/ide-probe.c | 32 ++++++++++++++++++++++++++++++++ drivers/ide/legacy/ali14xx.c | 28 +--------------------------- drivers/ide/legacy/dtc2278.c | 27 +-------------------------- drivers/ide/legacy/ht6560b.c | 28 +--------------------------- ...

Update IDE PMAC host driver to use drive->noprobe instead of hwif->noprobe and remove hwif->noprobe completely (it is always set to zero now). There should be no functional changes caused by this patch. Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com> --- drivers/ide/arm/icside.c | 1 - drivers/ide/ide-probe.c | 3 +-- drivers/ide/ide.c | 4 ---- drivers/ide/mips/swarm.c | 1 - drivers/ide/ppc/pmac.c ...

Especially when building with a separate object directory the following part of arch/cris/Makefile is not so nice since it creates links in the source directory: <-- snip --> # Create some links to make all tools happy $(SRC_ARCH)/.links: @rm -rf $(SRC_ARCH)/drivers @ln -sfn $(SARCH)/drivers $(SRC_ARCH)/drivers @rm -rf $(SRC_ARCH)/boot @ln -sfn $(SARCH)/boot $(SRC_ARCH)/boot @rm -rf $(SRC_ARCH)/lib @ln -sfn $(SARCH)/lib $(SRC_ARCH)/lib ...

I'm not claiming it needs it. But with this #include gcc can check that the prototypes in netlabel.h match the functions in netlabel.c The bugs this catches are rare, but when you have such a bug it can be cu Adrian -- "Is there not promise of rain?" Ling Tan asked suddenly out of the darkness. There had been need of rain for many days. "Only a promise," Lao Er said. Pearl S. Buck - Dragon Seed --

It doesn't _need_ the file in the strictest sense, the header file is just there for the core SELinux code to call into the NetLabel/SELinux glue code (what is in security/selinux/netlabel.c). However, if this is to conform better to kernel coding policy (I assume that is the case here?) then that is fine with me. -- paul moore linux security @ hp --

drivers/media/video/tda8290.c contains the following code: <-- snip --> ... static void tda8290_set_params(struct dvb_frontend *fe, struct analog_parameters *params) { ... unsigned char adc_sat, agc_stat, pll_stat; int i; set_audio(fe, params); if (priv->cfg.config) tuner_dbg("tda827xa config is 0x%02x\n", *priv->cfg.config); tuner_i2c_xfer_send(&priv->i2c_props, easy_mode, ...

This patch fixes an off-by-one error spotted by the Coverity checker. Signed-off-by: Adrian Bunk <bunk@kernel.org> --- f9829dd0d7dad9cbc08da3fceb105478e85c4137 foobar diff --git a/drivers/i2c/busses/i2c-amd756.c b/drivers/i2c/busses/i2c-amd756.c index 573abe4..2fa4318 100644 --- a/drivers/i2c/busses/i2c-amd756.c +++ b/drivers/i2c/busses/i2c-amd756.c @@ -335,7 +335,7 @@ static int __devinit amd756_probe(struct pci_dev *pdev, u8 temp; /* driver_data might come from user-space, so ...

This patch removes the !NO_TRUNCATE code that anyway required a manual editing of the code for being used. Signed-off-by: Adrian Bunk <bunk@kernel.org> --- This patch has been sent on: - 17 Feb 2008 fs/minix/inode.c | 3 --- fs/minix/minix.h | 6 ------ fs/minix/namei.c | 24 ------------------------ 3 files changed, 33 deletions(-) baff7307cff10a969567e2c7b81af14259170906 diff --git a/fs/minix/inode.c b/fs/minix/inode.c index 84f6242..523d737 100644 --- ...

This patch contains the following cleanups: - make the needlessly global file.c:configfs_add_file() static - #if 0 the unused inode.c:configfs_hash_and_remove() Signed-off-by: Adrian Bunk <bunk@kernel.org> --- This patch has been sent on: - 17 Feb 2008 fs/configfs/configfs_internal.h | 3 --- fs/configfs/file.c | 3 ++- fs/configfs/inode.c | 2 ++ 3 files changed, 4 insertions(+), 4 deletions(-) 651b7a33c0dce1790230fe8ba9b7445cd69c8bde diff --git ...

Every file should include the headers containing the externs for its global code. Signed-off-by: Adrian Bunk <bunk@kernel.org> --- eaeb567e2ce03ef3a7c5ed33e35def212ce2899a foobar diff --git a/security/selinux/ss/services.c b/security/selinux/ss/services.c index f374186..29fb216 100644 --- a/security/selinux/ss/services.c +++ b/security/selinux/ss/services.c @@ -40,6 +40,7 @@ #include <linux/sched.h> #include <linux/audit.h> #include <linux/mutex.h> +#include <linux/selinux.h> ...

Trying to compile the v850 port brings many compile errors, one of them exists since at least kernel 2.6.19. There also seems to be noone willing to bring this port back into a usable state. This patch therefore removes the v850 port. If anyone ever decides to revive the v850 port the code will still be available from older kernels, and it wouldn't be impossible for the port to reenter the kernel if it would become actively maintained again. Signed-off-by: Adrian Bunk ...

This patch makes the needlessly global scsi_end_bidi_request() static. Signed-off-by: Adrian Bunk <bunk@kernel.org> --- This patch has been sent on: - 13 Feb 2008 48e50a33ee11afb02eadc790ae2d9542b0805608 diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c index b12fb31..148c8b0 100644 --- a/drivers/scsi/scsi_lib.c +++ b/drivers/scsi/scsi_lib.c @@ -785,7 +785,7 @@ EXPORT_SYMBOL(scsi_release_buffers); * in req->data_len and req->next_rq->data_len. The upper-layer driver ...

The following functions can now become static: - rtc_interrupt() - rtc_get_rtc_time() Signed-off-by: Adrian Bunk <bunk@kernel.org> --- This patch has been sent on: - 13 Feb 2008 drivers/char/rtc.c | 5 +++-- include/linux/rtc.h | 2 -- 2 files changed, 3 insertions(+), 4 deletions(-) e642155f0065ec03a83ed95eef2a78f272e73e01 diff --git a/drivers/char/rtc.c b/drivers/char/rtc.c index 78b151c..39aba90 100644 --- a/drivers/char/rtc.c +++ b/drivers/char/rtc.c @@ -144,6 ...

yes on 64-bit. On 32-bit MK8 doesn't affect X86_HT. --

OK, you say "Don't mix two different semantics into one variable X86_HT", right? Thanks, Hiroshi Shimamoto --

he made X86_HT can be enabled when only Opteron supported is enabled. it should be stated in the comments. YH --

From: Hiroshi Shimamoto <h-shimamoto@ct.jp.nec.com> X86_HT is used for hyperthreading or multicore on 32-bit. The X86_HT on 64-bit is different from 32-bit, it means hyperthreading only. And X86_HT is not used on 64-bit except from cpu/initel_cacheinfo.c. Unify X86_HT for hyperthreading or multicore. Turn X86_HT on when X86_64 and SMP are enabled. Signed-off-by: Hiroshi Shimamoto <h-shimamoto@ct.jp.nec.com> --- arch/x86/Kconfig | 6 +++--- 1 files changed, 3 insertions(+), 3 ...

This patch includes page.h header into liker scripts that allow us to use PAGE_SIZE macro instead of numeric constant. To be able to include page.h into linker scripts page.h is needed for some modification - i.e. we need to use __ASSEMBLY__ and _AC macro Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com> --- WARNING: COMPLETELY UNTESTED !!! arch/um/kernel/dyn.lds.S | 7 ++++--- arch/um/kernel/uml.lds.S | 7 ++++--- include/asm-um/page.h | 17 +++++++++++------ 3 files ...

This patch includes page.h and thread_info.h headers into the linker script that allow us to use PAGE_SIZE and THREAD_SIZE macroses instead of numeric constants Also a few spaces removed Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com> --- WARNING: COMPLETELY UNTESTED !!! vmlinux.lds.S | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-) Index: ...

This patch includes page.h header into liker script that allow us to use PAGE_SIZE macro instead of numeric constant Also PERCPU macro is used instead of explicit section definition Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com> --- WARNING: COMPLETELY UNTESTED !!! vmlinux.lds.S | 22 ++++++++++------------ 1 file changed, 10 insertions(+), 12 deletions(-) Index: ...

This patch includes page.h header into liker script that allow us to use PAGE_SIZE macro instead of numeric constant Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com> --- WARNING: COMPLETELY UNTESTED !!! vmlinux-std.lds | 3 ++- vmlinux-sun3.lds | 7 ++++--- 2 files changed, 6 insertions(+), 4 deletions(-) Index: linux-2.6.git/arch/m68k/kernel/vmlinux-std.lds =================================================================== --- ...

This patch modifies page.h header to be able to include it into linker script and then we're able to use PAGE_SIZE in vmlinux.lds Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com> --- WARNING: COMPLETELY UNTESTED !!! arch/m68knommu/kernel/vmlinux.lds.S | 5 +++-- include/asm-m68knommu/page.h | 6 ++++-- 2 files changed, 7 insertions(+), 4 deletions(-) Index: ...

This series of patches is made in purpose to use PAGE_SIZE and THREAD_SIZE macroses in vmlinux linker scripts as much as possible instead of numeric constants. Please review. WARNING: the patches are UNTESTED COMPLETELY - I've no access to that architectures and errors is highly possible ;) If you have an ability to test this patches - please do it. What is most important on review - these macroses *must* be placed into appropriate places - check it please. ANY comments are ...

