Oleg Nesterov [Fri, 15 Jun 2012 15:43:28 +0000 (17:43 +0200)]
uprobes: Write_opcode()->__replace_page() can race with try_to_unmap()
write_opcode() gets old_page via get_user_pages() and then calls
__replace_page() which assumes that this old_page is still
mapped after pte_offset_map_lock().
This is not true if this old_page was already try_to_unmap()'ed,
and in this case everything __replace_page() does with old_page
is wrong. Just for example, put_page() is not balanced.
I think it is possible to teach __replace_page() to handle this
unlikely case correctly, but this patch simply changes it to use
page_check_address() and return -EAGAIN if it fails. The caller
should notice this error code and retry.
Note: write_opcode() asks for the cleanups, I'll try to do this
in a separate patch.
Signed-off-by: Oleg Nesterov <oleg@redhat.com> Acked-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com> Cc: Anton Arapov <anton@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/20120615154328.GA9571@redhat.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
Oleg Nesterov [Fri, 15 Jun 2012 15:43:25 +0000 (17:43 +0200)]
uprobes: __copy_insn() should ensure a_ops->readpage != NULL
__copy_insn() blindly calls read_mapping_page(), this will crash
the kernel if ->readpage == NULL, add the necessary check. For
example, hugetlbfs_aops->readpage is NULL. Perhaps we should
change read_mapping_page() instead.
Signed-off-by: Oleg Nesterov <oleg@redhat.com> Acked-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com> Cc: Anton Arapov <anton@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/20120615154325.GA9568@redhat.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
Oleg Nesterov [Fri, 15 Jun 2012 15:43:22 +0000 (17:43 +0200)]
uprobes: Valid_vma() should reject VM_HUGETLB
__replace_page() obviously can't work with the hugetlbfs
mappings, uprobe_register() will likely crash the kernel. Change
valid_vma() to check VM_HUGETLB as well.
As for PageTransHuge() no need to worry, vma->vm_file != NULL.
Signed-off-by: Oleg Nesterov <oleg@redhat.com> Acked-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com> Cc: Anton Arapov <anton@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/20120615154322.GA9561@redhat.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
David Rientjes [Fri, 8 Jun 2012 20:21:26 +0000 (13:21 -0700)]
mm, oom: fix badness score underflow
If the privileges given to root threads (3% of allowable memory) or a
negative value of /proc/pid/oom_score_adj happen to exceed the amount of
rss of a thread, its badness score overflows as a result of commit a7f638f999ff ("mm, oom: normalize oom scores to oom_score_adj scale only
for userspace").
Fix this by making the type signed and return 1, meaning the thread is
still eligible for kill, if the value is negative.
Reported-by: Dave Jones <davej@redhat.com> Acked-by: Oleg Nesterov <oleg@redhat.com> Signed-off-by: David Rientjes <rientjes@google.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Randy Dunlap [Fri, 8 Jun 2012 20:18:33 +0000 (13:18 -0700)]
sched/fair: fix lots of kernel-doc warnings
Fix lots of new kernel-doc warnings in kernel/sched/fair.c:
Warning(kernel/sched/fair.c:3625): No description found for parameter 'env'
Warning(kernel/sched/fair.c:3625): Excess function parameter 'sd' description in 'update_sg_lb_stats'
Warning(kernel/sched/fair.c:3735): No description found for parameter 'env'
Warning(kernel/sched/fair.c:3735): Excess function parameter 'sd' description in 'update_sd_pick_busiest'
Warning(kernel/sched/fair.c:3735): Excess function parameter 'this_cpu' description in 'update_sd_pick_busiest'
.. more warnings
Signed-off-by: Randy Dunlap <rdunlap@xenotime.net> Cc: Ingo Molnar <mingo@redhat.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
It incorrectly finds VGA connectors where none are attached, apparently
not noticing that nothing replied to the EDID queries, and happily using
the default EDID modes that have nothing to do with actual hardware.
That in turn then causes X to fall down to the lowest common
denominator, which is usually the default 1024x768 mode that is in the
default EDID and pretty much anything supports).
I'd suggest that if not relying on the HDP pin, the code should at least
check whether it gets valid EDID data back, rather than just assume
there's something on the VGA connector.
Cc: Dave Airlie <airlied@linux.ie> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Linus Torvalds [Fri, 8 Jun 2012 18:15:31 +0000 (11:15 -0700)]
Merge tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4
Pull ext4 bug fixes from Theodore Ts'o:
"This update contains two bug fixes, both destined for the stable tree.
Perhaps the most important is one which fixes ext4 when used with file
systems originally formatted for use with ext3, but then later
converted to take advantage of ext4."
* tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
ext4: don't set i_flags in EXT4_IOC_SETFLAGS
ext4: fix the free blocks calculation for ext3 file systems w/ uninit_bg
Linus Torvalds [Fri, 8 Jun 2012 18:06:01 +0000 (11:06 -0700)]
Merge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc
Pull powerpc fixes from Paul Mackerras:
"Two small fixes for powerpc:
- a fix for a regression since 3.2 that causes 4-second (or longer)
pauses
- a fix for a potential oops when loading kernel modules on 32-bit
embedded systems."
* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc:
powerpc: Fix kernel panic during kernel module load
powerpc/time: Sanity check of decrementer expiration is necessary
Linus Torvalds [Fri, 8 Jun 2012 18:04:06 +0000 (11:04 -0700)]
Merge tag 'upstream-3.5-rc2' of git://git.infradead.org/linux-ubifs
Pull UBI/UBIFS fixes from Artem Bityutskiy:
"Fix UBI and UBIFS - they refuse to work without debugfs. This was
broken by the 3.5-rc1 UBI/UBIFS changes when we removed the debugging
Kconfig switches.
Also, correct locking in 'ubi_wl_flush()' - it was extended to support
flushing a specific LEB in 3.5-rc1, and the locking was sub-optimal."
* tag 'upstream-3.5-rc2' of git://git.infradead.org/linux-ubifs:
UBI: correct ubi_wl_flush locking
UBIFS: fix debugfs-less systems support
UBI: fix debugfs-less systems support
We're chasing an elusive bug that Dave Jones can apparently reproduce
using his system call fuzzer tool, and that looks like some kind of
locking ordering problem on the directory i_mutex chain. Our i_mutex
locking is rather complex, and depends on the topological ordering of
the directories, which is why we have been very wary of splicing
directory entries around.
