I wanted to revert my v3.1 commit d0823576bf ("mm: pincer in
truncate_inode_pages_range"), to keep truncate_inode_pages_range() in
synch with shmem_undo_range(); but have stepped back - a change to
hole-punching in truncate_inode_pages_range() is a change to
hole-punching in every filesystem (except tmpfs) that supports it.
If there's a logical proof why no filesystem can depend for its own
correctness on the pincer guarantee in truncate_inode_pages_range() - an
instant when the entire hole is removed from pagecache - then let's
revisit later. But the evidence is that only tmpfs suffered from the
livelock, and we have no intention of extending hole-punch to ramfs. So
for now just add a few comments (to match or differ from those in
shmem_undo_range()), and fix one silliness noticed in d0823576bf4b...
Its "index == start" addition to the hole-punch termination test was
incomplete: it opened a way for the end condition to be missed, and the
loop go on looking through the radix_tree, all the way to end of file.
Fix that pessimization by resetting index when detected in inner loop.
Note that it's actually hard to hit this case, without the obsessive
concurrent faulting that trinity does: normally all pages are removed in
the initial trylock_page() pass, and this loop finds nothing to do. I
had to "#if 0" out the initial pass to reproduce bug and test fix.
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Sasha Levin <sasha.levin@oracle.com>
Cc: Konstantin Khlebnikov <koct9i@gmail.com>
Cc: Lukas Czerner <lczerner@redhat.com>
Cc: Dave Jones <davej@redhat.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
shmem_fault() is the actual culprit in trinity's hole-punch starvation,
and the most significant cause of such problems: since a page faulted is
one that then appears page_mapped(), needing unmap_mapping_range() and
i_mmap_mutex to be unmapped again.
But it is not the only way in which a page can be brought into a hole in
the radix_tree while that hole is being punched; and Vlastimil's testing
implies that if enough other processors are busy filling in the hole,
then shmem_undo_range() can be kept from completing indefinitely.
shmem_file_splice_read() is the main other user of SGP_CACHE, which can
instantiate shmem pagecache pages in the read-only case (without holding
i_mutex, so perhaps concurrently with a hole-punch). Probably it's
silly not to use SGP_READ already (using the ZERO_PAGE for holes): which
ought to be safe, but might bring surprises - not a change to be rushed.
shmem_read_mapping_page_gfp() is an internal interface used by
drivers/gpu/drm GEM (and next by uprobes): it should be okay. And
shmem_file_read_iter() uses the SGP_DIRTY variant of SGP_CACHE, when
called internally by the kernel (perhaps for a stacking filesystem,
which might rely on holes to be reserved): it's unclear whether it could
be provoked to keep hole-punch busy or not.
We could apply the same umbrella as now used in shmem_fault() to
shmem_file_splice_read() and the others; but it looks ugly, and use over
a range raises questions - should it actually be per page? can these get
starved themselves?
The origin of this part of the problem is my v3.1 commit d0823576bf
("mm: pincer in truncate_inode_pages_range"), once it was duplicated
into shmem.c. It seemed like a nice idea at the time, to ensure
(barring RCU lookup fuzziness) that there's an instant when the entire
hole is empty; but the indefinitely repeated scans to ensure that make
it vulnerable.
Revert that "enhancement" to hole-punch from shmem_undo_range(), but
retain the unproblematic rescanning when it's truncating; add a couple
of comments there.
Remove the "indices[0] >= end" test: that is now handled satisfactorily
by the inner loop, and mem_cgroup_uncharge_start()/end() are too light
to be worth avoiding here.
But if we do not always loop indefinitely, we do need to handle the case
of swap swizzled back to page before shmem_free_swap() gets it: add a
retry for that case, as suggested by Konstantin Khlebnikov; and for the
case of page swizzled back to swap, as suggested by Johannes Weiner.
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Sasha Levin <sasha.levin@oracle.com>
Suggested-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Konstantin Khlebnikov <koct9i@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Lukas Czerner <lczerner@redhat.com>
Cc: Dave Jones <davej@redhat.com>
Cc: <stable@vger.kernel.org> [3.1+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit f00cdc6df7 ("shmem: fix faulting into a hole while it's
punched") was buggy: Sasha sent a lockdep report to remind us that
grabbing i_mutex in the fault path is a no-no (write syscall may already
hold i_mutex while faulting user buffer).
