commit f5bf18fa22 upstream.
While testing AMS (Active Memory Sharing) / CMO (Cooperative Memory
Overcommit) on powerpc, we tripped the following:
kernel BUG at mm/bootmem.c:483!
cpu 0x0: Vector: 700 (Program Check) at [c000000000c03940]
pc: c000000000a62bd8: .alloc_bootmem_core+0x90/0x39c
lr: c000000000a64bcc: .sparse_early_usemaps_alloc_node+0x84/0x29c
sp: c000000000c03bc0
msr: 8000000000021032
current = 0xc000000000b0cce0
paca = 0xc000000001d80000
pid = 0, comm = swapper
kernel BUG at mm/bootmem.c:483!
enter ? for help
[c000000000c03c80] c000000000a64bcc
.sparse_early_usemaps_alloc_node+0x84/0x29c
[c000000000c03d50] c000000000a64f10 .sparse_init+0x12c/0x28c
[c000000000c03e20] c000000000a474f4 .setup_arch+0x20c/0x294
[c000000000c03ee0] c000000000a4079c .start_kernel+0xb4/0x460
[c000000000c03f90] c000000000009670 .start_here_common+0x1c/0x2c
This is
BUG_ON(limit && goal + size > limit);
and after some debugging, it seems that
goal = 0x7ffff000000
limit = 0x80000000000
and sparse_early_usemaps_alloc_node ->
sparse_early_usemaps_alloc_pgdat_section calls
return alloc_bootmem_section(usemap_size() * count, section_nr);
This is on a system with 8TB available via the AMS pool, and as a quirk
of AMS in firmware, all of that memory shows up in node 0. So, we end
up with an allocation that will fail the goal/limit constraints.
In theory, we could "fall-back" to alloc_bootmem_node() in
sparse_early_usemaps_alloc_node(), but since we actually have HOTREMOVE
defined, we'll BUG_ON() instead. A simple solution appears to be to
unconditionally remove the limit condition in alloc_bootmem_section,
meaning allocations are allowed to cross section boundaries (necessary
for systems of this size).
Johannes Weiner pointed out that if alloc_bootmem_section() no longer
guarantees section-locality, we need check_usemap_section_nr() to print
possible cross-dependencies between node descriptors and the usemaps
allocated through it. That makes the two loops in
sparse_early_usemaps_alloc_node() identical, so re-factor the code a
bit.
[akpm@linux-foundation.org: code simplification]
Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com>
Cc: Dave Hansen <haveblue@us.ibm.com>
Cc: Anton Blanchard <anton@au1.ibm.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ben Herrenschmidt <benh@kernel.crashing.org>
Cc: Robert Jennings <rcj@linux.vnet.ibm.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 1a5a9906d4 upstream.
In some cases it may happen that pmd_none_or_clear_bad() is called with
the mmap_sem hold in read mode. In those cases the huge page faults can
allocate hugepmds under pmd_none_or_clear_bad() and that can trigger a
false positive from pmd_bad() that will not like to see a pmd
materializing as trans huge.
It's not khugepaged causing the problem, khugepaged holds the mmap_sem
in write mode (and all those sites must hold the mmap_sem in read mode
to prevent pagetables to go away from under them, during code review it
seems vm86 mode on 32bit kernels requires that too unless it's
restricted to 1 thread per process or UP builds). The race is only with
the huge pagefaults that can convert a pmd_none() into a
pmd_trans_huge().
Effectively all these pmd_none_or_clear_bad() sites running with
mmap_sem in read mode are somewhat speculative with the page faults, and
the result is always undefined when they run simultaneously. This is
probably why it wasn't common to run into this. For example if the
madvise(MADV_DONTNEED) runs zap_page_range() shortly before the page
fault, the hugepage will not be zapped, if the page fault runs first it
will be zapped.
Altering pmd_bad() not to error out if it finds hugepmds won't be enough
to fix this, because zap_pmd_range would then proceed to call
zap_pte_range (which would be incorrect if the pmd become a
pmd_trans_huge()).
The simplest way to fix this is to read the pmd in the local stack
(regardless of what we read, no need of actual CPU barriers, only
compiler barrier needed), and be sure it is not changing under the code
that computes its value. Even if the real pmd is changing under the
value we hold on the stack, we don't care. If we actually end up in
zap_pte_range it means the pmd was not none already and it was not huge,
and it can't become huge from under us (khugepaged locking explained
above).