Of course, we really don't want to ever see aliased unconnected
directories anyway, so none of this should ever happen, but this revert
aims to basically get us back to a known older state.
Bruce points to some of the previous discussion at
It should be noted that it's possible that Dave's problems come from
other changes altohgether, including possibly just the fact that Dave
constantly is teachning his fuzzer new tricks. So what appears to be a
new bug could in fact be an old one that just gets newly triggered, but
reverting these patches as "still under heavy discussion" is the right
thing regardless.
Requested-by: Al Viro <viro@zeniv.linux.org.uk> Acked-by: J. Bruce Fields <bfields@fieldses.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Linus Torvalds [Fri, 8 Jun 2012 16:26:55 +0000 (09:26 -0700)]
Merge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 fixes from Ingo Molnar.
* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/nmi: Fix section mismatch warnings on 32-bit
x86/uv: Fix UV2 BAU legacy mode
x86/mm: Only add extra pages count for the first memory range during pre-allocation early page table space
x86, efi stub: Add .reloc section back into image
x86/ioapic: Fix NULL pointer dereference on CPU hotplug after disabling irqs
x86/reboot: Fix a warning message triggered by stop_other_cpus()
x86/intel/moorestown: Change intel_scu_devices_create() to __devinit
x86/numa: Set numa_nodes_parsed at acpi_numa_memory_affinity_init()
x86/gart: Fix kmemleak warning
x86: mce: Add the dropped timer interval init back
x86/mce: Fix the MCE poll timer logic
Linus Torvalds [Fri, 8 Jun 2012 16:14:46 +0000 (09:14 -0700)]
Merge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull perf fixes from Ingo Molnar:
"A bit larger than what I'd wish for - half of it is due to hw driver
updates to Intel Ivy-Bridge which info got recently released,
cycles:pp should work there now too, amongst other things. (but we
are generally making exceptions for hardware enablement of this type.)
There are also callchain fixes in it - responding to mostly
theoretical (but valid) concerns. The tooling side sports perf.data
endianness/portability fixes which did not make it for the merge
window - and various other fixes as well."
* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (26 commits)
perf/x86: Check user address explicitly in copy_from_user_nmi()
perf/x86: Check if user fp is valid
perf: Limit callchains to 127
perf/x86: Allow multiple stacks
perf/x86: Update SNB PEBS constraints
perf/x86: Enable/Add IvyBridge hardware support
perf/x86: Implement cycles:p for SNB/IVB
perf/x86: Fix Intel shared extra MSR allocation
x86/decoder: Fix bsr/bsf/jmpe decoding with operand-size prefix
perf: Remove duplicate invocation on perf_event_for_each
perf uprobes: Remove unnecessary check before strlist__delete
perf symbols: Check for valid dso before creating map
perf evsel: Fix 32 bit values endianity swap for sample_id_all header
perf session: Handle endianity swap on sample_id_all header data
perf symbols: Handle different endians properly during symbol load
perf evlist: Pass third argument to ioctl explicitly
perf tools: Update ioctl documentation for PERF_IOC_FLAG_GROUP
perf tools: Make --version show kernel version instead of pull req tag
perf tools: Check if callchain is corrupted
perf callchain: Make callchain cursors TLS
...
Linus Torvalds [Fri, 8 Jun 2012 16:12:21 +0000 (09:12 -0700)]
Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux
Pull drm intel and exynos fixes from Dave Airlie:
"A bunch of fixes for Intel and exynos, nothing too major, a new intel
PCI ID, and a fix for CRT detection."
* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
drm/i915: pch_irq_handler -> {ibx, cpt}_irq_handler
char/agp: add another Ironlake host bridge
drm/i915: fix up ivb plane 3 pageflips
drm/exynos: fixed blending for hdmi graphic layer
drm/exynos: Remove dummy encoder get_crtc operation implementation
drm/exynos: Keep a reference to frame buffer GEM objects
drm/exynos: Don't cast GEM object to Exynos GEM object when not needed
drm/exynos: DRIVER_BUS_PLATFORM is not a driver feature
drm/exynos: fixed size type.
drm/exynos: Use DRM_FORMAT_{NV12, YUV420} instead of DRM_FORMAT_{NV12M, YUV420M}
drm/i915: hold forcewake around ring hw init
drm/i915: Mark the ringbuffers as being in the GTT domain
drm/i915/crt: Do not rely upon the HPD presence pin
drm/i915: Reset last_retired_head when resetting ring
Linus Torvalds [Fri, 8 Jun 2012 16:10:35 +0000 (09:10 -0700)]
Merge tag 'moduleparam-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus
Pull minor module param fixes from Rusty Russell:
"One bugfix for multiple moduleparam levels, one removal of overzealous
printk."
* tag 'moduleparam-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus:
init: Drop initcall level output
module_param: stop double-calling parameters.
uprobes: Pass probed vaddr to arch_uprobe_analyze_insn()
On RISC architectures like powerpc, instructions are fixed size.
Instruction analysis on such platforms is just a matter of
(insn % 4). Pass the vaddr at which the uprobe is to be inserted so
that arch_uprobe_analyze_insn() can flag misaligned registration
requests.
WARNING: arch/x86/built-in.o(.data+0x5af0): Section mismatch in
reference from the variable test_nmi_ipi_callback_na.10451 to
the function .init.text:test_nmi_ipi_callback() [...]
WARNING: arch/x86/built-in.o(.data+0x5b04): Section mismatch in
reference from the variable nmi_unk_cb_na.10399 to the function
.init.text:nmi_unk_cb() The variable nmi_unk_cb_na.10399
references the function __init nmi_unk_cb() [...]
Both of these are attributed to the internal representation of
the nmiaction struct created during register_nmi_handler. The
reason for this is that those structs are not defined in the
init section whereas the rest of the code in nmi_selftest.c is.
To resolve this, I created a new #define,
register_nmi_handler_initonly, that tags the struct as
__initdata to resolve the mismatch. This #define should only be
used in rare situations where the register/unregister is called
during init of the kernel.
Big thanks to Jan Beulich for decoding this for me as I didn't
have a clue what was going on.
Steffen Rumler [Wed, 6 Jun 2012 14:37:17 +0000 (16:37 +0200)]
powerpc: Fix kernel panic during kernel module load
This fixes a problem which can causes kernel oopses while loading
a kernel module.
According to the PowerPC EABI specification, GPR r11 is assigned
the dedicated function to point to the previous stack frame.
In the powerpc-specific kernel module loader, do_plt_call()
(in arch/powerpc/kernel/module_32.c), GPR r11 is also used
to generate trampoline code.
