Commit 9c225f2655e3 ("vfs: atomic f_pos accesses as per POSIX") added
locking for file.f_pos access and in particular made concurrent read and
write not possible - now both those functions take f_pos lock for the
whole run, and so if e.g. a read is blocked waiting for data, write will
deadlock waiting for that read to complete.
This caused regression for stream-like files where previously read and
write could run simultaneously, but after that patch could not do so
anymore. See e.g. commit 581d21a2d02a ("xenbus: fix deadlock on writes
to /proc/xen/xenbus") which fixes such regression for particular case of
/proc/xen/xenbus.
The patch that added f_pos lock in 2014 did so to guarantee POSIX thread
safety for read/write/lseek and added the locking to file descriptors of
all regular files. In 2014 that thread-safety problem was not new as it
was already discussed earlier in 2006.
However even though 2006'th version of Linus's patch was adding f_pos
locking "only for files that are marked seekable with FMODE_LSEEK (thus
avoiding the stream-like objects like pipes and sockets)", the 2014
version - the one that actually made it into the tree as 9c225f2655e3 -
is doing so irregardless of whether a file is seekable or not.
The reason that it did so is, probably, that there are many files that
are marked non-seekable, but e.g. their read implementation actually
depends on knowing current position to correctly handle the read. Some
examples:
Despite that, many nonseekable_open users implement read and write with
pure stream semantics - they don't depend on passed ppos at all. And for
those cases where read could wait for something inside, it creates a
situation similar to xenbus - the write could be never made to go until
read is done, and read is waiting for some, potentially external, event,
for potentially unbounded time -> deadlock.
Besides xenbus, there are 14 such places in the kernel that I've found
with semantic patch (see below):
drivers/xen/evtchn.c:667:8-24: ERROR: evtchn_fops: .read() can deadlock .write()
drivers/isdn/capi/capi.c:963:8-24: ERROR: capi_fops: .read() can deadlock .write()
drivers/input/evdev.c:527:1-17: ERROR: evdev_fops: .read() can deadlock .write()
drivers/char/pcmcia/cm4000_cs.c:1685:7-23: ERROR: cm4000_fops: .read() can deadlock .write()
net/rfkill/core.c:1146:8-24: ERROR: rfkill_fops: .read() can deadlock .write()
drivers/s390/char/fs3270.c:488:1-17: ERROR: fs3270_fops: .read() can deadlock .write()
drivers/usb/misc/ldusb.c:310:1-17: ERROR: ld_usb_fops: .read() can deadlock .write()
drivers/hid/uhid.c:635:1-17: ERROR: uhid_fops: .read() can deadlock .write()
net/batman-adv/icmp_socket.c:80:1-17: ERROR: batadv_fops: .read() can deadlock .write()
drivers/media/rc/lirc_dev.c:198:1-17: ERROR: lirc_fops: .read() can deadlock .write()
drivers/leds/uleds.c:77:1-17: ERROR: uleds_fops: .read() can deadlock .write()
drivers/input/misc/uinput.c:400:1-17: ERROR: uinput_fops: .read() can deadlock .write()
drivers/infiniband/core/user_mad.c:985:7-23: ERROR: umad_fops: .read() can deadlock .write()
drivers/gnss/core.c:45:1-17: ERROR: gnss_fops: .read() can deadlock .write()
In addition to the cases above another regression caused by f_pos
locking is that now FUSE filesystems that implement open with
FOPEN_NONSEEKABLE flag, can no longer implement bidirectional
stream-like files - for the same reason as above e.g. read can deadlock
write locking on file.f_pos in the kernel.
FUSE's FOPEN_NONSEEKABLE was added in 2008 in a7c1b990f715 ("fuse:
implement nonseekable open") to support OSSPD. OSSPD implements /dev/dsp
in userspace with FOPEN_NONSEEKABLE flag, with corresponding read and
write routines not depending on current position at all, and with both
read and write being potentially blocking operations:
Corresponding libfuse example/test also describes FOPEN_NONSEEKABLE as
"somewhat pipe-like files ..." with read handler not using offset.
However that test implements only read without write and cannot exercise
the deadlock scenario:
I've actually hit the read vs write deadlock for real while implementing
my FUSE filesystem where there is /head/watch file, for which open
creates separate bidirectional socket-like stream in between filesystem
and its user with both read and write being later performed
simultaneously. And there it is semantically not easy to split the
stream into two separate read-only and write-only channels:
1. We can't change nonseekable_open to include &~FMODE_ATOMIC_POS -
doing so would break many in-kernel nonseekable_open users which
actually use ppos in read/write handlers.
2. Add stream_open() to kernel to open stream-like non-seekable file
descriptors. Read and write on such file descriptors would never use
nor change ppos. And with that property on stream-like files read and
write will be running without taking f_pos lock - i.e. read and write
could be running simultaneously.
3. With semantic patch search and convert to stream_open all in-kernel
nonseekable_open users for which read and write actually do not
depend on ppos and where there is no other methods in file_operations
which assume @offset access.
4. Add FOPEN_STREAM to fs/fuse/ and open in-kernel file-descriptors via
steam_open if that bit is present in filesystem open reply.
It was tempting to change fs/fuse/ open handler to use stream_open
instead of nonseekable_open on just FOPEN_NONSEEKABLE flags, but
grepping through Debian codesearch shows users of FOPEN_NONSEEKABLE,
and in particular GVFS which actually uses offset in its read and
write handlers
so if we would do such a change it will break a real user.
5. Add stream_open and FOPEN_STREAM handling to stable kernels starting
from v3.14+ (the kernel where 9c225f2655 first appeared).
This will allow to patch OSSPD and other FUSE filesystems that
provide stream-like files to return FOPEN_STREAM | FOPEN_NONSEEKABLE
in their open handler and this way avoid the deadlock on all kernel
versions. This should work because fs/fuse/ ignores unknown open
flags returned from a filesystem and so passing FOPEN_STREAM to a
kernel that is not aware of this flag cannot hurt. In turn the kernel
that is not aware of FOPEN_STREAM will be < v3.14 where just
FOPEN_NONSEEKABLE is sufficient to implement streams without read vs
write deadlock.
This patch adds stream_open, converts /proc/xen/xenbus to it and adds
semantic patch to automatically locate in-kernel places that are either
required to be converted due to read vs write deadlock, or that are just
safe to be converted because read and write do not use ppos and there
are no other funky methods in file_operations.
Regarding semantic patch I've verified each generated change manually -
that it is correct to convert - and each other nonseekable_open instance
left - that it is either not correct to convert there, or that it is not
converted due to current stream_open.cocci limitations.
The script also does not convert files that should be valid to convert,
but that currently have .llseek = noop_llseek or generic_file_llseek for
unknown reason despite file being opened with nonseekable_open (e.g.
drivers/input/mousedev.c)
If xace hardware reports a bad version number, the error handling code
in ace_setup() calls put_disk(), followed by queue cleanup. However, since
the disk data structure has the queue pointer set, put_disk() also
cleans and releases the queue. This results in blk_cleanup_queue()
accessing an already released data structure, which in turn may result
in a crash such as the following.
Fix the problem by setting the disk queue pointer to NULL before calling
put_disk(). A more comprehensive fix might be to rearrange the code
to check the hardware version before initializing data structures,
but I don't know if this would have undesirable side effects, and
it would increase the complexity of backporting the fix to older kernels.
Fixes: 74489a91dd43a ("Add support for Xilinx SystemACE CompactFlash interface") Acked-by: Michal Simek <michal.simek@xilinx.com> Signed-off-by: Guenter Roeck <linux@roeck-us.net> Signed-off-by: Jens Axboe <axboe@kernel.dk> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
At module load, if the selected home_node value is greater than
the available numa nodes, the system will crash in
__alloc_pages_nodemask() due to a bad paging request. Prevent this
user error crash by detecting the bad value, logging an error, and
setting g_home_node back to the default of NUMA_NO_NODE.
