Now races can happen between interrupt handler execution and PM runtime in
error handling code path in probe and in dwc3_omap_remove() which will lead
to system crash:
in probe:
...
err1:
pm_runtime_put_sync(dev);
^^ PM runtime can race with IRQ handler when deferred probing happening
due to extcon
pm_runtime_disable(dev);
return ret;
in dwc3_omap_remove:
...
dwc3_omap_disable_irqs(omap);
^^ IRQs are disabled in HW, but handler may still run
of_platform_depopulate(omap->dev);
pm_runtime_put_sync(&pdev->dev);
^^ PM runtime can race with IRQ handler
pm_runtime_disable(&pdev->dev);
return 0;
So, OMAP DWC3 IRQ need to be disabled before calling
pm_runtime_put() in probe and in dwc3_omap_remove().
Acked-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
The dev_config() routine in gadgetfs has a check that
dev->dev->bNumConfigurations is equal to 1, and then contains a
redundant line of code setting the value to 1. This patch removes the
unnecessary assignment.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Andrey Konovalov's fuzz testing of gadgetfs showed that we should
improve the driver's checks for valid configuration descriptors passed
in by the user. In particular, the driver needs to verify that the
wTotalLength value in the descriptor is not too short (smaller
than USB_DT_CONFIG_SIZE). And the check for whether wTotalLength is
too large has to be changed, because the driver assumes there is
always enough room remaining in the buffer to hold a device descriptor
(at least USB_DT_DEVICE_SIZE bytes).
This patch adds the additional check and fixes the existing check. It
may do a little more than strictly necessary, but one extra check
won't hurt.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
CC: Andrey Konovalov <andreyknvl@google.com>
CC: <stable@vger.kernel.org>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Andrey Konovalov reports that fuzz testing with syzkaller causes a
KASAN use-after-free bug report in gadgetfs:
BUG: KASAN: use-after-free in gadgetfs_setup+0x208a/0x20e0 at addr ffff88003dfe5bf2
Read of size 2 by task syz-executor0/22994
CPU: 3 PID: 22994 Comm: syz-executor0 Not tainted 4.9.0-rc7+ #16
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
ffff88006df06a18 ffffffff81f96aba ffffffffe0528500 1ffff1000dbe0cd6
ffffed000dbe0cce ffff88006df068f0 0000000041b58ab3 ffffffff8598b4c8
ffffffff81f96828 1ffff1000dbe0ccd ffff88006df06708 ffff88006df06748
Call Trace:
<IRQ> [ 201.343209] [< inline >] __dump_stack lib/dump_stack.c:15
<IRQ> [ 201.343209] [<ffffffff81f96aba>] dump_stack+0x292/0x398 lib/dump_stack.c:51
[<ffffffff817e4dec>] kasan_object_err+0x1c/0x70 mm/kasan/report.c:159
[< inline >] print_address_description mm/kasan/report.c:197
[<ffffffff817e5080>] kasan_report_error+0x1f0/0x4e0 mm/kasan/report.c:286
[< inline >] kasan_report mm/kasan/report.c:306
[<ffffffff817e562a>] __asan_report_load_n_noabort+0x3a/0x40 mm/kasan/report.c:337
[< inline >] config_buf drivers/usb/gadget/legacy/inode.c:1298
[<ffffffff8322c8fa>] gadgetfs_setup+0x208a/0x20e0 drivers/usb/gadget/legacy/inode.c:1368
[<ffffffff830fdcd0>] dummy_timer+0x11f0/0x36d0 drivers/usb/gadget/udc/dummy_hcd.c:1858
[<ffffffff814807c1>] call_timer_fn+0x241/0x800 kernel/time/timer.c:1308
[< inline >] expire_timers kernel/time/timer.c:1348
[<ffffffff81482de6>] __run_timers+0xa06/0xec0 kernel/time/timer.c:1641
[<ffffffff814832c1>] run_timer_softirq+0x21/0x80 kernel/time/timer.c:1654
[<ffffffff84f4af8b>] __do_softirq+0x2fb/0xb63 kernel/softirq.c:284
The cause of the bug is subtle. The dev_config() routine gets called
twice by the fuzzer. The first time, the user data contains both a
full-speed configuration descriptor and a high-speed config
descriptor, causing dev->hs_config to be set. But it also contains an
invalid device descriptor, so the buffer containing the descriptors is
deallocated and dev_config() returns an error.