This combination crashes the kernel, in the case where the compiler
chooses to use a helper function for saving GPRs on entry, and the
module loader has placed the .init.text section far away from the
.text section, meaning that it has to generate a trampoline for
functions in the .init.text section to call the GPR save helper.
Because the trampoline trashes r11, references to the stack frame
using r11 can cause an oops.
The fix just uses GPR r12 instead of GPR r11 for generating the
trampoline code. According to the statements from Freescale, this is
safe from an EABI perspective.
I've tested the fix for kernel 2.6.33 on MPC8541.
Cc: stable@vger.kernel.org Signed-off-by: Steffen Rumler <steffen.rumler.ext@nsn.com>
[paulus@samba.org: reworded the description] Signed-off-by: Paul Mackerras <paulus@samba.org>
Cliff Wickman [Thu, 7 Jun 2012 13:31:40 +0000 (08:31 -0500)]
x86/uv: Fix UV2 BAU legacy mode
The SGI Altix UV2 BAU (Broadcast Assist Unit) as used for
tlb-shootdown (selective broadcast mode) always uses UV2
broadcast descriptor format. There is no need to clear the
'legacy' (UV1) mode, because the hardware always uses UV2 mode
for selective broadcast.
But the BIOS uses general broadcast and legacy mode, and the
hardware pays attention to the legacy mode bit for general
broadcast. So the kernel must not clear that mode bit.
Dave Airlie [Fri, 8 Jun 2012 08:42:51 +0000 (09:42 +0100)]
Merge branch 'exynos-drm-fixes' of git://git.infradead.org/users/kmpark/linux-samsung into drm-fixes
* 'exynos-drm-fixes' of git://git.infradead.org/users/kmpark/linux-samsung:
drm/exynos: fixed blending for hdmi graphic layer
drm/exynos: Remove dummy encoder get_crtc operation implementation
drm/exynos: Keep a reference to frame buffer GEM objects
drm/exynos: Don't cast GEM object to Exynos GEM object when not needed
drm/exynos: DRIVER_BUS_PLATFORM is not a driver feature
drm/exynos: fixed size type.
drm/exynos: Use DRM_FORMAT_{NV12, YUV420} instead of DRM_FORMAT_{NV12M, YUV420M}
Dave Airlie [Fri, 8 Jun 2012 08:42:35 +0000 (09:42 +0100)]
Merge branch 'drm-intel-fixes' of git://people.freedesktop.org/~danvet/drm-intel into drm-fixes
* 'drm-intel-fixes' of git://people.freedesktop.org/~danvet/drm-intel:
drm/i915: pch_irq_handler -> {ibx, cpt}_irq_handler
char/agp: add another Ironlake host bridge
drm/i915: fix up ivb plane 3 pageflips
drm/i915: hold forcewake around ring hw init
drm/i915: Mark the ringbuffers as being in the GTT domain
drm/i915/crt: Do not rely upon the HPD presence pin
drm/i915: Reset last_retired_head when resetting ring
but they don't contain any info for the general user staring at dmesg.
I'm very doubtful the count of initcalls registered per level helps
anyone so drop that output completely.
Cc: Jim Cromie <jim.cromie@gmail.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: Jason Baron <jbaron@redhat.com> Signed-off-by: Borislav Petkov <borislav.petkov@amd.com> Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Rusty Russell [Fri, 8 Jun 2012 05:28:13 +0000 (14:58 +0930)]
module_param: stop double-calling parameters.
Commit 026cee0086fe1df4cf74691cf273062cc769617d "params:
<level>_initcall-like kernel parameters" set old-style module
parameters to level 0. And we call those level 0 calls where we used
to, early in start_kernel().
We also loop through the initcall levels and call the levelled
module_params before the corresponding initcall. Unfortunately level
0 is early_init(), so we call the standard module_param calls twice.
(Turns out most things don't care, but at least ubi.mtd does).
Change the level to -1 for standard module_param calls.
Paul Mackerras [Fri, 1 Jun 2012 08:13:43 +0000 (18:13 +1000)]
powerpc/time: Sanity check of decrementer expiration is necessary
This reverts 68568add2c ("powerpc/time: Remove unnecessary sanity check
of decrementer expiration"). We do need to check whether we have reached
the expiration time of the next event, because we sometimes get an early
decrementer interrupt, most notably when we set the decrementer to 1 in
arch_irq_work_raise(). The effect of not having the sanity check is that
if timer_interrupt() gets called early, we leave the decrementer set to
its maximum value, which means we then don't get any more decrementer
interrupts for about 4 seconds (or longer, depending on timebase
frequency). I saw these pauses as a consequence of getting a stray
hypervisor decrementer interrupt left over from exiting a KVM guest.
This isn't quite a straight revert because of changes to the surrounding
code, but it restores the same algorithm as was previously used.
Cc: stable@vger.kernel.org Acked-by: Anton Blanchard <anton@samba.org> Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
It's horribly and utterly broken for at least the following reasons:
- calling sync_mm_rss() from mmput() is fundamentally wrong, because
there's absolutely no reason to believe that the task that does the
mmput() always does it on its own VM. Example: fork, ptrace, /proc -
you name it.
- calling it *after* having done mmdrop() on it is doubly insane, since
the mm struct may well be gone now.
- testing mm against NULL before you call it is insane too, since a
NULL mm there would have caused oopses long before.
.. and those are just the three bugs I found before I decided to give up
looking for me and revert it asap. I should have caught it before I
even took it, but I trusted Andrew too much.
Cc: Konstantin Khlebnikov <khlebnikov@openvz.org> Cc: Markus Trippelsdorf <markus@trippelsdorf.de> Cc: Hugh Dickins <hughd@google.com> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Tao Ma [Thu, 7 Jun 2012 23:04:19 +0000 (19:04 -0400)]
ext4: don't set i_flags in EXT4_IOC_SETFLAGS
Commit 7990696 uses the ext4_{set,clear}_inode_flags() functions to
change the i_flags automatically but fails to remove the error setting
of i_flags. So we still have the problem of trashing state flags.
Fix this by removing the assignment.
Signed-off-by: Tao Ma <boyu.mt@taobao.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Cc: stable@kernel.org
Theodore Ts'o [Thu, 7 Jun 2012 22:56:06 +0000 (18:56 -0400)]
ext4: fix the free blocks calculation for ext3 file systems w/ uninit_bg
Ext3 filesystems that are converted to use as many ext4 file system
features as possible will enable uninit_bg to speed up e2fsck times.
