- 12 Oct, 2007 40 commits
-
-
git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6Linus Torvalds authored
* 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: [BNX2]: Refine napi poll loop. [TG3]: Refine napi poll loop.
-
Anton Altaparmakov authored
Big thanks go to Mathias Kolehmainen for reporting the bug, providing debug output and testing the patches I sent him to get it working. The fix was to stop calling ntfs_attr_set() at mount time as that causes balance_dirty_pages_ratelimited() to be called which on systems with little memory actually tries to go and balance the dirty pages which tries to take the s_umount semaphore but because we are still in fill_super() across which the VFS holds s_umount for writing this results in a deadlock. We now do the dirty work by hand by submitting individual buffers. This has the annoying "feature" that mounting can take a few seconds if the journal is large as we have clear it all. One day someone should improve on this by deferring the journal clearing to a helper kernel thread so it can be done in the background but I don't have time for this at the moment and the current solution works fine so I am leaving it like this for now. Signed-off-by:
Anton Altaparmakov <aia21@cantab.net> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-2.6-nmwLinus Torvalds authored
* git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-2.6-nmw: (51 commits) [DLM] block dlm_recv in recovery transition [DLM] don't overwrite castparam if it's NULL [GFS2] Get superblock a different way [GFS2] Don't try to remove buffers that don't exist [GFS2] Alternate gfs2_iget to avoid looking up inodes being freed [GFS2] Data corruption fix [GFS2] Clean up journaled data writing [GFS2] GFS2: chmod hung - fix race in thread creation [DLM] Make dlm_sendd cond_resched more [GFS2] Move inode deletion out of blocking_cb [GFS2] flocks from same process trip kernel BUG at fs/gfs2/glock.c:1118! [GFS2] Clean up gfs2_trans_add_revoke() [GFS2] Use slab operations for all gfs2_bufdata allocations [GFS2] Replace revoke structure with bufdata structure [GFS2] Fix ordering of dirty/journal for ordered buffer unstuffing [GFS2] Clean up ordered write code [GFS2] Move pin/unpin into lops.c, clean up locking [GFS2] Don't mark jdata dirty in gfs2_unstuffer_page() [GFS2] Introduce gfs2_remove_from_ail [GFS2] Correct lock ordering in unlink ...
-
Martin Schwidefsky authored
Signed-off-by:Martin Schwidefsky <schwidefsky@de.ibm.com>
-
Cornelia Huck authored
The early console device is special as it already has all needed data set and raw3215[0] allocated from con3215_init(). Therefore, it needs special treatment in the probe function so it can be matched with the 3215 driver which is already owning it. Signed-off-by:
Cornelia Huck <cornelia.huck@de.ibm.com> Signed-off-by:
-
Heiko Carstens authored
Make sure parameter list of the pfault token function is eight byte aligned. Otherwise we can get specification exceptions. Signed-off-by:
Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by:
-
Sam Ravnborg authored
Replace the hardcoded 4096 value with the PAGE_SIZE macro. Converted a few decimal numbers to readable hex numbers. Use of PAGE_SIZE required a small change to page.h to allow PAGE_SIZE to be used from assembler/linker scripts. Signed-off-by:
Sam Ravnborg <sam@ravnborg.org> Signed-off-by:
-
Sam Ravnborg authored
Introduce a consistent style in vmlinux.lds. This style is gradually being introduced for all archs. A few lables were moved inside the section definition so they are assigned the correct value of gcc decide to align the content to another address than the one . has. In the past this has fixed several bugs but for s390 it will not impact due to all the alignmnet already introduced. Stabs definitions are consolidated in asm-generic/vmlinux.lds.h This patch also introduce support for DWARF - without knowing if this makes sense for s390. Signed-off-by:
-
Heiko Carstens authored
Constraint for err is wrong since it is preinitialized and the code relies on it in case of an exception. Signed-off-by:
-
Michael Holzheu authored
Currently there are situations, where the tape medium state (loaded, unloaded, unknown) is not set correctly. The reason is that the medium association field is used to determine the medium state. This not always leads to correct results. To fix that we use another field of the medium sense data instead. Signed-off-by:
Michael Holzheu <holzheu@de.ibm.com> Signed-off-by:
-
Heiko Carstens authored
After assigning values to specific registers memset was called. This may clobber the contents of the used registers. To solve this extract the two used inline assemblies into small functions that don't call any functions. Signed-off-by:
-
Heiko Carstens authored
Prevent that modules get loaded at addresses below 4GB to prevent exchanging system call table entries. Signed-off-by:
-
Michael Holzheu authored
memcpy_real uses the mvcle instruction. This instruction alters all used registers (source, destination and 2 x count). Therefore we have to flag those registers as input/output registers (+d). In addition to that, we have to specify, that we read from memory designated by "src" and write to memory designated by "dest". Signed-off-by:
-
Robert P. J. Day authored
Signed-off-by:
Robert P. J. Day <rpjday@mindspring.com> Signed-off-by:
-
Michael Holzheu authored