Many of the sh CPU-types have their own plat_irq_setup() and
arch_init_clk_ops() functions, so these same (empty) functions in
arch/sh/boards/of-generic.c are not needed and cause build errors.
If there is some case where these empty functions are needed, they can
be retained by marking them as "__weak" while at the same time making
builds that do not need them succeed.
Fixes these build errors:
arch/sh/boards/of-generic.o: In function `plat_irq_setup':
(.init.text+0x134): multiple definition of `plat_irq_setup'
arch/sh/kernel/cpu/sh2/setup-sh7619.o:(.init.text+0x30): first defined here
arch/sh/boards/of-generic.o: In function `arch_init_clk_ops':
(.init.text+0x118): multiple definition of `arch_init_clk_ops'
arch/sh/kernel/cpu/sh2/clock-sh7619.o:(.init.text+0x0): first defined here
Link: http://lkml.kernel.org/r/9ee4e0c5-f100-86a2-bd4d-1d3287ceab31@infradead.org Signed-off-by: Randy Dunlap <rdunlap@infradead.org> Reported-by: kbuild test robot <lkp@intel.com> Cc: Takashi Iwai <tiwai@suse.de> Cc: Yoshinori Sato <ysato@users.sourceforge.jp> Cc: Rich Felker <dalias@libc.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
When mknod is used to create a block special file in hugetlbfs, it will
allocate an inode and kmalloc a 'struct resv_map' via resv_map_alloc().
inode->i_mapping->private_data will point the newly allocated resv_map.
However, when the device special file is opened bd_acquire() will set
inode->i_mapping to bd_inode->i_mapping. Thus the pointer to the
allocated resv_map is lost and the structure is leaked.
Programs to reproduce:
mount -t hugetlbfs nodev hugetlbfs
mknod hugetlbfs/dev b 0 0
exec 30<> hugetlbfs/dev
umount hugetlbfs/
resv_map structures are only needed for inodes which can have associated
page allocations. To fix the leak, only allocate resv_map for those
inodes which could possibly be associated with page allocations.
Commit 2d4f567103ff ("KVM: PPC: Introduce kvm_tmp framework") adds
kvm_tmp[] into the .bss section and then free the rest of unused spaces
back to the page allocator.
With DEBUG_PAGEALLOC=y, it will unmap those pages from kernel. As the
result, kmemleak scan will trigger a panic when it scans the .bss
section with unmapped pages.
This patch creates dedicated kmemleak objects for the .data, .bss and
potentially .data..ro_after_init sections to allow partial freeing via
the kmemleak_free_part() in the powerpc kvm_free_tmp() function.
Link: http://lkml.kernel.org/r/20190321171917.62049-1-catalin.marinas@arm.com Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> Reported-by: Qian Cai <cai@lca.pw> Acked-by: Michael Ellerman <mpe@ellerman.id.au> (powerpc) Tested-by: Qian Cai <cai@lca.pw> Cc: Paul Mackerras <paulus@samba.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Avi Kivity <avi@redhat.com> Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Radim Krcmar <rkrcmar@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
The tx ring buffer map when xmit and unmap when xmit done. So in
hnae_init_ring() did not map tx ring buffer, but in hnae_fini_ring()
have a unmap operation for tx ring buffer, which is already unmapped
when xmit done, than cause this WARNING.
The hnae_alloc_buffers() is called in hnae_init_ring(),
so the hnae_free_buffers() should be in hnae_fini_ring(), not in
hnae_free_desc().
In hnae_fini_ring(), adds a check is_rx_ring() as in hnae_init_ring().
When the ring buffer is tx ring, adds a piece of code to ensure that
the tx ring is unmap.
Signed-off-by: Yonglong Liu <liuyonglong@huawei.com> Signed-off-by: Peng Li <lipeng321@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
ICMP6 neighbor solicitation messages will be discard by the Hip06
chips, because of not setting forwarding pool. Enable promisc mode
has the same problem.
This patch fix the wrong forwarding table configs for the multicast
vague matching when enable promisc mode, and add forwarding pool
for the forwarding table.
Signed-off-by: Yonglong Liu <liuyonglong@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
Every time the call trace is not the same, but the overwrite address
is always the same:
Unable to handle kernel paging request at virtual address 0000000200000040
The root cause is, when write the reg XGMAC_MAC_TX_LF_RF_CONTROL_REG,
didn't use the io_base offset.
Signed-off-by: Yonglong Liu <liuyonglong@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
This patch is trying to fix the issue due to:
[27237.844750] BUG: KASAN: use-after-free in hns_nic_net_xmit_hw+0x708/0xa18[hns_enet_drv]
After hnae_queue_xmit() in hns_nic_net_xmit_hw(), can be
interrupted by interruptions, and than call hns_nic_tx_poll_one()
to handle the new packets, and free the skb. So, when turn back to
hns_nic_net_xmit_hw(), calling skb->len will cause use-after-free.
This patch update tx ring statistics in hns_nic_tx_poll_one() to
fix the bug.
Signed-off-by: Liubin Shu <shuliubin@huawei.com> Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com> Signed-off-by: Yonglong Liu <liuyonglong@huawei.com> Signed-off-by: Peng Li <lipeng321@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
When doing unwind_frame() in the context of pseudo nmi (need enable
CONFIG_ARM64_PSEUDO_NMI), reaching the bottom of the stack (fp == 0,
pc != 0), function on_sdei_stack() will return true while the sdei acpi
table is not inited in fact. This will cause a "NULL pointer dereference"
oops when going on.
When the number of sub-channels offered by Hyper-V is >= the number of CPUs
in the VM, calculate the correct number of sub-channels. The current code
produces one too many.
This scenario arises only when the number of CPUs is artificially
restricted (for example, with maxcpus=<n> on the kernel boot line), because
Hyper-V normally offers a sub-channel count < number of CPUs. While the
current code doesn't break, the extra sub-channel is unbalanced across the
CPUs (for example, a total of 5 channels on a VM with 4 CPUs).
Signed-off-by: Michael Kelley <mikelley@microsoft.com> Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com> Reviewed-by: Long Li <longli@microsoft.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
When compiling with -Wformat, clang emits the following warnings:
drivers/vfio/pci/vfio_pci.c:1601:5: warning: format specifies type
'unsigned short' but the argument has type 'unsigned int' [-Wformat]
vendor, device, subvendor, subdevice,
^~~~~~
drivers/vfio/pci/vfio_pci.c:1601:13: warning: format specifies type
'unsigned short' but the argument has type 'unsigned int' [-Wformat]
vendor, device, subvendor, subdevice,
^~~~~~
drivers/vfio/pci/vfio_pci.c:1601:21: warning: format specifies type
'unsigned short' but the argument has type 'unsigned int' [-Wformat]
vendor, device, subvendor, subdevice,
^~~~~~~~~
drivers/vfio/pci/vfio_pci.c:1601:32: warning: format specifies type
'unsigned short' but the argument has type 'unsigned int' [-Wformat]
vendor, device, subvendor, subdevice,
^~~~~~~~~
drivers/vfio/pci/vfio_pci.c:1605:5: warning: format specifies type
'unsigned short' but the argument has type 'unsigned int' [-Wformat]
vendor, device, subvendor, subdevice,
^~~~~~
drivers/vfio/pci/vfio_pci.c:1605:13: warning: format specifies type
'unsigned short' but the argument has type 'unsigned int' [-Wformat]
vendor, device, subvendor, subdevice,
^~~~~~
drivers/vfio/pci/vfio_pci.c:1605:21: warning: format specifies type
'unsigned short' but the argument has type 'unsigned int' [-Wformat]
vendor, device, subvendor, subdevice,
^~~~~~~~~
drivers/vfio/pci/vfio_pci.c:1605:32: warning: format specifies type
'unsigned short' but the argument has type 'unsigned int' [-Wformat]
vendor, device, subvendor, subdevice,
^~~~~~~~~
The types of these arguments are unconditionally defined, so this patch
updates the format character to the correct ones for unsigned ints.
