March 2010 - openSUSE Kernel - openSUSE Mailing Lists

Re: [opensuse-kernel] Kernel 2.6.34-rc2 hangs on one of three systems
by Larry Finger 26 Mar '10

26 Mar '10

On 03/26/2010 09:07 AM, Tejun Heo wrote: > > 2.6.34-rcX has a random memory corruption bug which is showing up as > various boot failures. Yinghai has a patch. > > http://thread.gmane.org/gmane.linux.kernel/963616/focus=964914 Thanks for the link. Nasty bug, but tidy patch. I love 1- or 2-liners. I note that this patch is not yet in Linus's tree. Larry -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org

1 0

[opensuse-kernel] Kernel for openSUSE 11.3 (was: Patch openSUSE Kernel for nouveau)
by Johannes Obermayr 13 Mar '10

13 Mar '10

Jeff Mahoney wrote: > On 03/05/2010 11:44 PM, Greg KH wrote: > >> On Fri, Mar 05, 2010 at 05:44:31PM +0100, Johannes Obermayr wrote: >>> >>> Hi, >>> >>> for fulfilling https://features.opensuse.org/307588 it is needed that >>> openSUSE Kernel gets patched with upstream patch: >>> >>> commit d5f3c90d4f3ad6b054f9855b7b69137b97bda131 >>> Author: Marcin Ko??cielnicki <koriakin@xxxxxxxx> >>> Date: Thu Feb 25 00:54:02 2010 +0000 >>> drm/nv50: Implement ctxprog/state generation. >>> This removes dependence on external firmware for NV50 generation cards. >>> If the generated ctxprogs don't work for you for some reason, please >>> report it. >>> Signed-off-by: Marcin Ko??cielnicki <koriakin@xxxxxxxx> >>> Signed-off-by: Ben Skeggs <bskeggs@xxxxxxxxxx> >>> >>> >>> Discussion started ... >>> >> >> This will get pulled in with the next kernel update, so no changes >> needed here. >> >> factory will have it soon. > > > Well that's something we should probably discuss. This was part of > 2.6.34-rc1 and the 11.3 release schedule puts RC1 pretty much exactly > where 2.6.34-final should be. If we get a lot of testing in with > 2.6.34-rc's then it's doable but tight. > > - -Jeff Linux 2.6.33 release: 24 Feb 2010 Development circle: ~90 days Linux 2.6.34 release: ~25 May 2010 openSUSE 11.3 Milestone 7: 21 May 2010 openSUSE 11.3 RC1: 17 Jun 2010 openSUSE 11.3 Final: 15 Jul 2010 I think we should deliver openSUSE 11.3 with Kernel 2.6.34. As Jeff said we should also deliver all 2.6.34-rc's for more testing. Kernel modesetting for AMD/ATI hardware would also be great ... (# CONFIG_DRM_RADEON_KMS is not set -> CONFIG_DRM_RADEON_KMS=y) Johannes -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org

5 6

[opensuse-kernel] v2.6.34 branch
by Jeff Mahoney 10 Mar '10

10 Mar '10

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi all - I've just created a new branch called v2.6.34. It contains the openSUSE patch set on top of 2.6.34-rc1. As per my usual process, I'll merge this into the master branch at rc2 or rc3. For now, you're welcome to test on this separate branch. It boots on my development node, but that's as much testing as I've given it. As usual for a first merge, Xen is disabled until those patches can be brought up to date. Also, for some time we've been running into issues where config options were changed during a version bump or for another reason but the documentation is lost. For single config value changes, it's generally (but not always) enough to look in the git log. For massive changes, like those associated with an -rc1 merge, that becomes very noisy, very quickly. So, for this merge, I created a doc/config-options.changes file that documents what config values were changed and why a certain choice was made. For drivers that are built as modules, there's no explanation needed. Michal - When you get a chance, can you set this up for building and publishing? Thanks! - -Jeff - -- Jeff Mahoney SUSE Labs -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.13 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org/ iEYEARECAAYFAkuW9VQACgkQLPWxlyuTD7IEjgCeIB2blIzmqVP28WPDdDsjsQMN g+QAnjpK5qoo7bLf/R8W/8Tii3KKdNws =EIt1 -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org

1 0

[opensuse-kernel] iwl3945 issue on openSUSE 11.3 M3
by Andi Sugandi 09 Mar '10

09 Mar '10

Hi listmates, I'm Acer and KDE 4 user, and this machine has wireless adapter: 03:00.0 Network controller: Intel Corporation PRO/Wireless 3945ABG [Golan] Network Connection (rev 02) openSUSE 11.2 does the right job to automatically bring the Wlan0 start (the led is on) right after KDE is up, but different on openSUSE 11.3 M3 (and M2), the led is off and (of course) the wireless connection seems disable (eth0 still works though). To solve this issue, I should first: 'service network restart' , and then 'modprobe -r iwl3945 && modprobe iwl3945' . My enabled repositories: http://download.opensuse.org/factory/repo/oss/ http://download.opensuse.org/factory/repo/non-oss/ The used kernel version: 2.6.33-5-desktop Does anyone have this issue too? Should i file a bug? Best regards, -- Andi Sugandi (Bandung) openSUSE Ambassador http://en.opensuse.org/User:Andisugandi -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org

