Alan Cox [Tue, 25 Oct 2016 20:42:13 +0000 (21:42 +0100)]
pdp11: introduce base PDP11 code
Far from complete
Alan Cox [Tue, 25 Oct 2016 20:41:15 +0000 (21:41 +0100)]
v68: more work on bugs and signals
Alan Cox [Tue, 25 Oct 2016 20:40:26 +0000 (21:40 +0100)]
v68: more fixing
Alan Cox [Tue, 25 Oct 2016 20:39:59 +0000 (21:39 +0100)]
68000: further fixes
Alan Cox [Fri, 21 Oct 2016 16:51:53 +0000 (17:51 +0100)]
init: don't read NULL pointers
Needed for the 68K port as it causes a trap
Alan Cox [Thu, 20 Oct 2016 22:13:14 +0000 (23:13 +0100)]
68000: Switch to gcc 6.2.0
Alan Cox [Thu, 20 Oct 2016 22:13:00 +0000 (23:13 +0100)]
68000: switch to gcc 6.2.0
Alan Cox [Thu, 20 Oct 2016 22:12:45 +0000 (23:12 +0100)]
68000: switch to gcc 6.2.0
Alan Cox [Thu, 20 Oct 2016 22:12:37 +0000 (23:12 +0100)]
v68: update instructions
Alan Cox [Thu, 20 Oct 2016 22:12:26 +0000 (23:12 +0100)]
syscall_exec32: remove some debug
Alan Cox [Thu, 20 Oct 2016 22:04:16 +0000 (23:04 +0100)]
68000: synchronous signal traps
Alan Cox [Thu, 20 Oct 2016 18:04:21 +0000 (19:04 +0100)]
lowlevel-68000: fixes for signals
Synchronous signal traps don't yet work but this takes us one step closer
Alan Cox [Thu, 20 Oct 2016 18:04:09 +0000 (19:04 +0100)]
p68000: fix vdso copy
Alan Cox [Thu, 20 Oct 2016 17:00:50 +0000 (18:00 +0100)]
README.md: update for 68K
Alan Cox [Thu, 20 Oct 2016 16:50:59 +0000 (17:50 +0100)]
process: note a problem to revisit
Alan Cox [Thu, 20 Oct 2016 16:49:29 +0000 (17:49 +0100)]
exec32: fixes
Firstly botch the stack value because elf2flt defaults to 4K which is usually
insufficient as in our case it includes brk() space. Just add 32K for now.
Also fix the argument aligmnet.
Alan Cox [Thu, 20 Oct 2016 16:48:23 +0000 (17:48 +0100)]
68000: fix tricks up
With this we seem to have nice working switching
Alan Cox [Thu, 20 Oct 2016 16:48:03 +0000 (17:48 +0100)]
p68000: add vdso, turn off signals for the moment
Alan Cox [Thu, 20 Oct 2016 16:46:57 +0000 (17:46 +0100)]
68000: add a sort of vdso over the headers
This lets us change the trap# and the like if we make syscalls call the stub
handler, and also lets us put the signal unwinder somewhere useful
Experimenting for now
Alan Cox [Thu, 20 Oct 2016 16:46:23 +0000 (17:46 +0100)]
68000: Fix irqrestore handler and tidy some logging
Alan Cox [Thu, 20 Oct 2016 16:45:56 +0000 (17:45 +0100)]
v7: add a Makefile for the v7 games
Alan Cox [Thu, 20 Oct 2016 16:45:09 +0000 (17:45 +0100)]
68000: add MWC makefile
Alan Cox [Thu, 20 Oct 2016 16:44:35 +0000 (17:44 +0100)]
libs: 68000 startup files
Alan Cox [Thu, 20 Oct 2016 16:43:57 +0000 (17:43 +0100)]
elf2flt: remove signal experiment
Alan Cox [Thu, 20 Oct 2016 16:43:29 +0000 (17:43 +0100)]
syscall68000: Special fork handling
Alan Cox [Wed, 19 Oct 2016 22:06:35 +0000 (23:06 +0100)]
init: oops commit the right version
Alan Cox [Wed, 19 Oct 2016 20:05:26 +0000 (21:05 +0100)]
init: trivial error path fix
Alan Cox [Wed, 19 Oct 2016 19:59:20 +0000 (20:59 +0100)]
68000: make fork a special case
By defining _fork() as trashing all registers we can implement multiple udata
pointer based 68K platforms without having to save the CPU registers on syscall
entry. Instead the child does a direct supervisor a7 fixup and rte to user space
with d0/d1 correct. The rest is handled by the wrapper.