These file systems will have a native ext3 layout of inode tables and
block allocation bitmaps (as opposed to ext4's flex_bg layout).
Unfortunately, in these cases, when first allocating a block in an
uninitialized block group, ext4 would incorrectly calculate the number
of free blocks in that block group, and then errorneously report that
the file system was corrupt:
EXT4-fs error (device vdd): ext4_mb_generate_buddy:741: group 30, 32254 clusters in bitmap, 32258 in gd
Linus Torvalds [Thu, 7 Jun 2012 22:05:43 +0000 (15:05 -0700)]
Merge branch 'akpm' (Andrew's fixups)
Merge random fixes from Andrew Morton.
* emailed from Andrew Morton <akpm@linux-foundation.org>: (11 patches)
mm: correctly synchronize rss-counters at exit/exec
btree: catch NULL value before it does harm
btree: fix tree corruption in btree_get_prev()
ipc: shm: restore MADV_REMOVE functionality on shared memory segments
drivers/platform/x86/acerhdf.c: correct Boris' mail address
c/r: prctl: drop VMA flags test on PR_SET_MM_ stack data assignment
c/r: prctl: add ability to get clear_tid_address
c/r: prctl: add minimal address test to PR_SET_MM
c/r: prctl: update prctl_set_mm_exe_file() after mm->num_exe_file_vmas removal
MAINTAINERS: whitespace fixes
shmem: replace_page must flush_dcache and others
mm: correctly synchronize rss-counters at exit/exec
mm->rss_stat counters have per-task delta: task->rss_stat. Before
changing task->mm pointer the kernel must flush this delta with
sync_mm_rss().
do_exit() already calls sync_mm_rss() to flush the rss-counters before
committing the rss statistics into task->signal->maxrss, taskstats,
audit and other stuff. Unfortunately the kernel does this before
calling mm_release(), which can call put_user() for processing
task->clear_child_tid. So at this point we can trigger page-faults and
task->rss_stat becomes non-zero again. As a result mm->rss_stat becomes
inconsistent and check_mm() will print something like this:
This patch moves sync_mm_rss() into mm_release(), and moves mm_release()
out of do_exit() and calls it earlier. After mm_release() there should
be no pagefaults.
Joern Engel [Thu, 7 Jun 2012 21:21:14 +0000 (14:21 -0700)]
btree: catch NULL value before it does harm
Storing NULL values in the btree is illegal and can lead to memory
corruption and possible other fun as well. Catch it on insert, instead
of waiting for the inevitable.
Signed-off-by: Joern Engel <joern@logfs.org> Signed-off-by: Roland Dreier <roland@purestorage.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Roland Dreier [Thu, 7 Jun 2012 21:21:13 +0000 (14:21 -0700)]
btree: fix tree corruption in btree_get_prev()
The memory the parameter __key points to is used as an iterator in
btree_get_prev(), so if we save off a bkey() pointer in retry_key and
then assign that to __key, we'll end up corrupting the btree internals
when we do eg
longcpy(__key, bkey(geo, node, i), geo->keylen);
to return the key value. What we should do instead is use longcpy() to
copy the key value that retry_key points to __key.
This can cause a btree to get corrupted by seemingly read-only
operations such as btree_for_each_safe.
[akpm@linux-foundation.org: avoid the double longcpy()] Signed-off-by: Roland Dreier <roland@purestorage.com> Acked-by: Joern Engel <joern@logfs.org> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Will Deacon [Thu, 7 Jun 2012 21:21:13 +0000 (14:21 -0700)]
ipc: shm: restore MADV_REMOVE functionality on shared memory segments
Commit 17cf28afea2a ("mm/fs: remove truncate_range") removed the
truncate_range inode operation in favour of the fallocate file
operation.
When using SYSV IPC shared memory segments, calling madvise with the
MADV_REMOVE advice on an area of shared memory will attempt to invoke
the .fallocate function for the shm_file_operations, which is NULL and
therefore returns -EOPNOTSUPP to userspace. The previous behaviour
would inherit the inode_operations from the underlying tmpfs file and
invoke truncate_range there.
This patch restores the previous behaviour by wrapping the underlying
fallocate function in shm_fallocate, as we do for fsync.
[hughd@google.com: use -ENOTSUPP in shm_fallocate()] Signed-off-by: Will Deacon <will.deacon@arm.com> Acked-by: Hugh Dickins <hughd@google.com> Signed-off-by: Hugh Dickins <hughd@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Cyrill Gorcunov [Thu, 7 Jun 2012 21:21:12 +0000 (14:21 -0700)]
c/r: prctl: drop VMA flags test on PR_SET_MM_ stack data assignment
In commit b76437579d13 ("procfs: mark thread stack correctly in
proc/<pid>/maps") the stack allocated via clone() is marked in
/proc/<pid>/maps as [stack:%d] thus it might be out of the former
mm->start_stack/end_stack values (and even has some custom VMA flags
set).
So to be able to restore mm->start_stack/end_stack drop vma flags test,
but still require the underlying VMA to exist.
As always note this feature is under CONFIG_CHECKPOINT_RESTORE and
requires CAP_SYS_RESOURCE to be granted.
c/r: prctl: update prctl_set_mm_exe_file() after mm->num_exe_file_vmas removal
A fix for commit b32dfe377102 ("c/r: prctl: add ability to set new
mm_struct::exe_file").
After removing mm->num_exe_file_vmas kernel keeps mm->exe_file until
final mmput(), it never becomes NULL while task is alive.
We can check for other mapped files in mm instead of checking
mm->num_exe_file_vmas, and mark mm with flag MMF_EXE_FILE_CHANGED in
order to forbid second changing of mm->exe_file.
Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org> Reviewed-by: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Matt Helsley <matthltc@us.ibm.com> Cc: Kees Cook <keescook@chromium.org> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Tejun Heo <tj@kernel.org> Cc: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Hugh Dickins [Thu, 7 Jun 2012 21:21:09 +0000 (14:21 -0700)]
shmem: replace_page must flush_dcache and others
Commit bde05d1ccd51 ("shmem: replace page if mapping excludes its zone")
is not at all likely to break for anyone, but it was an earlier version
from before review feedback was incorporated. Fix that up now.