Link: https://github.com/ClangBuiltLinux/linux/issues/378 Signed-off-by: Louis Taylor <louis@kragniz.eu> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com> Signed-off-by: Alex Williamson <alex.williamson@redhat.com> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
Since commit 6e2bd956936 ("i2c: omap: Use noirq system sleep pm ops to idle device for suspend")
on gta04 we have handle_twl4030_pih() called in situations where pm_runtime_get()
in i2c-omap.c returns -EACCES.
This happens when we wakeup via something behing twl4030 (powerbutton or rtc
alarm). This goes on for minutes until the system is finally resumed.
Disable the irq on suspend and enable it on resume to avoid
having i2c access problems when the irq registers are checked.
Fixes: 6e2bd956936 ("i2c: omap: Use noirq system sleep pm ops to idle device for suspend") Signed-off-by: Andreas Kemnade <andreas@kemnade.info> Tested-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Lee Jones <lee.jones@linaro.org> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
symlink body shouldn't be freed without an RCU delay. Switch debugfs to
->destroy_inode() and use of call_rcu(); free both the inode and symlink
body in the callback. Similar to solution for bpf, only here it's even
more obvious that ->evict_inode() can be dropped.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
free the symlink body after the same RCU delay we have for freeing the
struct inode itself, so that traversal during RCU pathwalk wouldn't step
into freed memory.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
This is log is harmful as it can trigger multiple times per packet. Delete
it.
Signed-off-by: Aaro Koskinen <aaro.koskinen@nokia.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
Packets without the last descriptor set should be dropped early. If we
receive a frame larger than the DMA buffer, the HW will continue using the
next descriptor. Driver mistakes these as individual frames, and sometimes
a truncated frame (without the LD set) may look like a valid packet.
This fixes a strange issue where the system replies to 4098-byte ping
although the MTU/DMA buffer size is set to 4096, and yet at the same
time it's logging an oversized packet.
Signed-off-by: Aaro Koskinen <aaro.koskinen@nokia.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
If we have error bits set, the discard_frame status will get overwritten
by checksum bit checks, which might set the status back to good one.
Fix by checking the COE status only if the frame is good.
Signed-off-by: Aaro Koskinen <aaro.koskinen@nokia.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
Currently, if we drop a packet, we exit from NAPI loop before the budget
is consumed. In some situations this will make the RX processing stall
e.g. when flood pinging the system with oversized packets, as the
errorneous packets are not dropped efficiently.
If we drop a packet, we should just continue to the next one as long as
the budget allows.
Signed-off-by: Aaro Koskinen <aaro.koskinen@nokia.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
Signed-off-by: Aaro Koskinen <aaro.koskinen@nokia.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
We always program the maximum DMA buffer size into the receive descriptor,
although the allocated size may be less. E.g. with the default MTU size
we allocate only 1536 bytes. If somebody sends us a bigger frame, then
memory may get corrupted.
Fix by using exact buffer sizes.
Signed-off-by: Aaro Koskinen <aaro.koskinen@nokia.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
So print full hwaddr in sysfs "bonding_slave/perm_hwaddr" as well.
Signed-off-by: Konstantin Khorenko <khorenko@virtuozzo.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
Traditionally, the PF (Physical Function) which resides on vport 0 was
the E-switch manager. Since the ECPF (Embedded CPU Physical Function),
which resides on vport 0xfffe, was introduced as the E-Switch manager,
the assumption that the E-switch manager is on vport 0 is incorrect.
Since the eswitch code already uses the actual vport value, all we
need is to always set other_vport=1.
The esw offloads structures share a union with the legacy mode structs.
Reset the offloads struct to zero in init to protect from null
assumptions made by the legacy mode code.
Signed-off-by: Roi Dayan <roid@mellanox.com> Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
We want to avoid leaking pointer info from xdp_frame (that is placed in
top of frame) like commit 6dfb970d3dbd ("xdp: avoid leaking info stored in
frame data on page reuse"), and followup commit 97e19cce05e5 ("bpf:
reserve xdp_frame size in xdp headroom") that reserve this headroom.
These changes also affected how cpumap constructed SKBs, as xdpf->headroom
size changed, the skb data starting point were in-effect shifted with 32
bytes (sizeof xdp_frame). This was still okay, as the cpumap frame_size
calculation also included xdpf->headroom which were reduced by same amount.
A bug was introduced in commit 77ea5f4cbe20 ("bpf/cpumap: make sure
frame_size for build_skb is aligned if headroom isn't"), where the
xdpf->headroom became part of the SKB_DATA_ALIGN rounding up. This
round-up to find the frame_size is in principle still correct as it does
not exceed the 2048 bytes frame_size (which is max for ixgbe and i40e),
but the 32 bytes offset of pkt_data_start puts this over the 2048 bytes
limit. This cause skb_shared_info to spill into next frame. It is a little
hard to trigger, as the SKB need to use above 15 skb_shinfo->frags[] as
far as I calculate. This does happen in practise for TCP streams when
skb_try_coalesce() kicks in.
KASAN can be used to detect these wrong memory accesses, I've seen:
BUG: KASAN: use-after-free in skb_try_coalesce+0x3cb/0x760
BUG: KASAN: wild-memory-access in skb_release_data+0xe2/0x250
Driver veth also construct a SKB from xdp_frame in this way, but is not
affected, as it doesn't reserve/deduct the room (used by xdp_frame) from
the SKB headroom. Instead is clears the pointers via xdp_scrub_frame(),
and allows SKB to use this area.
The fix in this patch is to do like veth and instead allow SKB to (re)use
the area occupied by xdp_frame, by clearing via xdp_scrub_frame(). (This
does kill the idea of the SKB being able to access (mem) info from this
area, but I guess it was a bad idea anyhow, and it was already killed by
the veth changes.)
Fixes: 77ea5f4cbe20 ("bpf/cpumap: make sure frame_size for build_skb is aligned if headroom isn't") Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
Currently, the rules for configuring search paths in Kbuild have
changed, this will lead some erros when compiling hns3 with the
following command:
make O=DIR M=drivers/net/ethernet/hisilicon/hns3
drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_cmd.c:11:10:
fatal error: hnae3.h: No such file or directory
This patch fix it by adding $(srctree)/ prefix to the serach paths.
Signed-off-by: Xi Wang <wangxi11@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
Similar to commit edfc3722cfef ("HID: quirks: Fix keyboard + touchpad on
Toshiba Click Mini not working"), the Lenovo Miix 630 has a combo
keyboard/touchpad device with vid:pid of 04F3:0400, which is shared with
Elan touchpads. The combo on the Miix 630 has an ACPI id of QTEC0001,
which is not claimed by the elan_i2c driver, so key on that similar to
what was done for the Toshiba Click Mini.
A memory save operation to 8-byte variable in RV32 is divided into
two sw instructions in the put_user macro. The current fixup returns
execution flow to the second sw instead of the one after it.
This patch fixes this fixup code according to the load access part.
Signed-off-by: Alan Kao<alankao@andestech.com> Cc: Greentime Hu <greentime@andestech.com> Cc: Vincent Chen <deanbo422@gmail.com> Signed-off-by: Palmer Dabbelt <palmer@sifive.com> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
The current check for WoL on i40e is broken. Code comment says only
magic packet is supported, so only check for that.
Fixes: 540a152da762 (i40e/ixgbe/igb: fail on new WoL flag setting WAKE_MAGICSECURE) Signed-off-by: Stefan Assmann <sassmann@kpanic.de> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
The ixgbe ignores errors returned from mdiobus_register() and leaves
adapter->mii_bus non-NULL and MDIO bus state as MDIOBUS_ALLOCATED.
This triggers a BUG from mdiobus_unregister() during ixgbe_remove() call.
Fixes: 8fa10ef01260 ("ixgbe: register a mdiobus") Signed-off-by: Ivan Vecera <ivecera@redhat.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
Commit 0ac30ce43323 ("i40e: fix up 32 bit timespec references",
2017-07-26) claims to be cleaning up references to 32-bit timespecs.