The disfunctionality of your dongle is a separate problem and I hope that the bluetooth wizards will help you to get this sucker running. The timer which is not deactivated before the data structure containing it is a simple bug, which needs to be addressed ASAP. Thanks again for your patience in tracking this down, tglx --

This makes the user_regset-based core dump code call user_regset writeback hooks when available. This is necessary groundwork to allow IA64 to set CORE_DUMP_USE_REGSET. CC: Shaohua Li <shaohua.li@intel.com> Signed-off-by: Roland McGrath <roland@redhat.com> --- fs/binfmt_elf.c | 15 +++++++++++++++ 1 files changed, 15 insertions(+), 0 deletions(-) diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c index 41a958a..eb7cca3 100644 --- a/fs/binfmt_elf.c +++ b/fs/binfmt_elf.c @@ -1424,6 ...

Priit's original patch used the LC_ALL trick. It is me to blame that it was replaced with [:alnum:] :-( Sam --

ok. I've queued up Roland's re-fix. We'll eventually converge to something everyone is happy with ;-) Ingo --

2008-2-28 ▼ 哇塞!!!!一片50 http://sam.yahoo-kiss.com/ 全年無休每日下午5點前下單-隔天即可宅配到貨 瘋狂大放送買兩套送ㄧ套 http://sam.yoyo-dvd.com/ 3:16:34 祔

It's inside his vfs git tree: http://git.kernel.org/?p=linux/kernel/git/viro/vfs-2.6.git;a=commit;h=49be4f8114e6ff0e... I'm fine with both approaches. - Davide --

> http://git.kernel.org/?p=linux/kernel/git/viro/vfs-2.6.git;a=commit;h=49be4f8114e6ff0e... > > I'm fine with both approaches. Both ways are OK with me too, although Al's change leaves the trap in the anon_inode_getfd() in that all users have to keep in mind the race against close() from another thread. Also Al's change moves all documentation to __anon_inode_getfd() and leaves anon_inode_getfd() undocumented, which is a little suboptimal. With Al's change the 2/2 ...

[CC-ing Al too] I believe Al changed the interface to not give out inode* and file*, *and* call fd_install() inside it. I'd slightly prefer Al version, although I don't see any major problems in this one too. - Davide --

> http://git.kernel.org/?p=linux/kernel/git/viro/vfs-2.6.git;a=commit;h=49be4f8114e6ff0e... Actually, looking closer at the kvm changes here, I think that create_vcpu_fd() needs the same treatment as kvm_dev_ioctl_create_vm() gets in the patch because of the race I mentioned in the changelog for my patch: otherwise kvm_vcpu_release() could drop the last reference to vcpu->kvm->filp before the get_file() gets an extra reference. I'm beginning to think that moving the ...

Hello, Currently at the start of a journal commit we loop through all of the buffers on the committing transaction and clear the b_modified flag (the flag that is set when a transaction modifies the buffer) under the j_list_lock. The problem is that everywhere else this flag is modified only under the jbd lock buffer flag, so it will race with a running transaction who could potentially set it, and have it unset by the committing transaction. This is also a big waste, you can have ...

huh, are you sure? The fixup-comments patch adds at least one char to the source file so the md5sums cannot be the same. -- Regards/Gruß, Boris. --

interdiff for the merged version diff -u b/drivers/ide/ide-cd.c b/drivers/ide/ide-cd.c --- b/drivers/ide/ide-cd.c +++ b/drivers/ide/ide-cd.c @@ -185,8 +185,7 @@ if (valid < 0) valid = 0; if (sector < get_capacity(info->disk) && - drive->probed_capacity - sector < 4 * 75) - + drive->probed_capacity - sector < 4 * 75) set_capacity(info->disk, sector); } } @@ -218,7 +217,7 @@ rq->data = sense; rq->cmd[0] = GPCMD_REQUEST_SENSE; - rq->cmd[4] = ...

Signed-off-by: Borislav Petkov <petkovbb@gmail.com> --- drivers/ide/ide-cd.c | 6 +++--- 1 files changed, 3 insertions(+), 3 deletions(-) diff --git a/drivers/ide/ide-cd.c b/drivers/ide/ide-cd.c index 8877b7b..4d39567 100644 --- a/drivers/ide/ide-cd.c +++ b/drivers/ide/ide-cd.c @@ -41,10 +41,10 @@ #include <scsi/scsi.h> /* For SCSI -> ATAPI command conversion */ -#include <asm/irq.h> -#include <asm/io.h> +#include <linux/irq.h> +#include <linux/io.h> #include ...

IDE is quite consistent with the above style but maybe it is just me being For this case there is a simple solution: fix ide-cd to use hwif->sg_table instead of having it to walk rq->bio's and the whole function will vanish. ;-) [ IOW We are putting way too much time into these coding style fixes and as the old code is replaced with the new one the coding style improves itself. ] Thanks, Bart --

Signed-off-by: Borislav Petkov <petkovbb@gmail.com> --- drivers/ide/ide-cd.c | 75 +++++++++++++++++++++++++++++++++---------------- 1 files changed, 50 insertions(+), 25 deletions(-) diff --git a/drivers/ide/ide-cd.c b/drivers/ide/ide-cd.c index bb501a4..84415cd 100644 --- a/drivers/ide/ide-cd.c +++ b/drivers/ide/ide-cd.c @@ -279,11 +279,11 @@ static void cdrom_end_request(ide_drive_t *drive, int uptodate) ide_end_request(drive, uptodate, nsectors); } -static void ...

Some of them are: WARNING: braces {} are not necessary for single statement blocks CHECK: multiple assignments should be avoided WARNING: printk() should include KERN_ facility level WARNING: no space between function name and open parenthesis '(' Signed-off-by: Borislav Petkov <petkovbb@gmail.com> --- drivers/ide/ide-cd.c | 46 ++++++++++++++++++++++------------------------ 1 files changed, 22 insertions(+), 24 deletions(-) diff --git a/drivers/ide/ide-cd.c b/drivers/ide/ide-cd.c index ...

interdiff for the merged version diff -u b/drivers/ide/ide-cd.c b/drivers/ide/ide-cd.c --- b/drivers/ide/ide-cd.c +++ b/drivers/ide/ide-cd.c @@ -296,7 +296,7 @@ struct request *rq = HWGROUP(drive)->rq; int stat, err, sense_key; - /* check for errors. */ + /* check for errors */ stat = ide_read_status(drive); if (stat_ret) @@ -305,7 +305,7 @@ if (OK_STAT(stat, good_stat, BAD_R_STAT)) return 0; - /* get the IDE error register. */ + /* get the IDE error register ...

So it omits other consoles (non-tty) intentionally? or does it include them even though the procfs filename contains "tty"? Anyway, I'd like to see something like this. Compare http://www.xenotime.net/linux/patches/consoles-list.patch --- ~Randy --

It only does consoles that have a tty_driver, as those are the ones I'd Reading that, unless I'm missing something, you'd lose the device name distinction (i.e., *which* serial port). Bill --

-- This is to officially inform you that(ATM Card Number;4278763100030014) has been accredited in your favour.Your Personal Identification Number is 2900. The ATM Card Value is $6.8MILLION USD Only. Contact:wallaceatmdesk@hotmail.com With the following,Your Name:Phone no:Address:Country --

-------------------------------------------------^^^^^^^^^^^^^^^^ The physical address looks strange: 0x47000EE000 Can you provide a full boot log please ? Thanks, tglx --

here's the updated patch. Ingo --------------> Subject: x86: check physical address range in ioremap From: Thomas Gleixner <tglx@linutronix.de> Date: Wed, 27 Feb 2008 20:57:40 +0100 Roland Dreier reported in http://lkml.org/lkml/2008/2/27/194 [ 8425.915139] BUG: unable to handle kernel paging request at ffffc20001a0a000 [ 8425.919087] IP: [<ffffffff8021dacc>] clflush_cache_range+0xc/0x25 [ 8425.919087] PGD 1bf80e067 PUD 1bf80f067 PMD 1bb497067 PTE 80000047000ee17b This is on a ...

That needs to be: + return addr < (1UL << boot_cpu_data.x86_phys_bits); Where is my brown paperbag ? I know that GCC _is_ stupid, but I never remember. tglx --

This one has 36bit physical address space. You can verify that via While the physical address of your ioremap is 47000ee000. 2^ 36 == 1000000000 ----> 47000ee000 So the fault is not very surprising. Unfortunately we do not check, whether physaddr is inside the valid physical address space. I whip up a patch to do that. Can you please instrument the driver to figure out where this information comes from ? Or maybe lspci tells already. Thanks, tglx --

> Where is my brown paperbag ? I know that GCC _is_ stupid, but I never > remember. Original patch actually worked fine and trapped the problem. I also instrumented the driver, and the problem is coming from the code: /* * Now we will check: * - SMBIOS checksum must be 0 * - intermediate anchor should be _DMI_ * - intermediate checksum should be 0 */ if ((bios_checksum(p, length)) && (strncmp((char *)eps->intermediate_anchor, "_DMI_", ...

What generation is the DL380 server? G5 I hope. Tom -----Original Message----- From: Roland Dreier [mailto:rdreier@cisco.com] Sent: Wednesday, February 27, 2008 11:08 AM To: linux-kernel@vger.kernel.org Cc: Ingo Molnar; Mingarelli, Thomas Subject: hpwdt oops in clflush_cache_range With an up-to-date git tree (head 7704a8b6) I'm seeing the oops below when the hpwdt driver gets loaded during boot on an HP DL380 system. I don't think I've seen any previous reports of this, and I guess ...