We tried a completely different approach (see following patch) but that
proved inadequate: good enough for a rational workload, but not good
enough against trinity - which forks off so many mappings of the object
that contention on i_mmap_mutex while hole-puncher holds i_mutex builds
into serious starvation when concurrent faults force the puncher to fall
back to single-page unmap_mapping_range() searches of the i_mmap tree.
So return to the original umbrella approach, but keep away from i_mutex
this time. We really don't want to bloat every shmem inode with a new
mutex or completion, just to protect this unlikely case from trinity.
So extend the original with wait_queue_head on stack at the hole-punch
end, and wait_queue item on the stack at the fault end.
This involves further use of i_lock to guard against the races: lockdep
has been happy so far, and I see fs/inode.c:unlock_new_inode() holds
i_lock around wake_up_bit(), which is comparable to what we do here.
i_lock is more convenient, but we could switch to shmem's info->lock.
This issue has been tagged with CVE-2014-4171, which will require commit
f00cdc6df7 and this and the following patch to be backported: we
suggest to 3.1+, though in fact the trinity forkbomb effect might go
back as far as 2.6.16, when madvise(,,MADV_REMOVE) came in - or might
not, since much has changed, with i_mmap_mutex a spinlock before 3.0.
Anyone running trinity on 3.0 and earlier? I don't think we need care.
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Sasha Levin <sasha.levin@oracle.com>
Tested-by: Sasha Levin <sasha.levin@oracle.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Konstantin Khlebnikov <koct9i@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Lukas Czerner <lczerner@redhat.com>
Cc: Dave Jones <davej@redhat.com>
Cc: <stable@vger.kernel.org> [3.1+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Ingo Korb reported that "repeated mapping of the same file on tmpfs
using remap_file_pages sometimes triggers a BUG at mm/filemap.c:202 when
the process exits".
He bisected the bug to d7c1755179 ("mm: implement ->map_pages for
shmem/tmpfs"), although the bug was actually added by commit
8c6e50b029 ("mm: introduce vm_ops->map_pages()").
The problem is caused by calling do_fault_around for a _non-linear_
fault. In this case pgoff is shifted and might become negative during
calculation.
Faulting around non-linear page-fault makes no sense and breaks the
logic in do_fault_around because pgoff is shifted.
Signed-off-by: Konstantin Khlebnikov <koct9i@gmail.com>
Reported-by: Ingo Korb <ingo.korb@tu-dortmund.de>
Tested-by: Ingo Korb <ingo.korb@tu-dortmund.de>
Cc: Hugh Dickins <hughd@google.com>
Cc: Sasha Levin <sasha.levin@oracle.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Ning Qu <quning@google.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: <stable@vger.kernel.org> [3.15.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When compiling a SH2A kernel (e.g. se7206_defconfig or rsk7203_defconfig)
using sh4-linux-gcc, linking fails with:
net/built-in.o: In function `__sk_run_filter':
net/core/filter.c:566: undefined reference to `__fpscr_values'
net/core/filter.c:269: undefined reference to `__fpscr_values'
...
net/built-in.o:net/core/filter.c:580: more undefined references to `__fpscr_values' follow
This happens because sh4-linux-gcc doesn't support the "-m2a-nofpu",
which is thus filtered out by "$(call cc-option, ...)".
As compiling using sh4-linux-gcc is useful for compile coverage, also
try passing "-m4-nofpu" (which is presumably filtered out when using a
real sh2a-linux toolchain) to disable the generation of FPU instructions
and references to __fpscr_values[].
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Cc: Guenter Roeck <linux@roeck-us.net>
Cc: Tony Breeds <tony@bakeyournoodle.com>
Cc: Alexei Starovoitov <ast@plumgrid.com>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Cc: Daniel Borkmann <dborkman@redhat.com>
Cc: Magnus Damm <magnus.damm@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
I triggered VM_BUG_ON() in vma_address() when I tried to migrate an
anonymous hugepage with mbind() in the kernel v3.16-rc3. This is
because pgoff's calculation in rmap_walk_anon() fails to consider
compound_order() only to have an incorrect value.