The second time dev_config() is called, the user data contains only a
full-speed config descriptor. But dev->hs_config still has the stale
pointer remaining from the first call, causing the routine to think
that there is a valid high-speed config. Later on, when the driver
dereferences the stale pointer to copy that descriptor, we get a
use-after-free access.
The fix is simple: Clear dev->hs_config if the passed-in data does not
contain a high-speed config descriptor.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Reported-by: Andrey Konovalov <andreyknvl@google.com>
Tested-by: Andrey Konovalov <andreyknvl@google.com>
CC: <stable@vger.kernel.org>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Andrey Konovalov reports that fuzz testing with syzkaller causes a
KASAN warning in gadgetfs:
BUG: KASAN: slab-out-of-bounds in dev_config+0x86f/0x1190 at addr ffff88003c47e160
Write of size 65537 by task syz-executor0/6356
CPU: 3 PID: 6356 Comm: syz-executor0 Not tainted 4.9.0-rc7+ #19
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
ffff88003c107ad8 ffffffff81f96aba ffffffff3dc11ef0 1ffff10007820eee
ffffed0007820ee6 ffff88003dc11f00 0000000041b58ab3 ffffffff8598b4c8
ffffffff81f96828 ffffffff813fb4a0 ffff88003b6eadc0 ffff88003c107738
Call Trace:
[< inline >] __dump_stack lib/dump_stack.c:15
[<ffffffff81f96aba>] dump_stack+0x292/0x398 lib/dump_stack.c:51
[<ffffffff817e4dec>] kasan_object_err+0x1c/0x70 mm/kasan/report.c:159
[< inline >] print_address_description mm/kasan/report.c:197
[<ffffffff817e5080>] kasan_report_error+0x1f0/0x4e0 mm/kasan/report.c:286
[<ffffffff817e5705>] kasan_report+0x35/0x40 mm/kasan/report.c:306
[< inline >] check_memory_region_inline mm/kasan/kasan.c:308
[<ffffffff817e3fb9>] check_memory_region+0x139/0x190 mm/kasan/kasan.c:315
[<ffffffff817e4044>] kasan_check_write+0x14/0x20 mm/kasan/kasan.c:326
[< inline >] copy_from_user arch/x86/include/asm/uaccess.h:689
[< inline >] ep0_write drivers/usb/gadget/legacy/inode.c:1135
[<ffffffff83228caf>] dev_config+0x86f/0x1190 drivers/usb/gadget/legacy/inode.c:1759
[<ffffffff817fdd55>] __vfs_write+0x5d5/0x760 fs/read_write.c:510
[<ffffffff817ff650>] vfs_write+0x170/0x4e0 fs/read_write.c:560
[< inline >] SYSC_write fs/read_write.c:607
[<ffffffff81803a5b>] SyS_write+0xfb/0x230 fs/read_write.c:599
[<ffffffff84f47ec1>] entry_SYSCALL_64_fastpath+0x1f/0xc2
Indeed, there is a comment saying that the value of len is restricted
to a 16-bit integer, but the code doesn't actually do this.
This patch fixes the warning. It replaces the comment with a
computation that forces the amount of data copied from the user in
ep0_write() to be no larger than the wLength size for the control
transfer, which is a 16-bit quantity.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Reported-by: Andrey Konovalov <andreyknvl@google.com>
Tested-by: Andrey Konovalov <andreyknvl@google.com>
CC: <stable@vger.kernel.org>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
In case EP disabled pass only EPCPLT command
to be handled. In other case we could hit
Bug like below.
BUG: unable to handle kernel NULL pointer dereference at 0000000000000003
IP:
[<ffffffff81673428>] dwc3_thread_interrupt+0x11c8/0x1790
while dep->endpoint.desc is NULL.
Signed-off-by: Janusz Dziedzic <januszx.dziedzic@linux.intel.com>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Andrey Konovalov reported that we were not properly checking the upper
limit before of a device configuration size before calling
memdup_user(), which could cause some problems.