3 4

[opensuse-kernel] KOTD double ships file
by Jan Engelhardt 08 Mar '10

08 Mar '10

Greetings. I am getting a warning from rpmlint that /boot/vmlinux-2.6.33-0-default.gz is packaged twice: once by kernel-default and kernel-default-devel. That should probably be corrected. -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org

1 0

[opensuse-kernel] novfs: replace MUTEX by sema functions
by Jan Engelhardt 08 Mar '10

08 Mar '10

Avoid old MUTEX macros and use sema instead, which works for both default and RT kernels. Signed-off-by: Jan Engelhardt <jengelh(a)medozas.de> --- fs/novfs/daemon.c | 9 +++++---- fs/novfs/inode.c | 8 +++++--- fs/novfs/profile.c | 5 +++-- fs/novfs/scope.c | 4 ++-- 4 files changed, 15 insertions(+), 11 deletions(-) Index: linux-2.6.33-rt/fs/novfs/daemon.c =================================================================== --- linux-2.6.33-rt.orig/fs/novfs/daemon.c +++ linux-2.6.33-rt/fs/novfs/daemon.c @@ -110,16 +110,17 @@ static atomic_t Daemon_Open_Count = ATOM static unsigned long Daemon_Command_Timeout = TIMEOUT_VALUE; -static DECLARE_MUTEX(DriveMapLock); +static struct semaphore DriveMapLock; static LIST_HEAD(DriveMapList); int novfs_max_iosize = PAGE_SIZE; -void novfs_daemon_queue_init() +void novfs_daemon_queue_init(void) { INIT_LIST_HEAD(&Daemon_Queue.list); spin_lock_init(&Daemon_Queue.lock); - init_MUTEX_LOCKED(&Daemon_Queue.semaphore); + sema_init(&DriveMapLock, 1); + sema_init(&Daemon_Queue.semaphore, 0); } void novfs_daemon_queue_exit(void) @@ -163,7 +164,7 @@ int Queue_Daemon_Command(void *request, que->status = QUEUE_SENDING; que->flags = 0; - init_MUTEX_LOCKED(&que->semaphore); + sema_init(&que->semaphore, 0); que->sequence = atomic_inc_return(&Sequence); Index: linux-2.6.33-rt/fs/novfs/inode.c =================================================================== --- linux-2.6.33-rt.orig/fs/novfs/inode.c +++ linux-2.6.33-rt/fs/novfs/inode.c @@ -298,11 +298,11 @@ static atomic_t novfs_Inode_Number = ATO struct dentry *novfs_root = NULL; char *novfs_current_mnt = NULL; -DECLARE_MUTEX(InodeList_lock); +struct semaphore InodeList_lock; LIST_HEAD(InodeList); -DECLARE_MUTEX(TimeDir_Lock); +struct semaphore TimeDir_Lock; uint64_t lastTime; char lastDir[PATH_MAX]; @@ -3766,7 +3766,7 @@ struct inode *novfs_get_inode(struct sup id->cntDC = 1; INIT_LIST_HEAD(&id->DirCache); - init_MUTEX(&id->DirCacheLock); + sema_init(&id->DirCacheLock, 1); id->FileHandle = 0; id->CacheFlag = 0; @@ -3971,6 +3971,8 @@ int __init init_novfs(void) { int retCode; + sema_init(&InodeList_lock, 1); + sema_init(&TimeDir_Lock, 1); lastDir[0] = 0; lastTime = get_nanosecond_time(); Index: linux-2.6.33-rt/fs/novfs/profile.c =================================================================== --- linux-2.6.33-rt.orig/fs/novfs/profile.c +++ linux-2.6.33-rt/fs/novfs/profile.c @@ -60,7 +60,7 @@ static struct proc_dir_entry *dbg_file = static struct proc_dir_entry *dentry_file = NULL; static struct proc_dir_entry *inode_file = NULL; -static DECLARE_MUTEX(LocalPrint_lock); +static struct semaphore LocalPrint_lock; static ssize_t User_proc_write_DbgBuffer(struct file *file, const char __user *buf, size_t nbytes, loff_t *ppos) { @@ -622,8 +622,9 @@ uint64_t get_nanosecond_time() return (retVal); } -void novfs_profile_init() +void novfs_profile_init(void) { + sema_init(&LocalPrint_lock, 1); if (novfs_procfs_dir) dbg_dir = novfs_procfs_dir; else Index: linux-2.6.33-rt/fs/novfs/scope.c =================================================================== --- linux-2.6.33-rt.orig/fs/novfs/scope.c +++ linux-2.6.33-rt/fs/novfs/scope.c @@ -633,8 +633,8 @@ char *novfs_scope_dget_path(struct dentr void novfs_scope_init(void) { INIT_LIST_HEAD(&Scope_List); - init_MUTEX(&Scope_Lock); - init_MUTEX_LOCKED(&Scope_Thread_Delay); + sema_init(&Scope_Lock, 1); + sema_init(&Scope_Thread_Delay, 0); kthread_run(Scope_Cleanup_Thread, NULL, "novfs_ST"); } -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org