If we use the CLEAR ipl option, reipl is faster, since then VM can release the memory, which has been paged out. Signed-off-by:
-
Heiko Carstens authored
Signed-off-by:
-
Ursula Braun authored
Signed-off-by:
Ursula Braun <braunu@de.ibm.com> Signed-off-by:
-
vignesh babu authored
Replacing n & (n - 1) for power of 2 check by is_power_of_2(n) Signed-off-by:
vignesh babu <vignesh.babu@wipro.com> Signed-off-by:
-
Heiko Carstens authored
According to C99 6.3.2.3 it's undefined what happens if a converted pointer is used to call a function whose type is not compatible with the pointed-to type. That's what the qdio code is doing, so fix it. Signed-off-by:
-
Heiko Carstens authored
slab cache debugging code has been changed so that we always get a minimum alignment of the alignment of a 64-integer. Since this is 8 on s390/s390x there is no need of ARCH_KMALLOC_MINALIGN anymore. Signed-off-by:
-
Christian Borntraeger authored
ext_int_info_t is no longer used in entry(64).S Instead do_extint is a C function that handles the hash search. As the structure is handled in C code, we can also remove the packed attribute to avoid alignment issues. (Currently there is no alignment problem in ext_int_info_t, even if packet) Signed-off-by:
Christian Borntraeger <borntraeger@de.ibm.com> Signed-off-by:
-
Heiko Carstens authored
Also removes a bunch of ^L in drivers/s390/cio/cmf.c Signed-off-by:
-
Christian Borntraeger authored
There is no need to assign "0" to "hops" twice. Remove one assigment. Signed-off-by:
-
Martin Schwidefsky authored
The termination condition of the loop that prints the operands of an instruction doesn't stop after the maximum of 6 operands. It continues with the operands of the next instruction format instead which create really long lines. Signed-off-by: -
Satyam Sharma authored
arch/s390/Kconfig tells us that CONFIG_APPLDATA_BASE is bool and hence can never be built modular. Given this, defining appldata_exit() function is pointless (and wasteful, actually). Remove all that. Previous patch annotated appldata_offline_cpu() as __cpuexit, but now with the __exit function appldata_exit() gone, the only callsite that references it is __cpuinit, so this function can also be __cpuinit, thereby saving space when HOTPLUG_CPU=n. Signed-off-by:
Satyam Sharma <satyam@infradead.org> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
-
Satyam Sharma authored
appldata_offline_cpu() is only called from __cpuinit-marked hotplug notifier callback and from the __exit-marked module_exit function, therefore candidate for __cpuexit. BTW the __exit module_exit function appldata_exit() of this driver fails to unregister_hotcpu_notifier() the notifier_block that was registered by appldata_init() during module startup. This will lead to oops if hotplug notification comes after module has been unloaded. Let's fix this by unregistering the notifier appropriately (before appldata_offline_cpu()'ing the CPUs). Signed-off-by:
-
Ralph Wuerthner authored
Symptom: zcrypt fails by setting all PCIXCC/CEX2C cards offline for a certain type of invalid keys. Problem: zcrypt does not handle rc=12/rs=769 request responses correctly Solution: modify convert_type86_ica() to handle these error codes correctly Signed-off-by:Ralph Wuerthner <rwuerthn@de.ibm.com> Signed-off-by:
-
Ralph Wuerthner authored
Resetting of a all queues within a domain requires that a domain must be selected first. Signed-off-by:
-
Ralph Wuerthner authored
Signed-off-by: -
Heiko Carstens authored
Signed-off-by:
-
Cornelia Huck authored
There was the possibilty that an action like ccw_device_set_offline() triggered by a device gone machine check might trigger a not oper event. Unfortunately, this could lead to the situation that we tried to unregister a subchannel twice: Once from the slow path evaluation, and once via the not oper event. Fix this by always using the same mechanism (css_schedule_eval()) for triggering the unregister. This makes sure that unregistration will only be done once. As an added bonus, it also simplyfies the code. Signed-off-by:
-
Cornelia Huck authored
Signed-off-by:
-
Cornelia Huck authored
Signed-off-by:
-
Cornelia Huck authored
Don't forget to set dev->groups for early ccw devices like the console device so the default attributes are created. Signed-off-by:
-
Cornelia Huck authored
Disable channel measurements for all ccw devices via the ccw bus's shutdown method. Clear residual cmf related information that may be in the schib when setting up a new subchannel. Signed-off-by:
-
Cornelia Huck authored
Introduce a shutdown method for the ccw bus that calls the driver specific shutdown method in struct ccw_driver. Switch zfcp to the new ccw_driver shutdown method. Signed-off-by:
-
Cornelia Huck authored
Make sure channel path measurements are disabled via a reboot notifier since we cannot rely on userspace to disable it again after enabling it. This is 1. clean and 2. makes sure that channel path measurements do not stay enabled after a reboot not involving a subsystem reset (which would lead to the channel subsystem scribbling onto random memory). Signed-off-by:
-
Cornelia Huck authored
Signed-off-by:
-
Cornelia Huck authored
s390-drivers is generated using the docbook comments. It should eventually supersede Documentation/s390/cds.txt. Signed-off-by:
-
Cornelia Huck authored
Comment a bunch of function in docbook style and convert existing comments on structures to docbook. Signed-off-by:
-