The actual contents of the commit make no sense, as it converts a call
to timespec64_add into timespec64_add_ns. This would seem ok, if (a) the
change was documented in the commit message, and (b) timespec64_add_ns
supported negative numbers.
timespec64_add_ns doesn't work with signed deltas, because the
implementation is based around iter_div_u64_rem. This change resulted in
a regression where i40e_ptp_adjtime would interpret small negative
adjustments as large positive additions, resulting in incorrect
behavior.
This commit doesn't appear to fix anything, is not well explained, and
introduces a bug, so lets just revert it.
Reverts: 0ac30ce43323 ("i40e: fix up 32 bit timespec references", 2017-07-26) Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Reviewed-by: Arnd Bergmann <arnd@arndb.de> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
When CONFIG_CFG80211 isn't enabled the compiler correcly warns about
'sinfo.pertid' may be unused. It can also happen for other error
conditions that it not warn about.
net/batman-adv/bat_v_elp.c: In function ‘batadv_v_elp_get_throughput.isra.0’:
include/net/cfg80211.h:6370:13: warning: ‘sinfo.pertid’ may be used
uninitialized in this function [-Wmaybe-uninitialized]
kfree(sinfo->pertid);
~~~~~^~~~~~~~
Rework so that we only release '&sinfo' if cfg80211_get_station returns
zero.
Fixes: 7d652669b61d ("batman-adv: release station info tidstats") Signed-off-by: Anders Roxell <anders.roxell@linaro.org> Signed-off-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
The batadv_hash_remove is a function which searches the hashtable for an
entry using a needle, a hashtable bucket selection function and a compare
function. It will lock the bucket list and delete an entry when the compare
function matches it with the needle. It returns the pointer to the
hlist_node which matches or NULL when no entry matches the needle.
The batadv_tt_global_free is not itself protected in anyway to avoid that
any other function is modifying the hashtable between the search for the
entry and the call to batadv_hash_remove. It can therefore happen that the
entry either doesn't exist anymore or an entry was deleted which is not the
same object as the needle. In such an situation, the reference counter (for
the reference stored in the hashtable) must not be reduced for the needle.
Instead the reference counter of the actually removed entry has to be
reduced.
Otherwise the reference counter will underflow and the object might be
freed before all its references were dropped. The kref helpers reported
this problem as:
refcount_t: underflow; use-after-free.
Fixes: 7683fdc1e886 ("batman-adv: protect the local and the global trans-tables with rcu") Reported-by: Martin Weinelt <martin@linuxlounge.net> Signed-off-by: Sven Eckelmann <sven@narfation.org> Acked-by: Antonio Quartulli <a@unstable.cc> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
The batadv_hash_remove is a function which searches the hashtable for an
entry using a needle, a hashtable bucket selection function and a compare
function. It will lock the bucket list and delete an entry when the compare
function matches it with the needle. It returns the pointer to the
hlist_node which matches or NULL when no entry matches the needle.
The batadv_tt_local_remove is not itself protected in anyway to avoid that
any other function is modifying the hashtable between the search for the
entry and the call to batadv_hash_remove. It can therefore happen that the
entry either doesn't exist anymore or an entry was deleted which is not the
same object as the needle. In such an situation, the reference counter (for
the reference stored in the hashtable) must not be reduced for the needle.
Instead the reference counter of the actually removed entry has to be
reduced.
Otherwise the reference counter will underflow and the object might be
freed before all its references were dropped. The kref helpers reported
this problem as:
refcount_t: underflow; use-after-free.
Fixes: ef72706a0543 ("batman-adv: protect tt_local_entry from concurrent delete events") Signed-off-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
The batadv_hash_remove is a function which searches the hashtable for an
entry using a needle, a hashtable bucket selection function and a compare
function. It will lock the bucket list and delete an entry when the compare
function matches it with the needle. It returns the pointer to the
hlist_node which matches or NULL when no entry matches the needle.
The batadv_bla_del_claim is not itself protected in anyway to avoid that
any other function is modifying the hashtable between the search for the
entry and the call to batadv_hash_remove. It can therefore happen that the
entry either doesn't exist anymore or an entry was deleted which is not the
same object as the needle. In such an situation, the reference counter (for
the reference stored in the hashtable) must not be reduced for the needle.
Instead the reference counter of the actually removed entry has to be
reduced.
Otherwise the reference counter will underflow and the object might be
freed before all its references were dropped. The kref helpers reported
this problem as:
e = f(...);
... when != of_node_put(e)
when != x = e
when != e = x
when any
if (<+...of_device_is_available(e)...+>) {
... when != of_node_put(e)
(
return e;
|
+ of_node_put(e);
return ...;
)
}
// </smpl>
Fixes: e0c827aca0730 ("drm/omap: Populate DSS children in omapdss driver") Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr> Signed-off-by: Tony Lindgren <tony@atomide.com> Signed-off-by: Sasha Levin (Microsoft) <sashal@kernel.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
sh_rtc_read_alarm_value() returns 0xff when querying a hardware alarm
field that is not enabled. __rtc_read_alarm() validates the received
alarm values, and fills in missing fields when needed.
While 0xff is handled fine for the year, month, and day fields, and
corrected as considered being out-of-range, this is not the case for the
hour, minute, and second fields, where -1 is expected for missing
fields.
Fix this by returning -1 instead, as this value is handled fine for all
fields.
If we encounter a failure during suspend where this RTC was programmed
to wakeup the system from suspend, but that wakeup couldn't be
configured because the system didn't support wakeup interrupts, we'll
run into the following warning:
Unbalanced IRQ 166 wake disable
WARNING: CPU: 7 PID: 3071 at kernel/irq/manage.c:669 irq_set_irq_wake+0x108/0x278
This happens because the suspend process isn't aborted when the RTC
fails to configure the wakeup IRQ. Instead, we continue suspending the
system and then another suspend callback fails the suspend process and
"unwinds" the previously suspended drivers by calling their resume
callbacks. When we get back to resuming this RTC driver, we'll call
disable_irq_wake() on an IRQ that hasn't been configured for wake.
Let's just fail suspend/resume here if we can't configure the system to
wake and the user has chosen to wakeup with this device. This fixes this
warning and makes the code more robust in case there are systems out
there that can't wakeup from suspend on this line but the user has
chosen to do so.
There is a race condition that could happen if hid_debug_rdesc_show()
is running while hdev is in the process of going away (device removal,
system suspend, etc) which could result in NULL pointer dereference:
BUG: unable to handle kernel paging request at 0000000783316040
CPU: 1 PID: 1512 Comm: getevent Tainted: G U O 4.19.20-quilt-2e5dc0ac-00029-gc455a447dd55 #1
RIP: 0010:hid_dump_device+0x9b/0x160
Call Trace:
hid_debug_rdesc_show+0x72/0x1d0
seq_read+0xe0/0x410
full_proxy_read+0x5f/0x90
__vfs_read+0x3a/0x170
vfs_read+0xa0/0x150
ksys_read+0x58/0xc0
__x64_sys_read+0x1a/0x20
do_syscall_64+0x55/0x110
entry_SYSCALL_64_after_hwframe+0x49/0xbe
Grab driver_input_lock to make sure the input device exists throughout the
whole process of dumping the rdesc.
create_singlethread_workqueue may fail and return NULL. The fix checks if it is
NULL to avoid NULL pointer dereference. Also, the fix moves the call of
create_singlethread_workqueue earlier to avoid resource-release issues.
The rk3328-roc-cc board exhibits tx stability issues with large packets,
as does the rock64 board, which was fixed with this patch
https://patchwork.kernel.org/patch/10178969/
A similar patch was merged for the rk3328-roc-cc here
https://patchwork.kernel.org/patch/10804863/
but it doesn't include the tx/rx_delay tweaks, and I find that they
help with an issue where large transfers would bring the ethernet
link down, causing a link reset regularly.