* shmem_replace_page must flush_dcache_page after copy_highpage [akpm]
* Expand comment on why shmem_unuse_inode needs page_swapcount [akpm]
* Remove excess of VM_BUG_ONs from shmem_replace_page [wangcong]
* Check page_private matches swap before calling shmem_replace_page [hughd]
* shmem_replace_page allow for unexpected race in radix_tree lookup [hughd]
Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Cong Wang <xiyou.wangcong@gmail.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Cc: Alan Cox <alan@lxorguk.ukuu.org.uk> Cc: Stephane Marchesin <marcheu@chromium.org> Cc: Andi Kleen <andi@firstfloor.org> Cc: Dave Airlie <airlied@gmail.com> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Rob Clark <rob.clark@linaro.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Jordan Justen [Thu, 7 Jun 2012 16:05:21 +0000 (09:05 -0700)]
x86, efi stub: Add .reloc section back into image
Some UEFI firmware will not load a .efi with a .reloc section
with a size of 0.
Therefore, we create a .efi image with 4 main areas and 3 sections.
1. PE/COFF file header
2. .setup section (covers all setup code following the first sector)
3. .reloc section (contains 1 dummy reloc entry, created in build.c)
4. .text section (covers the remaining kernel image)
To make room for the new .setup section data, the header
bugger_off_msg had to be shortened.
Reported-by: Henrik Rydberg <rydberg@euromail.se> Signed-off-by: Jordan Justen <jordan.l.justen@intel.com> Link: http://lkml.kernel.org/r/1339085121-12760-1-git-send-email-jordan.l.justen@intel.com Tested-by: Lee G Rosenbaum <lee.g.rosenbaum@intel.com> Tested-by: Henrik Rydberg <rydberg@euromail.se> Cc: Matt Fleming <matt.fleming@intel.com> Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Linus Torvalds [Thu, 7 Jun 2012 16:06:54 +0000 (09:06 -0700)]
Merge tag 'parisc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/parisc-2.6
Pull PARISC fixes from James Bottomley:
"This is a set of three bug fixes for minor build breakages that got
introduced just before 3.5-rc1 was released."
* tag 'parisc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/parisc-2.6:
[PARISC] fix code to find libgcc
[PARISC] fix compile break in use of lib/strncopy_from_user.c
[PARISC] fix missing TAINT_WARN problem
Commit "62f38455 UBI: modify ubi_wl_flush function to clear work queue for a lnum"
takes the 'work_sem' semaphore in write mode for the entire loop, which is not
very good because it will block other workers for potentially long time. We do
not need to have it in write mode - read mode is enough, and we do not need to
hole it over the entire loop. So this patch turns changes the locking: takes
'work_sem' in read mode and pushes it down to the loop.
Commit "f70b7e5 UBIFS: remove Kconfig debugging option" broke UBIFS and it
refuses to initialize if debugfs (CONFIG_DEBUG_FS) is disabled. I incorrectly
assumed that debugfs files creation function will return success if debugfs
is disabled, but they actually return -ENODEV. This patch fixes the issue.
Reported-by: Paul Parsons <lost.distance@yahoo.com> Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com> Tested-by: Paul Parsons <lost.distance@yahoo.com>
Commit "aa44d1d UBI: remove Kconfig debugging option" broke UBI and it
refuses to initialize if debugfs (CONFIG_DEBUG_FS) is disabled. I incorrectly
assumed that debugfs files creation function will return success if debugfs
is disabled, but they actually return -ENODEV. This patch fixes the issue.
Reported-by: Paul Parsons <lost.distance@yahoo.com> Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com> Tested-by: Paul Parsons <lost.distance@yahoo.com>
Cougar/Panther Point redefine the bits in SDEIIR pretty completely.
This function is just debugging, but if we're debugging we probably want
to be told accurate things instead of lies.
I'm told Lynx Point changes this yet more, but I have no idea how...
Note from Eugeni's review:
"For the record and for future enabling efforts, for LPT, bits 28-31
and 1-14 are gone since CPT/PPT (e.g., those must be zero). And there
is the bit 15 as a new addition, but we are not using it yet and
probably won't be using in foreseeable future."
Signed-off-by: Adam Jackson <ajax@redhat.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=35103 Reviewed-by: Eugeni Dodonov <eugeni.dodonov@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Linus Torvalds [Wed, 6 Jun 2012 17:47:15 +0000 (10:47 -0700)]
Merge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux
Pull ACPI and Power Management changes from Len Brown.
This does an evil merge to fix up what I think is a mismerge by Len to
the gma500 driver, and restore it to the mainline state.
In that driver, both branches had commented out the call to
acpi_video_register(), and Len resolved the merge to that commented-out
version.
However, in mainline, further changes by Alan (commit d839ede47a56:
"gma500: opregion and ACPI" to be exact) had re-enabled the ACPI video
registration, so the current state of the driver seems to want it.
Alan is apparently still feeling the effects of partying with the Queen,
so he didn't reply to my query, but I'll do the evil merge anyway.
* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux:
ACPI: fix acpi_bus.h build warnings when ACPI is not enabled
drivers: acpi: Fix dependency for ACPI_HOTPLUG_CPU
tools/power turbostat: fix IVB support
tools/power turbostat: fix un-intended affinity of forked program
ACPI video: use after input_unregister_device()
gma500: don't register the ACPI video bus
acpi_video: Intel video is not always i915
acpi_video: fix leaking PCI references
ACPI: Ignore invalid _PSS entries, but use valid ones
ACPI battery: only refresh the sysfs files when pertinent information changes
Linus Torvalds [Wed, 6 Jun 2012 17:45:21 +0000 (10:45 -0700)]
Merge tag 'rdma-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband
Pull InfiniBand/RDMA fixes from Roland Dreier:
"All in hardware drivers:
- Fix crash in cxgb4
- Fixes to new ocrdma driver
- Regression fixes for mlx4"
* tag 'rdma-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
IB/mlx4: Fix max_wqe capacity reported from query device
mlx4_core: Fix setting VL_cap in mlx4_SET_PORT wrapper flow
IB/mlx4: Fix EQ deallocation in legacy mode
RDMA/cxgb4: Fix crash when peer address is 0.0.0.0
RDMA/ocrdma: Remove unnecessary version.h includes
RDMA/ocrdma: Fix signaled event for SRQ_LIMIT_REACHED
RDMA/ocrdma: Correct queue free count math
Sagi Grimberg [Thu, 24 May 2012 13:08:08 +0000 (16:08 +0300)]
IB/mlx4: Fix max_wqe capacity reported from query device
1. Limit the max number of WQEs per QP reported when querying the
device, so that ib_create_qp() will not fail for a QP size that the
device claimed to support due to additional headroom WQEs being
allocated.