Last commit for unistd was not correct, it only had a partial update of syscall numbers for __NR_timerfd_settime and __NR_timerfd_gettime. Also, NR_syscalls was not incremented for the new syscalls. Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com> --- include/asm-cris/unistd.h | 6 +++--- 1 files changed, 3 insertions(+), 3 deletions(-) diff --git a/include/asm-cris/unistd.h b/include/asm-cris/unistd.h index 007cb16..76398ef 100644 --- a/include/asm-cris/unistd.h +++ ...

Function __copy_user_zeroing in arch/lib/usercopy.c had the wrong parameter set as __user, and in include/asm-cris/uaccess.h, it was not set at all for some of the calling functions. This will cut the number of warnings quite dramatically when using sparse. While we're here, remove useless CVS log and correct confusing typo. --- arch/cris/arch-v10/lib/usercopy.c | 2 +- arch/cris/arch-v32/lib/usercopy.c | 2 +- include/asm-cris/uaccess.h | 53 ...

The SIGCONT-handling related comment in handle_stop_signal() states: * If there is a handler for SIGCONT, we must make * sure that no thread returns to user mode before * we post the signal, in case it was the only * thread eligible to run the signal handler--then * it must not do anything between resuming and * running the handler. With the ...

Hi, this patch add a hardware watchdog device driver. It supports the 16 bit watchdog timer on superio chip it8712f-j, it 8712f-k, it8716f and it8726f. Today it was tested under vanilla 2.6.24.3, vanilla 2.6.24.2 and Mandriva with 2.6.24.2 and 2.6.24 kernel. Regards, Oliver

That's due to commit ce28b9 x86: fix vsyscall wreckage The proc_handler initialization is simply dropped for this table, but AFAIS from the commit comment this was done simply to disable the vsyscall_gtod_data.sysctl_enabled tuning. Did I miss something? --

Thanks. Can confirm that cherry-pick of this fix from that tree fixed it: commit ed19f1cd69aa82619e2def4e11dfece8f9fe7e31 Author: Thomas Gleixner <tglx@linutronix.de> Date: Wed Feb 27 09:39:52 2008 +0100 x86: restore vsyscall64 prochandler --

Reviewed-by: Sam Ravnborg <sam@ravnborg.org> And I also build tested the patch and checked the resulting .lds files. And they were semantically equal. Sam --

This makes the Xen console just work. Before, you had to ask for it on the kernel command line with console=hvc0 Signed-off-by: Markus Armbruster <armbru@redhat.com> --- diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c index 49e5358..df63185 100644 --- a/arch/x86/xen/enlighten.c +++ b/arch/x86/xen/enlighten.c @@ -25,6 +25,7 @@ #include <linux/mm.h> #include <linux/page-flags.h> #include <linux/highmem.h> +#include <linux/console.h> #include ...

Many (most of) sysctls do not have a per-container sense. E.g. kernel.print_fatal_signals, vm.panic_on_oom, net.core.netdev_budget and so on and so forth. Besides, tuning then from inside a container is not even secure. On the other hand, hiding them completely from the container's tasks sometimes causes user-space to stop working. When developing net sysctl, the common practice was to duplicate a table and drop the write bits in table->mode, but this approach was not very elegant, lead to ...

The do_sysctl_strategy isn't used outside kernel/sysctl.c, so this can be static and without a prototype in header. Besides, move this one and parse_table() above their callers and drop the forward declarations of the latter call. One more "besides" - fix two checkpatch warnings: space before a ( and an extra space at the end of a line. Signed-off-by: Pavel Emelyanov <xemul@openvz.org> --- include/linux/sysctl.h | 5 -- kernel/sysctl.c | 144 ...

Hi Linus, Raised by: http://bugzilla.kernel.org/show_bug.cgi?id=10095 , there is the question of whether we want to separate the env+arg arrays from the stack proper. Currently these arrays are considered part of the stack, and RLIMIT_STACK includes them. However POSIX does not specify it must be so. The complaint is that sysconf(_SC_ARG_MAX) returns a hard coded value (which is not obtained from the kernel) and might, depending on the RLIMIT_STACK setting, be invalid. POSIX disallows ...

TLB flushing at least on x86-64 should be already well optimized on its With cpusets and isolation etc. it is the normal case. -Andi --

That's possible, but it is slower and less scalable on my 8-way, and I suspect it might become even slower than the generic code on larger Why? --

*vmalloc* TLB flushing. void flush_tlb_all(void) { on_each_cpu(do_flush_tlb_all, NULL, 1, 1); } Of course we could use a new vector for it and speed it up a lot more, but after my vmalloc improvements I think that would be a waste of a Oh really? Coming from what callers? --

Hi, This isn't finished yet, however I'd just like to ask for comments. Design a new smp_call_function calling convention. This significantly improves scalability of smp_call_function_single, and slightly improves scalability and performance of smp_call_function; especially in the cases where wait=0. The non-wait case allocates function call data dynamically, so the caller is able to return before all other CPUs have copied out the data into their stack as is the case today. The ...

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=463669 I have the same problem - it is not debian specific. Did the semantics of kill() change with the new kernel? I thought as long as something is setuid, even with capability stuff around the setuid programs just get _all_ capabilities and would keep working. I did a good search and found many people with the problem, but no solutions except going back to 2.6.23.x kernels. I guess you'll flame me, but at least include a link to the ...

Would it be this commit that fixed this issue? commit 094972840f2e7c1c6fc9e1a97d817cc17085378e Author: Serge E. Hallyn <serue@us.ibm.com> Date: Sat Feb 23 15:23:33 2008 -0800 file capabilities: simplify signal check I see stable was CCed on it, so I guess it will be included in the next point release, but distributions may want to apply it earlier. --

Why would we flame you? I'll just apologize as I think it's my fault, and ask you to please try the newest available kernel where I believe it should be fixed. thanks, -serge --

My first try went in this direction. But at a second look I thought that the names ANY (255.255.255.255) and NONE (0.0.0.0) are indirect leading to confusion. The interpretation of the zero addresses in the DHCPDISCOVER packet is in the kind of 'no address' and not of 'any addressee'. So the meaning and usage of INADDR_ANY and ZERO may be totally different in this file. In my opinion it is necessary to distinguish between INADDR_ANY and the ZERO address which ist requested to be in the ...

On Wed, 27 Feb 2008 14:11:55 +0100 Just delete the lines and the address will be zero because of earlier memset(). --

From my personal point of view it is important to document the requirement of the two addresses to be 0.0.0.0 either through the code itself or through a comment which states clear the side effect of memset'ting the structure. Both roads lead to Rome :-) Marcel -- Marcel Wappler Bridgeco AG CH-8600 Dübendorf Switzerland --

doesnt apply to sched-devel.git - what's the plan here? Ingo --

--

This should be a do_each_thread() { } while_each_thread(); iteration loop; cgroups are thread oriented, not task. --

May I suggest: rm -rf / /me runs --

Subject: sched: rt-group: fixup schedulability constraints calculation div64_64() has a different calling convention than do_div() :/ fix a few untidies while were here; sysctl_sched_rt_period may overflow due to that multiplication, so cast to u64 first. Also that RUNTIME_INF juggling makes little sense although its an effective NOP. Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> --- kernel/sched.c | 10 +++------- 1 file changed, 3 insertions(+), 7 deletions(-) Index: ...

This patch fixes the following build error introduced by commit a79d8e93d300adb84cccc38ac396cfb118c238ad and reported by Olaf Hering: <-- snip --> ... LD .tmp_vmlinux1 arch/powerpc/sysdev/built-in.o: In function `of_add_fixed_phys': fsl_soc.c:(.init.text+0xd34): undefined reference to `fixed_phy_add' make: *** [.tmp_vmlinux1] Error 1 <-- snip --> Signed-off-by: Adrian Bunk <adrian.bunk@movial.fi> --- 47a94f1d74e448d39f34f0ad37f8b4f69412ee13 diff --git ...

these many special conditions in the Kconfig space are way too ugly. (Also, i can see no reason why for example hibernation wouldnt be possible on rdc321x, it's a pure software concept ...) Ingo --

no. Just leave those things alone. Obviously, if you dont initiate hibernation it wont be done - and you dont have to build it in either. But there's no fundamental reason that it _couldnt_ work - if a suitable device was attached. Dont put such hardcoded dependencies into the kernel ... Ingo --

Current tun/tap driver sets also net device's hw address when asked to change character device's hw address. This is a good idea, but it misses RTLN-locking, resulting following error message in 2.6.25-rc3's inetdev_event() function: RTNL: assertion failed at net/ipv4/devinet.c (1050) Attached patch fixes this problem. Signed-off-by: Kim B. Heino <Kim.Heino@bluegiga.com> --- linux-2.6.25-rc3/drivers/net/tun.c.foo 2008-02-27 10:06:28.000000000 +0200 +++ ...

From: Jon Schindler <jkschind@gmail.com> As part of the kernel janitor projected listed at http://kernelnewbies.org/KernelJanitors/Todo/ApiChanges I have removed some deprecated api calls. The deprecated init_module and cleanup_module functions have been removed from the following ...