To minimize the latency of timer interrupts as observed by the guest,
KVM adjusts the values it programs into the host timers to account for
the host's overhead of programming and handling the timer event. In
the event that the adjustments are too aggressive, i.e. the timer fires
earlier than the guest expects, KVM busy waits immediately prior to
entering the guest.
Currently, KVM manually converts the delay from nanoseconds to clock
cycles. But, the conversion is done in the guest's time domain, while
the delay occurs in the host's time domain. This is perfectly ok when
the guest and host are using the same TSC ratio, but if the guest is
using a different ratio then the delay may not be accurate and could
wait too little or too long.
When the guest is not using the host's ratio, convert the delay from
guest clock cycles to host nanoseconds and use ndelay() instead of
__delay() to provide more accurate timing. Because converting to
nanoseconds is relatively expensive, e.g. requires division and more
multiplication ops, continue using __delay() directly when guest and
host TSCs are running at the same ratio.
Cc: Liran Alon <liran.alon@oracle.com> Cc: Wanpeng Li <wanpengli@tencent.com> Cc: stable@vger.kernel.org Fixes: 3b8a5df6c4dc6 ("KVM: LAPIC: Tune lapic_timer_advance_ns automatically") Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
The introduction of adaptive tuning of lapic timer advancement did not
allow for the scenario where userspace would want to disable adaptive
tuning but still employ timer advancement, e.g. for testing purposes or
to handle a use case where adaptive tuning is unable to settle on a
suitable time. This is epecially pertinent now that KVM places a hard
threshold on the maximum advancment time.
Rework the timer semantics to accept signed values, with a value of '-1'
being interpreted as "use adaptive tuning with KVM's internal default",
and any other value being used as an explicit advancement time, e.g. a
time of '0' effectively disables advancement.
Note, this does not completely restore the original behavior of
lapic_timer_advance_ns. Prior to tracking the advancement per vCPU,
which is necessary to support autotuning, userspace could adjust
lapic_timer_advance_ns for *running* vCPU. With per-vCPU tracking, the
module params are snapshotted at vCPU creation, i.e. applying a new
advancement effectively requires restarting a VM.
Dynamically updating a running vCPU is possible, e.g. a helper could be
added to retrieve the desired delay, choosing between the global module
param and the per-VCPU value depending on whether or not auto-tuning is
(globally) enabled, but introduces a great deal of complexity. The
wrapper itself is not complex, but understanding and documenting the
effects of dynamically toggling auto-tuning and/or adjusting the timer
advancement is nigh impossible since the behavior would be dependent on
KVM's implementation as well as compiler optimizations. In other words,
providing stable behavior would require extremely careful consideration
now and in the future.
Given that the expected use of a manually-tuned timer advancement is to
"tune once, run many", use the vastly simpler approach of recognizing
changes to the module params only when creating a new vCPU.
Automatically adjusting the globally-shared timer advancement could
corrupt the timer, e.g. if multiple vCPUs are concurrently adjusting
the advancement value. That could be partially fixed by using a local
variable for the arithmetic, but it would still be susceptible to a
race when setting timer_advance_adjust_done.
And because virtual_tsc_khz and tsc_scaling_ratio are per-vCPU, the
correct calibration for a given vCPU may not apply to all vCPUs.
Furthermore, lapic_timer_advance_ns is marked __read_mostly, which is
effectively violated when finding a stable advancement takes an extended
amount of timer.
Opportunistically change the definition of lapic_timer_advance_ns to
a u32 so that it matches the style of struct kvm_timer. Explicitly
pass the param to kvm_create_lapic() so that it doesn't have to be
exposed to lapic.c, thus reducing the probability of unintentionally
using the global value instead of the per-vCPU value.
If guest sets MSR_IA32_TSCDEADLINE to value such that in host
time-domain it's shorter than lapic_timer_advance_ns, we can
reach a case that we call hrtimer_start() with expiration time set at
the past.
Because lapic_timer.timer is init with HRTIMER_MODE_ABS_PINNED, it
is not allowed to run in softirq and therefore will never expire.
To avoid such a scenario, verify that deadline expiration time is set on
host time-domain further than (now + lapic_timer_advance_ns).
A future patch can also consider adding a min_timer_deadline_ns module parameter,
similar to min_timer_period_us to avoid races that amount of ns it takes
to run logic could still call hrtimer_start() with expiration timer set
at the past.
To minimize the latency of timer interrupts as observed by the guest,
KVM adjusts the values it programs into the host timers to account for
the host's overhead of programming and handling the timer event. Now
that the timer advancement is automatically tuned during runtime, it's
effectively unbounded by default, e.g. if KVM is running as L1 the
advancement can measure in hundreds of milliseconds.
Disable timer advancement if adaptive tuning yields an advancement of
more than 5000ns, as large advancements can break reasonable assumptions
of the guest, e.g. that a timer configured to fire after 1ms won't
arrive on the next instruction. Although KVM busy waits to mitigate the
case of a timer event arriving too early, complications can arise when
shifting the interrupt too far, e.g. kvm-unit-test's vmx.interrupt test
will fail when its "host" exits on interrupts as KVM may inject the INTR
before the guest executes STI+HLT. Arguably the unit test is "broken"
in the sense that delaying a timer interrupt by 1ms doesn't technically
guarantee the interrupt will arrive after STI+HLT, but it's a reasonable
assumption that KVM should support.
Furthermore, an unbounded advancement also effectively unbounds the time
spent busy waiting, e.g. if the guest programs a timer with a very large
delay.
5000ns is a somewhat arbitrary threshold. When running on bare metal,
which is the intended use case, timer advancement is expected to be in
the general vicinity of 1000ns. 5000ns is high enough that false
positives are unlikely, while not being so high as to negatively affect
the host's performance/stability.
Note, a future patch will enable userspace to disable KVM's adaptive
tuning, which will allow priveleged userspace will to specifying an
advancement value in excess of this arbitrary threshold in order to
satisfy an abnormal use case.
Cc: Liran Alon <liran.alon@oracle.com> Cc: Wanpeng Li <wanpengli@tencent.com> Cc: stable@vger.kernel.org Fixes: 3b8a5df6c4dc6 ("KVM: LAPIC: Tune lapic_timer_advance_ns automatically") Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
The syzkaller fuzzer reported a bug in the USB hub driver which turned
out to be caused by a negative runtime-PM usage counter. This allowed
a hub to be runtime suspended at a time when the driver did not expect
it. The symptom is a WARNING issued because the hub's status URB is
submitted while it is already active:
URB 0000000031fb463e submitted while active
WARNING: CPU: 0 PID: 2917 at drivers/usb/core/urb.c:363
The negative runtime-PM usage count was caused by an unfortunate
design decision made when runtime PM was first implemented for USB.
At that time, USB class drivers were allowed to unbind from their
interfaces without balancing the usage counter (i.e., leaving it with
a positive count). The core code would take care of setting the
counter back to 0 before allowing another driver to bind to the
interface.
Later on when runtime PM was implemented for the entire kernel, the
opposite decision was made: Drivers were required to balance their
runtime-PM get and put calls. In order to maintain backward
compatibility, however, the USB subsystem adapted to the new
implementation by keeping an independent usage counter for each
interface and using it to automatically adjust the normal usage
counter back to 0 whenever a driver was unbound.
This approach involves duplicating information, but what is worse, it
doesn't work properly in cases where a USB class driver delays
decrementing the usage counter until after the driver's disconnect()
routine has returned and the counter has been adjusted back to 0.
Doing so would cause the usage counter to become negative. There's
even a warning about this in the USB power management documentation!
As it happens, this is exactly what the hub driver does. The
kick_hub_wq() routine increments the runtime-PM usage counter, and the
corresponding decrement is carried out by hub_event() in the context
of the hub_wq work-queue thread. This work routine may sometimes run
after the driver has been unbound from its interface, and when it does
it causes the usage counter to go negative.
It is not possible for hub_disconnect() to wait for a pending
hub_event() call to finish, because hub_disconnect() is called with
the device lock held and hub_event() acquires that lock. The only
feasible fix is to reverse the original design decision: remove the
duplicate interface-specific usage counter and require USB drivers to
balance their runtime PM gets and puts. As far as I know, all
existing drivers currently do this.