All we need is to enforce that there is no way anymore that in a code
path like below, pmd_trans_huge can be false, but pmd_none_or_clear_bad
can run into a hugepmd. The overhead of a barrier() is just a compiler
tweak and should not be measurable (I only added it for THP builds). I
don't exclude different compiler versions may have prevented the race
too by caching the value of *pmd on the stack (that hasn't been
verified, but it wouldn't be impossible considering
pmd_none_or_clear_bad, pmd_bad, pmd_trans_huge, pmd_none are all inlines
and there's no external function called in between pmd_trans_huge and
pmd_none_or_clear_bad).
if (pmd_trans_huge(*pmd)) {
if (next-addr != HPAGE_PMD_SIZE) {
VM_BUG_ON(!rwsem_is_locked(&tlb->mm->mmap_sem));
split_huge_page_pmd(vma->vm_mm, pmd);
} else if (zap_huge_pmd(tlb, vma, pmd, addr))
continue;
/* fall through */
}
if (pmd_none_or_clear_bad(pmd))
Because this race condition could be exercised without special
privileges this was reported in CVE-2012-1179.
The race was identified and fully explained by Ulrich who debugged it.
I'm quoting his accurate explanation below, for reference.
====== start quote =======
mapcount 0 page_mapcount 1
kernel BUG at mm/huge_memory.c:1384!
At some point prior to the panic, a "bad pmd ..." message similar to the
following is logged on the console:
mm/memory.c:145: bad pmd ffff8800376e1f98(80000000314000e7).
The "bad pmd ..." message is logged by pmd_clear_bad() before it clears
the page's PMD table entry.
143 void pmd_clear_bad(pmd_t *pmd)
144 {
-> 145 pmd_ERROR(*pmd);
146 pmd_clear(pmd);
147 }
After the PMD table entry has been cleared, there is an inconsistency
between the actual number of PMD table entries that are mapping the page
and the page's map count (_mapcount field in struct page). When the page
is subsequently reclaimed, __split_huge_page() detects this inconsistency.
1381 if (mapcount != page_mapcount(page))
1382 printk(KERN_ERR "mapcount %d page_mapcount %d\n",
1383 mapcount, page_mapcount(page));
-> 1384 BUG_ON(mapcount != page_mapcount(page));
The root cause of the problem is a race of two threads in a multithreaded
process. Thread B incurs a page fault on a virtual address that has never
been accessed (PMD entry is zero) while Thread A is executing an madvise()
system call on a virtual address within the same 2 MB (huge page) range.
virtual address space
.---------------------.
| |
| |
.-|---------------------|
| | |
| | |<-- B(fault)
| | |
2 MB | |/////////////////////|-.
huge < |/////////////////////| > A(range)
page | |/////////////////////|-'
| | |
| | |
'-|---------------------|
| |
| |
'---------------------'
- Thread A is executing an madvise(..., MADV_DONTNEED) system call
on the virtual address range "A(range)" shown in the picture.
sys_madvise
// Acquire the semaphore in shared mode.
down_read(¤t->mm->mmap_sem)
...
madvise_vma
switch (behavior)
case MADV_DONTNEED:
madvise_dontneed
zap_page_range
unmap_vmas
unmap_page_range
zap_pud_range
zap_pmd_range
//
// Assume that this huge page has never been accessed.
// I.e. content of the PMD entry is zero (not mapped).
//
if (pmd_trans_huge(*pmd)) {
// We don't get here due to the above assumption.
}
//
// Assume that Thread B incurred a page fault and
.---------> // sneaks in here as shown below.
| //
| if (pmd_none_or_clear_bad(pmd))
| {
| if (unlikely(pmd_bad(*pmd)))
| pmd_clear_bad
| {
| pmd_ERROR
| // Log "bad pmd ..." message here.
| pmd_clear
| // Clear the page's PMD entry.
| // Thread B incremented the map count
| // in page_add_new_anon_rmap(), but
| // now the page is no longer mapped
| // by a PMD entry (-> inconsistency).
| }
| }
|
v
- Thread B is handling a page fault on virtual address "B(fault)" shown
in the picture.
...
do_page_fault
__do_page_fault
// Acquire the semaphore in shared mode.
down_read_trylock(&mm->mmap_sem)
...
handle_mm_fault
if (pmd_none(*pmd) && transparent_hugepage_enabled(vma))
// We get here due to the above assumption (PMD entry is zero).
do_huge_pmd_anonymous_page
alloc_hugepage_vma
// Allocate a new transparent huge page here.
...