Now this isn't perfect and we probably need to adjust the rules a bit if
anyone is doing clever register resident relative code because if you take a
signal handler just at the wrong moment.
I guess as we have to save A5, you'll have to use A5 8) for Amiga style
resident code if we ever get that far.
Alan Cox [Wed, 19 Oct 2016 19:53:48 +0000 (20:53 +0100)]
68000: add all the pieces for the signal return path helper
Alan Cox [Wed, 19 Oct 2016 19:53:06 +0000 (20:53 +0100)]
syscall_exec32: Multiple fixes
With these in place it now seems to work
Alan Cox [Wed, 19 Oct 2016 19:51:31 +0000 (20:51 +0100)]
syscall_proc: 32bit clean
Alan Cox [Wed, 19 Oct 2016 19:50:39 +0000 (20:50 +0100)]
simple: Allow udata to be more than one sector
We need this for 68K
Alan Cox [Wed, 19 Oct 2016 19:50:11 +0000 (20:50 +0100)]
swap: 32bit clean
Alan Cox [Wed, 19 Oct 2016 19:49:30 +0000 (20:49 +0100)]
v68: clean up config.h
Alan Cox [Wed, 19 Oct 2016 19:49:06 +0000 (20:49 +0100)]
v68: task switching
All the bits needed for pure swap
Alan Cox [Wed, 19 Oct 2016 19:48:19 +0000 (20:48 +0100)]
v68: fix up the low level code, add pre-emption and signals
Alan Cox [Wed, 19 Oct 2016 19:48:03 +0000 (20:48 +0100)]
v68: remove more fluff
Alan Cox [Wed, 19 Oct 2016 19:47:00 +0000 (20:47 +0100)]
68000: first cut at signal handling
This needs a fixed position user space helper which will follow in the lib
patches
Alan Cox [Wed, 19 Oct 2016 19:46:26 +0000 (20:46 +0100)]
v68: remove un-needed fluff
Alan Cox [Wed, 19 Oct 2016 19:46:00 +0000 (20:46 +0100)]
68000: for 32bit platforms we need a codebase field
Alan Cox [Wed, 19 Oct 2016 19:45:37 +0000 (20:45 +0100)]
tty: Use ugeti/uputi for int * ioctls
Alan Cox [Wed, 19 Oct 2016 19:44:56 +0000 (20:44 +0100)]
cpu: introduce uputi/ugeti
There are syscalls that use an int *, and also ioctls. For those we can't
use uputw/getw as we were. Add new helper defines.
Alan Cox [Tue, 18 Oct 2016 22:06:33 +0000 (23:06 +0100)]
Merge branch 'master' of https://github.com/EtchedPixels/FUZIX
Alan Cox [Tue, 18 Oct 2016 22:02:24 +0000 (23:02 +0100)]
68000: elf2flt linker rules
Alan Cox [Tue, 18 Oct 2016 22:02:06 +0000 (23:02 +0100)]
68000: Library fixes
Alan Cox [Tue, 18 Oct 2016 22:01:43 +0000 (23:01 +0100)]
68000: Application build files
Alan Cox [Tue, 18 Oct 2016 21:59:52 +0000 (22:59 +0100)]
syscall_exec32: various fixes/hacks to get it to work for testing only
Alan Cox [Tue, 18 Oct 2016 21:59:40 +0000 (22:59 +0100)]
68000: build fixes
Alan Cox [Tue, 18 Oct 2016 21:59:12 +0000 (22:59 +0100)]
68000: set traps properly
Alan Cox [Tue, 18 Oct 2016 21:58:57 +0000 (22:58 +0100)]
68000: syscall path debug
Alan Cox [Wed, 12 Oct 2016 18:37:56 +0000 (19:37 +0100)]
utils: build for M68K
Alan Cox [Wed, 12 Oct 2016 18:37:31 +0000 (19:37 +0100)]
sync: fix compile error
sync() is void and now correctly defined so return sync() fails.
