Alan Cox [Sat, 10 Mar 2018 21:43:23 +0000 (21:43 +0000)]
vile: add d and c functionality
Alan Cox [Fri, 9 Mar 2018 21:31:15 +0000 (21:31 +0000)]
68000: add more app makefiles
Alan Cox [Fri, 9 Mar 2018 21:30:25 +0000 (21:30 +0000)]
68000: update some Makefiles
Alan Cox [Fri, 9 Mar 2018 21:29:23 +0000 (21:29 +0000)]
68000: Library makefiles
Alan Cox [Fri, 9 Mar 2018 20:56:16 +0000 (20:56 +0000)]
malloc: tidy formatting to match core code standard
Alan Cox [Fri, 9 Mar 2018 18:07:55 +0000 (18:07 +0000)]
vile: add some of the needed bits for modify tracking etc
Plus some other msall fixups
Alan Cox [Fri, 9 Mar 2018 14:23:44 +0000 (14:23 +0000)]
vile: hook into build
Until we get rid of curses it's a bit large
Alan Cox [Mon, 5 Mar 2018 20:31:18 +0000 (20:31 +0000)]
6809: update rules to build startrek FP
No idea of it works yet but it does now build
Alan Cox [Mon, 5 Mar 2018 20:29:55 +0000 (20:29 +0000)]
6809: Update the library build
Build a libm and don't confuse with libc
Build the double libary (or most of it)
Alan Cox [Mon, 5 Mar 2018 20:29:00 +0000 (20:29 +0000)]
sqrt: fix SNaN
Alan Cox [Mon, 5 Mar 2018 20:28:36 +0000 (20:28 +0000)]
round: fix types for shifts
Alan Cox [Mon, 5 Mar 2018 20:28:05 +0000 (20:28 +0000)]
modf: replace MUSL version with Sun one for portability
Alan Cox [Mon, 5 Mar 2018 20:27:47 +0000 (20:27 +0000)]
frexp: replace MUSL with Sun version for portability
Alan Cox [Mon, 5 Mar 2018 20:27:07 +0000 (20:27 +0000)]
fabs: make friendly with non uint64_t
Alan Cox [Mon, 5 Mar 2018 20:26:54 +0000 (20:26 +0000)]
expm1: remove old STRICT_ASSIGN macro
Alan Cox [Mon, 5 Mar 2018 20:26:26 +0000 (20:26 +0000)]
copysign: make it friendly to non 64bit capable platforms
Alan Cox [Mon, 5 Mar 2018 20:26:05 +0000 (20:26 +0000)]
cbrt: make it friendly with non 64bit capable
Alan Cox [Mon, 5 Mar 2018 20:25:58 +0000 (20:25 +0000)]
atanh: fix build
Alan Cox [Mon, 5 Mar 2018 20:24:48 +0000 (20:24 +0000)]
fpclassify: rewrite in terms of macros
Alan Cox [Mon, 5 Mar 2018 20:24:30 +0000 (20:24 +0000)]
math,h: Fix mismatched atof declaration
Alan Cox [Mon, 5 Mar 2018 19:29:47 +0000 (19:29 +0000)]
rename: fix leak I hope
Rename needs some more thinking about
Alan Cox [Mon, 5 Mar 2018 19:29:21 +0000 (19:29 +0000)]
vile: further design note
EtchedPixels [Mon, 5 Mar 2018 18:47:38 +0000 (18:47 +0000)]
Merge pull request #592 from k0gaMSX/master
Improve msx2 port
Roberto E. Vargas Caballero [Mon, 26 Feb 2018 18:54:35 +0000 (18:54 +0000)]
kernel: Initialize dirty and time flags of bb
The buffer cache is not located in bss in some configurations,
and it means that the memory is not initialized when the system
boots. When emulators allocate the memory, it is initialized
to 0 by the operating system, but it doesn't happen in actual
hardware.
Roberto E. Vargas Caballero [Tue, 13 Feb 2018 22:22:52 +0000 (22:22 +0000)]
start: Avoid infinite loop when BOOTDEVICE is defined
When BOOTDEVICE is defined get_root_dev() was always returning
the same device entering in an infinite loop with the only
effect of scroll out all the previous messages.
Roberto E. Vargas Caballero [Tue, 13 Feb 2018 22:18:28 +0000 (22:18 +0000)]
msx2: Use DEVTTY for kernel messages
Commit
3616d6d added support for multiple vt in msx2,
but in the process it changed the console to the debug
port of openmsx. Since this is not a device present in
actual computers, any error in the boot process was
silently ignored. This patch reverts the console to
TTYDEV and at the same time it prints all the messages
to the debug port of openmsx.