__do_huge_pmd_anonymous_page
...
spin_lock(&mm->page_table_lock)
...
page_add_new_anon_rmap
// Here we increment the page's map count (starts at -1).
atomic_set(&page->_mapcount, 0)
set_pmd_at
// Here we set the page's PMD entry which will be cleared
// when Thread A calls pmd_clear_bad().
...
spin_unlock(&mm->page_table_lock)
The mmap_sem does not prevent the race because both threads are acquiring
it in shared mode (down_read). Thread B holds the page_table_lock while
the page's map count and PMD table entry are updated. However, Thread A
does not synchronize on that lock.
====== end quote =======
[akpm@linux-foundation.org: checkpatch fixes]
Reported-by: Ulrich Obergfell <uobergfe@redhat.com>
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Hugh Dickins <hughd@google.com>
Cc: Dave Jones <davej@redhat.com>
Acked-by: Larry Woodman <lwoodman@redhat.com>
Acked-by: Rik van Riel <riel@redhat.com>
Cc: Mark Salter <msalter@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 89e984e2c2 upstream.
An iser target may send iscsi NO-OP PDUs as soon as it marks the iSER
iSCSI session as fully operative. This means that there is window
where there are no posted receive buffers on the initiator side, so
it's possible for the iSER RC connection to break because of RNR NAK /
retry errors. To fix this, rely on the flags bits in the login
request to have FFP (0x3) in the lower nibble as a marker for the
final login request, and post an initial chunk of receive buffers
before sending that login request instead of after getting the login
response.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 41c7f74242 upstream.
Currently, the RTC code does not disable the alarm in the hardware.
This means that after a sequence such as the one below (the files are in the
RTC sysfs), the box will boot up after 2 minutes even though we've
asked for the alarm to be turned off.
# echo $((`cat since_epoch`)+120) > wakealarm
# echo 0 > wakealarm
# poweroff
Fix this by disabling the alarm when there are no timers to run.
The original version of this patch was reverted. This version
disables the irq directly instead of setting a disabled timer
in the future.
Cc: John Stultz <john.stultz@linaro.org>
Signed-off-by: Rabin Vincent <rabin.vincent@stericsson.com>
[Merged in the second revision from Rabin]
Signed-off-by: John Stultz <john.stultz@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit a09b659cd6 upstream.
In 2008, commit 0c5d1eb77a ("genirq: record trigger type") modified the
way set_irq_type() handles the 'no trigger' condition. However, this has
an adverse effect on PCMCIA support on Intel StrongARM and probably PXA
platforms.
PCMCIA has several status signals on the socket which can trigger
interrupts; some of these status signals depend on the card's mode
(whether it is configured in memory or IO mode). For example, cards have
a 'Ready/IRQ' signal: in memory mode, this provides an indication to
PCMCIA that the card has finished its power up initialization. In IO
mode, it provides the device interrupt signal. Other status signals
switch between on-board battery status and loud speaker output.
In classical PCMCIA implementations, where you have a specific socket
controller, the controller provides a method to mask interrupts from the
socket, and importantly ignore any state transitions on the pins which
correspond with interrupts once masked. This masking prevents unwanted
events caused by the removal and application of socket power being
forwarded.
However, on platforms where there is no socket controller, the PCMCIA
status and interrupt signals are routed to standard edge-triggered GPIOs.
These GPIOs can be configured to interrupt on rising edge, falling edge,
or never. This is where the problems start.
Edge triggered interrupts are required to record events while disabled via
the usual methods of {free,request,disable,enable}_irq() to prevent
problems with dropped interrupts (eg, the 8390 driver uses disable_irq()
to defer the delivery of interrupts). As a result, these interfaces can
not be used to implement the desired behaviour.
The side effect of this is that if the 'Ready/IRQ' GPIO is disabled via
disable_irq() on suspend, and enabled via enable_irq() after resume, we
will record the state transitions caused by powering events as valid
interrupts, and foward them to the card driver, which may attempt to
access a card which is not powered up.
This leads delays resume while drivers spin in their interrupt handlers,
and complaints from drivers before they realize what's happened.
Moreover, in the case of the 'Ready/IRQ' signal, this is requested and
freed by the card driver itself; the PCMCIA core has no idea whether the
interrupt is requested, and, therefore, whether a call to disable_irq()
would be valid. (We tried this around 2.4.17 / 2.5.1 kernel era, and
ended up throwing it out because of this problem.)
Therefore, it was decided back in around 2002 to disable the edge
triggering instead, resulting in all state transitions on the GPIO being
ignored. That's what we actually need the hardware to do.
