| Age | Commit message (Collapse) | Author |
|---|
|
triggered for a reply-to rule. It turns out this case has been using
uninitialised memory as if it were a valid pf pool.
As the rest of the function assumes a valid pool for now just return.
Problem reported by RD Thrush.
ok jung@ mikeb@
|
|
avoid a potential deadlock.
ok miod@, deraadt@
|
|
ok pirofti@
|
|
|
|
Eliminate the goto that I tripped on.
problem noted by Mark Latimer (mark.latimer (at) gmail.com)
ok miod@ millert@
|
|
Thanks to Patrick Wild for noticing that!
ok mpi@, deraadt@.
|
|
ok deraadt@
|
|
ok miod@, deraadt@
|
|
No binary change.
|
|
to advance in time. This is needed to properly recover, when
interrupts are blocked for too long. Fixes permanent distortion on MP
systems.
ok deraadt
|
|
faultbuf. But 1/ sr was only restored for machine check exceptions, and 2/ the
way it was saved was unsafe if interrupts were enabled, and could cause %r2
to be lost.
Discussing this with deraadt@ at the end of c2k15, this was probably needed
for the old VI boards which were the target of the original powerpc port,
came with a worse-than-Genesi openfirmware. Since then, machine check
exceptions have been unheard of; or, if they happen, they do not need the
status register to be restored.
ok mpi@ deraadt@
|
|
on some laptops. Mailing lists threads mention x220 and Asus machines.
Reverting for now.
|
|
simple and gross fashion, but the least impact on other parts...
|
|
stated here as I have been asked to be polite in this commit message.
ok deraadt@
|
|
ok reyk, with deraadt@'s permission
|
|
Originally from pelikan@, recent prodding from bluhm@ and jsg@.
ok pelikan@ millert@ bluhm@
|
|
- use simpler data structures (makes the code shorter).
- if multiple interrupts are missed, call audio(4) as many times as we
missed interrupts (unless the pointer wrapped, which this is
undetectable, but upper layer handles it)
- fetch the link fifos size in the interrupt handler, as, according to
the spec it may change dynamically.
- certain devices expose the last offset of the link fifo size instead
of the fifo size itself; workaround this by incrementing the
reported link fifo size if it's odd. Fixes recovery after missed
interrupts on asus eeepcs, maybe others.
ok deraadt, help from armani@
|
|
trying to build.
|
|
directly. Also protect non mp-safe functions while at it.
ok mpi@.
|
|
read and validate disklabels on GPT partitioned disks. Makes it much
more likely GPT partitioned disks with non-512-byte sectors will
work.
Part of the larger reverted GPT work.
ok deraadt@
|
|
thanks!
|
|
pointers to wrap, recover by detecting and compensating
for the missed interrupts. Fixes certain audio hangs on
MP machines.
with help from armani@, typos fixed by Alexey Suslikov
|
|
returned by AUDIO_GETxOFFS and AUDIO_xERROR ioctls.
ok deraadt
|
|
obvious idiom.
ok bluhm@ jsing@
|
|
|
|
tests reported an EMSGSIZE error although the sent message was not
too large. The way backpressure was implemented for unix domain
sockets confused the check in sosend().
Unix domain sockets append data only to the recv buffer. To report
the amount of content to the sender, the high watermark of the send
buffer was reduced. This happend for SOCK_STREAM and SOCK_SEQPACKET.
Sosend checks wether atomic chunks could ever fit into the send
buffer which is limited by the high watermark. This happens for
SOCK_DGRAM and SOCK_SEQPACKET. For SOCK_SEQPACKET the combination
of these mechanisms resulted in an EMSGSIZE error when the buffer
got filled. This also happened when space could be created by
reading from the other end in contradiction to the semantics of
EMSGSIZE.
Do not emulate a send buffer that has no space. It is better to
fill the buffer with fake data than to reduce its size. Thus the
high watermark always contains the real value. When disconnecting,
reset the counters. Otherwise the socket layer would try to flush
non existing data in the send buffer.
Tested by jeremy@ with a C program and the ruby tests.
OK markus@ jeremy@
|
|
divert->addr means that the rule has a divert-reply option. With
divert-to the divert->addr is always set. The divert-reply rules
should compare the original packet with the socket addresses. So
skip the step that compares the inp->inp_laddr with the empty
divert->addr.
OK mikeb@
|
|
packets with a different ICMP ID are sent over the same raw socket,
multiple states should be created. Put a similar check into
rip6_output() like in the IPv4 case.
OK mikeb@
|
|
information and for some generic CPUID/MSR-level initialization. This
changes the attach/match path a bit.
Hypervisor informatiom may include a version number, for example:
pvbus0 at mainbus0: Hyper-V 6.3.9600
OK mlarkin@
|
|
sendmsg(), and recvmsg(). For cmsghdr, the len, level, and type are always
shown, and for SOL_SOCKET,SCM_RIGHTS the fd numbers being passed are shown.
ok millert@ deraadt@
|
|
a coredump is happening. This improves behaviour while threaded.
ok semarie
|
|
|
|
non-blocking. This improves performance on systems that have more than
a few cores and where there is a heavy IPI load. Currently the only
place where coalescing must not happen is rendezvous processing, but
there invocations are serialized by the rendezvous mutex.
ok miod@
|
|
if TAME_ABORT is setted.
ok deraadt@
|
|
ok jsing@
|
|
ok deraadt@
|
|
problem noted by landry@
ok dlg@
|
|
a buf. Otherwise stale data might cause a successful I/O to be
seen as a failed I/O.
ok tedu@
|
|
behind a bridge, if the SRM didn't pick an interrupt line for them.
|
|
querying the 6845 registers, and preserve it if it is larger than 80x25 yet
plausible.
This is necessary to properly operate on the Alphabook 1 where the vga display
is running in a 100x37 textmode.
Other platforms are not affected.
|
|
|
|
|
|
multiply the block count by the block size to get the bytes count
because they won't wrap correctly.
|
|
try and queue it for the stack and cause a use after free.
found by maxime villard and brainy
|
|
Requested by deraadt@
|
|
Found the hardway by benoit@.
|
|
|
|
from martin natano
|
|
OK mlarkin@
|
|
Tested on VMware-GENERIC-i386.
OK mlarkin@
|