Roberto E. Vargas Caballero [Sun, 11 Feb 2018 22:07:53 +0000 (22:07 +0000)]
Make the Makefiles more portable
- BSD make doesn't support the GNU extensions, and in that
systems the GNU make is called gmake. Seeting the value of
MAKE directly to make break the compilation in that systems.
- In the same way, in BSD systems yacc is not aliased to byacc,
and that explicit rule is not needed at all because POSIX
make already has a default rule to generate c files from y
files.
Alan Cox [Sun, 25 Feb 2018 19:42:33 +0000 (19:42 +0000)]
binfmt: update notes on plans
Alan Cox [Sun, 25 Feb 2018 18:49:21 +0000 (18:49 +0000)]
fweep: more tidying
EtchedPixels [Sun, 25 Feb 2018 18:48:44 +0000 (18:48 +0000)]
Merge pull request #591 from beretta42/nop-explain
Explain reason for NOPs in SD driver
Alan Cox [Sun, 25 Feb 2018 18:46:33 +0000 (18:46 +0000)]
libm,6809: add a helper set for the double ops we can't inline
Alan Cox [Sun, 25 Feb 2018 18:37:09 +0000 (18:37 +0000)]
libm: 6809 now handles double as well as float
Need to write a few small extra helpers to match the inline ones using uint64_t
Alan Cox [Sun, 25 Feb 2018 17:29:42 +0000 (17:29 +0000)]
6809: build libm (sort of)
There are some outstanding problems to fix
- We only build for float, but the C type promotion rules at this point then
mess us up and turn the float into a double for varargs (blame the C spec)
- Several things use __NaN but that breaks on 6809 really and now we've got
SDCC fixed we can probably switch to using 0/0 etc as we did before.
Brett Gordon [Sun, 25 Feb 2018 14:25:38 +0000 (09:25 -0500)]
Explain reason for NOPs in SD driver
Alan Cox [Sun, 25 Feb 2018 00:57:44 +0000 (00:57 +0000)]
Merge branch 'master' of https://github.com/EtchedPixels/FUZIX
Alan Cox [Sun, 25 Feb 2018 00:57:35 +0000 (00:57 +0000)]
games: update Makefiles
Alan Cox [Sun, 25 Feb 2018 00:45:57 +0000 (00:45 +0000)]
fweep: tidying
- 16bit for writes
- trim text buffer to a sane size
- formatting glitches
- fix up alpha in a way that the compiler doesn't include both
Alan Cox [Sun, 25 Feb 2018 00:39:58 +0000 (00:39 +0000)]
fweep: indent
Alan Cox [Sun, 25 Feb 2018 00:37:02 +0000 (00:37 +0000)]
fweep: work in progress
Needs a lot more thinning down to be useful anywhere but 68000
EtchedPixels [Sat, 24 Feb 2018 20:16:34 +0000 (20:16 +0000)]
Merge pull request #589 from beretta42/sd7
coco3: implement sd card.
EtchedPixels [Sat, 24 Feb 2018 20:15:36 +0000 (20:15 +0000)]
Merge pull request #588 from beretta42/net-fix
net native: u_count is reset after calls to readi/writei now.
Brett Gordon [Sun, 29 Oct 2017 04:05:51 +0000 (00:05 -0400)]
coco3: implement sd card.
Brett Gordon [Sat, 24 Feb 2018 15:20:03 +0000 (10:20 -0500)]
net native: u_count is reset after calls to readi/writei now.
