Alan Cox [Sat, 20 Feb 2016 20:59:33 +0000 (20:59 +0000)]
syscall_net: note another area to look at later
Alan Cox [Sat, 20 Feb 2016 20:33:15 +0000 (20:33 +0000)]
z80pack: switch over to net_native so we can begin debugging it
Alan Cox [Sat, 20 Feb 2016 20:23:34 +0000 (20:23 +0000)]
net_native: note a future bug risk
Alan Cox [Sat, 20 Feb 2016 19:34:54 +0000 (19:34 +0000)]
devsys: Make the devsys code identing consistent
Alan Cox [Sat, 20 Feb 2016 19:34:25 +0000 (19:34 +0000)]
devsys: wire up the network hooks
Alan Cox [Sat, 20 Feb 2016 19:33:41 +0000 (19:33 +0000)]
inode: remove dead socket checks
Alan Cox [Sat, 20 Feb 2016 19:33:04 +0000 (19:33 +0000)]
net: Further tidying up
Remove various obsolete bogisities as well
Alan Cox [Sat, 20 Feb 2016 18:45:20 +0000 (18:45 +0000)]
net: tidy up stuff we don't want in kernel
Alan Cox [Sat, 20 Feb 2016 18:44:47 +0000 (18:44 +0000)]
net_native: put header in the right spot
Alan Cox [Sat, 20 Feb 2016 18:41:37 +0000 (18:41 +0000)]
ps: update to reflect kernel state removals
EtchedPixels [Sat, 20 Feb 2016 18:09:12 +0000 (18:09 +0000)]
Merge pull request #318 from retrodeluxe/fuzix-msx2
msx2: fix build dependency
geijoenr [Sat, 20 Feb 2016 17:20:02 +0000 (17:20 +0000)]
msx2: fix build dependency
bootrom.bin is generated along with fuzix.ihx
Alan Cox [Fri, 19 Feb 2016 23:02:01 +0000 (23:02 +0000)]
networking: first cut at the native TCP/IP stack interface
This is far from finished or even building yet, but now has the right sort
of form. For the user space side the current plan is to plug it into
something like uIP, LwIP or Harry Kalogirou's ktcp for ELKS.
In theory the API is such that beyond "what fits" it shouldn't matter whether
we bind a teeny/tiny stack like A J Kroll's microcontroller TCP/IP or something
a bit more refined.
Alan Cox [Fri, 19 Feb 2016 23:01:35 +0000 (23:01 +0000)]
TODO: update
Alan Cox [Wed, 17 Feb 2016 21:47:47 +0000 (21:47 +0000)]
exec: add core dump support
This adds some minimalistic core dump support to the 16bit code base.
Alan Cox [Wed, 17 Feb 2016 18:13:28 +0000 (18:13 +0000)]
doexit: tidy up types
Alan Cox [Wed, 17 Feb 2016 18:08:21 +0000 (18:08 +0000)]
process: clean up signal handling logic and finish off SIGSTOP/CONT
Alan Cox [Wed, 17 Feb 2016 17:33:11 +0000 (17:33 +0000)]
README.md: small update
Alan Cox [Wed, 17 Feb 2016 17:32:34 +0000 (17:32 +0000)]
README: update kernel README a bit
Alan Cox [Sat, 13 Feb 2016 21:36:39 +0000 (21:36 +0000)]
net: add the missing handling for conflicting addresses and for lookups
Alan Cox [Sat, 13 Feb 2016 21:21:57 +0000 (21:21 +0000)]
devsys: low protection makes no sense so remove
Alan Cox [Sat, 13 Feb 2016 21:17:40 +0000 (21:17 +0000)]
flat_mem: start adjusting atarist as a prototype user of flat_mem
Alan Cox [Sat, 13 Feb 2016 20:47:54 +0000 (20:47 +0000)]
flat_mem: more tidying up
Alan Cox [Sat, 13 Feb 2016 20:34:15 +0000 (20:34 +0000)]
waitpid: add more job control bits
Our SIGSTOP handling still isn't quite right for the case we get a stop while
already sleeping.