2. Limit qp resources accepted for ib_create_qp() to the limits
reported in ib_query_device(). In kernel space, make sure that the
limits returned to the caller following qp creation also lie within
the reported device limits. For userspace, report as before, and do
adjustment in libmlx4 (so as not to break ABI).
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Sagi Grimberg <sagig@mellanox.co.il> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
Jack Morgenstein [Thu, 24 May 2012 13:08:09 +0000 (16:08 +0300)]
mlx4_core: Fix setting VL_cap in mlx4_SET_PORT wrapper flow
Commit 096335b3f983 ("mlx4_core: Allow dynamic MTU configuration for
IB ports") modifies the port VL setting. This exposes a bug in
mlx4_common_set_port(), where the VL cap value passed in (inside the
command mailbox) is incorrectly zeroed-out:
mlx4_SET_PORT modifies the VL_cap field (byte 3 of the mailbox).
Since the SET_PORT command is paravirtualized on the master as well as
on the slaves, mlx4_SET_PORT_wrapper() is invoked on the master. This
calls mlx4_common_set_port() where mailbox byte 3 gets overwritten by
code which should only set a single bit in that byte (for the reset
qkey counter flag) -- but instead overwrites the entire byte.
The result is that when running in SR-IOV mode, the VL_cap will be set
to zero -- fix this.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
Linus Torvalds [Wed, 6 Jun 2012 16:49:28 +0000 (09:49 -0700)]
Merge tag 'md-3.5-fixes' of git://neil.brown.name/md
Pull two md fixes from NeilBrown:
"One sparse-warning fix, one bugfix for 3.4-stable"
* tag 'md-3.5-fixes' of git://neil.brown.name/md:
md: raid1/raid10: fix problem with merge_bvec_fn
lib/raid6: fix sparse warnings in recovery functions
Linus Torvalds [Wed, 6 Jun 2012 16:47:57 +0000 (09:47 -0700)]
Merge tag 'iommu-fixes-3.5-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu
Pull IOMMU fixes from Joerg Roedel:
"Two patches are in here which fix AMD IOMMU specific issues. One
patch fixes a long-standing warning on resume because the
amd_iommu_resume function enabled interrupts. The other patch fixes a
deadlock in an error-path of the page-fault request handling code of
the IOMMU driver.
* tag 'iommu-fixes-3.5-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu:
iommu/amd: Fix deadlock in ppr-handling error path
iommu/amd: Cache pdev pointer to root-bridge
Chris Metcalf [Wed, 6 Jun 2012 15:21:44 +0000 (11:21 -0400)]
tile: remove cpu_idle_on_new_stack
This routine isn't used unless CONFIG_HOMECACHE is enabled, which
isn't even available as a public configuration option yet.
Since it no longer links correctly in 3.4, just remove it for now.
Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
Andi Kleen [Wed, 6 Jun 2012 00:56:48 +0000 (17:56 -0700)]
perf/x86: Don't assume there can be only 4 PEBS events
On Sandy Bridge in non HT mode there are 8 counters available.
Since every counter can write a PEBS record assuming there are
4 max is incorrect. Use the reported counter number -- with an
upper limit for a static array -- instead.
Also I made the warning messages a bit more informational.
Vince Weaver [Thu, 1 Mar 2012 22:28:14 +0000 (17:28 -0500)]
perf/x86: Use rdpmc() rather than rdmsr() when possible in the kernel
The rdpmc instruction is faster than the equivelant rdmsr call,
so use it when possible in the kernel.
The perfctr kernel patches did this, after extensive testing showed
rdpmc to always be faster (One can look in etc/costs in the perfctr-2.6
package to see a historical list of the overhead).
I have done some tests on a 3.2 kernel, the kernel module I used
was included in the first posting of this patch:
[ It's probably possible (and desirable) to do this without
requiring a new field in the hw_perf_event structure, but
the fixed events make this tricky. ]
Peter Zijlstra [Mon, 14 May 2012 13:25:34 +0000 (15:25 +0200)]
perf/x86: Fix wrmsrl() debug wrapper
Move the wrmslr() debug wrapper to the common header now that all the
include games are gone. Also clean it up a bit to avoid multiple
evaluation of the argument.
Peter Zijlstra [Tue, 5 Jun 2012 08:26:43 +0000 (10:26 +0200)]
perf/x86: Enable/Add IvyBridge hardware support
Implement rudimentary IVB perf support. The SDM states its identical
to SNB with exception of the exact event tables, but a quick look
suggests they're similar enough.
Peter Zijlstra [Tue, 5 Jun 2012 13:30:31 +0000 (15:30 +0200)]
perf/x86: Fix Intel shared extra MSR allocation
Zheng Yan reported that event group validation can wreck event state
when Intel extra_reg allocation changes event state.
Validation shouldn't change any persistent state. Cloning events in
validate_{event,group}() isn't really pretty either, so add a few
special cases to avoid modifying the event state.
The code is restructured to minimize the special case impact.
Reported-by: Zheng Yan <zheng.z.yan@linux.intel.com> Acked-by: Stephane Eranian <eranian@google.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1338903031.28282.175.camel@twins Signed-off-by: Ingo Molnar <mingo@kernel.org>
Oleg Nesterov [Tue, 29 May 2012 19:30:08 +0000 (21:30 +0200)]
uprobes: Kill uprobes_srcu/uprobe_srcu_id
Kill the no longer needed uprobes_srcu/uprobe_srcu_id code.
It doesn't really work anyway. synchronize_srcu() can only
synchronize with the code "inside" the
srcu_read_lock/srcu_read_unlock section, while
uprobe_pre_sstep_notifier() does srcu_read_lock() _after_ we
already hit the breakpoint.
I guess this probably works "in practice". synchronize_srcu() is
slow and it implies synchronize_sched(), and the probed task
enters the non- preemptible section at the start of exception
handler. Still this is not right at least in theory, and
task->uprobe_srcu_id blows task_struct.
Oleg Nesterov [Tue, 29 May 2012 19:29:47 +0000 (21:29 +0200)]
uprobes: Teach handle_swbp() to rely on "is_swbp" rather than uprobes_srcu
Currently handle_swbp() assumes that it can't race with
unregister, so it roughly does:
if (find_uprobe(vaddr))
process_uprobe();
else
send_sig(SIGTRAP);
This relies on the not-really-working uprobes_srcu code we are
going to remove, see the next patch.