Alan Cox [Fri, 23 Feb 2018 23:29:28 +0000 (23:29 +0000)]
startrek: add a decent version of it
Downsize is it uses floats and is about 40K. Defloating and debloating would
I think get it under our 32K preferred target
Alan Cox [Fri, 23 Feb 2018 22:56:10 +0000 (22:56 +0000)]
math.h: correct prototype of powf
Alan Cox [Fri, 23 Feb 2018 22:53:54 +0000 (22:53 +0000)]
libc: Compliance bugs
Missing a prototype for powf()
Missing RAND_MAX definition required by ISO C
Alan Cox [Fri, 23 Feb 2018 19:19:47 +0000 (19:19 +0000)]
Merge branch 'master' of https://github.com/EtchedPixels/FUZIX
Alan Cox [Fri, 23 Feb 2018 19:19:13 +0000 (19:19 +0000)]
pdp11: we need setjmp_pdp11.S committed
(not .s)
EtchedPixels [Fri, 23 Feb 2018 19:17:27 +0000 (19:17 +0000)]
Merge pull request #587 from beretta42/fsck2
build: add fsck -y passing to buildscript
Alan Cox [Fri, 23 Feb 2018 15:24:23 +0000 (15:24 +0000)]
vile: add half screen scroll vi commands
Alan Cox [Fri, 23 Feb 2018 00:45:43 +0000 (00:45 +0000)]
vile: a VI Like Editor - WIP
Not yet ready for Fuzix as we probably want to avoid stdio and curses
(but maybe not - need to see on size). Either way our cursors lacks some
of the funky features this uses at the moment I believe (eg keypad)
Alan Cox [Wed, 21 Feb 2018 21:44:57 +0000 (21:44 +0000)]
localtim: should automatically call tzset
Alan Cox [Wed, 21 Feb 2018 21:28:26 +0000 (21:28 +0000)]
localtim/gmtime: correct which one is applying corrections
Still need to sort the actual correction logic out
Alan Cox [Wed, 21 Feb 2018 21:21:04 +0000 (21:21 +0000)]
ucp: fix bugs
1. Remove mount/umount support as they don't work properly anyway
2. Fux read/write out of space etc reporting
3. Fix creat crash when there is no space to make the new name
Alan Cox [Wed, 21 Feb 2018 20:44:46 +0000 (20:44 +0000)]
rmdir: report path correctly
Alan Cox [Wed, 21 Feb 2018 20:40:49 +0000 (20:40 +0000)]
link: don't allow superuser to link to directories
Alan Cox [Wed, 21 Feb 2018 20:35:48 +0000 (20:35 +0000)]
link: remove root power to link to a directory
It's never used, it makes a nasty mess and it breaks the locking
Alan Cox [Wed, 21 Feb 2018 20:32:26 +0000 (20:32 +0000)]
mkdir: Fix error recover path
This tried to unlink random crap not the correct path. Broken when I tidied
up the path walking logic
Alan Cox [Wed, 21 Feb 2018 20:31:36 +0000 (20:31 +0000)]
flock: LEVEL_0 update
Alan Cox [Wed, 21 Feb 2018 20:30:41 +0000 (20:30 +0000)]
levels: LEVEL_0 makes no sense except as a uaccess check shortener now
For anything else we are probably better using overlays for stuff like
mkdir and rmdir than being incompatible
Alan Cox [Wed, 21 Feb 2018 20:30:30 +0000 (20:30 +0000)]
unlink: lock right node
Alan Cox [Wed, 21 Feb 2018 20:24:01 +0000 (20:24 +0000)]
fs: fixes from testing
1. Don't allow rmdir .
2. Don't allow rmdir of a directory with a file in it (we stopped only subdirs)
3. Don't consider unreferenced inodes when checking if a remount is valid
4. Fix the case of
process A
cd /tmp/foo
process B
rmdir /tmp/foo
process A
ls . (stale entry)
Alan Cox [Wed, 21 Feb 2018 20:23:42 +0000 (20:23 +0000)]
filesys: introduce a routine to check if a directory is empty of files
Alan Cox [Wed, 21 Feb 2018 20:22:47 +0000 (20:22 +0000)]
unlink: will need to hold lock over unlinki
Alan Cox [Wed, 21 Feb 2018 20:22:27 +0000 (20:22 +0000)]
65c816: fix kprintf
Alan Cox [Wed, 21 Feb 2018 14:14:54 +0000 (14:14 +0000)]
devio: Fix build with IDUMP, minor tweaks for error handling
Alan Cox [Wed, 21 Feb 2018 13:32:49 +0000 (13:32 +0000)]
devio: First cut at putting into place sleeping I/O support
This allows the locking frameworks needed to let block devices sleep, but not
to do I/O from interrupts. The latter is much more complicated and it's not
clear that is useful except on bigger systems which need a different buffer
cache anyway
Alan Cox [Wed, 21 Feb 2018 11:08:29 +0000 (11:08 +0000)]
kernel: pathbuf is dead remove the remnants
Alan Cox [Tue, 20 Feb 2018 22:14:33 +0000 (22:14 +0000)]
kernel: minimal patches and defines ready for parallel fs access support
This puts in place the pieces we will need to hook up in order to support
sleeping file system accesses. We also need to add 'D' state to the scheduler
but that is needed anyway to fix some other corner cases.