So set the upper limit to PAGE_SIZE * 4, which should be good enough for
all devices.
Reported-by: Andrey Konovalov <andreyknvl@google.com>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
The stop_activity() routine in dummy-hcd is supposed to unlink all
active requests for every endpoint, among other things. But it
doesn't handle ep0. As a result, fuzz testing can generate a WARNING
like the following:
WARNING: CPU: 0 PID: 4410 at drivers/usb/gadget/udc/dummy_hcd.c:672 dummy_free_request+0x153/0x170
Modules linked in:
CPU: 0 PID: 4410 Comm: syz-executor Not tainted 4.9.0-rc7+ #32
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
ffff88006a64ed10 ffffffff81f96b8a ffffffff41b58ab3 1ffff1000d4c9d35
ffffed000d4c9d2d ffff880065f8ac00 0000000041b58ab3 ffffffff8598b510
ffffffff81f968f8 0000000041b58ab3 ffffffff859410e0 ffffffff813f0590
Call Trace:
[< inline >] __dump_stack lib/dump_stack.c:15
[<ffffffff81f96b8a>] dump_stack+0x292/0x398 lib/dump_stack.c:51
[<ffffffff812b808f>] __warn+0x19f/0x1e0 kernel/panic.c:550
[<ffffffff812b831c>] warn_slowpath_null+0x2c/0x40 kernel/panic.c:585
[<ffffffff830fcb13>] dummy_free_request+0x153/0x170 drivers/usb/gadget/udc/dummy_hcd.c:672
[<ffffffff830ed1b0>] usb_ep_free_request+0xc0/0x420 drivers/usb/gadget/udc/core.c:195
[<ffffffff83225031>] gadgetfs_unbind+0x131/0x190 drivers/usb/gadget/legacy/inode.c:1612
[<ffffffff830ebd8f>] usb_gadget_remove_driver+0x10f/0x2b0 drivers/usb/gadget/udc/core.c:1228
[<ffffffff830ec084>] usb_gadget_unregister_driver+0x154/0x240 drivers/usb/gadget/udc/core.c:1357
This patch fixes the problem by iterating over all the endpoints in
the driver's ep array instead of iterating over the gadget's ep_list,
which explicitly leaves out ep0.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Reported-by: Andrey Konovalov <andreyknvl@google.com>
CC: <stable@vger.kernel.org>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Reserved1 is documented as expected to be set to 0, but this test fails
when it it set to 0. Reverse the condition.
Signed-off-by: Vincent Pelletier <plr.vincent@gmail.com>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
When FUNCTIONFS_EVENTFD flag is set, __ffs_data_got_descs reads a 32bits,
little-endian value right after the fixed structure header, and passes it
to eventfd_ctx_fdget. Document this.
Also, rephrase a comment to be affirmative about the role of string
descriptor at index 0. Ref: USB 2.0 spec paragraph "9.6.7 String", and
also checked to still be current in USB 3.0 spec paragraph "9.6.9 String".
Signed-off-by: Vincent Pelletier <plr.vincent@gmail.com>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
By convention (according to doc) if function does not provide
get_alt() callback composite framework should assume that it has only
altsetting 0 and should respond with error if host tries to set
other one.
After commit dd4dff8b03 ("USB: composite: Fix bug: should test
set_alt function pointer before use it")
we started checking set_alt() callback instead of get_alt().
This check is useless as we check if set_alt() is set inside
usb_add_function() and fail if it's NULL.
Let's fix this check and move comment about why we check the get
method instead of set a little bit closer to prevent future false
fixes.
Fixes: dd4dff8b03 ("USB: composite: Fix bug: should test set_alt function pointer before use it")
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Krzysztof Opasiak <k.opasiak@samsung.com>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
usb_get_dr_mode() expects the device-property to be spelled
"dr_mode" not "dr-mode".
Spelling it properly fixes the following warning showing up in dmesg:
[ 8704.500545] dwc3 dwc3.2.auto: Configuration mismatch. dr_mode forced to gadget
Signed-off-by: Hans de Goede <hdegoede@redhat.com
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Now that we're handling so many transfers at a time
and for some dwc3 revisions LPM events *must* be
enabled, we can fall into a situation where too many
events fire and we start receiving Overflow events.