EtchedPixels [Wed, 12 Oct 2016 10:33:29 +0000 (11:33 +0100)]
Merge pull request #444 from beretta42/keyboard
coco3: define KEY_ROWS to compile vt ioctls
EtchedPixels [Wed, 12 Oct 2016 10:29:34 +0000 (11:29 +0100)]
Merge pull request #443 from beretta42/sync
sync: gcc complains about sync being void
Brett Gordon [Tue, 11 Oct 2016 14:23:42 +0000 (10:23 -0400)]
coco3: define KEY_ROWS to compile vt ioctls
Brett Gordon [Tue, 11 Oct 2016 14:16:14 +0000 (10:16 -0400)]
sync: gcc complains about sync being void
Alan Cox [Tue, 11 Oct 2016 12:00:18 +0000 (13:00 +0100)]
Merge branch 'master' of https://github.com/EtchedPixels/FUZIX
Alan Cox [Tue, 11 Oct 2016 12:00:01 +0000 (13:00 +0100)]
marksman: fix a small newline handling bug
Alan Cox [Tue, 11 Oct 2016 11:56:36 +0000 (12:56 +0100)]
68000: Add the syscalls and libraries
This is sufficient to build the C and supporting libraries plus the syscalls
for 68000. We can't load and run a 68K binary yet but this gets us closer as
we can now begin to build some test binaries.
EtchedPixels [Mon, 10 Oct 2016 20:07:23 +0000 (21:07 +0100)]
Merge pull request #440 from beretta42/bank_vid3
coco3: shuffle memory to let us actually use kernel memory past 0xc000
EtchedPixels [Mon, 10 Oct 2016 20:06:57 +0000 (21:06 +0100)]
Merge pull request #441 from beretta42/keyboard
simple util to set console key repeat rates
Brett Gordon [Sun, 9 Oct 2016 15:18:23 +0000 (11:18 -0400)]
kbdrate.c: simple sanity.
Brett Gordon [Sun, 9 Oct 2016 14:28:27 +0000 (10:28 -0400)]
vt.c: fix ioctl: flipped src/dest in uget
Brett Gordon [Sun, 9 Oct 2016 14:27:21 +0000 (10:27 -0400)]
add util to set vt's keyboard's rates
Brett Gordon [Sat, 8 Oct 2016 16:32:28 +0000 (12:32 -0400)]
coco3: make sure to reset mmu when done copying!
Brett Gordon [Sat, 8 Oct 2016 15:54:10 +0000 (11:54 -0400)]
coco3: bank 6 copy: use *correct* end ptr
Brett Gordon [Sat, 8 Oct 2016 15:01:43 +0000 (11:01 -0400)]
coco3: copy 8k bank at cpu $c000 to new bank freeing 0x6 for init
EtchedPixels [Mon, 3 Oct 2016 09:02:12 +0000 (10:02 +0100)]
Merge pull request #438 from beretta42/video_blip
coco3: init bug: skip scanning screen memory
EtchedPixels [Mon, 3 Oct 2016 09:01:36 +0000 (10:01 +0100)]
Merge pull request #437 from beretta42/video
coco3: redirect tty queueing
Brett Gordon [Sun, 2 Oct 2016 22:22:26 +0000 (18:22 -0400)]
coco3: whitespace
Brett Gordon [Sun, 2 Oct 2016 20:57:08 +0000 (16:57 -0400)]
coco3: init bug: skip scanning screen memory
Brett Gordon [Sun, 2 Oct 2016 20:24:52 +0000 (16:24 -0400)]
coco3: add prototypes for queue access functions
Brett Gordon [Sun, 2 Oct 2016 19:36:49 +0000 (15:36 -0400)]
coco3: use indirect queues
EtchedPixels [Sun, 25 Sep 2016 15:32:22 +0000 (16:32 +0100)]
Merge pull request #436 from
dfffffff/master
fixed compile error: added ssize_t, changed size_t to unsigned
dfffffff [Sun, 25 Sep 2016 12:02:07 +0000 (08:02 -0400)]
fixed compile error: added ssize_t, changed size_t to unsigned
EtchedPixels [Sat, 24 Sep 2016 19:59:57 +0000 (20:59 +0100)]
Merge pull request #435 from erkinalp/patch-3
cut: fix output
Erkin Alp Güney [Sat, 24 Sep 2016 15:57:27 +0000 (18:57 +0300)]
oops, one letter missing
Erkin Alp Güney [Sat, 24 Sep 2016 15:55:24 +0000 (18:55 +0300)]
fix output
replace printf() with just a string by a puts()
no need to go on formatting just to print unmodified string
Alan Cox [Sat, 24 Sep 2016 15:18:01 +0000 (16:18 +0100)]
syscalls: more manual pages
Alan Cox [Sat, 24 Sep 2016 15:16:24 +0000 (16:16 +0100)]
syscalls: sync is void
Alan Cox [Sat, 24 Sep 2016 13:19:01 +0000 (14:19 +0100)]
man2: add a couple more entries
One by one...