This patch introduces page_to_pgoff(), which gets the page's offset in
PAGE_CACHE_SIZE.
Kirill pointed out that page cache tree should natively handle
hugepages, and in order to make hugetlbfs fit it, page->index of
hugetlbfs page should be in PAGE_CACHE_SIZE. This is beyond this patch,
but page_to_pgoff() contains the point to be fixed in a single function.
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Hillf Danton <dhillf@gmail.com>
Cc: Naoya Horiguchi <nao.horiguchi@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit 079148b919 ("coredump: factor out the setting of PF_DUMPCORE")
cleaned up the setting of PF_DUMPCORE by removing it from all the
linux_binfmt->core_dump() and moving it to zap_threads().But this ended
up clearing all the previously set flags. This causes issues during
core generation when tsk->flags is checked again (eg. for PF_USED_MATH
to dump floating point registers). Fix this.
Signed-off-by: Silesh C V <svellattu@mvista.com>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Cc: Mandeep Singh Baines <msb@chromium.org>
Cc: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull networking fixes from David Miller:
1) Null termination fix in dns_resolver got the pointer dereferncing
wrong, fix from Ben Hutchings.
2) ip_options_compile() has a benign but real buffer overflow when
parsing options. From Eric Dumazet.
3) Table updates can crash in netfilter's nftables if none of the state
flags indicate an actual change, from Pablo Neira Ayuso.
4) Fix race in nf_tables dumping, also from Pablo.
5) GRE-GRO support broke the forwarding path because the segmentation
state was not fully initialized in these paths, from Jerry Chu.
6) sunvnet driver leaks objects and potentially crashes on module
unload, from Sowmini Varadhan.
7) We can accidently generate the same handle for several u32
classifier filters, fix from Cong Wang.
8) Several edge case bug fixes in fragment handling in xen-netback,
from Zoltan Kiss.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (21 commits)
ipv4: fix buffer overflow in ip_options_compile()
batman-adv: fix TT VLAN inconsistency on VLAN re-add
batman-adv: drop QinQ claim frames in bridge loop avoidance
dns_resolver: Null-terminate the right string
xen-netback: Fix pointer incrementation to avoid incorrect logging
xen-netback: Fix releasing header slot on error path
xen-netback: Fix releasing frag_list skbs in error path
xen-netback: Fix handling frag_list on grant op error path
net_sched: avoid generating same handle for u32 filters
net: huawei_cdc_ncm: add "subclass 3" devices
net: qmi_wwan: add two Sierra Wireless/Netgear devices
wan/x25_asy: integer overflow in x25_asy_change_mtu()
net: ppp: fix creating PPP pass and active filters
net/mlx4_en: cq->irq_desc wasn't set in legacy EQ's
sunvnet: clean up objects created in vnet_new() on vnet_exit()
r8169: Enable RX_MULTI_EN for RTL_GIGA_MAC_VER_40
net-gre-gro: Fix a bug that breaks the forwarding path
netfilter: nf_tables: 64bit stats need some extra synchronization
netfilter: nf_tables: set NLM_F_DUMP_INTR if netlink dumping is stale
netfilter: nf_tables: safe RCU iteration on list when dumping
...
Pull sparc fix from David Miller:
"Need to hook up the new renameat2 system call"
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc:
sparc: Hook up renameat2 syscall.
Pull IDE fixes from David Miller:
- fix interrupt registry for some Atari IDE chipsets.
- adjust Kconfig dependencies for x86_32 specific chips.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/ide:
ide: Fix SC1200 dependencies
ide: Fix CS5520 and CS5530 dependencies
m68k/atari - ide: do not register interrupt if host->get_lock is set
Pull trace fix from Steven Rostedt:
"Tony Luck found that using the "uptime" trace clock that uses jiffies
as a counter was converted to nanoseconds (silly), and after 1 hour 11
minutes and 34 seconds, this monotonic clock would wrap, causing havoc
with the tracing system and making the clock useless.