Let's do what XHCI does and allocate a full page for
the Event Ring, this will avoid any future issues.
Cc: <stable@vger.kernel.org> # v4.9
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
commit 0416e494ce ("usb: dwc3: ep0: correct cache
sync issue in case of ep0_bounced") introduced a bug
where we would leak DMA resources which would cause
us to starve the system of them resulting in failing
DMA transfers.
Fix the bug by making sure that we always unmap EP0
requests since those are *always* mapped.
Fixes: 0416e494ce ("usb: dwc3: ep0: correct cache
sync issue in case of ep0_bounced")
Cc: <stable@vger.kernel.org>
Tested-by: Tomasz Medrek <tomaszx.medrek@intel.com>
Reported-by: Janusz Dziedzic <januszx.dziedzic@linux.intel.com>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Let's call dwc3_ep0_prepare_one_trb() explicitly
because there are occasions where we will need more
than one TRB to handle an EP0 transfer.
A follow-up patch will fix one bug related to
multiple-TRB Data Phases when it comes to
mapping/unmapping requests for DMA.
Cc: <stable@vger.kernel.org>
Reported-by: Janusz Dziedzic <januszx.dziedzic@linux.intel.com>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
For now this is just a cleanup patch, no functional
changes. We will be using the new function to fix a
bug introduced long ago by commit 0416e494ce
("usb: dwc3: ep0: correct cache sync issue in case
of ep0_bounced") and further worsened by commit
c0bd5456a4 ("usb: dwc3: ep0: handle non maxpacket
aligned transfers > 512")
Cc: <stable@vger.kernel.org>
Reported-by: Janusz Dziedzic <januszx.dziedzic@linux.intel.com>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
The current default for gadget DMA descriptor results on bcm2835 in a
unnecessary error message:
Invalid value 1 for param gadget-dma-desc
So fix this by using hw->dma_desc_enable as default value.
Fixes: dec4b55677 ("usb: dwc2: gadget: Add descriptor DMA parameter")
Acked-by: John Youn <johnyoun@synopsys.com>
Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
The current defaults for DMA results on a non-DMA platform in a unnecessary
error message:
Invalid value 0 for param gadget-dma
So fix this by using dma_capable as default value.
Fixes: 9962b62f1b ("usb: dwc2: Deprecate g-use-dma binding")
Acked-by: John Youn <johnyoun@synopsys.com>
Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
According to the Devicetree ePAPR [1] the datatypes u8 and u16 are
not defined. So using device_property_read_u16() would result in
a partial read of a 32-bit big-endian integer which is not intended.
So we better read the complete 32-bit value. This fixes a regression
on bcm2835 where the values for g-rx-fifo-size and g-np-tx-fifo-size
always read as zero:
Invalid value 0 for param g-rx-fifo-size
Invalid value 0 for param g-np-tx-fifo-size
[1] - http://elinux.org/images/c/cf/Power_ePAPR_APPROVED_v1.1.pdf
Fixes: 05ee799f20 ("usb: dwc2: Move gadget settings into core_params")
Acked-by: John Youn <johnyoun@synopsys.com>
Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Since commit "usb: dwc2: Improve handling of host and device hwparams" the
host mode specific hardware parameter aren't initialized in peripheral mode
from the register settings anymore. So we better do not set them in this
case which avoids the following warnings on bcm2835:
256 invalid for host_nperio_tx_fifo_size. Check HW configuration.
512 invalid for host_perio_tx_fifo_size. Check HW configuration.
Fixes: 55e1040e42 ("usb: dwc2: Improve handling of host and device hwparams")
Acked-by: John Youn <johnyoun@synopsys.com>
Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Pull DAX updates from Dan Williams:
"The completion of Jan's DAX work for 4.10.
As I mentioned in the libnvdimm-for-4.10 pull request, these are some
final fixes for the DAX dirty-cacheline-tracking invalidation work
that was merged through the -mm, ext4, and xfs trees in -rc1. These
patches were prepared prior to the merge window, but we waited for
4.10-rc1 to have a stable merge base after all the prerequisites were
merged.
