Will Sowerbutts [Sat, 31 Jan 2015 00:07:36 +0000 (00:07 +0000)]
n8vem-mark4: devsdspi - save another two bytes! collect the whole set!
Will Sowerbutts [Sat, 31 Jan 2015 00:02:05 +0000 (00:02 +0000)]
n8vem-mark4: devsdspi - save a byte
Alan Cox [Fri, 30 Jan 2015 23:39:04 +0000 (23:39 +0000)]
socz80: move to the new issue sd driver
With this in place we mount the root file system and blow up trying to start
init.
Alan Cox [Fri, 30 Jan 2015 23:27:16 +0000 (23:27 +0000)]
tty: fix typo
Alan Cox [Fri, 30 Jan 2015 21:59:20 +0000 (21:59 +0000)]
socz80: fix tty data port definition
With tx interrupts off we can now get some vague impression of sanity.
Alan Cox [Fri, 30 Jan 2015 21:59:03 +0000 (21:59 +0000)]
socz80: next step of fixing tty problems
Alan Cox [Fri, 30 Jan 2015 21:07:24 +0000 (21:07 +0000)]
socz80: Various fixes for the mm and interrupts
Still definitely not there yet.
Alan Cox [Fri, 30 Jan 2015 17:51:55 +0000 (17:51 +0000)]
z80: make the outfoo routines preserve everything
They are mostly used for debug, so having them eat AF is just too good a way to
have them muck up the debug work.
Alan Cox [Fri, 30 Jan 2015 16:52:36 +0000 (16:52 +0000)]
socz80: Bring the basic bootstrap code up to speed and use Will's .com file maker
This gets us a fuzix.com which correctly probes the SD card, then explodes when
you hit a key at the bootdev prompt
Alan Cox [Fri, 30 Jan 2015 00:42:05 +0000 (00:42 +0000)]
z80: fix all the other .lnk files
Alan Cox [Fri, 30 Jan 2015 00:37:58 +0000 (00:37 +0000)]
trs80: fix link rules for file move
Alan Cox [Fri, 30 Jan 2015 00:30:57 +0000 (00:30 +0000)]
68000: clean up most of the 16bitisms and sign stuff
Mostly this just tidies up all the introduced types. Various things are made
explicitly unsigned that were not called out before and which gcc rightly warned
on.
The 32bit execve needs are quite different, so we split into two execve() designs
rather than make a mess of the already quite nasty 8bit one.
Alan Cox [Fri, 30 Jan 2015 00:30:19 +0000 (00:30 +0000)]
malloc: We need a heap allocator for flat model systems
This isn't intended as a general purpose kernel malloc.
Alan Cox [Fri, 30 Jan 2015 00:27:58 +0000 (00:27 +0000)]
flat: initial flat memory support code
Alan Cox [Fri, 30 Jan 2015 00:27:31 +0000 (00:27 +0000)]
process: fix bracketing
Alan Cox [Fri, 30 Jan 2015 00:27:10 +0000 (00:27 +0000)]
syscall_proc: fix 16bit ism on signals
Alan Cox [Fri, 30 Jan 2015 00:26:53 +0000 (00:26 +0000)]
tty: fix warnings for types and arguments
Alan Cox [Fri, 30 Jan 2015 00:26:24 +0000 (00:26 +0000)]
link/mkdir: error link counter overrun
Alan Cox [Fri, 30 Jan 2015 00:25:41 +0000 (00:25 +0000)]
68000: base enabling pieces
Alan Cox [Fri, 30 Jan 2015 00:24:36 +0000 (00:24 +0000)]
usermem: support for flat binaries
Alan Cox [Fri, 30 Jan 2015 00:24:17 +0000 (00:24 +0000)]
vt: fix argument typing ready for 32bit
Alan Cox [Fri, 30 Jan 2015 00:24:03 +0000 (00:24 +0000)]
swap.c: fix unused argument warning
Alan Cox [Fri, 30 Jan 2015 00:23:38 +0000 (00:23 +0000)]
tty/vt: fix prototypes for ioctl
Alan Cox [Fri, 30 Jan 2015 00:23:05 +0000 (00:23 +0000)]
devsys: fix types
This doesn't matter for 16bit but will shortly
Alan Cox [Fri, 30 Jan 2015 00:22:42 +0000 (00:22 +0000)]
devsys: fix warnings caused by unused parameters
Alan Cox [Fri, 30 Jan 2015 00:22:23 +0000 (00:22 +0000)]
devio: fix warnings caused by unused parameters
Alan Cox [Fri, 30 Jan 2015 00:22:10 +0000 (00:22 +0000)]
6809: add used macro
Alan Cox [Fri, 30 Jan 2015 00:21:38 +0000 (00:21 +0000)]
cpu-z80: introduce used() macro
SDCC wants the reverse behaviour to everyone else
Alan Cox [Fri, 30 Jan 2015 00:21:13 +0000 (00:21 +0000)]
cpu-6502: introduce used() macro
So we can tidy up all the unused argument warnings
Alan Cox [Fri, 30 Jan 2015 00:20:34 +0000 (00:20 +0000)]
68000: very initial pieces to work on a first 680x0 port
This is at the 'gee it compiled' stage.