Some drivers (such as the vub300 MMC driver) expect usb_string() to
return a properly NUL-terminated string, even when an error occurs.
(In fact, vub300's probe routine doesn't bother to check the return
code from usb_string().) When the driver goes on to use an
unterminated string, it leads to kernel errors such as
stack-out-of-bounds, as found by the syzkaller USB fuzzer.
An out-of-range string index argument is not at all unlikely, given
that some devices don't provide string descriptors and therefore list
0 as the value for their string indexes. This patch makes
usb_string() return a properly terminated empty string along with the
-EINVAL error code when an out-of-range index is encountered.
And since a USB string index is a single-byte value, indexes >= 256
are just as invalid as values of 0 or below.
Change the validation of number_of_packets in get_pipe to compare the
number of packets to a fixed maximum number of packets allowed, set to
be 1024. This number was chosen due to it being used by other drivers as
well, for example drivers/usb/host/uhci-q.c
Background/reason:
The get_pipe function in stub_rx.c validates the number of packets in
isochronous mode and aborts with an error if that number is too large,
in order to prevent malicious input from possibly triggering large
memory allocations. This was previously done by checking whether
pdu->u.cmd_submit.number_of_packets is bigger than the number of packets
that would be needed for pdu->u.cmd_submit.transfer_buffer_length bytes
if all except possibly the last packet had maximum length, given by
usb_endpoint_maxp(epd) * usb_endpoint_maxp_mult(epd). This leads to an
error if URBs with packets shorter than the maximum possible length are
submitted, which is allowed according to
Documentation/driver-api/usb/URB.rst and occurs for example with the
snd-usb-audio driver.
The syzkaller USB fuzzer identified a failure mode in which dummy-hcd
would never give back an unlinked URB. This causes usb_kill_urb() to
hang, leading to WARNINGs and unkillable threads.
In dummy-hcd, all URBs are given back by the dummy_timer() routine as
it scans through the list of pending URBS. Failure to give back URBs
can be caused by failure to start or early exit from the scanning
loop. The code currently has two such pathways: One is triggered when
an unsupported bus transfer speed is encountered, and the other by
exhausting the simulated bandwidth for USB transfers during a frame.
This patch removes those two paths, thereby allowing all unlinked URBs
to be given back in a timely manner. It adds a check for the bus
speed when the gadget first starts running, so that dummy_timer() will
never thereafter encounter an unsupported speed. And it prevents the
loop from exiting as soon as the total bandwidth has been used up (the
scanning loop continues, giving back unlinked URBs as they are found,
but not transferring any more data).
Thanks to Andrey Konovalov for manually running the syzkaller fuzzer
to help track down the source of the bug.
The syzkaller USB fuzzer spotted a slab-out-of-bounds bug in the
ds2490 driver. This bug is caused by improper use of the altsetting
array in the usb_interface structure (the array's entries are not
always stored in numerical order), combined with a naive assumption
that all interfaces probed by the driver will have the expected number
of altsettings.
The bug can be fixed by replacing references to the possibly
non-existent intf->altsetting[alt] entry with the guaranteed-to-exist
intf->cur_altsetting entry.
The syzkaller USB fuzzer found a general-protection-fault bug in the
yurex driver. The fault occurs when a device has been unplugged; the
driver's interrupt-URB handler logs an error message referring to the
device by name, after the device has been unregistered and its name
deallocated.
This problem is caused by the fact that the interrupt URB isn't
cancelled until the driver's private data structure is released, which
can happen long after the device is gone. The cure is to make sure
that the interrupt URB is killed before yurex_disconnect() returns;
this is exactly the sort of thing that usb_poison_urb() was meant for.
Some ASUS models like Q325UAR with ALC295 codec requires the same
fixup that has been applied to ALC294 codec. Just copy the entry with
the pin matching to cover ALC295 too.
[ Note: this is a further correction / addition of the previous
pin-based quirks for Dell machines; another entry for ALC236 with
the d-mic pin 0x12 and an entry for ALC295 -- tiwai ]
Fixes: b26e36b7ef36 ("ALSA: hda/realtek - add two more pin configuration sets to quirk table") Signed-off-by: Kailang Yang <kailang@realtek.com> Cc: <stable@vger.kernel.org> Signed-off-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
Multiple users have reported their Synaptics touchpad has stopped
working between v4.20.1 and v4.20.2 when using SMBus interface.
The culprit for this appeared to be commit c5eb1190074c ("PCI / PM: Allow
runtime PM without callback functions") that fixed the runtime PM for
i2c-i801 SMBus adapter. Those Synaptics touchpad are using i2c-i801
for SMBus communication and testing showed they are able to get back
working by preventing the runtime suspend of adapter.
Normally when i2c-i801 SMBus adapter transmits with the client it resumes
before operation and autosuspends after.
However, if client requires SMBus Host Notify protocol, what those
Synaptics touchpads do, then the host adapter must not go to runtime
suspend since then it cannot process incoming SMBus Host Notify commands
the client may send.
Fix this by keeping I2C/SMBus adapter active in case client requires
Host Notify.
Reported-by: Keijo Vaara <ferdasyn@rocketmail.com> Link: https://bugzilla.kernel.org/show_bug.cgi?id=203297 Fixes: c5eb1190074c ("PCI / PM: Allow runtime PM without callback functions") Cc: stable@vger.kernel.org # v4.20+ Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com> Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Tested-by: Keijo Vaara <ferdasyn@rocketmail.com> Signed-off-by: Wolfram Sang <wsa@the-dreams.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
The way of getting private imx_i2c_struct in i2c_imx_clk_notifier_call()
is incorrect, should use clk_change_nb element to get correct address
and avoid below kernel dump during POST_RATE_CHANGE notify by clk
framework:
The I2C host driver for SynQuacer fails to populate the of_node and
ACPI companion fields of the struct i2c_adapter it instantiates,
resulting in enumeration of the subordinate I2C bus to fail.
Fixes: 0d676a6c4390 ("i2c: add support for Socionext SynQuacer I2C controller") Cc: <stable@vger.kernel.org> # v4.19+ Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Signed-off-by: Wolfram Sang <wsa@the-dreams.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
We need to dereference the directory to get its parent to
be able to rename it, so it's clearly not safe to try to
do this with ERR_PTR() pointers. Skip in this case.
It seems that this is most likely what was causing the
report by syzbot, but I'm not entirely sure as it didn't
come with a reproducer this time.
Since the migration of the driver to stop using the legacy
->select_chip() hook, there is nothing deselecting the target anymore,
thus the selection is not forced at the next access. Ensure the ND_RUN
bit and the interrupts are always in a clean state.
Cc: Daniel Mack <daniel@zonque.org> Cc: stable@vger.kernel.org Fixes: b25251414f6e00 ("mtd: rawnand: marvell: Stop implementing ->select_chip()") Suggested-by: Boris Brezillon <boris.brezillon@collabora.com> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com> Tested-by: Daniel Mack <daniel@zonque.org> Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com> Signed-off-by: Richard Weinberger <richard@nod.at> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
The commit fc3a2fcaa1ba ("mwifiex: use atomic bitops to represent
adapter status variables") had a fairly straightforward bug in it. It
contained this bit of diff:
- if (!adapter->is_suspended) {
+ if (test_bit(MWIFIEX_IS_SUSPENDED, &adapter->work_flags)) {
As you can see the patch missed the "!" when converting to the atomic
bitops. This meant that the resume hasn't done anything at all since
that commit landed and suspend/resume for mwifiex SDIO cards has been
totally broken.
After fixing this mwifiex suspend/resume appears to work again, at
least with the simple testing I've done.