He converted that clock to use jiffies_64 and made it into a counter
instead of nanosecond conversions, and displayed the clock with the
straight jiffy count, which works much better than it did in the past"
* tag 'trace-fixes-v3.16-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace:
tracing: Fix wraparound problems in "uptime" trace clock
Antonio Quartulli says:
====================
pull request [net]: batman-adv 20140721
here you have two fixes that we have been testing for quite some time
(this is why they arrived a bit late in the rc cycle).
Patch 1) ensures that BLA packets get dropped and not forwarded to the
mesh even if they reach batman-adv within QinQ frames. Forwarding them
into the mesh means messing up with the TT database of other nodes which
can generate all kind of unexpected behaviours during route computation.
Patch 2) avoids a couple of race conditions triggered upon fast VLAN
deletion-addition. Such race conditions are pretty dangerous because
they not only create inconsistencies in the TT database of the nodes
in the network, but such scenario is also unrecoverable (unless
nodes are rebooted).
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull media fixes from Mauro Carvalho Chehab:
"A series of driver fixes:
- fix DVB-S tuning with tda1071
- fix tuner probe on af9035 when the device has a bad eeprom
- some fixes for the new si2168/2157 drivers
- one Kconfig build fix (for omap4iss)
- fixes at vpif error path
- don't lock saa7134 ioctl at driver's base core level, as it now
uses V4L2 and VB2 locking schema
- fix audio at hdpvr driver
- fix the aspect ratio at the digital timings table
- one new USB ID (at gspca_pac7302): Genius i-Look 317 webcam"
* 'v4l_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media:
[media] gspca_pac7302: Add new usb-id for Genius i-Look 317
[media] tda10071: fix returned symbol rate calculation
[media] tda10071: fix spec inversion reporting
[media] tda10071: add missing DVB-S2/PSK-8 FEC AUTO
[media] tda10071: force modulation to QPSK on DVB-S
[media] hdpvr: fix two audio bugs
[media] davinci: vpif: missing unlocks on error
[media] af9035: override tuner id when bad value set into eeprom
[media] saa7134: use unlocked_ioctl instead of ioctl
[media] media: v4l2-core: v4l2-dv-timings.c: Cleaning up code wrong value used in aspect ratio
[media] si2168: firmware download fix
[media] si2157: add one missing parenthesis
[media] si2168: add one missing parenthesis
[media] staging: tighten omap4iss dependencies
Pull block fixes from Jens Axboe:
"Final block fixes for 3.16
Four small fixes that should go into 3.16, have been queued up for a
bit and delayed due to vacation and other euro duties. But here they
are. The pull request contains:
- Fix for a reported crash with shared tagging on SCSI from Christoph
- A regression fix for drbd. From Lars Ellenberg.
- Hooking up the compat ioctl for BLKZEROOUT, which requires no
translation. From Mikulas.
- A fix for a regression where we woud crash on queue exit if the
root_blkg is gone/not there. From Tejun"
* 'for-linus' of git://git.kernel.dk/linux-block:
block: provide compat ioctl for BLKZEROOUT
blkcg: don't call into policy draining if root_blkg is already gone
drbd: fix regression 'out of mem, failed to invoke fence-peer helper'
block: don't assume last put of shared tags is for the host
Pull libata fixes from Tejun Heo:
"Late libata fixes.
The most important one is from Kevin Hao which makes sure that libata
only allocates tags inside the max tag number the controller supports.
libata always had this problem but the recent tag allocation change
and addition of support for sata_fsl which only supports queue depth
of 16 exposed the issue.