Alan Cox [Tue, 27 Jan 2015 23:54:15 +0000 (23:54 +0000)]
6502/z80: finish fixing up signal types
This now passes both compilers. 6809 still needs re-aligning
Alan Cox [Tue, 27 Jan 2015 23:47:45 +0000 (23:47 +0000)]
6502: correct typecasts for the break limit
Alan Cox [Tue, 27 Jan 2015 23:44:38 +0000 (23:44 +0000)]
Merge branch 'master' of https://github.com/EtchedPixels/FUZIX
EtchedPixels [Tue, 27 Jan 2015 23:27:24 +0000 (23:27 +0000)]
Merge pull request #91 from geijoenr/master
update megasd driver for rawflag and some small fixes including sdcc 3.4.2 fixes
geijoenr [Tue, 27 Jan 2015 21:41:36 +0000 (21:41 +0000)]
msx2: update megasd transfer functions for rawflag=1
geijoenr [Tue, 27 Jan 2015 18:41:59 +0000 (18:41 +0000)]
msx2: fix memory sizing
no need to write a byte on every page, specially
after we have already copied the kernel on it.
geijoenr [Mon, 19 Jan 2015 15:04:30 +0000 (15:04 +0000)]
sdcc: fix some type warnings for sdcc 3.4.2
Alan Cox [Mon, 26 Jan 2015 22:54:35 +0000 (22:54 +0000)]
types: set the type definitions ready for non 8bit 16bit addressed
Alan Cox [Mon, 26 Jan 2015 22:53:32 +0000 (22:53 +0000)]
cpu: add uarg_t
Alan Cox [Mon, 26 Jan 2015 22:53:07 +0000 (22:53 +0000)]
bank code; Add comments and information to these
EtchedPixels [Mon, 26 Jan 2015 22:12:37 +0000 (22:12 +0000)]
Merge pull request #89 from willsowerbutts/fixdd
Fix for unmounting filesystem, fix for dd count= option, change bf_free to tristate for unfreeable superblocks
Will Sowerbutts [Mon, 26 Jan 2015 21:31:23 +0000 (21:31 +0000)]
Kernel: Change bf_free to tristate -- free, busy, superblock.
Superblocks are never freed by bfree(). This allows the superblock to be
read by user space from the raw block device without being freed until
the filesystem is later unmounted.
Will Sowerbutts [Mon, 26 Jan 2015 21:30:52 +0000 (21:30 +0000)]
Kernel: Bug fix - do not zero the superblock magic number on unmount.
Will Sowerbutts [Sun, 25 Jan 2015 22:53:20 +0000 (22:53 +0000)]
Fix count= option to dd
Alan Cox [Sun, 25 Jan 2015 23:19:58 +0000 (23:19 +0000)]
cpu-z80/cpu.h: add a warning about __COMMONMEM
Its an exciting new way to blow yourself up...
EtchedPixels [Sun, 25 Jan 2015 23:18:43 +0000 (23:18 +0000)]
Merge pull request #87 from willsowerbutts/blkdirect
blkdev and friends: Update to support rawflag=1
Will Sowerbutts [Sun, 25 Jan 2015 22:16:07 +0000 (22:16 +0000)]
devsd: Remove unused memory, fix return without releasing CS
Will Sowerbutts [Sun, 25 Jan 2015 22:10:18 +0000 (22:10 +0000)]
mbr: Reset blk_op.nblock on each transfer
Will Sowerbutts [Sun, 25 Jan 2015 22:01:43 +0000 (22:01 +0000)]
n8vem-mark4: Update SD transfer functions for rawflag=1
Will Sowerbutts [Sun, 25 Jan 2015 22:00:34 +0000 (22:00 +0000)]
devsd: Convert to new blkdev API and support rawflag=1
Will Sowerbutts [Sun, 25 Jan 2015 21:57:32 +0000 (21:57 +0000)]
devide: Convert to new blkdev API and support rawflag=1
Will Sowerbutts [Sun, 25 Jan 2015 21:55:30 +0000 (21:55 +0000)]
blkdev: Support rawflag=1, pass parameters using global variables rather
than function call parameters.