Alan Cox [Sat, 13 Feb 2016 20:08:54 +0000 (20:08 +0000)]
flat_mem: bring into line with prototypes
Alan Cox [Sat, 13 Feb 2016 20:08:35 +0000 (20:08 +0000)]
memory: bring more files into line with new prototypes
Alan Cox [Sat, 13 Feb 2016 20:08:02 +0000 (20:08 +0000)]
memory: fix some of the types for 32bit address spacing
Alan Cox [Thu, 11 Feb 2016 20:00:26 +0000 (20:00 +0000)]
usermem: add MISALIGNED
Word sized misaligned user space memory accesses can now be faulted if
desired by the platform
Alan Cox [Thu, 11 Feb 2016 20:00:06 +0000 (20:00 +0000)]
net_ip: note an oversight
Alan Cox [Thu, 11 Feb 2016 19:59:40 +0000 (19:59 +0000)]
flat: fix stupid bug I noticed
May well kick this code out shortly anyway
Alan Cox [Thu, 11 Feb 2016 19:56:40 +0000 (19:56 +0000)]
flat_mem: import the code based upon MAPUX
Why invent when you can recycle 8)
This hopefully forms the basis of the bits we need for supporting devices
with a flat memory space, while handling fork and memory compaction and other
horrors a little bit.
It's a starting point. There are lots of things that would improve the
performance, and it can't handle multiple memory ranges yet.
EtchedPixels [Thu, 4 Feb 2016 19:59:07 +0000 (19:59 +0000)]
Merge pull request #316 from beretta42/section
coco3: coco3.s: fix code sections as per Issue: #310.
Brett Gordon [Thu, 4 Feb 2016 19:16:51 +0000 (14:16 -0500)]
coco3: coco3.s: fix code sections as per Issue: #310.
Alan Cox [Mon, 1 Feb 2016 15:11:52 +0000 (15:11 +0000)]
init: add reload support
Alan Cox [Mon, 1 Feb 2016 15:09:15 +0000 (15:09 +0000)]
syscalls: assume the shared header will be placed in include/sys
EtchedPixels [Mon, 25 Jan 2016 20:29:37 +0000 (20:29 +0000)]
Merge pull request #313 from davidgiven/alignement1
Fix alignment and sizing issues
David Given [Mon, 25 Jan 2016 20:14:53 +0000 (21:14 +0100)]
Fix typo.
EtchedPixels [Sun, 24 Jan 2016 15:39:46 +0000 (15:39 +0000)]
Merge pull request #315 from davidgiven/priorities
Data corruption in waitpid()
David Given [Sat, 23 Jan 2016 23:51:29 +0000 (00:51 +0100)]
Fix data corruption issue.
David Given [Thu, 21 Jan 2016 20:37:35 +0000 (21:37 +0100)]
Move struct _uzistat into its own structure, shared between the kernel and
userspace. stcpy() now copies this safely (avoiding compiler packing gotchas).
David Given [Thu, 21 Jan 2016 19:42:31 +0000 (20:42 +0100)]
Undo binary-breaking struct _uzistat rearrangement.
Alan Cox [Thu, 21 Jan 2016 11:55:13 +0000 (11:55 +0000)]
Merge branch 'master' of https://github.com/EtchedPixels/FUZIX
Alan Cox [Thu, 21 Jan 2016 11:53:52 +0000 (11:53 +0000)]
syscall_proc: document that dofork might do two things
On banked platforms it'll probably return twice for now anyway. On some others
the child will return to user space directly on a different kernel stack.
EtchedPixels [Thu, 21 Jan 2016 11:41:15 +0000 (11:41 +0000)]
Merge pull request #314 from davidgiven/malloc
Rewrite of kernel memory allocator
David Given [Tue, 19 Jan 2016 22:45:54 +0000 (23:45 +0100)]
ints aren't necessarily two bytes long!
David Given [Wed, 20 Jan 2016 22:32:12 +0000 (23:32 +0100)]
Fix alignment issues primarily with the malloc() alignment payload.
David Given [Tue, 19 Jan 2016 20:46:44 +0000 (21:46 +0100)]
Fix 32-bit alignment issues.
David Given [Wed, 20 Jan 2016 22:25:39 +0000 (23:25 +0100)]
Don't use the 16-bit definitions if the compiler's provided 32-bit ones.
Alan Cox [Sat, 16 Jan 2016 19:35:50 +0000 (19:35 +0000)]
net: knock out some remaining bugs and print the AT command properly
We now pass the "sufficiently rigged demo" test case.
Alan Cox [Sat, 16 Jan 2016 15:27:50 +0000 (15:27 +0000)]
syscall_net: hooks for reading from a socket
This needs some rethinking for native sockets but we'll get there when we
deal with those.