Fixes: fc3a2fcaa1ba ("mwifiex: use atomic bitops to represent adapter status variables") Cc: <stable@vger.kernel.org> Signed-off-by: Douglas Anderson <dianders@chromium.org> Reviewed-by: Brian Norris <briannorris@chromium.org> Signed-off-by: Kalle Valo <kvalo@codeaurora.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
We introduced a bug that prevented this old device from
working. The driver would simply not be able to complete
the INIT flow while spewing this warning:
In case of 'L1_CACHE_SHIFT != 6' we define dummy assembly macroses
PREALLOC_INSTR and PREFETCHW_INSTR without arguments. However
we pass arguments to them in code which cause build errors.
Fix that.
As the comment notes, the return codes for TSYNC and NEW_LISTENER
conflict, because they both return positive values, one in the case of
success and one in the case of error. So, let's disallow both of these
flags together.
While this is technically a userspace break, all the users I know
of are still waiting on me to land this feature in libseccomp, so I
think it'll be safe. Also, at present my use case doesn't require
TSYNC at all, so this isn't a big deal to disallow. If someone
wanted to support this, a path forward would be to add a new flag like
TSYNC_AND_LISTENER_YES_I_UNDERSTAND_THAT_TSYNC_WILL_JUST_RETURN_EAGAIN,
but the use cases are so different I don't see it really happening.
Finally, it's worth noting that this does actually fix a UAF issue: at the
end of seccomp_set_mode_filter(), we have:
if (flags & SECCOMP_FILTER_FLAG_NEW_LISTENER) {
if (ret < 0) {
listener_f->private_data = NULL;
fput(listener_f);
put_unused_fd(listener);
} else {
fd_install(listener, listener_f);
ret = listener;
}
}
out_free:
seccomp_filter_free(prepared);
But if ret > 0 because TSYNC raced, we'll install the listener fd and then
free the filter out from underneath it, causing a UAF when the task closes
it or dies. This patch also switches the condition to be simply if (ret),
so that if someone does add the flag mentioned above, they won't have to
remember to fix this too.
Reported-by: syzbot+b562969adb2e04af3442@syzkaller.appspotmail.com Fixes: 6a21cc50f0c7 ("seccomp: add a return code to trap to userspace") CC: stable@vger.kernel.org # v5.0+ Signed-off-by: Tycho Andersen <tycho@tycho.ws> Signed-off-by: Kees Cook <keescook@chromium.org> Acked-by: James Morris <jamorris@linux.microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
Joerg Roedel [Fri, 7 Jun 2019 17:47:30 +0000 (13:47 -0400)]
iommu/amd: Set exclusion range correctly
BugLink: https://bugs.launchpad.net/bugs/1823037
The exlcusion range limit register needs to contain the
base-address of the last page that is part of the range, as
bits 0-11 of this register are treated as 0xfff by the
hardware for comparisons.
So correctly set the exclusion range in the hardware to the
last page which is _in_ the range.
Fixes: b2026aa2dce44 ('x86, AMD IOMMU: add functions for programming IOMMU MMIO space') Signed-off-by: Joerg Roedel <jroedel@suse.de>
(cherry picked from commit 3c677d206210f53a4be972211066c0f1cd47fe12 ) Signed-off-by: Jeffrey Lane <jeffrey.lane@canonical.com> Acked-by: Stefan Bader <stefan.bader@canonical.com> Acked-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Khalid Elmously <khalid.elmously@canonical.com>
Colin Ian King [Wed, 19 Jun 2019 13:30:32 +0000 (14:30 +0100)]
mm/page_idle.c: fix oops because end_pfn is larger than max_pfn
BugLink: https://bugs.launchpad.net/bugs/1833410
Currently the calcuation of end_pfn can round up the pfn number to more
than the actual maximum number of pfns, causing an Oops. Fix this by
ensuring end_pfn is never more than max_pfn.
This can be easily triggered when on systems where the end_pfn gets
rounded up to more than max_pfn using the idle-page stress-ng stress test:
Link: http://lkml.kernel.org/r/20190618124352.28307-1-colin.king@canonical.com Fixes: 33c3fc71c8cf ("mm: introduce idle page tracking") Signed-off-by: Colin Ian King <colin.king@canonical.com>
(cherry picked from commit d96d6145d9796d5f1eac242538d45559e9a23404 linux-next) Reviewed-by: Andrew Morton <akpm@linux-foundation.org> Acked-by: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mike Rapoport <rppt@linux.vnet.ibm.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
(cherry picked from commit d96d6145d9796d5f1eac242538d45559e9a23404 linux-next) Signed-off-by: Colin Ian King <colin.king@canonical.com> Acked-by: Kamal Mostafa <kamal@canonical.com> Acked-by: Andrea Righi <andrea.righi@canonical.com> Signed-off-by: Khalid Elmously <khalid.elmously@canonical.com>
BugLink: https://bugs.launchpad.net/bugs/1833484
Dell machines come with goodix touchpad IC suffer from the double click
issue if the Designware I2C adapter enters runtime suspend.
It's because the goodix re-assert the interrupt if host doesn't read the
data within 100ms and desiginware takes a longer time to wake up from
runtime suspend. In the case, it got a second interrupt during
resuming, so it thinks it's a double click.
There is no simple way to fix this, it's a firmware issue and goodix
agrees to fix this in their firmware on next release, but this issue is
still affects the machines that don't come with an updated firmware. So,
add a quirk to mark those machines and avoid the designware to enter
runtime suspend.
Signed-off-by: AceLan Kao <acelan.kao@canonical.com> Acked-by: Seth Forshee <seth.forshee@canonical.com> Acked-by: Timo Aaltonen <tjaalton@ubuntu.com> Signed-off-by: Khalid Elmously <khalid.elmously@canonical.com>
ath10k_mac_vif_chan() always returns an error for the given vif
during system-wide resume which reliably triggers two WARN_ON()s
in ath10k_bss_info_changed() and they are not particularly
useful in that code path, so drop them.
Tested: QCA6174 hw3.2 PCI with WLAN.RM.2.0-00180-QCARMSWPZ-1
Tested: QCA6174 hw3.2 SDIO with WLAN.RMH.4.4.1-00007-QCARMSWP-1
Fixes: cd93b83ad927 ("ath10k: support for multicast rate control") Fixes: f279294e9ee2 ("ath10k: add support for configuring management packet rate") Cc: stable@vger.kernel.org Reviewed-by: Brian Norris <briannorris@chromium.org> Tested-by: Brian Norris <briannorris@chromium.org> Tested-by: Claire Chang <tientzu@chromium.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Kalle Valo <kvalo@codeaurora.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
debugfs can now report an error code if something went wrong instead of
just NULL. So if the return value is to be used as a "real" dentry, it
needs to be checked if it is an error before dereferencing it.
This is now happening because of ff9fb72bc077 ("debugfs: return error
values, not NULL"). If multiple iwlwifi devices are in the system, this
can cause problems when the driver attempts to create the main debugfs
directory again. Later on in the code we fail horribly by trying to
dereference a pointer that is an error value.
Reported-by: Laura Abbott <labbott@redhat.com> Reported-by: Gabriel Ramirez <gabriello.ramirez@gmail.com> Cc: Johannes Berg <johannes.berg@intel.com> Cc: Emmanuel Grumbach <emmanuel.grumbach@intel.com> Cc: Luca Coelho <luciano.coelho@intel.com> Cc: Intel Linux Wireless <linuxwifi@intel.com> Cc: Kalle Valo <kvalo@codeaurora.org> Cc: stable <stable@vger.kernel.org> # 5.0 Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
The line6 driver uses a lot of USB buffers off of the stack, which is
not allowed on many systems, causing the driver to crash on some of
them. Fix this up by dynamically allocating the buffers with kmalloc()
which allows for proper DMA-able memory.
The size checks in vmx_nested_state are wrong because the calculations
are made based on the size of a pointer to a struct kvm_nested_state
rather than the size of a struct kvm_nested_state.