Will Sowerbutts [Sun, 25 Jan 2015 21:54:10 +0000 (21:54 +0000)]
monitor-z80: Print newline before banner
Will Sowerbutts [Sun, 25 Jan 2015 21:52:34 +0000 (21:52 +0000)]
New COMMON_MEMORY macro to put code into COMMONMEM segment
Will Sowerbutts [Sun, 25 Jan 2015 20:16:49 +0000 (20:16 +0000)]
Kernel: syscall_exec: Convert block number to off_t before shifting,
otherwise we're restricted to accessing only the first 64KB of the disk.
Alan Cox [Sun, 25 Jan 2015 20:46:15 +0000 (20:46 +0000)]
binmunge: add option so we can stub every relocation rather than use RST
We need this for machines where some inconsiderate designer left ROM at address
zero.
Alan Cox [Sun, 25 Jan 2015 20:43:56 +0000 (20:43 +0000)]
zx128: fix silly bug with !IDE
Alan Cox [Sun, 25 Jan 2015 20:43:23 +0000 (20:43 +0000)]
syscall_exec: unsigned type for the basepage check
Otherwise we break on 0x8000 +
Alan Cox [Sun, 25 Jan 2015 20:42:43 +0000 (20:42 +0000)]
zx128: first pass minor fixes
Still need to sort the IDE out, and more importantly get the banked linker working
Alan Cox [Sun, 25 Jan 2015 20:28:10 +0000 (20:28 +0000)]
zx128: fix devtty header
Lots more needed to start getting this port on the road usefully
Alan Cox [Sun, 25 Jan 2015 19:56:06 +0000 (19:56 +0000)]
msx2: fix reference to unused symbol as the linker now cares
Alan Cox [Sun, 25 Jan 2015 19:47:09 +0000 (19:47 +0000)]
pcw8256: remove unused symbol from tricks.s
Alan Cox [Sun, 25 Jan 2015 17:52:25 +0000 (17:52 +0000)]
z80pack: add raw I/O offset check
Alan Cox [Sun, 25 Jan 2015 17:51:57 +0000 (17:51 +0000)]
nc100: fix silly error in raw check
Alan Cox [Sun, 25 Jan 2015 17:48:01 +0000 (17:48 +0000)]
nc100: fix devtty.h breakage from vt improvements
Alan Cox [Sun, 25 Jan 2015 17:47:01 +0000 (17:47 +0000)]
tricks.s: remove reference to unused symbol - new linker cares
Alan Cox [Sun, 25 Jan 2015 17:46:51 +0000 (17:46 +0000)]
nc100: fix tty table bug
Alan Cox [Sun, 25 Jan 2015 17:46:27 +0000 (17:46 +0000)]
devlpr: update comment - 1uS required for strobe
Alan Cox [Sun, 25 Jan 2015 17:46:10 +0000 (17:46 +0000)]
nc100,devrd: check lengths are valid, update comment
Alan Cox [Sun, 25 Jan 2015 17:38:06 +0000 (17:38 +0000)]
syscall_exec: further work - mostly to persuade sdcc to generate passable code
Alan Cox [Sun, 25 Jan 2015 17:37:36 +0000 (17:37 +0000)]
trs80: update to support direct to user space I/O
Alan Cox [Sun, 25 Jan 2015 17:37:03 +0000 (17:37 +0000)]
kernel.h: inode lengths are unsigned, tell the compiler this
Alan Cox [Sun, 25 Jan 2015 17:36:51 +0000 (17:36 +0000)]
tgl6502: fix old comment
Alan Cox [Sun, 25 Jan 2015 16:09:04 +0000 (16:09 +0000)]
syscall_exec: start cleaning up
Throw out the magic CP/M support - this can be better done by attaching helper code
to the CP/M binary that does an alloca and loads the emulator into the space then
runs it to move the first bytes back right, set up CP/M and go.
Alan Cox [Sun, 25 Jan 2015 16:08:01 +0000 (16:08 +0000)]
lowlevel-z80: fix asm error: missing #
Alan Cox [Sun, 25 Jan 2015 16:07:10 +0000 (16:07 +0000)]
fixedbank: don't import unused and undefined symbols
The newer linker errors this
Alan Cox [Sun, 25 Jan 2015 15:57:44 +0000 (15:57 +0000)]
z80/6502: final (hopefully) alternate magic bytes for relocatables
Alan Cox [Sun, 25 Jan 2015 15:54:08 +0000 (15:54 +0000)]
z80: define PROGLOAD in the asm side of things, also tidy PROG_BASE to match C PROGBASE
This gives us the foundation to do relocation handling by passing the start of user
code its address in IY
Alan Cox [Sun, 25 Jan 2015 15:46:56 +0000 (15:46 +0000)]
SIG_DFL: note must be zero, ready for execve tidy
Alan Cox [Sun, 25 Jan 2015 15:45:38 +0000 (15:45 +0000)]
tgl6502: set up ready to allow relocatable binaries.