The commit above changes this behaviour; it explicitly prevents the 'no
trigger' state being selected.
The reason that request_irq() does not accept the 'no trigger' state is
for compatibility with existing drivers which do not provide their desired
triggering configuration. The set_irq_type() function is 'new' and not
used by non-trigger aware drivers.
Therefore, revert this change, and restore previously working platforms
back to their former state.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Cc: linux@arm.linux.org.uk
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 7b60a18da3 upstream.
The queue handling in the udev daemon assumes that the events are
ordered.
Before this patch uevent_seqnum is incremented under sequence_lock,
than an event is send uner uevent_sock_mutex. I want to say that code
contained a window between incrementing seqnum and sending an event.
This patch locks uevent_sock_mutex before incrementing uevent_seqnum.
v2: delete sequence_lock, uevent_seqnum is protected by uevent_sock_mutex
v3: unlock the mutex before the goto exit
Thanks for Kay for the comments.
Signed-off-by: Andrew Vagin <avagin@openvz.org>
Tested-By: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit a9b89e2567 upstream.
Driver rtl8192ce when used with the RTL8188CE device would start at about
20 Mbps on a 54 Mbps connection, but quickly drop to 1 Mbps. One of the
symptoms is that the AP would need to retransmit each packet 4 of 5 times
before the driver would acknowledge it. Recovery is possible only by
unloading and reloading the driver. This problem was reported at
https://bugzilla.redhat.com/show_bug.cgi?id=770207.
The problem is due to a missing update of the gain setting.
Signed-off-by: Jingjun Wu <jingjun_wu@realsil.com.cn>
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 093ea2d3a7 upstream.
A MCS7820 device supports two serial ports and a MCS7840 device supports
four serial ports. Both devices use the same driver, but the attach function
in driver was unable to correctly handle the port numbers for MCS7820
device. This problem has been fixed in this patch and this fix has been
verified on x86 Linux kernel 3.2.9 with both MCS7820 and MCS7840 devices.
Signed-off-by: Donald Lee <donald@asix.com.tw>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit a5360a53a7 upstream.
This patch updates the cp210x driver to support CP210x multiple
interface devices devices from Silicon Labs. The existing driver
always sends control requests to interface 0, which is hardcoded in
the usb_control_msg function calls. This only allows for single
interface devices to be used, and causes a bug when using ports on an
interface other than 0 in the multiple interface devices.
Here are the changes included in this patch:
- Updated the device list to contain the Silicon Labs factory default
VID/PID for multiple interface CP210x devices
- Created a cp210x_port_private struct created for each port on
startup, this struct holds the interface number
- Added a cp210x_release function to clean up the cp210x_port_private
memory created on startup
- Modified usb_get_config and usb_set_config to get a pointer to the
cp210x_port_private struct, and use the interface number there in the
usb_control_message wIndex param
Signed-off-by: Preston Fick <preston.fick@silabs.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 6d161b99f8 upstream.
This patch adds new device IDs to the ftdi_sio module to support
the new Sealevel SeaLINK+8 2038-ROHS device.
Signed-off-by: Scott Dial <scott.dial@scientiallc.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit c192c8e71a upstream.
Gobi 1000 devices have a different port layout, which wasn't respected
by the current driver, and thus it grabbed the QMI/net port. In the
near future we'll be attaching another driver to the QMI/net port for
these devices (cdc-wdm and qmi_wwan) so make sure the qcserial driver
doesn't claim them. This patch also prevents qcserial from binding to
interfaces 0 and 1 on 1K devices because those interfaces do not
respond.
Signed-off-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit e90fc3cb08 upstream.
When build i.mx platform with imx_v6_v7_defconfig, and after adding
USB Gadget support, it has below build error:
CC drivers/usb/host/fsl-mph-dr-of.o
drivers/usb/host/fsl-mph-dr-of.c: In function 'fsl_usb2_device_register':
drivers/usb/host/fsl-mph-dr-of.c:97: error: 'struct pdev_archdata'
has no member named 'dma_mask'
It has discussed at: http://www.spinics.net/lists/linux-usb/msg57302.html
For PowerPC, there is dma_mask at struct pdev_archdata, but there is
no dma_mask at struct pdev_archdata for ARM. The pdev_archdata is
related to specific platform, it should NOT be accessed by
cross platform drivers, like USB.
The code for pdev_archdata should be useless, as for PowerPC,
it has already gotten the value for pdev->dev.dma_mask at function
arch_setup_pdev_archdata of arch/powerpc/kernel/setup-common.c.