Quoting Jan on the overall changes in these patches:
"So I'd like all these 6 patches to go for rc2. The first three
patches fix invalidation of exceptional DAX entries (a bug which
is there for a long time) - without these patches data loss can
occur on power failure even though user called fsync(2). The other
three patches change locking of DAX faults so that ->iomap_begin()
is called in a more relaxed locking context and we are safe to
start a transaction there for ext4"
These have received a build success notification from the kbuild
robot, and pass the latest libnvdimm unit tests. There have not been
any -next releases since -rc1, so they have not appeared there"
* 'libnvdimm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm:
ext4: Simplify DAX fault path
dax: Call ->iomap_begin without entry lock during dax fault
dax: Finish fault completely when loading holes
dax: Avoid page invalidation races and unnecessary radix tree traversals
mm: Invalidate DAX radix tree entries only if appropriate
ext2: Return BH_New buffers for zeroed blocks
- A merge error on my part broke the DocBook build. I've requisitioned
one of tglx's frozen sharks for appropriate disciplinary action and
resolved to be more careful about testing the DocBook stuff as long as
it's still around.
- Fix an error in unaligned-memory-access.txt
-----BEGIN PGP SIGNATURE-----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=W4e5
-----END PGP SIGNATURE-----
Merge tag 'docs-4.10-rc1-fix' of git://git.lwn.net/linux
Pull documentation fixes from Jonathan Corbet:
"Two small fixes:
- A merge error on my part broke the DocBook build. I've
requisitioned one of tglx's frozen sharks for appropriate
disciplinary action and resolved to be more careful about testing
the DocBook stuff as long as it's still around.
- Fix an error in unaligned-memory-access.txt"
* tag 'docs-4.10-rc1-fix' of git://git.lwn.net/linux:
Documentation/unaligned-memory-access.txt: fix incorrect comparison operator
docs: Fix build failure
Pull crypto fix from Herbert Xu:
"This fixes a boot failure on some platforms when crypto self test is
enabled along with the new acomp interface"
* 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
crypto: testmgr - Use heap buffer for acomp test input
mm/filemap.c: In function 'clear_bit_unlock_is_negative_byte':
mm/filemap.c:933:9: error: too few arguments to function 'test_bit'
return test_bit(PG_waiters);
^~~~~~~~
Fixes: b91e1302ad ('mm: optimize PageWaiters bit use for unlock_page()')
Signed-off-by: Olof Johansson <olof@lixom.net>
Brown-paper-bag-by: Linus Torvalds <dummy@duh.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In commit 6290602709 ("mm: add PageWaiters indicating tasks are
waiting for a page bit") Nick Piggin made our page locking no longer
unconditionally touch the hashed page waitqueue, which not only helps
performance in general, but is particularly helpful on NUMA machines
where the hashed wait queues can bounce around a lot.
However, the "clear lock bit atomically and then test the waiters bit"
sequence turns out to be much more expensive than it needs to be,
because you get a nasty stall when trying to access the same word that
just got updated atomically.
On architectures where locking is done with LL/SC, this would be trivial
to fix with a new primitive that clears one bit and tests another
atomically, but that ends up not working on x86, where the only atomic
operations that return the result end up being cmpxchg and xadd. The
atomic bit operations return the old value of the same bit we changed,
not the value of an unrelated bit.
On x86, we could put the lock bit in the high bit of the byte, and use
"xadd" with that bit (where the overflow ends up not touching other
bits), and look at the other bits of the result. However, an even
simpler model is to just use a regular atomic "and" to clear the lock
bit, and then the sign bit in eflags will indicate the resulting state
of the unrelated bit #7.
So by moving the PageWaiters bit up to bit #7, we can atomically clear
the lock bit and test the waiters bit on x86 too. And architectures
with LL/SC (which is all the usual RISC suspects), the particular bit
doesn't matter, so they are fine with this approach too.
This avoids the extra access to the same atomic word, and thus avoids
the costly stall at page unlock time.
The only downside is that the interface ends up being a bit odd and
specialized: clear a bit in a byte, and test the sign bit. Nick doesn't
love the resulting name of the new primitive, but I'd rather make the
name be descriptive and very clear about the limitation imposed by
trying to work across all relevant architectures than make it be some
generic thing that doesn't make the odd semantics explicit.
