Expect that drivers call into the network stack with the net epoch
entered. This has already been the fact since early 2020. The net
interrupts, that are marked with INTR_TYPE_NET, were entering epoch
since 511d1afb6b. For the taskqueues there is NET_TASK_INIT() and
all drivers that were known back in 2020 we marked with it in
6c3e93cb5a. However in e87c494015 we took conservative approach
and preferred to opt-in rather than opt-out for the epoch.
This change not only reverts e87c494015 but adds a safety belt to
avoid panicing with INVARIANTS if there is a missed driver. With
INVARIANTS we will run in_epoch() check, print a warning and enter
the net epoch. A driver that prints can be quickly fixed with the
IFF_NEEDSEPOCH flag, but better be augmented to properly enter the
epoch itself.
Note on TCP LRO: it is a backdoor to enter the TCP stack bypassing
some layers of net stack, ignoring either old IFF_KNOWSEPOCH or the
new IFF_NEEDSEPOCH. But the tcp_lro_flush_all() asserts the presence
of network epoch. Indeed, all NIC drivers that support LRO already
provide the epoch, either with help of INTR_TYPE_NET or just running
NET_EPOCH_ENTER() in their code.
Reviewed by: zlei, gallatin, erj
Differential Revision: https://reviews.freebsd.org/D39510
Handling of the CLOSE_RANGE_UNSHARE flag is not implemented due to
difference in fd unsharing mechanism in the Linux and FreeBSD.
Reviewed by: mjg
Differential revision: https://reviews.freebsd.org/D39398
MFC after: 2 weeks
For further use in linux_ioctl_socket add bsd_to_linux_ifflags() which
accepts FreeBSD interface flags and translate it to Linux if flags.
Reviewed by: melifaro
Differential Revision: https://reviews.freebsd.org/D38873
Split the linux_ioctl_socket() function on two counterparts, where
the linux_ioctl_socket_ifreq() intended to use in a code path which
requires the struct ifreq manipulation, i.e., translating in/out
values of the struct, while the linux_ioctl_socket() function is left
as is, it calls sys_ioctl() without touching in/out values.
Due to structures ifreq, sockaddr difference between FreeBSD and Linux
the linux_ioctl_socket_ifreq() calls kern_ioctl() directly, converting
in and out values to FreeBSD and to Linux accordingly.
Finally, modify the ifname_linux_to_bsd() to return error code, not
an unstable reference to the interface.
Reviewed by: melifaro
Differential Revision: https://reviews.freebsd.org/D38794
For now we are using mixed names to access struct ifreq members, some
of Linux (ifr_name, ifr_ifindex), others of FreeBSD. To avoid conflicts
switch to use FreeBSD names.
Reviewed by: jhibbits
Differential Revision: https://reviews.freebsd.org/D38792
Enter the net epoch when traversing a list of interfaces. For that
split the ifname_linux_to_bsd() function on two counterparts, where
the ifname_linux_to_ifp() intended to use in epoch, while the
ifname_linux_to_bsd() intended to be a self-contained.
Until the linux_ioctl_coket() function is refactored, the
ifname_linux_to_bsd() temporarily returns interface outside
of the net epoch.
Reviewed by: melifaro
Differential Revision: https://reviews.freebsd.org/D38790
Migrate linux and linprocfs to use the IfAPI interfaces instead of
direct ifnet accesses.
The code initially writed by jhibbits@, and adapted by me to 3ab3c9c2.
Reviewed by: jhibbits
Differential Revision: https://reviews.freebsd.org/D38735
Netlink should translate a FreeBSD interface name to a Linux
interface name.
Reviewed by: melifaro
Differential Revision: https://reviews.freebsd.org/D38715
MFC after: 3 days
We have some amount of interface names translation functions which are
differs by bugs implementation. Consolidates it in a one place.
Fixup loopback interface names translation and use ifnet methods and
accessors, where possible.
Reviewed by: melifaro
Differential Revision: https://reviews.freebsd.org/D38714
MFC after: 3 days
X-MFC with: 32fdc75fe7
To avoid confusing people, rename linux_timer.h to linux_time.h,
as linux_timer.c is the implementation of timer syscalls only,
while linux_time.c contains implementation of all stuff declared
in linux_time.h.
MFC after: 2 weeks
MI linux.[c|h] are the module independent in terms of the Linux emulation
layer (ie, intended for both ISA - 32 & 64 bit), analogue of MD linux.h.
There must be a code here that cannot be placed into the corresponding by
common sense MI source and header files, i.e., code is machine independent,
but ISA dependent.
For the use_real_names knob, the code must be placed into the
linux_socket.[c|h], however linux_socket is ISA dependent.
MFC after: 2 weeks
This obsolete system call is not supported by glibc. In ancient libc
versions (before glibc 2.0), uselib() was used to load the shared
libraries with names found in an array of names in the binary.
On Linux, since 3.15, this system call is available only when
the kernel is configured with the CONFIG_USELIB option.
It doesn't look like anyone needs this syscall for others Linuxulators,
so move it to the corresponding MD Linuxulator.
MFC after: 2 weeks
Include sys/sysent.h directly where it needed. The linux_util.h included
in a most source files of the Linuxulator, avoid collecting a rarely used
includes here.
MFC after: 2 weeks
Include vm headers directly where they needed. The linux_util.h included
in a most source files of the Linuxulator, avoid collecting a rarely used
includes here.
MFC after: 2 weeks
COMPAT_LINUX32 option is defined for case when building 32-bit Linuxulator
for the 64-bit host. Usage of __ELF_WORD_SIZE is wrong here as it is equal to 32
on i386 too.
MFC after: 1 week
Rework the routines to convert a native statfs structure (with fixed-size 64-bit
counters) to a Linux statfs structure (with long-sized counters) for 32-bit apps.
Instead of following Linux and return an EOVERFLOW error from statfs() family of
syscalls when actual fs stat value(s) are large enough to not fit into 32 bits,
apply scale logics used by FreeBSD to convert a 5.x statfs structure to a 4.x
statfs structure.
For more details see cc479dda.
Tested by: glebius
MFC after: 1 week
The system call returns the head of the robust futex list. The list head is stored
in the location pointed to by the head argument. When copying data between address
spaces use proper head storage size as it depends on an emulated ABI.
PR: 267616
MFC after: 3 days
For some of these Clang produced a warning that "a function declaration
without a prototype is deprecated in all versions of C". In other cases
the function defintion used () which did not match the header
declaration, which used (void).
Sponsored by: The FreeBSD Foundation
Add the glue code to support netlink in Linuxolator.
linux_common(4) now depends on netlink(4).
All netlink protocol constants are consistent with the Linux version.
However, certain OS-specific constants such as AF_INET6, interface
flags or default routing table id, are different between FreeBSD and
Linux. Thus, it may be needed to rewrite some message parts or even
rewrite the whole message, adding or removing some TLVs. The core
netlink implementation code provides efficient rewriting callbacks
which Linuxolator now uses.
Reviewed by: dchagin
Differential Revision: https://reviews.freebsd.org/D36361
MFC after: 2 months
