In order to ensure a specific owner and mode when the local socket
filesystem endpoint is instantiated, we need to be able to call
fchmod() and fchown() on a socket fd between socket() and bind().
This is because until we call bind(), there is no filesystem inode
for the socket yet.
We now have these API's in <Kernel/Random.h>:
- get_fast_random_bytes(u8* buffer, size_t buffer_size)
- get_good_random_bytes(u8* buffer, size_t buffer_size)
- get_fast_random<T>()
- get_good_random<T>()
Internally they both use x86 RDRAND if available, otherwise they fall
back to the same LCG we had in RandomDevice all along.
The main purpose of this patch is to give kernel code a way to better
express its needs for random data.
Randomness is something that will require a lot more work, but this is
hopefully a step in the right direction.
The new PCI subsystem is initialized during runtime.
PCI::Initializer is supposed to be called during early boot, to
perform a few tests, and initialize the proper configuration space
access mechanism. Kernel boot parameters can be specified by a user to
determine what tests will occur, to aid debugging on problematic
machines.
After that, PCI::Initializer should be dismissed.
PCI::IOAccess is a class that is derived from PCI::Access
class and implements PCI configuration space access mechanism via x86
IO ports.
PCI::MMIOAccess is a class that is derived from PCI::Access
and implements PCI configurtaion space access mechanism via memory
access.
The new PCI subsystem also supports determination of IO/MMIO space
needed by a device by checking a given BAR.
In addition, Every device or component that use the PCI subsystem has
changed to match the last changes.
The majority of the time in NetworkTask was being spent in allocating
and deallocating KBuffers for each incoming packet.
We'll now keep up to 100 buffers around and reuse them for new packets
if the next incoming packet fits in an old buffer. This is pretty
naively implemented but definitely cuts down on time spent here.
Since stream sockets don't actually need to deliver packets-at-a-time
data in recvfrom(), they can just buffer the payload bytes instead.
This avoids keeping one KBuffer per incoming packet in the receive
queue, which was a big performance issue in ProtocolServer.
This code is definitely not perfect and is something we should keep
improving over time.
Using int was a mistake. This patch changes String, StringImpl,
StringView and StringBuilder to use size_t instead of int for lengths.
Obviously a lot of code needs to change as a result of this.
This patch adds these I/O counters to each thread:
- (Inode) file read bytes
- (Inode) file write bytes
- Unix socket read bytes
- Unix socket write bytes
- IPv4 socket read bytes
- IPv4 socket write bytes
These are then exposed in /proc/all and seen in SystemMonitor.
This defaults to 1500 for all adapters, but LoopbackAdapter increases
it to 65536 on construction.
If an IPv4 packet is larger than the MTU, we'll need to break it into
smaller fragments before transmitting it. This part is a FIXME. :^)
After a socket has disconnected, we shouldn't return -EAGAIN. Instead
we should allow userspace to read/recvfrom the socket until its packet
queue has been exhausted.
At that point, we now return 0, signalling EOF.
It might be even better to start returning -ENOTCONN after signalling
EOF once. I'm not sure how that should work, needs looking into.
This reverts commit 1cca5142af.
This appears to be causing intermittent triple-faults and I don't know
why yet, so I'll just revert it to keep the tree in decent shape.
Background: DoubleBuffer is a handy buffer class in the kernel that
allows you to keep writing to it from the "outside" while the "inside"
reads from it. It's used for things like LocalSocket and PTY's.
Internally, it has a read buffer and a write buffer, but the two will
swap places when the read buffer is exhausted (by reading from it.)
Before this patch, it was internally implemented as two Vector<u8>
that we would swap between when the reader side had exhausted the data
in the read buffer. Now instead we preallocate a large KBuffer (64KB*2)
on DoubleBuffer construction and use that throughout its lifetime.
This removes all the kmalloc heap traffic caused by DoubleBuffers :^)
Make sure we don't move accepted sockets to the Completed setup state
until we've actually constructed a FileDescription for them.
This is important, since this state transition will trigger connect()
to unblock on the client side, and the client may try writing to the
socket right away.
This makes DNS lookups way more reliable since we don't just fail to
write() right after connect()ing to LookupServer sometimes. :^)
This was causing connect() to unblock immediately for local sockets,
since that's exactly what ConnectBlocker checks for.
Instead, just move to SetupState::Completed when it's accept()ed.
If we can't already read when we enter recvfrom() on a LocalSocket,
we'll now block the current thread until we can.
Also added a buffer_for(FileDescription&) helper so that the client
and server can share some of the code. :^)
Made getsockopt() and setsockopt() virtual so we can handle them in the
various Socket subclasses. The subclasses map kinda nicely to "levels".
This will allow us to implement things like "traceroute", although..
I spent some time trying to do that, but then hit a wall when it turned
out that the user-mode networking in QEMU doesn't preserve TTL in the
ICMP packets passing through.
This approach is a bit naiive - whenever we send a packet out, we
check to see if there are any other packets we should try to send.
This works well enough for a busy connection but not very well for a
quiet one. Ideally we would check for not-acked packets on some kind
of timer, and use the length of this not-acked list as feedback to
throttle the writes coming from userspace.
This allows us to take advantage of unsolicited ARP replies, such as
those that are emitted by many systems after their network interfaces
are enabled, or after their DHCP client sets their IP.
This also makes us a bit more vulnerable to ARP flooding, but we need
some kind of eviction strategy anyway, so we can deal with that later.
An incoming socket should only be considered connected after a
program has received it from accept(). Before that point, it's only
"half" open, and it might not ever actually be served to a program.
Socket::accept is where m_connected is correctly set.