So this introduces the new architecture primitive
clear_bit_unlock_is_negative_byte();
and adds the trivial implementation for x86. We have a generic
non-optimized fallback (that just does a "clear_bit()"+"test_bit(7)"
combination) which can be overridden by any architecture that can do
better. According to Nick, Power has the same hickup x86 has, for
example, but some other architectures may not even care.
All these optimizations mean that my page locking stress-test (which is
just executing a lot of small short-lived shell scripts: "make test" in
the git source tree) no longer makes our page locking look horribly bad.
Before all these optimizations, just the unlock_page() costs were just
over 3% of all CPU overhead on "make test". After this, it's down to
0.66%, so just a quarter of the cost it used to be.
(The difference on NUMA is bigger, but there this micro-optimization is
likely less noticeable, since the big issue on NUMA was not the accesses
to 'struct page', but the waitqueue accesses that were already removed
by Nick's earlier commit).
Acked-by: Nick Piggin <npiggin@gmail.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Bob Peterson <rpeterso@redhat.com>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Andrew Lutomirski <luto@kernel.org>
Cc: Andreas Gruenbacher <agruenba@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull crypto fix from Herbert Xu:
"This fixes a hash corruption bug in the marvell driver"
* 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
crypto: marvell - Copy IVDIG before launching partial DMA ahash requests
Pull networking fixes from David Miller:
1) Various ipvlan fixes from Eric Dumazet and Mahesh Bandewar.
The most important is to not assume the packet is RX just because
the destination address matches that of the device. Such an
assumption causes problems when an interface is put into loopback
mode.
2) If we retry when creating a new tc entry (because we dropped the
RTNL mutex in order to load a module, for example) we end up with
-EAGAIN and then loop trying to replay the request. But we didn't
reset some state when looping back to the top like this, and if
another thread meanwhile inserted the same tc entry we were trying
to, we re-link it creating an enless loop in the tc chain. Fix from
Daniel Borkmann.
3) There are two different WRITE bits in the MDIO address register for
the stmmac chip, depending upon the chip variant. Due to a bug we
could set them both, fix from Hock Leong Kweh.
4) Fix mlx4 bug in XDP_TX handling, from Tariq Toukan.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net:
net: stmmac: fix incorrect bit set in gmac4 mdio addr register
r8169: add support for RTL8168 series add-on card.
net: xdp: remove unused bfp_warn_invalid_xdp_buffer()
openvswitch: upcall: Fix vlan handling.
ipv4: Namespaceify tcp_tw_reuse knob
net: korina: Fix NAPI versus resources freeing
net, sched: fix soft lockup in tc_classify
net/mlx4_en: Fix user prio field in XDP forward
tipc: don't send FIN message from connectionless socket
ipvlan: fix multicast processing
ipvlan: fix various issues in ipvlan_process_multicast()
In the actual implementation ether_addr_equal function tests for equality to 0
when returning. It seems in commit 0d74c4 it is somehow overlooked to change
this operator to reflect the actual function.
Signed-off-by: Cihangir Akturk <cakturk@gmail.com>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
The 80211.tmpl DocBook file was removed in commit 819bf59376 ("docs-rst:
sphinxify 802.11 documentation"), but the 80211.xml target was re-added to
the Makefile by commit 7ddedebb03 ("ALSA: doc: ReSTize
writing-an-alsa-driver document"), leading to a failure when building the
documentation:
*** No rule to make target 'Documentation/DocBook/80211.xml', needed by
'Documentation/DocBook/80211.aux.xml'.
cc: stable@vger.kernel.org
Signed-off-by: John Brooks <john@fastquake.com>
Mea-culpa-by: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
Fixing the gmac4 mdio write access to use MII_GMAC4_WRITE only instead of
OR together with MII_WRITE.
Signed-off-by: Kweh, Hock Leong <hock.leong.kweh@intel.com>
Acked-By: Joao Pinto <jpinto@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This chip is the same as RTL8168, but its device id is 0x8161.