Alan Cox [Sun, 25 Jan 2015 00:24:06 +0000 (00:24 +0000)]
execve: use direct user mappings
If your disk devices don't support this then you'll need to enable the
CONFIG_LEGACY_EXEC option and fix the target. (This probably will get renamed
because if we get to stuff like 68000 it may be we have a cache big enough to
make execve via it useful)
This makes execve much faster, way faster on slow memory devices. The remaining
performance killer is the lack of vfork.
Alan Cox [Sun, 25 Jan 2015 00:23:50 +0000 (00:23 +0000)]
tgl6502: enable raw I/O on the rom driver
Alan Cox [Sun, 25 Jan 2015 00:05:09 +0000 (00:05 +0000)]
bank16k_low: remove debug
Alan Cox [Sat, 24 Jan 2015 23:40:34 +0000 (23:40 +0000)]
socz80: definitely a "don't spin" platform
Alan Cox [Sat, 24 Jan 2015 23:39:03 +0000 (23:39 +0000)]
tty: allow -1 return from waitready to indicate the driver knows sleeping is good
Alan Cox [Sat, 24 Jan 2015 23:34:26 +0000 (23:34 +0000)]
Merge branch 'master' of https://github.com/EtchedPixels/FUZIX
EtchedPixels [Sat, 24 Jan 2015 23:34:15 +0000 (23:34 +0000)]
Merge pull request #84 from willsowerbutts/fixdf
Fix strerror, df
Will Sowerbutts [Sat, 24 Jan 2015 23:13:10 +0000 (23:13 +0000)]
mbr: Fix stray > at end of output with four primary partitions, no extended partitions
Will Sowerbutts [Sat, 24 Jan 2015 23:03:08 +0000 (23:03 +0000)]
filesystem: Add /usr/lib/liberror.txt expected by strerror()
Will Sowerbutts [Sat, 24 Jan 2015 23:04:30 +0000 (23:04 +0000)]
Applications: Make df work
Alan Cox [Sat, 24 Jan 2015 21:23:18 +0000 (21:23 +0000)]
bank16k_low: Introduce 16K low common banking logic and use it
We need a different algorithm because we must keep the low page safe not the high
one. The standard bank16k grows by inserting pages before the existing repeating
top map so that common is preserved in high space. The 6502 has common in low
space so we must instead preserve the first page mapping.
Alan Cox [Sat, 24 Jan 2015 21:22:19 +0000 (21:22 +0000)]
exec: introduce MAPBASE as distinguished from PROGBASE
PROGBASE is where the code starts
MAPBASE is where the memory mapping begins
This is needed for systems with a low mapped common
Alan Cox [Sat, 24 Jan 2015 21:21:42 +0000 (21:21 +0000)]
chmem: allow for 6502 and Z80
Alan Cox [Sat, 24 Jan 2015 16:03:33 +0000 (16:03 +0000)]
Merge branch 'master' of https://github.com/EtchedPixels/FUZIX
Alan Cox [Sat, 24 Jan 2015 16:03:05 +0000 (16:03 +0000)]
tgl6502: kill stray space
Alan Cox [Sat, 24 Jan 2015 16:01:44 +0000 (16:01 +0000)]
tgl6502: switch to 16K banking
No swapping yet (and swap for 16K bank needs some core tweaks). Tested with shell
and login paths but not heavily.
Alan Cox [Sat, 24 Jan 2015 16:01:26 +0000 (16:01 +0000)]
devlpr: use need_resched
EtchedPixels [Sat, 24 Jan 2015 13:18:27 +0000 (13:18 +0000)]
Merge pull request #82 from willsowerbutts/rootfs
Add missing /tmp to root filesystem. /etc/mtab removed by init.
Alan Cox [Sat, 24 Jan 2015 13:07:10 +0000 (13:07 +0000)]
proc: Add need_resched()
Use this to make the tty code behave better
Will Sowerbutts [Sat, 24 Jan 2015 00:23:49 +0000 (00:23 +0000)]
Add missing /tmp to root filesystem. /etc/mtab removed by init.