Yes we'll get to it right away... Oh wait... Dave. --

From: Jon Schindler <jkschind@gmail.com> As part of the kernel janitor projected listed at http://kernelnewbies.org/KernelJanitors/Todo/ApiChanges I have removed some deprecated api calls. The deprecated init_module and cleanup_module functions have been removed from the following ...

--

- no need to check err == 0 - the last return value of cgroup_add_file() is ignored Signed-off-by: Li Zefan <lizf@cn.fujitsu.com> --- kernel/cpuset.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/kernel/cpuset.c b/kernel/cpuset.c index 3e296ed..1c5b2d2 100644 --- a/kernel/cpuset.c +++ b/kernel/cpuset.c @@ -1482,10 +1482,10 @@ static int cpuset_populate(struct cgroup_subsys *ss, struct cgroup *cont) if ((err = cgroup_add_file(cont, ss, &cft_spread_slab)) < ...

core fix is in x86.git#testing http://git.kernel.org/?p=linux/kernel/git/x86/linux-2.6-x86.git;a=commitdiff;h=ae2627a... it will make sure that apicid to node mapping is right. esp don't have SRAT or wrong SRAT, and your system have apicid lifting YH --

On Wed, 27 Feb 2008 17:03:00 +1100 Stephen Rothwell <sfr@canb.auug.org.au> = The astute will notice that the pata tree has been renamed to ide as this better reflects reality (according to Andrew). --=20 Cheers, Stephen Rothwell sfr@canb.auug.org.au http://www.canb.auug.org.au/~sfr/

Hi Jeff, Excellent. Thanks for that. --=20 Cheers, Stephen Rothwell sfr@canb.auug.org.au http://www.canb.auug.org.au/~sfr/

Hi Stephen, please add: git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4.git next This contains a subset of the patches in the ext4 patch set which we are explicitly planning on pushing to Linus at the next merge window. It actually probably wouldn't hurt to pull the entire quilt series into linux-next, since we don't export any interfaces that are used by other linux trees, and that would give us better heads up for conflicts in patches still in development conflicting ...

I keep the labeled networking patches for the upcoming (in this case 2.6.26) kernel here: * git://git.infradead.org/users/pcmoore/lblnet-2.6_testing Patches for the current kernel (in this case 2.6.25) live here: * git://git.infradead.org/users/pcmoore/lblnet-2.6 Both are currently empty. -- paul moore linux security @ hp --

Thanks, sorry for not sending mail sooner. -- paul moore linux security @ hp --

Hi Ted, Thanks, the less conflicts I have to fix the better. :-) I guess over time patches will migrate into your next branch or they will appear there after the next Linus release. --=20 Cheers, Stephen Rothwell sfr@canb.auug.org.au

Hmm, perhaps this is just hand-waving but why don't you have a kmalloc_reserve() function in SLUB that does the accounting properly? Pekka --

Yes, but alas there is no room to store such information in kmalloc(). That is, in a sane way. I think it was Daniel Phillips who suggested encoding it in the return pointer by flipping the low bit - but that is Yes, it would require a massive overhaul of quite a few things. I agree, Like mempools, we could add a wrapper with a mem_reserve and waitqueue inside, strip __GFP_WAIT, try, see if the reservation allows, and wait if not. I haven't yet done such a wrapper because it wasn't ...

Hi Peter, Thanks... the tree thing is starting to make sense, and I'm not confused my __mem_reserve_add any more :-) I've been having a closer read of some of the code that I skimmed over before and I have some more questions. 1/ I note there is no way to tell if memory returned by kmalloc is from the emergency reserve - which contrasts with alloc_page which does make that information available through page->reserve. This seems a slightly unfortunate aspect of the interface. ...

Make printk boot_delay use loops_per_jiffy directly, declare preset_lpj as static, and set lpj value to loops_per_jiffy in lpj_setup. Signed-off-by: Dave Young <hidave.darkstar@gmail.com> --- init/calibrate.c | 6 +++--- kernel/printk.c | 15 ++++----------- 2 files changed, 7 insertions(+), 14 deletions(-) diff -upr linux/init/calibrate.c linux.new/init/calibrate.c --- linux/init/calibrate.c 2008-02-27 10:54:15.000000000 +0800 +++ linux.new/init/calibrate.c 2008-02-27 ...

Make boot_delay use loops_per_jiffy directly, this does not change the original behaviour if apply the first patch before it. Signed-off-by: Dave Young <hidave.darkstar@gmail.com> --- kernel/printk.c | 15 ++++----------- 1 file changed, 4 insertions(+), 11 deletions(-) diff -upr linux/kernel/printk.c linux.new/kernel/printk.c --- linux/kernel/printk.c 2008-02-26 16:54:23.000000000 +0800 +++ linux.new/kernel/printk.c 2008-02-26 16:59:02.000000000 +0800 @@ -173,24 +173,14 @@ ...

Make preset_lpj static. Give preset lpj value to loops_per_jiffy in lpj_setup. Signed-off-by: Dave Young <hidave.darkstar@gmail.com> --- init/calibrate.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff -upr linux/init/calibrate.c linux.new/init/calibrate.c --- linux/init/calibrate.c 2008-02-27 10:54:15.000000000 +0800 +++ linux.new/init/calibrate.c 2008-02-27 10:54:26.000000000 +0800 @@ -9,10 +9,11 @@ #include <linux/init.h> #include <linux/timex.h> -unsigned ...

That's very interesting. It's only 64 bytes, but still, it's not supposed to have any different effect. Could you distill a test case for the GCC folks and file it in their bugzilla? -- Intel are signing my paycheques ... these opinions are still mine "Bill, look, we understand that you're interested in selling us this operating system, but compare it to ours. We can't possibly take such a retrograde step." --

When DEBUG is not defined, pr_debug and dev_dbg and some other local debugging functions are specified as: "inline __attribute__((format (printf, x, y)))" This is done to validate printk arguments when not debugging. Converting these functions to macros or statement expressions "do { if (0) printk(fmt, ##arg); } while (0)" or "({ if (0) printk(fmt, ##arg); 0; }) makes at least gcc 4.2.2 produce smaller objects. This has the additional benefit of allowing the optimizer to avoid calling ...

$ gcc --version gcc (GCC) 4.2.2 20071128 (prerelease) (4.2.2-3.1mdv2008.0) It's definitely odd. The .o size changes are inconsistent. Some get bigger, some get smaller. The versioning ones I understand but I have no idea why changes in drivers/ or mm/ or net/ exist. I think it's gcc optimization changes, but dunno... Any good ideas? $ git reset --hard HEAD is now at 7704a8b... Merge branch 'for-linus' of git://oss.sgi.com:8090/xfs/xfs-2.6 $ make mrproper ; make defconfig ; make > ...

allyesconfig should be able to capture any text savings that this patch My x86_64 defconfig with gcc 4.0.3 had no difference in text size after applying your patch, yet the same config on gcc 4.1.2 did: text data bss dec hex filename 5386112 846328 719560 6952000 6a1440 vmlinux.before 5386048 846328 719560 6951936 6a1400 vmlinux.after --

When I did the same comparisons on my x86_64 defconfig with gcc 4.1.3, I What's interesting about this is that it doesn't appear to be related to your change (static inline function to macro definition). It appears to be simply removing the static inline function. The only reference to __simple_attr_check_format() in either the x86 or x86_64 defconfig is via DEFINE_SIMPLE_ATTRIBUTE() in fs/debugfs/file.c. If you remove the only reference to it: diff --git a/include/linux/fs.h ...

Hi. Thanks for resolving this guys. -- Evgeniy Polyakov --

Here's a series of twelve patches to introduce down_killable(). I expect the first ten to be uncontroversial, and easy to apply. The last two ... I'm still testing. More testers would be welcome, particularly people who have performance test-suites they wouldn't want to see regress. The way I go about introducing down_killable is to replace all the carefully hand-tuned assembler with a generic C implementation. One of the annoying things about this is that the header file is ...

Hello, this patch allows to prevent linux from using the ram below PHYSICAL_START. The "reserved RAM" can be mapped by virtualization software with to create a 1:1 mapping between guest physical (bus) address and host physical (bus) address. This will allow pci passthrough with DMA for the guest with current production hardware that misses VT-d. The only detail to take care of is the ram marked "reserved RAM failed". The virtualization software must create for the guest an e820 map ...

Something for Gautham and the USB people to scratch their heads over. --

Same question. -- paul moore linux security @ hp --

Hi!, My pleasure. You'll find them consistent in the next send. [Same release_ctx() consistency point] Nice spot, will do. 's' will be changed too, context is already used as 'ctx' in the whole file anyway. Thanks a lot for such a deep review :). I'll modify spotted problems as suggested, reorder the patch in a non compilation-breaking order, and keep you informed. Modified patches will be sent with the new ones that will complete the separation not to add too much noise in ...

For what is worth, I seem to suffer from the very same problem here in a different setting: trying to run a SUSE kernel-head i586 build of 2.6.24.1, which oopses reliable during initrd. BUG: unable to handle kernel NULL pointer dereference at virtual address 00000040 printing eip: c011f96c *pde = 00000000 Oops: 0000 [#1] SMP last sysfs file: /block/sde/sde1/dev Modules linked in: sata_sil24 libata 3w_9xxx sd_mod scsi_mod Pid: 538, comm: udev Not tainted (2.6.24.1-35.1-default #1) EIP: ...