7 21

[opensuse-kernel] Patch openSUSE Kernel for nouveau
by Johannes Obermayr 07 Mar '10

07 Mar '10

Hi, for fulfilling https://features.opensuse.org/307588 it is needed that openSUSE Kernel gets patched with upstream patch: commit d5f3c90d4f3ad6b054f9855b7b69137b97bda131 Author: Marcin Kościelnicki <koriakin(a)0x04.net> Date: Thu Feb 25 00:54:02 2010 +0000 drm/nv50: Implement ctxprog/state generation. This removes dependence on external firmware for NV50 generation cards. If the generated ctxprogs don't work for you for some reason, please report it. Signed-off-by: Marcin Kościelnicki <koriakin(a)0x04.net> Signed-off-by: Ben Skeggs <bskeggs(a)redhat.com> Discussion started ... Thanks. Johannes -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org

3 2

[opensuse-kernel] novfs type mismatch warning
by Jan Engelhardt 06 Mar '10

06 Mar '10

I picked up this warning in KOTD: fs/novfs/inode.c: In function ■novfs_a_write_begin■: fs/novfs/inode.c:2462:8: warning: passing argument 4 of ■novfs_read_pages■ from incompatible pointer type fs/novfs/vfs.h:307:12: note: expected ■size_t *■ but argument is of type ■unsigned int *■ int novfs_a_write_begin(unsigned len) in inode.c calls novfs_read_pages(...&len...) which expects size_t instead. [I really wonder whether anybody check-compiled novfs on 64-bit at all.] -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org

2 1

[opensuse-kernel] 4k sector size problems in linux
by Stephen Shaw 02 Mar '10

02 Mar '10

I was approach by daniel while at the recent scale8x conference and was asked to get in touch with follow novell engineers that work on our partitioning tool with regard to 4k sectors size problems. Daniel is with western digital and was hoping to work with us to make sure these issues are resolved. I'll let him go into more detail. Thanks Stephen -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org

4 3

Re: [opensuse-kernel] 4k sector size problems in linux
by Tejun Heo 02 Mar '10

02 Mar '10

Hello, On 03/02/2010 11:44 AM, Daniel Taylor wrote: > If I can provide some test hardware, I'll need an NDA. > > Is there someone who has the appropriate lab and can sign one? Richardson has been providing test drives to me and I've forwarded some to Arvin Schnell and now that I have more 4k (512 logical) drives I can forward more to any Novell people. If anyone wants some, please let me know. > FYI, here's the summary I have been providing (some of which may, finally, > be obsolete): > > ... > Three flavors of drives with 4096-byte sectors: 4096 physical/logical; 4096 > physical/512 logical, LBA offset 0 within physical; 4096 physical/512 > logical, LBA offset 1 within the physical. Around here they are often > referred to as AF-4K, AF-0, and AF-1, respectively. This is complicated by > two facters, drive capacities larger that 2TiB, and MBR vs GPT. Are you guys planning on shipping 4k logical ATA drives? Seriously? > The problem is not usually the kernel (although I have one case that I'm > debugging now which seems to be a kernel issue). > > The partitioning/installation tools do not work well with two of the three > flavors. Currently, parted explicity complains about AF-4K, and really does > not work. Sfdisk cannot handle AF-0 and AF-1 for drives larger than 2TiB. > > The default installers do not check whether a drive is AF-0 or AF-1, and > tend to use obsolete BIOS alignments (track-aligned for a track with 63 > sectors/track), which work well enough for AF-1, but utterly destroy > performance with AF-0, since the drives are continually performing > read-modify-write on two physical sectors for each file system block write. > The problem is that the file system blocks are mis-aligned with regard to > the drive physical sectors. > ... > > One more new thing, there also appear to be drives that actually have 4K > sectors, but report 512 physical and logical. Those appear to be AF-0. It > might be a good idea for the installers to assume 4K sectors on all > late-model drives. If it's just a glitch on certain early models, easier workaround might be blacklisting those drives in the driver so that the kernel can report the correct physical sector size and alignment via sysfs to userland. Thanks. -- tejun -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org

1 0