Signed-off-by: Chun-Hao Lin <hau@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
After commit 73b62bd085 ("virtio-net:
remove the warning before XDP linearizing"), there's no users for
bpf_warn_invalid_xdp_buffer(), so remove it. This is a revert for
commit f23bc46c30.
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: John Fastabend <john.r.fastabend@intel.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Networking stack accelerate vlan tag handling by
keeping topmost vlan header in skb. This works as
long as packet remains in OVS datapath. But during
OVS upcall vlan header is pushed on to the packet.
When such packet is sent back to OVS datapath, core
networking stack might not handle it correctly. Following
patch avoids this issue by accelerating the vlan tag
during flow key extract. This simplifies datapath by
bringing uniform packet processing for packets from
all code paths.
Fixes: 5108bbaddc ("openvswitch: add processing of L3 packets").
CC: Jarno Rajahalme <jarno@ovn.org>
CC: Jiri Benc <jbenc@redhat.com>
Signed-off-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Different namespaces might have different requirements to reuse
TIME-WAIT sockets for new connections. This might be required in
cases where different namespace applications are in place which
require TIME_WAIT socket connections to be reduced independently
of the host.
Signed-off-by: Haishuang Yan <yanhaishuang@cmss.chinamobile.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Christopher Covington reported a crash on aarch64 on recent Fedora
kernels:
kernel BUG at ./include/linux/scatterlist.h:140!
Internal error: Oops - BUG: 0 [#1] PREEMPT SMP
Modules linked in:
CPU: 2 PID: 752 Comm: cryptomgr_test Not tainted 4.9.0-11815-ge93b1cc #162
Hardware name: linux,dummy-virt (DT)
task: ffff80007c650080 task.stack: ffff800008910000
PC is at sg_init_one+0xa0/0xb8
LR is at sg_init_one+0x24/0xb8
...
[<ffff000008398db8>] sg_init_one+0xa0/0xb8
[<ffff000008350a44>] test_acomp+0x10c/0x438
[<ffff000008350e20>] alg_test_comp+0xb0/0x118
[<ffff00000834f28c>] alg_test+0x17c/0x2f0
[<ffff00000834c6a4>] cryptomgr_test+0x44/0x50
[<ffff0000080dac70>] kthread+0xf8/0x128
[<ffff000008082ec0>] ret_from_fork+0x10/0x50
The test vectors used for input are part of the kernel image. These
inputs are passed as a buffer to sg_init_one which eventually blows up
with BUG_ON(!virt_addr_valid(buf)). On arm64, virt_addr_valid returns
false for the kernel image since virt_to_page will not return the
correct page. Fix this by copying the input vectors to heap buffer
before setting up the scatterlist.
Reported-by: Christopher Covington <cov@codeaurora.org>
Fixes: d7db7a882d ("crypto: acomp - update testmgr with support for acomp")
Signed-off-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Now that dax_iomap_fault() calls ->iomap_begin() without entry lock, we
can use transaction starting in ext4_iomap_begin() and thus simplify
ext4_dax_fault(). It also provides us proper retries in case of ENOSPC.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Currently ->iomap_begin() handler is called with entry lock held. If the
filesystem held any locks between ->iomap_begin() and ->iomap_end()
(such as ext4 which will want to hold transaction open), this would cause
lock inversion with the iomap_apply() from standard IO path which first
calls ->iomap_begin() and only then calls ->actor() callback which grabs
entry locks for DAX (if it faults when copying from/to user provided
buffers).
Fix the problem by nesting grabbing of entry lock inside ->iomap_begin()
- ->iomap_end() pair.
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
The only case when we do not finish the page fault completely is when we
are loading hole pages into a radix tree. Avoid this special case and
finish the fault in that case as well inside the DAX fault handler. It
will allow us for easier iomap handling.
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Currently dax_iomap_rw() takes care of invalidating page tables and
evicting hole pages from the radix tree when write(2) to the file
happens. This invalidation is only necessary when there is some block
allocation resulting from write(2). Furthermore in current place the
invalidation is racy wrt page fault instantiating a hole page just after
we have invalidated it.
So perform the page invalidation inside dax_iomap_actor() where we can
do it only when really necessary and after blocks have been allocated so
nobody will be instantiating new hole pages anymore.