*msr_safe_*() can only trap #GP's, which would be manifest as oopses if you used the unsafe versions. If the CPU *hangs* when touching an MSR, nothing we can do in software will help. -hpa --

Instead, start to encourage newcomers to help us fix bugs. Thanks. -- Stefan Richter -=====-==--- --=- ==-== http://arcgraph.de/sr/ --

The proposed two space change is ugly. Can someone NAK it? -sb (The elisp guy who added the linux c-style) --

I think you do misunderstand something. The "linux" emacs mode as I know it does for the following example 1: function(arg1, 2: arg2); in line 2 two TABs and one space. Whereas the proposal in discussion wants to have there one TAB and nine spaces. Thanks Guennadi --- Guennadi Liakhovetski --

Really nitpick, but (since it is already in the patch) wouldn't it be Thumbs up thanks Richard Knutsson --

Then why do we define rules that we know that nobody will (be reasonably able to) adhere to? I can see the reasoning behind using spaces instead of tabs in continuation lines and in in-line indentation like int x = 1; void *c = NULL; or #define X 1 #define XY 2 although I personally perfer TABs there too. But as long as we do not want to actually impose that, I don't think we should define that. Whereas an explicit rule requesting to use TABs and not spaces for ...

Not sure about the PI-reference in that context, but the rest IMO looks good. --

Not sure if it can be counted as an overwhelming need, just actually defining the variable-wide character to be use "properly". Saw the way through CodeStyle as the correct way to get the "replace 8-space alignment to tab" nulled (then fix/change checkpatch.pl). Actually, I see this as an quite simple question: is tab _the indention-marker_? If not, then why the refusal of spaces indents (or mix)? --

I agree and a NAK is a vote for not changing anything. Thanks. -sb --

Unless I'm misunderstanding something completely: Add "indent-tabs-mode: nil" in a "Local Variables:" section. But since Linux-style has that since ages (if not ever), that should be set for am emacs'ens linux-mode anyway. Bernd -- Firmix Software GmbH http://www.firmix.at/ mobil: +43 664 4416156 fax: +43 1 7890849-55 Embedded Linux Development and Services --

i'm wondering, have you seen this happen in practice? We use sys_set_thread_area() for every new task started up. I guess we havent seen problems in the field yet because this early during startup tasks do not normally receive signals? (or if they do they are fatal and no user signal context is used.) btw., gcc 4.2.3 doesnt do it due to CONFIG_FRAME_POINTERS=y here, and most distros turn on frame pointers. btw., this whole thing of us having to notice such tail-optimization ...

That sounds like a good idea to me. Thanks, Roland --

Already merged and sent to davem... Thanks, John -- John W. Linville linville@tuxdriver.com --

I spent some time thinking about this one. The original code is ok if led_out is always either 0x01 or 0x00. But what if it is eg 0xc0? Then the negation amounts to the negation of a nonzero number, so the result is 0. So the result of the bit and is 0. So the last bit is not flipped. But I don't know what is the range of led_out. If it is always 1 or 0, then why bother with the bit and? julia --

Indeed. Applied, thanks. --

Indeed, 2.2.1 fixes this issue for me. -- Stefan Richter -=====-==--- --=- ==-== http://arcgraph.de/sr/ --

s/sysfs attribute/module parameter/ Sorry, too early in the morning for me. --

Hi, Ok, so a sysfs attribute would be sufficient, I suppose? I don't have the hardware though, so I don't even know if the jumbo frame code actually works at this point hence the RFC. So I'll wait for someone to test this before sending a new patch. Thanks for the review! Pekka --

(cc's added) --

This is not 3ware-specific, but kernel 2.6.24 has new per-device write throttling that might help with the responsiveness issue: http://kernelnewbies.org/LinuxChanges#head-92340ffcec39e7c2a09fd933243fb18eda57f1b4 http://lwn.net/Articles/245600/ Also, check to see if the 3ware controller has a background initialize or verify in progress, since that will obviously slow things down until it is complete. Tony --

Actually, it's normal for pdflush to spawn up to 8 threads when you're dirtying memory faster than it can be written to disk. Load going to 4 Does the machine become unresponsive during the single-threaded test, or only when doing the rsync? -- Chris --

The machine becomes sluggish also when I write directly to the raid array. A simple dd if=3D/dev/zero of=3Dtmpfile Single threaded throughput seems to be ok (140M/s). The problem is that the machine becomes unresponsive. Andre --=20 The only person who always got his work done by Friday was Robinson Crusoe

Yes, but we tried both 2.6.24 and 2.6.25-rc, so Peter's new write-throttling code doesn't seem to help much in our situtation. I'll play a bit with the various /proc/sys/vm/* knobs to see if that makes That's certainly not the case. Thanks Andre --=20 The only person who always got his work done by Friday was Robinson Crusoe

(cc linux-usb) --

Well, if you think it is resonable to pass u64 into function, which is supposed to sleep no more than several cpu cycles. I do not want to Absolutely. -- Evgeniy Polyakov --

well found. Something like this: --- a/include/linux/delay.h~a +++ a/include/linux/delay.h @@ -7,10 +7,12 @@ * Delay routines, using a pre-computed "loops_per_jiffy" value. */ -extern unsigned long loops_per_jiffy; +#include <linux/kernel.h> #include <asm/delay.h> +extern unsigned long loops_per_jiffy; + /* * Using udelay() for intervals greater than a few milliseconds can * risk overflow for high loops_per_jiffy (high bogomips) machines. The @@ -32,7 +34,11 @@ ...

On Tue, Feb 26, 2008 at 01:21:00PM +0100, Martin Michlmayr wrote: References to __divdi3 / __udivdi3 are becoming a somewhat regular bug. I've created a patch to add these to the kernel but I'd rather not push it unless I have to. 64-bit operations but especially divisions are slow on 32-bit hardware so undefined references serve as an important detector. Ralf --

This is similar to my patch, which didn't fix the problem. Adrian already fixed it. --

Ah! My apologies, I managed to misunderstand that. -- Matthew Garrett | mjg59@srcf.ucam.org --

In theory we should be able to get this kind of information from ACPI, but I've never seriously looked at that. I'm not sure if anyone has ... Thanks, Rafael --

I suspect nv_adma_register_mode failing may be the key. I don't know why this happens. It could be the timeout needs to be longer but I've tried increasing it significantly without apparent effect. (Essentially it's waiting for the status to go to the IDLE state, and then clears the GO bit and waits for LEGACY to be set. In this case both waits timed out.) --

Just to make sure I didn't goofed up the bisection... I bisected between v2.6.24 and v2.6.25-rc2. In the midst of the bisection, make install decided to call the new version 2.6.24-rc7-gXXXX. Is that ok? I figured rc7+delta was before 24-final, hence outside the bisection? After that event I got the normal series of good god bad good... So I figured we were on the right track anyway... /A --

On Wed, Feb 27, 2008 at 12:33 AM, Andrew Morton # CONFIG_CLASSIC_RCU is not set 2.6.25-rc3 with some bluetooth debug code of mine --

[Empty message]

You can use the latest iwlwifi driver with kernel 2.6.24 when using the compat-wireless project (http://linuxwireless.org/en/users/Download). You can also load the driver with debugging (modprobe iwl4965 debug=0x43fff) to get more info about what happens during module unload. Reinette --

hm, intersting. but sysctl parameter is more better, i think. OK, I'll add it at next post. --

On Tue, 26 Feb 2008 22:18:38 +0100 I think it doesn't make sense for fairness. IMHO, this functionality is an unfair one in nature. While someone is reclaiming pages, other processes can get a newly reclaimed page without calling try_to_free_page. For high-priority processes, 1. avoiding diving into try_to_free_pages if it's congested. 2. just waiting for that someone relcaim pages and grab it ASAP maybe good for quick work. Thanks, -Kame --

I think sysctl should be interesting. The config option provides good documentation, but it is static in nature (requires reboot to change). I wish we could have the best of both worlds. -- Warm Regards, Balbir Singh Linux Technology Center IBM, ISTL --

Agghhh, sorry ;-) Hmmm may be, we don't need perfectly fair. because try_to_free_page() is unfair mechanism. but I will test use wake_up() instead wake_up_all(). it makes so so fair order if no performance regression happend. Thanks very useful comment. - kosaki --

You mentioned CONFIG_NUM_RECLAIM_THREADS_PER_CPU and not CONFIG_NUM_RECLAIM_THREADS_PER_NODE. The advantage with syscalls is that even if we get the thing wrong, the system administrator has an alternative. Please look through the existing sysctl's and you'll see what I mean. What is wrong with providing the flexibility that comes with sysctl? We cannot possibly think of all situations and come up with the right answer for a heuristic. Why not come up with a default and let everyone use what ...

That is why the proposal was made to make this a static configuration option, such as CONFIG_NUM_RECLAIM_THREADS_PER_NODE, that will handle both The benchmark that have been posted suggest that memory locality is more important than lock contention, as I've already mentioned. David --

We lack node hotplug, so the dependence on the number of system nodes in the equation is static and can easily be defined at compile-time. I agree that the maximum number of parallel reclaim threads should be a function of cpus, so you can easily make it that by adding callback functions for cpu hotplug events. Perhaps a better alternative than creating a set of heuristics and setting a user-defined maximum on the number of concurrent reclaim threads is to configure the number of ...