With this patch we rely on the result of
is_swbp_at_addr(bp_vaddr) if find_uprobe() fails.
If is_swbp == 1, then we hit the normal int3, we should send
SIGTRAP.
If is_swbp == 0, we raced with uprobe_unregister(), we simply
restart this insn again.
The "difficult" case is is_swbp == -EFAULT, when we can't read
this memory. In this case I think we should restart too, and
this is more correct compared to the current code which sends
SIGTRAP.
Ignoring ENOMEM/etc from get_user_pages(), this can only happen
if another thread unmaps this memory before find_active_uprobe()
takes mmap_sem. It would be better to pretend it was unmapped
before this insn was executed, restart, and get SIGSEGV.
Oleg Nesterov [Tue, 29 May 2012 19:29:28 +0000 (21:29 +0200)]
uprobes: Change register_for_each_vma() to take mm->mmap_sem for writing
Change register_for_each_vma() to take mm->mmap_sem for writing.
This is a bit unfortunate but hopefully not too bad, this is the
slow path anyway.
This is needed to ensure that find_active_uprobe() can not race
with uprobe_register() which adds the new bp at the same
bp_vaddr, after find_uprobe() fails and before
is_swbp_at_addr_fast() checks the memory.
IOW, this is needed to ensure that if find_active_uprobe()
returns NULL but is_swbp == true, we can safely assume that it
was the "normal" int3 and we should send SIGTRAP.
There is another reason for this change. We are going to replace
uprobes_state->count with MMF_ flags set by register/unregister
and cleared by find_active_uprobe(), and set/clear shouldn't
race with each other.
Oleg Nesterov [Tue, 29 May 2012 19:29:14 +0000 (21:29 +0200)]
uprobes: Teach find_active_uprobe() to provide the "is_swbp" info
A separate patch to simplify the review, and for the
documentation.
The patch adds another "int *is_swbp" argument to
find_active_uprobe(), so far its only caller doesn't use this
info.
With this patch find_active_uprobe() additionally does:
- if find_vma() + ->vm_start check fails, *is_swbp = -EFAULT
- otherwise, if valid_vma() + find_uprobe() fails, it holds
the result of is_swbp_at_addr(), can be negative too. The
latter is only possible if we raced with another thread
which did munmap/etc after we hit this bp.
IOW. If find_active_uprobe(&is_swbp) returns NULL, the caller
can look at is_swbp to figure out whether the current insn is bp
or not, or detect the race with another thread if it is
negative.
Note: I think that performance-wise this change is fine. This
adds is_swbp_at_addr(), but only if we raced with
uprobe_unregister() or if we hit the "normal" int3 but this mm
has uprobes as well. And even in this case the slow
read_opcode() path is very unlikely, this insn recently
triggered do_int3(), __copy_from_user_inatomic() shouldn't fail
in the likely case.
Oleg Nesterov [Tue, 29 May 2012 19:28:57 +0000 (21:28 +0200)]
uprobes: Introduce find_active_uprobe() helper
No functional changes. Move the "find uprobe" code from
handle_swbp() to the new helper, find_active_uprobe().
Note: with or without this change, the find-active-uprobe logic
is not exactly right. We can race with another thread which
unmaps the memory with the valid uprobe before we take
mm->mmap_sem. We can't find this uprobe simply because
find_vma() fails. In this case we wrongly assume that this trap
was not caused by uprobe and send the erroneous SIGTRAP. See the
next changes.
Oleg Nesterov [Tue, 29 May 2012 19:27:59 +0000 (21:27 +0200)]
uprobes: Change read_opcode() to use FOLL_FORCE
set_orig_insn()->read_opcode() should not fail if the probed
task did mprotect() after uprobe_register(), change it to use
FOLL_FORCE. Without FOLL_WRITE this doesn't have any "side"
effect but allows to read the !VM_READ memory.
There is another reason for this change, we are going to use
is_swbp_at_addr() from handle_swbp() which can race with another
thread doing mprotect().
Without this patch, applications with two different stack
regions (eg: native stack vs JIT stack) get truncated
callchains even when RBP chaining is present. GDB shows proper
stack traces and the frame pointer chaining is intact.
This patch disables the (fp < RSP) check, hoping that other checks
in the code save the day for us. In our limited testing, this
didn't seem to break anything.
In the long term, we could potentially have userspace advise
the kernel on the range of valid stack addresses, so we don't
spend a lot of time unwinding from bogus addresses.
Signed-off-by: Arun Sharma <asharma@fb.com> CC: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Stephane Eranian <eranian@google.com> Cc: Namhyung Kim <namhyung.kim@lge.com> Cc: Tom Zanussi <tzanussi@gmail.com> Cc: linux-kernel@vger.kernel.org Cc: linux-perf-users@vger.kernel.org Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1334961696-19580-2-git-send-email-asharma@fb.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
It does not get processed because sched_domain_level_max is 0 at the
time that setup_relax_domain_level() is run.
Simply accept the value as it is, as we don't know the value of
sched_domain_level_max until sched domain construction is completed.
Fix sched_relax_domain_level in cpuset. The build_sched_domain() routine calls
the set_domain_attribute() routine prior to setting the sd->level, however,
the set_domain_attribute() routine relies on the sd->level to decide whether
idle load balancing will be off/on.
Peter Zijlstra [Tue, 5 Jun 2012 08:26:43 +0000 (10:26 +0200)]
perf/x86: Enable/Add IvyBridge hardware support
Implement rudimentary IVB perf support. The SDM states its identical
to SNB with exception of the exact event tables, but a quick look
suggests they're similar enough.
Peter Zijlstra [Tue, 5 Jun 2012 13:30:31 +0000 (15:30 +0200)]
perf/x86: Fix Intel shared extra MSR allocation
Zheng Yan reported that event group validation can wreck event state
when Intel extra_reg allocation changes event state.
Validation shouldn't change any persistent state. Cloning events in
validate_{event,group}() isn't really pretty either, so add a few
special cases to avoid modifying the event state.
The code is restructured to minimize the special case impact.
Reported-by: Zheng Yan <zheng.z.yan@linux.intel.com> Acked-by: Stephane Eranian <eranian@google.com> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1338903031.28282.175.camel@twins Signed-off-by: Ingo Molnar <mingo@kernel.org>
if you look at that code you'll find the double_lock_balance() in
question is the one in find_lock_lowest_rq() [yay for inlining].