Reported-by: Felix Wilhelm <fwilhelm@google.com> Signed-off-by: Jim Mattson <jmattson@google.com> Reviewed-by: Drew Schmitt <dasch@google.com> Reviewed-by: Marc Orr <marcorr@google.com> Reviewed-by: Peter Shier <pshier@google.com> Reviewed-by: Krish Sadhukhan <krish.sadhukhan@oracle.com> Fixes: 8fcc4b5923af5de58b80b53a069453b135693304 Cc: stable@ver.kernel.org Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
KVM's recent bug fix to update %rip after emulating I/O broke userspace
that relied on the previous behavior of incrementing %rip prior to
exiting to userspace. When running a Windows XP guest on AMD hardware,
Qemu may patch "OUT 0x7E" instructions in reaction to the OUT itself.
Because KVM's old behavior was to increment %rip before exiting to
userspace to handle the I/O, Qemu manually adjusted %rip to account for
the OUT instruction.
Arguably this is a userspace bug as KVM requires userspace to re-enter
the kernel to complete instruction emulation before taking any other
actions. That being said, this is a bit of a grey area and breaking
userspace that has worked for many years is bad.
Pre-increment %rip on OUT to port 0x7e before exiting to userspace to
hack around the issue.
Fixes: 45def77ebf79e ("KVM: x86: update %rip after emulating IO") Reported-by: Simon Becherer <simon@becherer.de> Reported-and-tested-by: Iakov Karpov <srid@rkmail.ru> Reported-by: Gabriele Balducci <balducci@units.it> Reported-by: Antti Antinoja <reader@fennosys.fi> Cc: stable@vger.kernel.org Cc: Takashi Iwai <tiwai@suse.com> Cc: Jiri Slaby <jslaby@suse.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
Fragments may contain data from other records so we have to account
for that when we calculate the destination and max length of copy we
can perform. Note that 'offset' is the offset within the message,
so it can't be passed as offset within the frag..
Here skb_store_bits() would have realised the call is wrong and
simply not copy data.
Fixes: 4799ac81e52a ("tls: Add rx inline crypto offload") Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com> Reviewed-by: John Hurley <john.hurley@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
There is no guarantee the record starts before the skb frags.
If we don't check for this condition copy amount will get
negative, leading to reads and writes to random memory locations.
Familiar hilarity ensues.
Fixes: 4799ac81e52a ("tls: Add rx inline crypto offload") Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com> Reviewed-by: John Hurley <john.hurley@netronome.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
In bnxt_rx_pkt(), if the driver encounters BD errors, it will recycle
the buffers and jump to the end where the uninitailized variable "len"
is referenced. Fix it by adding a new jump label that will skip
the length update. This is the most correct fix since the length
may not be valid when we get this type of error.
Fixes: 6a8788f25625 ("bnxt_en: add support for software dynamic interrupt moderation") Reported-by: Nathan Chancellor <natechancellor@gmail.com> Cc: Nathan Chancellor <natechancellor@gmail.com> Signed-off-by: Michael Chan <michael.chan@broadcom.com> Reviewed-by: Nathan Chancellor <natechancellor@gmail.com> Tested-by: Nathan Chancellor <natechancellor@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
In an earlier commit that fixes the number of stats contexts to
reserve for the RDMA driver, we added a function parameter to pass in
the number of stats contexts to all the relevant functions. The passed
in parameter should have been used to set the enables field of the
firmware message.
Fixes: 780baad44f0f ("bnxt_en: Reserve 1 stat_ctx for RDMA driver.") Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
If driver determines that extended TX port statistics are not supported
or allocation of the data structure fails, make sure to pass 0 TX stats
size to firmware to disable it. The firmware returned TX stats size should
also be set to 0 for consistency. This will prevent
bnxt_get_ethtool_stats() from accessing the NULL TX stats pointer in
case there is mismatch between firmware and driver.
Fixes: 36e53349b60b ("bnxt_en: Add additional extended port statistics.") Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
If we encounter errors during open and proceed to clean up,
bnxt_hwrm_ring_free() may crash if the rings we try to free have never
been allocated. bnxt_cp_ring_for_rx() or bnxt_cp_ring_for_tx()
may reference pointers that have not been allocated.
Fix it by checking for valid fw_ring_id first before calling
bnxt_cp_ring_for_rx() or bnxt_cp_ring_for_tx().
Fixes: 2c61d2117ecb ("bnxt_en: Add helper functions to get firmware CP ring ID.") Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
In the bnxt_init_one() error path, short FW command request memory
is not freed. This patch fixes it.
Fixes: e605db801bde ("bnxt_en: Support for Short Firmware Message") Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com> Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
The driver builds a list of multicast addresses and sends it to the
firmware when the driver's ndo_set_rx_mode() is called. In rare
cases, the firmware can fail this call if internal resources to
add multicast addresses are exhausted. In that case, we should
try the call again by setting the ALL_MCAST flag which is more
guaranteed to succeed.
Fixes: c0c050c58d84 ("bnxt_en: New Broadcom ethernet driver.") Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
syzbot was able to crash host by sending UDP packets with a 0 payload.
TCP does not have this issue since we do not aggregate packets without
payload.
Since dev_gro_receive() sets gso_size based on skb_gro_len(skb)
it seems not worth trying to cope with padded packets.
BUG: KASAN: slab-out-of-bounds in skb_gro_receive+0xf5f/0x10e0 net/core/skbuff.c:3826
Read of size 16 at addr ffff88808893fff0 by task syz-executor612/7889
Memory state around the buggy address: ffff88808893fe80: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc ffff88808893ff00: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
>ffff88808893ff80: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
^ ffff888088940000: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ffff888088940080: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
Fixes: e20cf8d3f1f7 ("udp: implement GRO for plain UDP sockets.") Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Paolo Abeni <pabeni@redhat.com> Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
Currently, the UDP GRO code path does bad things on some edge
conditions - Aggregation can happen even on packet with different
lengths.
Fix the above by rewriting the 'complete' condition for GRO
packets. While at it, note explicitly that we allow merging the
first packet per burst below gso_size.
Reported-by: Sean Tong <seantong114@gmail.com> Fixes: e20cf8d3f1f7 ("udp: implement GRO for plain UDP sockets.") Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
Richard and Bruno both reported that my commit added a bug,
and Bruno was able to determine the problem came when a segment
wih a FIN packet was coalesced to a prior one in tcp backlog queue.
It turns out the header prediction in tcp_rcv_established()
looks back to TCP headers in the packet, not in the metadata
(aka TCP_SKB_CB(skb)->tcp_flags)
The fast path in tcp_rcv_established() is not supposed to
handle a FIN flag (it does not call tcp_fin())
Therefore we need to make sure to propagate the FIN flag,
so that the coalesced packet does not go through the fast path,
the same than a GRO packet carrying a FIN flag.
While we are at it, make sure we do not coalesce packets with
RST or SYN, or if they do not have ACK set.
Many thanks to Richard and Bruno for pinpointing the bad commit,
and to Richard for providing a first version of the fix.
Fixes: 4f693b55c3d2 ("tcp: implement coalescing on backlog queue") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: Richard Purdie <richard.purdie@linuxfoundation.org> Reported-by: Bruno Prémont <bonbons@sysophe.eu> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
A recent commit returns an error if icmp is used as the ip-proto for
IPv6 fib rules. Update fib_rule_tests to send ipv6-icmp instead of icmp.
Fixes: 5e1a99eae8499 ("ipv4: Add ICMPv6 support when parse route ipproto") Signed-off-by: David Ahern <dsahern@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
Packet send checks that msg_name is at least sizeof sockaddr_ll.
Packet recv must return at least this length, so that its output
can be passed unmodified to packet send.
This ceased to be true since adding support for lladdr longer than
sll_addr. Since, the return value uses true address length.
Always return at least sizeof sockaddr_ll, even if address length
is shorter. Zero the padding bytes.
Change v1->v2: do not overwrite zeroed padding again. use copy_len.
Fixes: 0fb375fb9b93 ("[AF_PACKET]: Allow for > 8 byte hardware addresses.") Suggested-by: David Laight <David.Laight@aculab.com> Signed-off-by: Willem de Bruijn <willemb@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Connor Kuehl <connor.kuehl@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
projects / mirror_ubuntu-disco-kernel.git / log