Alan Cox [Sat, 24 Sep 2016 13:18:09 +0000 (14:18 +0100)]
marksman: allow the use of \ on line ends for 'newline' suppression
Alan Cox [Sat, 24 Sep 2016 13:17:25 +0000 (14:17 +0100)]
read/write: error oversized requests
We could allow it but it's undefined behaviour in the standard and in some
of our driver code, so force a nice clear error.
EtchedPixels [Fri, 23 Sep 2016 16:34:03 +0000 (17:34 +0100)]
Merge pull request #433 from
dfffffff/master
fixed compile error
oops
dfffffff [Fri, 23 Sep 2016 15:59:04 +0000 (11:59 -0400)]
fixed compile error
EtchedPixels [Fri, 23 Sep 2016 13:36:47 +0000 (14:36 +0100)]
Merge pull request #432 from beretta42/nocpm
util: patchcpm no longer exists to install
Brett Gordon [Fri, 23 Sep 2016 13:26:51 +0000 (09:26 -0400)]
util: patchcpm no longer exists to install
Alan Cox [Fri, 23 Sep 2016 11:36:16 +0000 (12:36 +0100)]
marksman: table formatting with width correctly handled
Alan Cox [Thu, 22 Sep 2016 21:52:02 +0000 (22:52 +0100)]
close: update man page for stuff we fixed
Alan Cox [Thu, 22 Sep 2016 21:50:20 +0000 (22:50 +0100)]
open: fix truncating on read only open
Also note another point where we need to fix up refcount handling for FIFO
nodes
Alan Cox [Thu, 22 Sep 2016 21:50:06 +0000 (22:50 +0100)]
filesys.c: add a TODO
Alan Cox [Thu, 22 Sep 2016 21:49:52 +0000 (22:49 +0100)]
syscall_other: fix up various rename problems
Alan Cox [Thu, 22 Sep 2016 21:48:40 +0000 (22:48 +0100)]
libc: add dummies for fsync and fdatasync
We don't need syscalls for these on smaller boxes (and it's surprisingly hard
to get right) so just wrap sync. That way a future bigger box that does need
fsync/fdatasync can be done properly.
Alan Cox [Thu, 22 Sep 2016 21:48:14 +0000 (22:48 +0100)]
sys/types.h: add major/minor/makedev trio
Alan Cox [Thu, 22 Sep 2016 21:42:57 +0000 (22:42 +0100)]
man2: a beginning
And some test material for marksman
Alan Cox [Thu, 22 Sep 2016 21:11:56 +0000 (22:11 +0100)]
marksman: fix various buglets
Still need to sort out width() for tables to work properly
EtchedPixels [Thu, 22 Sep 2016 14:25:33 +0000 (15:25 +0100)]
Merge pull request #430 from beretta42/level2_3
bring coco3 up to level 2
Brett Gordon [Thu, 22 Sep 2016 14:12:34 +0000 (10:12 -0400)]
coco3: bring platform up to level 2 w/ coredumps + networking
Brett Gordon [Thu, 22 Sep 2016 14:10:53 +0000 (10:10 -0400)]
syscall_net: various bug fixes.
Brett Gordon [Thu, 22 Sep 2016 14:08:58 +0000 (10:08 -0400)]
syscall_exec: gcc doesn't like subtracting a ptr from a int