Alan Cox [Sat, 16 Jan 2016 15:26:16 +0000 (15:26 +0000)]
grp.h: setgroups is a syscall it shouldn't be here
Fixes the Dragon NX build failure. Reported by Tony Jewell
Alan Cox [Wed, 13 Jan 2016 18:39:55 +0000 (18:39 +0000)]
cpu-z80: add ntohl
Alan Cox [Wed, 13 Jan 2016 18:38:54 +0000 (18:38 +0000)]
net_at: Loopback doesn't work on this interface
Alan Cox [Wed, 13 Jan 2016 18:37:54 +0000 (18:37 +0000)]
syscall_net: fix write, local addressing
At this point our socket syscall side can step through everything setting up
our socket and writing to it. Read needs reworking.
Alan Cox [Wed, 13 Jan 2016 18:37:38 +0000 (18:37 +0000)]
net: add ECONNRESET
Alan Cox [Tue, 12 Jan 2016 23:56:32 +0000 (23:56 +0000)]
net_at: knock out a few more bugs and missing bits
We can now make a connection, and close it again. Error handling doesn't seem
quite right but not sure of the root cause yet.
Alan Cox [Tue, 12 Jan 2016 23:38:12 +0000 (23:38 +0000)]
net: fix more bugs
close() now works
connect() correctly calls into the net implementation but net_at hangs
Alan Cox [Tue, 12 Jan 2016 23:21:21 +0000 (23:21 +0000)]
syscall: getsockaddrs is a 3 argument function
Alan Cox [Tue, 12 Jan 2016 23:20:41 +0000 (23:20 +0000)]
libc: syscall Makefile
Alan Cox [Tue, 12 Jan 2016 23:19:51 +0000 (23:19 +0000)]
syscall_net: Fix bugs in socket() call
At this point we can actually create and close sockets. Not yet tried doing
anything with them!
Alan Cox [Tue, 12 Jan 2016 23:19:19 +0000 (23:19 +0000)]
libc: add syscall bindings for network layer
Alan Cox [Tue, 12 Jan 2016 23:18:36 +0000 (23:18 +0000)]
net: socket types are zero based
Alan Cox [Tue, 12 Jan 2016 23:18:12 +0000 (23:18 +0000)]
kernel: Finish updating syscall names and fix a wrong error define
Alan Cox [Tue, 12 Jan 2016 23:17:56 +0000 (23:17 +0000)]
net: Must move the socket into unconnected
Alan Cox [Tue, 12 Jan 2016 00:00:02 +0000 (00:00 +0000)]
liberror: fix missed entry
Alan Cox [Mon, 11 Jan 2016 19:23:08 +0000 (19:23 +0000)]
libc: synchronize error codes with the kernel
Alan Cox [Mon, 11 Jan 2016 19:09:31 +0000 (19:09 +0000)]
net_at: fix stupid bug
Alan Cox [Mon, 11 Jan 2016 19:07:12 +0000 (19:07 +0000)]
net_at: header for silly stack
Alan Cox [Mon, 11 Jan 2016 19:06:40 +0000 (19:06 +0000)]
z80pack: add this lot to the build
Alan Cox [Mon, 11 Jan 2016 19:05:48 +0000 (19:05 +0000)]
z80pack: add a network socket simulation helper
This just uses the 3rd network virtual serial. There's a small tweak to the
cpmsim source also needed to report bit 2 as "carrier"
Alan Cox [Mon, 11 Jan 2016 19:04:08 +0000 (19:04 +0000)]
net_at: silly test stack
The first bits of a "network" stack which implements a single offload socket
driven via AT commands. It's easier to debug this and then tackle stuff like
the WizNet5200 offloads and perhaps Drivewire for offload stacks.
Alan Cox [Mon, 11 Jan 2016 19:03:40 +0000 (19:03 +0000)]
net: shuffle things around and add syscall to protocol stack bindings
Alan Cox [Mon, 11 Jan 2016 19:01:55 +0000 (19:01 +0000)]
kernel: a couple of error numbers we'll need for networking
Alan Cox [Mon, 11 Jan 2016 19:01:41 +0000 (19:01 +0000)]
cpu: add a standardised BIG_ENDIAN define
David Given [Sun, 10 Jan 2016 18:18:54 +0000 (19:18 +0100)]
Rewrote from scratch, as the old version had show-stopping bugs (it never
correctly added the initial memory pool!) and I couldn't understand the
code.