Hans de Goede agreed to become the maintainer of libahci_platform
which is under higher than usual development pressure from all the new
controllers popping up from the ARM world"
* 'for-3.16-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/libata:
ahci: add support for the Promise FastTrak TX8660 SATA HBA (ahci mode)
drivers/ata/pata_ep93xx.c: use signed int type for result of platform_get_irq()
libata: EH should handle AMNF error condition as a media error
libata: support the ata host which implements a queue depth less than 32
MAINTAINERS: Add Hans de Goede as ahci-platform maintainer
Pull kvm fixes from Paolo Bonzini:
"These are mostly PPC changes for 3.16-new things. However, there is
an x86 change too and it is a regression from 3.14. As it only
affects nested virtualization and there were other changes in this
area in 3.16, I am not nominating it for 3.15-stable"
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
KVM: x86: Check for nested events if there is an injectable interrupt
KVM: PPC: RTAS: Do byte swaps explicitly
KVM: PPC: Book3S PR: Fix ABIv2 on LE
KVM: PPC: Assembly functions exported to modules need _GLOBAL_TOC()
PPC: Add _GLOBAL_TOC for 32bit
KVM: PPC: BOOK3S: HV: Use base page size when comparing against slb value
KVM: PPC: Book3E: Unlock mmu_lock when setting caching atttribute
Pull s390 fixes from Martin Schwidefsky:
"A couple of last minute bug fixes for 3.16, including a fix for ptrace
to close a hole which allowed a user space program to write to the
kernel address space"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
s390: fix restore of invalid floating-point-control
s390/zcrypt: improve device probing for zcrypt adapter cards
s390/ptrace: fix PSW mask check
s390/MSI: Use standard mask and unmask funtions
s390/3270: correct size detection with the read-partition command
s390: require mvcos facility, not tod clock steering facility
The "uptime" trace clock added in:
commit 8aacf017b0
tracing: Add "uptime" trace clock that uses jiffies
has wraparound problems when the system has been up more
than 1 hour 11 minutes and 34 seconds. It converts jiffies
to nanoseconds using:
(u64)jiffies_to_usecs(jiffy) * 1000ULL
but since jiffies_to_usecs() only returns a 32-bit value, it
truncates at 2^32 microseconds. An additional problem on 32-bit
systems is that the argument is "unsigned long", so fixing the
return value only helps until 2^32 jiffies (49.7 days on a HZ=1000
system).
Avoid these problems by using jiffies_64 as our basis, and
not converting to nanoseconds (we do convert to clock_t because
user facing API must not be dependent on internal kernel
HZ values).
Link: http://lkml.kernel.org/p/99d63c5bfe9b320a3b428d773825a37095bf6a51.1405708254.git.tony.luck@intel.com
Cc: stable@vger.kernel.org # 3.10+
Fixes: 8aacf017b0 "tracing: Add "uptime" trace clock that uses jiffies"
Signed-off-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
When a VLAN interface (on top of batX) is removed and
re-added within a short timeframe TT does not have enough
time to properly cleanup. This creates an internal TT state
mismatch as the newly created softif_vlan will be
initialized from scratch with a TT client count of zero
(even if TT entries for this VLAN still exist). The
resulting TT messages are bogus due to the counter / tt
client listing mismatch, thus creating inconsistencies on
every node in the network
To fix this issue destroy_vlan() has to not free the VLAN
object immediately but it has to be kept alive until all the
TT entries for this VLAN have been removed. destroy_vlan()
still removes the sysfs folder so that the user has the
feeling that everything went fine.
If the same VLAN is re-added before the old object is free'd,
then the latter is resurrected and re-used.
Implement such behaviour by increasing the reference counter
of a softif_vlan object every time a new local TT entry for
such VLAN is created and remove the object from the list
only when all the TT entries have been destroyed.
Signed-off-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Since bridge loop avoidance only supports untagged or simple 802.1q
tagged VLAN claim frames, claim frames with stacked VLAN headers (QinQ)
should be detected and dropped. Transporting the over the mesh may cause
problems on the receivers, or create bogus entries in the local tt
tables.
Reported-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Simon Wunderlich <simon@open-mesh.com>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Signed-off-by: Antonio Quartulli <antonio@meshcoding.com>
*_result[len] is parsed as *(_result[len]) which is not at all what we
want to touch here.
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Fixes: 84a7c0b1db ("dns_resolver: assure that dns_query() result is null-terminated")
Signed-off-by: David S. Miller <davem@davemloft.net>
Hugh Dickins
Konstantin Khlebnikov
Geert Uytterhoeven
Minchan Kim
Naoya Horiguchi
Silesh C V
Linus Torvalds
David S. Miller
Eric Dumazet
Tony Luck
Antonio Quartulli
Simon Wunderlich
Ben Hutchings