Tested-by: Ramneek Mehresh <ramneek.mehresh@freescale.com>
Signed-off-by: Peter Chen <peter.chen@freescale.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit c5cc5ed866 upstream.
When loading g_ether gadget, there is below message:
Backtrace:
[<80012248>] (dump_backtrace+0x0/0x10c) from [<803cb42c>] (dump_stack+0x18/0x1c)
r7:00000000 r6:80512000 r5:8052bef8 r4:80513f30
[<803cb414>] (dump_stack+0x0/0x1c) from [<8000feb4>] (show_regs+0x44/0x50)
[<8000fe70>] (show_regs+0x0/0x50) from [<8004c840>] (__schedule_bug+0x68/0x84)
r5:8052bef8 r4:80513f30
[<8004c7d8>] (__schedule_bug+0x0/0x84) from [<803cd0e4>] (__schedule+0x4b0/0x528)
r5:8052bef8 r4:809aad00
[<803ccc34>] (__schedule+0x0/0x528) from [<803cd214>] (_cond_resched+0x44/0x58)
[<803cd1d0>] (_cond_resched+0x0/0x58) from [<800a9488>] (dma_pool_alloc+0x184/0x250)
r5:9f9b4000 r4:9fb4fb80
[<800a9304>] (dma_pool_alloc+0x0/0x250) from [<802a8ad8>] (fsl_req_to_dtd+0xac/0x180)
[<802a8a2c>] (fsl_req_to_dtd+0x0/0x180) from [<802a8ce4>] (fsl_ep_queue+0x138/0x274)
[<802a8bac>] (fsl_ep_queue+0x0/0x274) from [<7f004328>] (composite_setup+0x2d4/0xfac [g_ether])
[<7f004054>] (composite_setup+0x0/0xfac [g_ether]) from [<802a9bb4>] (fsl_udc_irq+0x8dc/0xd38)
[<802a92d8>] (fsl_udc_irq+0x0/0xd38) from [<800704f8>] (handle_irq_event_percpu+0x54/0x188)
[<800704a4>] (handle_irq_event_percpu+0x0/0x188) from [<80070674>] (handle_irq_event+0x48/0x68)
[<8007062c>] (handle_irq_event+0x0/0x68) from [<800738ec>] (handle_level_irq+0xb4/0x138)
r5:80514f94 r4:80514f40
[<80073838>] (handle_level_irq+0x0/0x138) from [<8006ffa4>] (generic_handle_irq+0x38/0x44)
r7:00000012 r6:80510b1c r5:80529860 r4:80512000
[<8006ff6c>] (generic_handle_irq+0x0/0x44) from [<8000f4c4>] (handle_IRQ+0x54/0xb4)
[<8000f470>] (handle_IRQ+0x0/0xb4) from [<800085b8>] (tzic_handle_irq+0x64/0x94)
r9:412fc085 r8:00000000 r7:80513f30 r6:00000001 r5:00000000
r4:00000000
[<80008554>] (tzic_handle_irq+0x0/0x94) from [<8000e680>] (__irq_svc+0x40/0x60)
The reason of above dump message is calling dma_poll_alloc with can-schedule
mem_flags at atomic context.
To fix this problem, below changes are made:
- fsl_req_to_dtd doesn't need to be protected by spin_lock_irqsave,
as struct usb_request can be access at process context. Move lock
to beginning of hardware visit (fsl_queue_td).
- Change the memory flag which using to allocate dTD descriptor buffer,
the memory flag can be from gadget layer.
It is tested at i.mx51 bbg board with g_mass_storage, g_ether, g_serial.
Signed-off-by: Peter Chen <peter.chen@freescale.com>
Acked-by: Li Yang <leoli@freescale.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 711c68b3c0 upstream.
We must not allow the input buffer length to change while we're
shuffling the buffer contents. We also mustn't clear the WDM_READ
flag after more data might have arrived. Therefore move both of these
into the spinlocked region at the bottom of wdm_read().
When reading desc->length without holding the iuspin lock, use
ACCESS_ONCE() to ensure the compiler doesn't re-read it with
inconsistent results.
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Tested-by: Bjørn Mork <bjorn@mork.no>
Cc: Oliver Neukum <oliver@neukum.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 548dd4b6da upstream.
Do not report errors in write path if port is used as a console as this
may trigger the same error (and error report) resulting in a loop.
Reported-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: Johan Hovold <jhovold@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
