On my Debian gcc -m32 is hopelessly broken.
Using clang fixes at least arch 386.
Arch arm is still broken b/c clang does not like
some of kernel arm inline assemly constraints.
It's /dev/watch_queue not /dev/fd_watch_queue.
Don't know if it was renamed, or alwys wrong, but does not matter much,
now it's /dev/watch_queue.
Also attach v4l ioctls to /dev/swradio,radio,vbi,cec.
It seems that _some_ of them are applicable to these devices as well.
If we have way too many programs to send (more than 100000),
cap total number to 100000 and give up sending all.
Otherwise new managers will never chew all this on a busy hub.
They should have been detected by "same binary" logic.
But the problem is that we may use different compilers
for different commits and they switch exactly at release commits.
So we can build the release with a differnet compiler than the
rest of commits and then obviously it won't be "same binary".
Detect release commits separately.
Update #1271
3/4 added tests currently fail.
The problem is that we don't collect results
from initial bisection range detection.
As the result we won't detect "same binary"
for release and first commit in a release.
Update #1271
Linux v5.5 extends struct clone_args for clone3() by two additional
parameters:
@set_tid: Pointer to an array of type *pid_t. The size
of the array is defined using @set_tid_size.
This array is used to select PIDs/TIDs for
newly created processes. The first element in
this defines the PID in the most nested PID
namespace. Each additional element in the array
defines the PID in the parent PID namespace of
the original PID namespace. If the array has
less entries than the number of currently
nested PID namespaces only the PIDs in the
corresponding namespaces are set.
@set_tid_size: This defines the size of the array referenced
in @set_tid. This cannot be larger than the
kernel's limit of nested PID namespaces.
Cc: Adrian Reber <areber@redhat.com>
Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>
gVisor doesn't support netdevsim and NETLINK_GENERIC and
without this fix, sys-executor always fails:
E openat(AT_FDCWD, /sys/bus/netdevsim/new_device, ...)
X openat(AT_FDCWD, /sys/bus/netdevsim/new_device, ...) = 0x0 errno=2
E socket(AF_NETLINK, SOCK_RAW|0x0, NETLINK_GENERIC)
X socket(AF_NETLINK, SOCK_RAW|0x0, NETLINK_GENERIC) = 0x0 errno=93
E write(0x2 host:[3], "socket(AF_NETLINK) failed\n", 0x1a)
X write(0x2 host:[3], ..., 0x1a) = 0x1a (2.767µs)
E write(0x2 host:[3], " (errno 93)\n", 0xc)
X write(0x2 host:[3], ..., 0xc) = 0xc (2.729µs)
E exit_group(0x43)
X exit_group(0x43) = 0x0 (900ns)
Cc: Jiri Pirko <jiri@mellanox.com>
Fixes: f350e2dc1f ("executor: rename devlink port netdevices to defined names")
Signed-off-by: Andrei Vagin <avagin@google.com>
Add new crypto algorithms from linux-next (future v5.5):
- BLAKE2b and BLAKE2s
- "-arm" implementations of ChaCha and XChaCha
- "-arm" and "-neon" implementations of Poly1305
Regenerate consts on linux-next next-20191125.
DEVLINK_ATTR_NETNS_* consts were chifted by 1.
__BPF_FUNC_MAX_ID increased as usual.
CRYPTO_ALG_TYPE_BLKCIPHER, CRYPTO_ALG_TYPE_ABLKCIPHER were removed,
replace them with CRYPTO_ALG_TYPE_SKCIPHER.
These used to be required to build Fuchsia's Go toolchain to build
syz-fuzzer to run on Fuchsia. However, Fuchsia now uses HostFuzzer
mode, so only syz-executor (written in C) needs to run on Fuchsia; all
Go code runs on the Host OS (currently Linux).
stdbool.h is required by initialize_devlink_ports(): true define.
fcntl.h is required by initialize_devlink_pci(): O_RDONLY define.
Signed-off-by: Denis Efremov <efremov@linux.com>
An upcoming patch for Linux will change the error reporting pattern for
general protection faults such that the colon doesn't necessarily come
immediately after the string "general protection fault" (see
https://lore.kernel.org/lkml/20191118142144.GC6363@zn.tnic/).
Change the pattern in syzkaller before that happens.
Note that this is not necessarily the final format; in particular, the
ordering of the KASAN note and the "general protection fault" line might
swap.