On Tue, 26 Feb 2008 23:56:39 -0800 (PST) Hmm, do you know there is already zone-hotplug ? ;) (Means, onlining new memory in new zone increase the # of zones. I agree that what is the best is based on benchmark. I like per-node, now. I hear that there is also I/O bottle-neck for page reclaiming, at last. Thanks, -Kame --

One more thing, I would request you to add default heuristics (number of reclaimers), based on the number of cpus in the system. Letting people tuning it is fine, but defaults should be related to number of cpus, nodes and zones on the system. Zones can be reaped in parallel per node and cpus allow threads to run in parallel. So please use that to come up with good defaults, instead of a number like "3". -- Warm Regards, Balbir Singh Linux Technology Center IBM, ISTL --

Things are changing, with memory hot-add remove, CPU hotplug , the topology can change and is no longer static. One can create fake NUMA nodes on the fly using a boot option as well. Since we're talking of parallel reclaims, I think it's a function of CPUs and Nodes. I'd rather keep it as a sysctl with a good default value based on the topology. If we end up getting it wrong, the system administrator has a choice. That is better than expecting him/her to recompile the kernel and boot that. ...

I agree with Balbir, just turn it into a sysctl, its easy enough to do, and those who need it will thank you for it instead of curse you for hard coding it. --

sorry, I don't understand yet. I think my patch is already function of CPUs and Nodes. per zone limit indicate propotional #cpus and #nodes. please tell me the topology that per zone limit doesn't works so good. I think boot option and sysctl should be used only while -mm for get various feedback. end up, we should select more better default, and remove sysctl. - kosaki --

No, but I thought that we had already been thru this... --- ~Randy --

From: Greg KH <gregkh@suse.de> Greg, look more carefully, it's because of the patch us networking folks asked you yesterday to revert. --

Marcel/Dave, can you or someone else followup with the bug reportor to get their dongle working now that the OOPS problem is fixed? Let's not lose track of this bug, thanks. --

can you please wait for a confirmation from Quel that this fixes it. My ACK is based on that he confirms that it fixes it for sure. Regards Marcel --

That won't work. Variables at beginning of function take only ~600 bytes, There's "noinline" attribute in linux/compiler.h (compiler-gcc.h actually) Nice idea. I'll try 2nd and 4th approaches. Marcin Slusarz --

David Miller writes: > From: Mikael Pettersson <mikpe@it.uu.se> > Date: Tue, 26 Feb 2008 09:55:50 +0100 > > > Minor update: rc2-git7 has the slow initial console behaviour, > > but successfully switches to the framebuffer. rc2-git8 however > > hangs in the console handover. So I'll bisect git7->git8 next. > > Between the VT layer registering it's console and the atyfb > driver initializing we get a crash, and it happens on all > sparc64 systems. It is caused by this commit and I ...

I would have preferred not to. The hassle is that we need to figure out, whether it works or not _before_ any user space program can use the interfaces. We could omit the check for archs where the Ok. Thanks, tglx --

David Miller writes: > From: David Miller <davem@davemloft.net> > Date: Tue, 26 Feb 2008 16:49:00 -0800 (PST) > > [ Thomas, forgot to CC: you earlier, changeset > a0c1e9073ef7428a14309cba010633a6cd6719ea ("futex: runtime enable pi > and robust functionality") broke sparc64. ] > > > From: Mikael Pettersson <mikpe@it.uu.se> > > Date: Tue, 26 Feb 2008 09:55:50 +0100 > > > > > Minor update: rc2-git7 has the slow initial console behaviour, > > > but successfully switches to the ...

From: Thomas Gleixner <tglx@linutronix.de> No, I'm saying it would be better not to take faults purposefully in the kernel address space. We don't have a usable user address space setup at this point in the boot, so using USER_DS would be even worse. I think I'll just add a different version of the sanity check to this sparc64 code later on, one that will take into consideration this KERNEL_DS case because I can see how it could be useful in other Crap, I'll need to push my fix there ...

Thanks for bisecting this. Please, don't forget to CC the author of the commit, whenever you finish your bisection (this time it is Ingo). Added to CC. -- Jiri Kosina SUSE Labs --

Yes, that was it, thanks!! - Ted --

Nope. Please stop pedalling complete fiction --

Right. Must have been somewhere else. -- Krzysztof Halasa --

How do I know revision of git-sched.patch of 2.6.25-rc2-mm1? rc2-mm1 + that patch doesn't boot too. stop at the same point ;) --

I'm not sure. The core moves the device to dpm_active only after ->resume() has run. Thus, if ->resume() registers new children, the ordering of I'm not sure if we need to do it. It's always been like this, so the current drivers' ->suspend() and ->resume() don't unregister the device they're called Yes, it does. Thanks, Rafael --

.../... Yup, old problem, I think I've said a long time ago that it's the reponsibility of bus drivers (such as USB khub) to stop issuing device additions when suspend is in progress. It might be possible to still do No, I think the global notifier is plenty for now. Later on, we can re-add early-suspend, late-resume, or we can finally turn the whole thing into a recursive call tree where the parents are the ones calling the children suspend :-) But for now, I think the global notifier ...

Nice presentation ... thanks. I agree with your patch to fix the MPOL_PREFERRED | MPOL_F_RELATIVE_NODES case. -- I won't rest till it's the best ... Programmer, Linux Scalability Paul Jackson <pj@sgi.com> 1.940.382.4214 --

It was already a 16-bit value, that is unchanged, so no space savings is intended. It is not possible to use the enum here because an additional unsigned short member to hold the mode flags is added to this struct later. The increase in size of an int compared to an unsigned short would make the entire struct bigger; the alignment requirement of the union no longer prevents that. After this patchset, it is 24 bytes in size while using unsigned short compared to 32 bytes while using ...

The point here is to have a 16 bit value? There are no space savins due to the alignment requirement of the union. Isnt it possible to use the enum here? If not then what is the point of the enum? --

Cool, this is precisely what I'd expect and was hoping to see. thanks, --

If I get into kconfig hacking again I will try to look into fixing kconfig so we can have all Kconfig files in one hirachy. This would gives us useful information such as dependencies on unknow symbols and the like. Statistics as you describe could be relevant to find bad Kconfig design but I see this much more as a check thing for bad design than something to make it easier to reach more configs with 100 randconfigs. Anyway - this is just vapourware until someone starts to hack on ...

On Wed, 2008-02-27 at 11:54 +0000, Nuno Tavares wrote: FYI, DELL often changes its hardware for the same model number. Xav --

using HCI_OP_SETUP_SYNC_CONN doesn't mean eSCO. It is perfectly fine to request SCO links via that command. The difference here is Inside the kernel it is not called BlueZ :) It simply is the Bluetooth subsystem and in case the HCI core. Regards Marcel --

so the patch is fine, but the description of what this patch is doing, why it is doing it this way and what it fixes is fully missing. Please update it with a proper description. Regards Marcel --

On Mon, 25 Feb 2008 02:41:40 +0100 Actually, that patch doesn't fix the real problem -- it only makes the Oops output a bit more sensible. Ben, did you manage to reproduce it with that patch applied? Haavard --

On Sun, Feb 24, 2008 at 4:34 PM, Stephen Neuendorffer Acked-by: Grant Likely <grant.likely@secretlab.ca> Josh, can you please pick this up? Thanks, g. -- Grant Likely, B.Sc., P.Eng. Secret Lab Technologies Ltd. --

Ah, davej has already updated to -git1. Excellent. Thanks! Kyle --

Hi Andi, Can you provide the complete 'raw' e820 info? This is at least not complete, and also might be after the sanitation. Do you mean that (1) there is usable RAM somewhere between 0xa0000 and 0xd2000? Or that (2) this second line should be RAM, not reserved? Case (1) would surpise me, because I expect the VGA adapter (which I assume is there...) to occupy 0xa0000 to 0xc0000 for the framebuffer. Also, your case would be a lot stronger if there were a line that explicitly indicated ...

So you've basically wasted a lot of memory with no concrete breakage known ...Please reconsider. -Andi --

It's not needed, the e820 maps are always correct for modern systems in The i386 kernel did this always, but I intentionally removed it from the 64bit kernel because all the modern BIOS seem to correctly report holes in this area. Only didn't do it on i386 because there were some concerns of very old systems not doing this correctly. My suspicion is that modern Windows systems rely on this, that is why BIOSes typically get it correct now. I think it should be only undone if you have a ...

Oh, sorry, I got lost in the set of patches in the message. To be clear, my little 4-patch uid-ns-signal patchset can simply be updated to make the cap_task_kill() uid check into if (task_user_equiv(current, p) But Eric if you simply drop cap_task_kill() (don't make it return 0, just drop the function and go back to not setting task_kill in the capability_security_ops) I'll ack that. Else I'll write the patch thursday. At this point the only thing that will be denied by cap_task_kill() but ...

I haven't done too much looking at what the tool on that site does, but I was under the impression that I should be able to do: echo mem > /sys/power/state and then bring the system back up and have it work. If that is not expected to work, then I won't bother trying it again. My goal in all this is to help test the kernel, and provide feedback and testing for features that are expected to work. If this is expected to work, then I would rather provide testing for patches to the kernel ...

And here is the trace from the kernel with the revert and patch: Feb 26 20:42:14 alekhine kernel: [ 278.487844] PM: Syncing filesystems ... done. Feb 26 20:42:40 alekhine kernel: [ 278.523599] Freezing user space processes ... (elapsed 0.00 seconds) done. Feb 26 20:42:40 alekhine kernel: [ 278.525885] Freezing remaining freezable tasks ... (elapsed 0.00 seconds) done. Feb 26 20:42:40 alekhine kernel: [ 278.526196] ACPI: Preparing to enter system sleep state S3 Feb 26 20:42:40 alekhine ...