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Currently invalidate_inode_pages2_range() and invalidate_mapping_pages()
just delete all exceptional radix tree entries they find. For DAX this
is not desirable as we track cache dirtiness in these entries and when
they are evicted, we may not flush caches although it is necessary. This
can for example manifest when we write to the same block both via mmap
and via write(2) (to different offsets) and fsync(2) then does not
properly flush CPU caches when modification via write(2) was the last
one.
Create appropriate DAX functions to handle invalidation of DAX entries
for invalidate_inode_pages2_range() and invalidate_mapping_pages() and
wire them up into the corresponding mm functions.
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
So far we did not return BH_New buffers from ext2_get_blocks() when we
allocated and zeroed-out a block for DAX inode to avoid racy zeroing in
DAX code. This zeroing is gone these days so we can remove the
workaround.
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
If mce_device_init() fails then the mce device pointer is NULL and the
AMD mce code happily dereferences it.
Add a sanity check.
Reported-by: Markus Trippelsdorf <markus@trippelsdorf.de>
Reported-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The attempt to prevent overwriting an active state resulted in a
disaster which effectively disables all dynamically allocated hotplug
states.
Cleanup the mess.
Fixes: dc280d9362 ("cpu/hotplug: Prevent overwriting of callbacks")
Reported-by: Markus Trippelsdorf <markus@trippelsdorf.de>
Reported-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit beb0babfb7 ("korina: disable napi on close and restart")
introduced calls to napi_disable() that were missing before,
unfortunately this leaves a small window during which NAPI has a chance
to run, yet we just freed resources since korina_free_ring() has been
called:
Fix this by disabling NAPI first then freeing resource, and make sure
that we also cancel the restart task before doing the resource freeing.
Fixes: beb0babfb7 ("korina: disable napi on close and restart")
Reported-by: Alexandros C. Couloumbis <alex@ozo.com>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Shahar reported a soft lockup in tc_classify(), where we run into an
endless loop when walking the classifier chain due to tp->next == tp
which is a state we should never run into. The issue only seems to
trigger under load in the tc control path.
What happens is that in tc_ctl_tfilter(), thread A allocates a new
tp, initializes it, sets tp_created to 1, and calls into tp->ops->change()
with it. In that classifier callback we had to unlock/lock the rtnl
mutex and returned with -EAGAIN. One reason why we need to drop there
is, for example, that we need to request an action module to be loaded.
This happens via tcf_exts_validate() -> tcf_action_init/_1() meaning
after we loaded and found the requested action, we need to redo the
whole request so we don't race against others. While we had to unlock
rtnl in that time, thread B's request was processed next on that CPU.
Thread B added a new tp instance successfully to the classifier chain.
When thread A returned grabbing the rtnl mutex again, propagating -EAGAIN
and destroying its tp instance which never got linked, we goto replay
and redo A's request.
This time when walking the classifier chain in tc_ctl_tfilter() for
checking for existing tp instances we had a priority match and found
the tp instance that was created and linked by thread B. Now calling
again into tp->ops->change() with that tp was successful and returned
without error.
tp_created was never cleared in the second round, thus kernel thinks
that we need to link it into the classifier chain (once again). tp and
*back point to the same object due to the match we had earlier on. Thus
for thread B's already public tp, we reset tp->next to tp itself and
link it into the chain, which eventually causes the mentioned endless
loop in tc_classify() once a packet hits the data path.
Fix is to clear tp_created at the beginning of each request, also when
we replay it. On the paths that can cause -EAGAIN we already destroy
the original tp instance we had and on replay we really need to start
from scratch. It seems that this issue was first introduced in commit
12186be7d2 ("net_cls: fix unconfigured struct tcf_proto keeps chaining
and avoid kernel panic when we use cls_cgroup").
Fixes: 12186be7d2 ("net_cls: fix unconfigured struct tcf_proto keeps chaining and avoid kernel panic when we use cls_cgroup")
Reported-by: Shahar Klein <shahark@mellanox.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Cc: Cong Wang <xiyou.wangcong@gmail.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Tested-by: Shahar Klein <shahark@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