Now find_lock_lowest_rq() has a bug.. it fails to use
double_unlock_balance() in one exit path, if this results in a retry in
push_rt_task() we'll call double_lock_balance() again, at which point
we'll run into said lockdep confusion.
Reported-by: Roland Dreier <roland@kernel.org> Acked-by: Steven Rostedt <rostedt@goodmis.org> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1337282386.4281.77.camel@twins Signed-off-by: Ingo Molnar <mingo@kernel.org>
Alex Shi [Wed, 6 Jun 2012 06:52:51 +0000 (14:52 +0800)]
sched/numa: Load balance between remote nodes
Commit cb83b629b ("sched/numa: Rewrite the CONFIG_NUMA sched
domain support") removed the NODE sched domain and started checking
if the node distance in SLIT table is farther than REMOTE_DISTANCE,
if so, it will lose the load balance chance at exec/fork/wake_affine
points.
But actually, even the node distance is farther than REMOTE_DISTANCE.
Modern CPUs also has QPI like connections, which ensures that memory
access is not too slow between nodes. So the above change in behavior
on NUMA machine causes a performance regression on various benchmarks:
hackbench, tbench, netperf, oltp, etc.
This patch will recover the scheduler behavior to old mode on all my
Intel platforms: NHM EP/EX, WSM EP, SNB EP/EP4S, and thus fixes the
perfromance regressions. (all of them just have 2 kinds distance, 10, 21)
Tomoki Sekiyama [Mon, 28 May 2012 09:09:18 +0000 (18:09 +0900)]
x86/ioapic: Fix NULL pointer dereference on CPU hotplug after disabling irqs
In current Linux, percpu variable `vector_irq' is not cleared on
offlined cpus while disabling devices' irqs. If the cpu that has
the disabled irqs in vector_irq is hotplugged,
__setup_vector_irq() hits invalid irq vector and may crash.
The root cause seems to be the
default_send_IPI_mask_allbutself_phys() takes quite some time (I
measured it could be several ms) to complete sending NMIs to all
the other 23 CPUs, and for HZ=250/1000 system, the time is long
enough for a timer interrupt to happen, which will in turn
trigger to kick load balance to a stopped CPU and cause this
warning in native_smp_send_reschedule().
So disabling the local irq before stop_other_cpu() can fix this
problem (tested 25 times reboot ok), and it is fine as there
should be nobody caring the timer interrupt in such reboot
stage.
The latest 3.4 kernel slightly changes this behavior by sending
REBOOT_VECTOR first and only send NMI_VECTOR if the REBOOT_VCTOR
fails, and this patch is still needed to prevent the problem.
Signed-off-by: Feng Tang <feng.tang@intel.com> Acked-by: Don Zickus <dzickus@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/20120530231541.4c13433a@feng-i7 Signed-off-by: Ingo Molnar <mingo@kernel.org>
x86/intel/moorestown: Change intel_scu_devices_create() to __devinit
The allmodconfig hits:
WARNING: vmlinux.o(.text+0x6553d): Section mismatch in
reference from the function intel_scu_devices_create() to the
function .devinit.text: spi_register_board_info()
[...]
This patch marks intel_scu_devices_create() as devinit because
it only calls a devinit function, spi_register_board_info().
Signed-off-by: Sebastian Andrzej Siewior <sebastian@breakpoint.cc> Cc: Alan Cox <alan@linux.intel.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Mika Westerberg <mika.westerberg@linux.intel.com> Cc: Samuel Ortiz <sameo@linux.intel.com> Cc: Feng Tang <feng.tang@intel.com> Link: http://lkml.kernel.org/r/20120531212025.GA8519@breakpoint.cc Signed-off-by: Ingo Molnar <mingo@kernel.org>
The reason is that the bit of node 2 was not set at
numa_nodes_parsed. numa_nodes_parsed is set by only
acpi_numa_processor_affinity_init /
acpi_numa_x2apic_affinity_init. Thus even if hot-added memory
which is same PXM as hot-added CPU is written in ACPI SRAT
Table, if the hot-added CPU is not written in ACPI SRAT table,
numa_nodes_parsed is not set.
But according to ACPI Spec Rev 5.0, it says about ACPI SRAT
table as follows: This optional table provides information that
allows OSPM to associate processors and memory ranges, including
ranges of memory provided by hot-added memory devices, with
system localities / proximity domains and clock domains.
It means that ACPI SRAT table only provides information for CPUs
present at boot time and for memory including hot-added memory.
So hot-added memory is written in ACPI SRAT table, but hot-added
CPU is not written in it. Thus numa_nodes_parsed should be set
by not only acpi_numa_processor_affinity_init /
acpi_numa_x2apic_affinity_init but also
acpi_numa_memory_affinity_init for the case.
Additionally, if system has cpuless memory node,
acpi_numa_processor_affinity_init /
acpi_numa_x2apic_affinity_init cannot set numa_nodes_parseds
since these functions cannot find cpu description for the node.
In this case, numa_nodes_parsed needs to be set by
acpi_numa_memory_affinity_init.
x86/decoder: Fix bsr/bsf/jmpe decoding with operand-size prefix
Fix the x86 instruction decoder to decode bsr/bsf/jmpe with
operand-size prefix (66h). This fixes the test case failure
reported by Linus, attached below.
bsf/bsr/jmpe have a special encoding. Opcode map in
Intel Software Developers Manual vol2 says they have
TZCNT/LZCNT variants if it has F3h prefix. However, there
is no information if it has other 66h or F2h prefixes.
Current instruction decoder supposes that those are
bad instructions, but it actually accepts at least
operand-size prefixes.
H. Peter Anvin further explains:
" TZCNT/LZCNT are F3 + BSF/BSR exactly because the F2 and
F3 prefixes have historically been no-ops with most instructions.
This allows software to unconditionally use the prefixed versions
and get TZCNT/LZCNT on the processors that have them if they don't
care about the difference. "
This fixes errors reported by test_get_len:
Warning: arch/x86/tools/test_get_len found difference at <em_bsf>:ffffffff81036d87
Warning: ffffffff81036de5: 66 0f bc c2 bsf %dx,%ax
Warning: objdump says 4 bytes, but insn_get_length() says 3
Warning: arch/x86/tools/test_get_len found difference at <em_bsr>:ffffffff81036ea6
Warning: ffffffff81036f04: 66 0f bd c2 bsr %dx,%ax
Warning: objdump says 4 bytes, but insn_get_length() says 3
Warning: decoded and checked 13298882 instructions with 2 warnings
projects / karo-tx-linux.git / log