Not if you want video to come back on most hardware - the kernel has no idea how to reinitialise your graphics card. s2ram and similar tools attempt to deal with this by reprogramming it from userspace. On the other hand, if it did work on the same hardware with earlier kernels, then that's a bug. -- Matthew Garrett | mjg59@srcf.ucam.org --

To suspend to ram, I do Ctrl-Alt-1 to switch from the GDM login screen to the console, and then: /etc/init.d/gdm stop echo mem > /sys/power/state I have also tried it from X in the past (using the Ubuntu quit menu option), with just about the same result, as I recall. This time I figured I would try to make it easier for the suspend code and try it from the console. Should I have been doing something else? -- Kevin Winchester --

I think it's enough now. I tested boot_delay yestoday, and found the result is different with before version. Even with lpj preset the delay is very slow, 5-10 seconds for every printk. Maybe this is what davej said. --

redundant 'break' --

I can't say I really like Alexey's patches, but I guess they are OK. I think I like the "b43_pci_bridge as a module" option a little better, but as Michael illustrated there are problems with that approach too. I'll apply the patches. John -- John W. Linville linville@tuxdriver.com --

Hi Nikola, Please put support@areca.com.tw in the loop. I am sure Areca support, Kevin, has taken over your case. If you like, please let him know your configuration and operations to synchronize both sides. Thank you for your patience and sorry for your inconvenience, -----Original Message----- From: Zan Lynx [mailto:zlynx@acm.org] Sent: Wednesday, February 27, 2008 5:04 AM To: Nikola Ciprich Cc: Andrew Morton; linux-kernel@vger.kernel.org; linux-scsi@vger.kernel.org; Nick Cheng; Erich ...

On Fri, 22 Feb 2008 17:12:17 -0800 For what reason? --

On Fri, 22 Feb 2008 10:14:15 -0800 Unrelated to your change, but... We prefer min_t over casts like this (not sure why, but let's be consistent) The reason for this cast is that all this code is using `int' for the transfer count (arguably wrong - should use size_t, but that propagates all the way to tty_operations.write) but cyclades has #define SERIAL_XMIT_SIZE (min(PAGE_SIZE, 4096)) which is unsigned on most architectures. A possibly-sane fix for all of this is to use min_t ...

On Wed, 27 Feb 2008 15:38:07 -0800 Fix sparse, I'd say. Probably hard to do, but that warning really isn't useful. --

Well, we can at least get rid of the warnings when min/max are nested. I sent you a patch earlier today that uses different temp variables in min/max/min_t/max_t to avoid this. min/min and max/max nesting will always see this. I also introduced clamp()/clamp_t() in that patch min3()? No _good_ ideas I'm afraid. Cheers, Harvey --

There are several reasons to not do a seperate module. First one being: People won't load it and complain about a regression. Yeah, stupid stuff like that happens all the time. That's the reason we SELECT the ssb code instead of using DEPENDS ON. People will otherwise not enable it and report regressions. And please don't say this won't happen. It _did_ happen when b44-PCI was a seperate kconfig option. People reported regressions, although there were none. Second one being: It's overkill ...

See the comment in b43_pci_bridge.c -- Greetings Michael. --

Thanks, Nick. Acked-by: Serge Hallyn <serue@us.ibm.com> --

Any decision made on what to do about this one? Mike or I can repost the per cpu allocator against mm? The fix by Eric could be used in the interim for 2.6.24? --

SLAB_HWCACHE_ALIGN *is* effectively a maybe align flag given the above code. If we all agree then we could change this to have must have semantics? It has the potential of enlarging objects for small caches. SLAB_HWCACHE_ALIGN has an effect that varies according to the alignment requirements of the architecture that the kernel is build on. We may be in for some surprises if we change this. --

Acked-by: Christoph Lameter <clameter@sgi.com> --

Bugger, went splat... forgot preformat for patchlet insert. <quiltuple checks> Fix CD burning regression introduced by 6b00769fe1502b4ad97bb327ef7ac971b208bfb5. raw_data_len must be updated to reflect residual data upon IO completion because it is used by blk_complete_sghdr_rq() to set hdr->resid which eventually becomes visible to userland. Signed-off-by: Mike Galbraith <efault@gmx.de> diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c index ba21d97..7a6f784 ...

I would love to get an answer as to what data_len (and of course raw_data_len) should be set to AFTER the command completes, which is what is going on here. I can see the above being correct -- scsi_get_resid() returns the length of the left-over data after the command is processed -- but I am mainly curious why setting [raw_]data_len matters after I/O completion. Jeff --

Hm. After rummaging around some more in both kernel and userland, I think this patchlet is not only functional, but (random accident) technically correct. What the heck, let's see if it flies... snippet from userland: /* * Return the residual DMA count for last command. * If this count is < 0, then a DMA overrun occured. */ EXPORT int scg_getresid(scgp) SCSI *scgp; { return (scgp->scmd->resid); } This function is used all over the place in cdrecord to determine transfer ...

SGkgSWxwbywKCk9uIFdlZCwgRmViIDI3LCAyMDA4IGF0IDEwOjI5IEFNLCBJbHBvIErDpHJ2aW5l bgo8aWxwby5qYXJ2aW5lbkBoZWxzaW5raS5maT4gd3JvdGU6Cj4gSSBkaWQgaW4gdGhlIHBhc3Qg YSBkZWJ1ZyBwYXRjaAo+ICB0aGF0IHZlcmlmaWVzIFRDUCdzIHdyaXRlIHF1ZXVlIHN0YXRlIGJ5 IHRoZSBoYXJkIHdheSwgaWUuLCBieSBicnV0ZWZvcmNlCj4gIHdhbGtpbmcgb2Z0ZW4gZW5vdWdo IHRvIGNhdGNoIGluY29uc2lzdGVuY2llcyBlYXJseSBlbm91Z2ggdG8gZmluZCBvdXQKPiAgdGhl IHJvb3QgY2F1c2UuCgpBcmUgeW91IHRhbGtpbmcgYWJvdXQgdGhpcyBvbmU6Cmh0dHA6Ly9tYXJj LmluZm8vP2w9bGludXgtbmV0ZGV2Jm09MTE5NDgy ...

Could you remove the OHCI1394_HCControl_postedWriteEnable flag from fw-ohci.c::ohci_enable() and test without and with the endianess patch? -- Stefan Richter -=====-==--- --=- ==-== http://arcgraph.de/sr/ --

Sure, added to the todo list along with making sure things survive bus resets. -- Jarod Wilson jwilson@redhat.com --

This is a multipart MIME message. --==_Exmh_1204139192_52520 Content-Type: text/plain; charset=us-ascii I think it exploded again, for some other reason... I just gave it a try just for grins-n-giggles. The resulting kernel got loaded by grub, the screen cleared, and about 5 seconds later it rebooted. Never got as far as putting up penguin icons. Tried again with netconsole, early_printk=vga, and initcall_debug, and it *still* didn't live long enough to produce anything resembling ...

I dream of being a card carrying member of the club but apparently Or their own profiling infrastructure! OK I'm done being a pain, time to be constructive. It's becoming clear we have some work to do in order to make oprofile easier to use. I've been involved with the project from the early days, and it's hard to be objective when it comes to usability issues. Off the top of my head, there are a number of reasons I think it makes sense to use the existing oprofile kernel code instead ...

Ok so it somehow works slowly with GRU and you are happy with it. What Would it not be better to have a solution that fits all instead of hacking something in now and then having to modify it later? Hmmm.. There were earlier discussions of changing the anon vma lock to a rw lock because of contention issues in large systems. Maybe we can just generally switch the locks taken while walking rmaps to semaphores? That would still require to put the invalidate outside of the pte ...

--

Hello, I hope this will can be considered final for .25 and be merged. Risk is zero, the only discussion here is to make an API that will last forever, functionality-wise all these patches provides zero risk and zero overhead when MMU_NOTIFIER=n. This last patch covers KVM and GRU and hopefully all other non-blocking users optimally, and the below API will hopefully last forever (but even if it lasts just for .25 and .26 is changed that's fine with us, it's a kernel _internal_ API anyway, ...

The skeletons shows how to do most of it using a spinlock and a counter. --

Yes, this one does. But Greg's original idea, that disabling CONFIG_SYSFS_DEPRECATED would eliminate one of the warnings doesn't hold, all three of them are still there (after your patches): kobject (c7c84700): tried to init an initialized object, something is seriously wrong. Pid: 1, comm: swapper Not tainted 2.6.25-rc2-testpm-dirty #40 [<c01ea105>] kobject_init+0x75/0x80 [<c025093e>] device_initialize+0x1e/0x90 [<c02510bb>] device_register+0xb/0x20 [<c01f3df8>] ...

On Wed, 27 Feb 2008 14:46:05 +0100 (CET), I'm afraid I can't help there, as this looks PCI-related... Greg: Attached are the fixes against your tree with proper description and s-o-b. Driver core: Fix cleanup when failing device_add(). -Don't call cleanup_device_parent() if we didn't call setup_parent(). -dev->kobj.parent may be NULL when cleanup_device_parent() is called, so we need to handle glue_dir == NULL in cleanup_glue_dir(). Signed-off-by: Cornelia Huck ...