Alan Cox [Sun, 10 Jan 2016 16:23:58 +0000 (16:23 +0000)]
net: hook together all the networking kernel bits into z80pack
This doesn't give us anything remotely useful yet but means it's possible
to start testing out the syscall paths a bit and seeing what is needed
to finish the job off
Alan Cox [Sun, 10 Jan 2016 12:34:22 +0000 (12:34 +0000)]
makeversion: pass the level 2 flag via uname
Alan Cox [Sun, 10 Jan 2016 12:33:31 +0000 (12:33 +0000)]
ps: adaptions for level 2 handling
Actually more generally fix it to be far less fragile about struct sizes
Alan Cox [Sat, 9 Jan 2016 21:18:23 +0000 (21:18 +0000)]
z80fixedbank: remove old debug
Alan Cox [Sat, 9 Jan 2016 21:15:22 +0000 (21:15 +0000)]
z80pack: fix wakeup on console 2 and console 3
Fixes the "must telnet before starting fuzix" bug
Alan Cox [Sat, 9 Jan 2016 21:00:32 +0000 (21:00 +0000)]
tty: Fix O_NOCTTY handling
Finally we get the right process groups in the right places and init does
what we expect
Alan Cox [Sat, 9 Jan 2016 20:45:01 +0000 (20:45 +0000)]
init: don't acquire a controlling tty
We want getty to be controlling /dev/tty1 so it that the shell can do job
control on it once that all works.
Alan Cox [Sat, 9 Jan 2016 20:44:46 +0000 (20:44 +0000)]
level2: update headers to match job control changes
Alan Cox [Sat, 9 Jan 2016 20:44:29 +0000 (20:44 +0000)]
process: update signal handling
SIGTTIN/TTOU are not default fatal
Alan Cox [Sat, 9 Jan 2016 20:44:14 +0000 (20:44 +0000)]
tty: job control needs minor passing
Alan Cox [Sat, 9 Jan 2016 20:43:33 +0000 (20:43 +0000)]
level2: clear the tty on a setsid
Alan Cox [Sat, 9 Jan 2016 20:43:08 +0000 (20:43 +0000)]
start: Make sure we don't acquire a controlling tty before running init
Alan Cox [Sat, 9 Jan 2016 20:42:41 +0000 (20:42 +0000)]
level2: fix various job control logics
Alan Cox [Sat, 9 Jan 2016 20:41:24 +0000 (20:41 +0000)]
z80pack: bring the tty layer into compliance with current code
Alan Cox [Sat, 9 Jan 2016 19:38:31 +0000 (19:38 +0000)]
level2: fix first couple of bits
Alan Cox [Sat, 9 Jan 2016 18:30:48 +0000 (18:30 +0000)]
libs: link in opendir_r/closedir_r on Z80
Alan Cox [Sat, 9 Jan 2016 16:46:40 +0000 (16:46 +0000)]
Makefile: hook in the various level 2 files
Alan Cox [Sat, 9 Jan 2016 16:46:00 +0000 (16:46 +0000)]
z80pack: make this a level2 platform so we can begin test/debug
Alan Cox [Sat, 9 Jan 2016 16:45:24 +0000 (16:45 +0000)]
level2: update the remaining bits to get a level2 library
Alan Cox [Sat, 9 Jan 2016 16:44:11 +0000 (16:44 +0000)]
syscall_name: use the right count
With this the new syscall tool can build and the libc with the level 2 feature
set
Alan Cox [Sat, 9 Jan 2016 15:27:13 +0000 (15:27 +0000)]
level2: plumb together all the level2 work and make it actually build
Alan Cox [Tue, 29 Dec 2015 23:30:23 +0000 (23:30 +0000)]
level2: add hooks to allow level 2 platforms to provide their own path buffers
(If you go over 512 bytes the standard tmpbuf/brelse solution won't work, so
for bigger systems malloc may be preferred)
Alan Cox [Tue, 29 Dec 2015 23:24:44 +0000 (23:24 +0000)]
process: fix dumb error in last commit
Alan Cox [Tue, 29 Dec 2015 23:18:26 +0000 (23:18 +0000)]
process: add initial SIGCHLD handling
Alan Cox [Tue, 29 Dec 2015 22:53:54 +0000 (22:53 +0000)]
tty: bring comments up to date