No, didn't help, same 3 stack dumps, only offsets / addresses a bit different (CONFIG_SYSFS_DEPRECATED enabled): kobject (c7c35900): tried to init an initialized object, something is seriously wrong. Pid: 1, comm: swapper Not tainted 2.6.25-rc2-testpm-dirty #37 [<c01ea0c5>] kobject_init+0x75/0x80 [<c025072e>] device_initialize+0x1e/0x90 [<c0250f7b>] device_register+0xb/0x20 [<c01f3db8>] pci_bus_add_devices+0x98/0x140 [<c030add7>] ? pcibios_scan_root+0x27/0xa0 [<c03f69d0>] ...

I also put the config at http://home.arcor.de/g.liakhovetski/ap400/config-2.6.25-gregkh-0, the panic is produced by just disabling CONFIG_SYSFS_DEPRECATED. Thanks Guennadi --- Guennadi Liakhovetski --

Also, using anticipatory IO scheduler seems to be the best option for an array with lots of seeks and random reads and writes (quite surprisingly, closely followed by NOOP - both were behaving much better than deadline or CFQ). Here are some numbers I posted to BackupPC mailing list: http://thread.gmane.org/gmane.comp.sysutils.backup.backuppc.general/14009 -- Tomasz Chmielewski http://wpkg.org --

Ok, I found the cause of the oops. Some of radio tuner code was expecting a struct bttv_fh to be allocated but this wasn't done in radio_open. So it would dereference an invalid data structure, causing a hang for me and an oops for Bongani. I also had to add support for the radio tuner to some shared functions. Patches to follow. Robert --

Signed-off-by: Robert Fitzsimons <robfitz@273k.net> --- drivers/media/video/bt8xx/bttv-driver.c | 5 +++-- 1 files changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/media/video/bt8xx/bttv-driver.c b/drivers/media/video/bt8xx/bttv-driver.c index 5404fcc..817a961 100644 --- a/drivers/media/video/bt8xx/bttv-driver.c +++ b/drivers/media/video/bt8xx/bttv-driver.c @@ -1990,7 +1990,7 @@ static int bttv_g_frequency(struct file *file, void *priv, if (0 != err) return err; ...

On Wed, 27 Feb 2008 01:47:29 +0000 Maybe, but the proper way would be to use just one open for both radio and video, like cx88. This driver violates V4L2 API, since the spec says that opening /dev/radio will select radio, by default, but it is possible to listen video also on that interface (the opposite is valid also for /dev/video). I'll apply the fixes, for now. The better would be if you could try to use the same approach present on cx88. Cheers, Mauro --

Yes, I do test compilations of all of my patches. But I don't have an overview of all development trees for all cu Adrian -- "Is there not promise of rain?" Ling Tan asked suddenly out of the darkness. There had been need of rain for many days. "Only a promise," Lao Er said. Pearl S. Buck - Dragon Seed --

I'm not sure about a complete git installation, but I should at least be able to find the last working 2.6.23-gitX or 2.6.24-rcX-gitY tree. (Assuming that these are the intermediate steps between 2.6.23 and 2.6.24.) Cheers, Chris __________________________________________________________ Sent from Yahoo! Mail. A Smarter Inbox. http://uk.docs.yahoo.com/nowyoucan.html --

Yes, that's correct. The patch doesn't apply cleanly to -rc3 and, That would be very helpful. -- Beauty is in the eye of the beerholder. --

On Tue, 26 Feb 2008 22:36:19 +0000 Known and fixed. The fix isn't in 2.6.25-rc3 yet, though. Fix: commit 32720eae675d08990e97bffbf71a31382599cc8a Author: Stefano Brivio <stefano.brivio@polimi.it> Date: Tue Jan 29 20:29:16 2008 +0100 rc80211-pid: fix rate adjustment -- Ciao Stefano --

Already done, should hit upstream soon. --

Hi. I saw it this morning with 2.6.23.14. It was find for ages, and seems to be okay again now. Not sure what triggered it though. nigel@laptop:/usr/src/redhat/BUILD/kernel-2.6.23/build-2.6.23.x86_64$ grep SCHED .config CONFIG_IOSCHED_NOOP=y CONFIG_IOSCHED_AS=y CONFIG_IOSCHED_DEADLINE=y CONFIG_IOSCHED_CFQ=y CONFIG_DEFAULT_IOSCHED="cfq" CONFIG_SCHED_SMT=y CONFIG_SCHED_MC=y # CONFIG_NET_SCHED is not set CONFIG_USB_EHCI_TT_NEWSCHED=y CONFIG_SCHED_DEBUG=y CONFIG_SCHEDSTATS=y (Old ...

Yes but that is only for used for statistics which can be racy. Note that the VM event statistics also can be racy. --

Yeah that would determine if its the fastpath code. Note that the new fastpath code may expose bugs that so far have been tolerated by the slab allocators. If a store of a NULL to the first word of an object occurs then this will be tolerated in 2.6.24 because slub will take it as the end of the object list and just forget about the remaining objects. The fastpath code will terminate with a NULL pointer deref. Enabling full debugging may get rid of the race due to the slowdown. It could ...

I don't rely on the pte locking in #v7, exactly to satisfy GRU Calls are still in atomic context until we change the i_mmap_lock to a mutex under a CONFIG_XPMEM, or unless we boost mm_users, drop the lock and restart the loop at every different mm. In any case those changes should be under CONFIG_XPMEM IMHO given desktop users definitely don't need this (regular non-blocking mmu notifiers in my patch are all what GRU will perform the same with #v7 or V8. --

The patch has to satisfy RDMA, XPMEM, GRU and KVM. I keep hearing that we have a KVM only solution that works 100% (which makes me just switch ignore the rest of the argument because 100% solutions usually do not Ok so you are now getting away from keeping the refcount elevated? That It would have helped if you would have repeated my answers that you had already gotten before. You knew I was on vacation.... --

truncate is handled by the range invalidates. This is special code to deal The situation that you are imagining has already been dealt with by the earlier patches. This is only to allow sleeping while unmapping individual pages. --

This removal has bells and whistles and printk() since 2005... Dominik, what exactly is the problem why we can't remove it now? cu Adrian -- "Is there not promise of rain?" Ling Tan asked suddenly out of the darkness. There had been need of rain for many days. "Only a promise," Lao Er said. Pearl S. Buck - Dragon Seed --

Hi Nick, With this patch, I'm getting "bad data" errors while running the connectathon tests from a Solaris NFS client to my Linux server. (The failing test writes a "big" file (1MB) and reads it back, verifying the data read back is correct.) Let me know if I can provide more details. K.C. --

Hi, I agree. Because the original xfs_freeze doesn't have the timeout feature, I think my freeze ioctl had better not do the timeout in case of specifying 1 as ioctl's parameter. So I have modified my freeze ioctl not to set the timeout in case of specifying 1 or 0 as below. (I have attached the modified patch in this mail.) int ioctl(int fd, int FIFREEZE, long *timeval) fd: The file descriptor of the mountpoint FIFREEZE: The request code for the freeze timeval: The ...

Sure. I'm going to send another revision in the next day or so. (Only comsmetic changes wrt the previous version.) I'll cc you. -- Arthur --

Michael, Please take a look at the following patch to getrusage.2. This is the first time I have edited a manpage, so I hope I have done it correctly! Also, the RUSAGE_THREAD patch is currently in -mm, but not in mainline yet. Hoping that it will make it, I have put a line in the patch that it is supported from 2.6.25 onwards. Thanks, Sripathi. PS: I fixed spelling error in Ulrich's mail id in the CC list. Signed-off-by: Sripathi Kodi <sripathik@in.ibm.com> --- ...

x86.git#testing should be solid with numa and nodes without memory. YH --

Hi, My problem is that I haven't been able to set up a NUMA box where some of the nodes have no memory. The one box we have, the serial port doesn't work and it's in a lab far, far away. Everything else, it works fine. Mike --

Why do we now have to declare this static ? Can we pass it as a pointer Hrm, the semantic of STOPMACHINE_RUN is a bit weird : - The CPU where the do_stop thread runs will first execute (alone) the callback. - Then, all the other CPUs will execute the callback concurrently. Given that you use a "started" boolean in the callback, which is ok as long as there is no concurrent modification (correct given the current semantic, but fragile), I would tend to document that the first time ...

I guess it's time to get back to this problem. I don't want to make PNP insert resources only at the root. That would avoid the ALSA problem because the enclosing PCI resource is allocated first, but it also precludes us from dealing with lots of valid ACPI information. ACPI can describe layers of bridges and devices behind them. It don't think we should throw away that information just to solve this problem. And I don't really want to add IORESOURCE_INSERT. It's probably a useful ...

To clarify, the key appears to be file* plus the user-space integer that I think fork(2) should not be in the above list. fork(2) duplicates the kernel's fd, but the user-space integer that represents the fd remains the same, so you will get EEXIST if you try to add the fd that was duplicated by fork. Chris --

In fact original issue (wich was reported by me) stil exist # uname -a Linux ts52 2.6.25-rc3 #8 SMP Wed Feb 27 11:41:29 MSK 2008 x86_64 x86_64 x86_64 GNU/Linux # blockdev --getsz /dev/sda7 208782 # blockdev --getbsz /dev/sda7 4096 # dd if=/dev/zero of=/dev/sda7 bs=1k seek=104390 count=1 dd: writing `/dev/sda7': Input/output error 1+0 records in 0+0 records out 0 bytes (0 B) copied, 0.000918887 seconds, 0.0 kB/s Issue from the bug was triggered in __generic_make_request() ...