Alan Cox [Tue, 20 Feb 2018 18:31:19 +0000 (18:31 +0000)]
pg: Add the late V7 "pg" command
Alan Cox [Tue, 20 Feb 2018 18:31:03 +0000 (18:31 +0000)]
cpio: some tidying
Really we should switch to ascii format cpio
Alan Cox [Tue, 20 Feb 2018 14:24:54 +0000 (14:24 +0000)]
From: Tormod Volden <debian.tormod@gmail.com>
Signed-off-by: Tormod Volden <debian.tormod@gmail.com>
---
Kernel/platform-dragon-nx32/Makefile | 1 +
Kernel/tools/lw-checkmap | 12 +++++++++---
2 files changed, 10 insertions(+), 3 deletions(-)
diff --git a/Kernel/platform-dragon-nx32/Makefile b/Kernel/platform-dragon-nx32/Makefile
index
efea7ab..
8a40526 100644
--- a/Kernel/platform-dragon-nx32/Makefile
+++ b/Kernel/platform-dragon-nx32/Makefile
@@ -60,3 +60,4 @@ image:
devsd.o devsd_discard.o spi.o devscsi.o devscsi_discard.o \
scsi_tc3.o discard.o
../tools/pad256 ../fuzix.bin
+ ../tools/lw-checkmap ../fuzix.map
diff --git a/Kernel/tools/lw-checkmap b/Kernel/tools/lw-checkmap
index
f98a552..
9402b2f 100755
--- a/Kernel/tools/lw-checkmap
+++ b/Kernel/tools/lw-checkmap
@@ -1,8 +1,14 @@
:
# Simple check of map file generated by lwlink (LWTOOLS)
-grep ^Section "$1" | sort -k5 | awk '{
+grep ^Section "$1" | sort -k5 | awk '
+ BEGIN{ errs=0 }
+
+ {
st=strtonum("0x"$6);
if (en<st) {printf "free 0x%x-0x%x = %i\n",en,st,st-en};
- if (en>st) {printf "overlap at 0x%x-0x%x!\n",st,en};
+ if (en>st) {printf "overlap at 0x%x-0x%x!\n",st,en; errs++};
en=st+strtonum("0x"$8);
-}'
+ }
+
+ END{ if (errs>0) {exit 1}; }
+'
--
2.7.4
Brett Gordon [Tue, 20 Feb 2018 04:08:11 +0000 (23:08 -0500)]
build: add fsck -y passing to buildscript
Alan Cox [Sun, 18 Feb 2018 14:15:03 +0000 (14:15 +0000)]
pdp11: Add another missing piece and turn on library building
Alan Cox [Sun, 18 Feb 2018 14:03:35 +0000 (14:03 +0000)]
pdp11: setjmp structure
Alan Cox [Sun, 18 Feb 2018 14:03:18 +0000 (14:03 +0000)]
68000: fix longjmp code
Alan Cox [Sun, 18 Feb 2018 14:02:20 +0000 (14:02 +0000)]
pdp11: setjmp and longjmp
Alan Cox [Sun, 18 Feb 2018 14:01:56 +0000 (14:01 +0000)]
pdp11: udivhi3 helper
Alan Cox [Sun, 18 Feb 2018 14:01:34 +0000 (14:01 +0000)]
pdp11: sort out more of the low level bits
Alan Cox [Sun, 18 Feb 2018 00:32:17 +0000 (00:32 +0000)]
pdp11: set correct header type for binaries
We'll probably just go a.out but for now
Alan Cox [Sat, 17 Feb 2018 22:39:43 +0000 (22:39 +0000)]
syscall_other: bracketing
Alan Cox [Sat, 17 Feb 2018 22:39:13 +0000 (22:39 +0000)]
pdp11: more platform updating
Not yet a complete usable machine
Alan Cox [Sat, 17 Feb 2018 22:38:14 +0000 (22:38 +0000)]
pdp11: sketch out more of the needed base code
Alan Cox [Sat, 17 Feb 2018 22:37:30 +0000 (22:37 +0000)]
libs: PDP11 syscall directory
Alan Cox [Sat, 17 Feb 2018 22:36:57 +0000 (22:36 +0000)]
pdp11: add syscall generator code
Alan Cox [Sat, 17 Feb 2018 22:35:43 +0000 (22:35 +0000)]
inet_aton: use long type otherwise it breaks on 16bit
Alan Cox [Mon, 12 Feb 2018 23:31:59 +0000 (23:31 +0000)]
cpio: reformat
Alan Cox [Mon, 12 Feb 2018 00:42:21 +0000 (00:42 +0000)]
cpio: add first cut at porting cpio over
Needs work yet
Alan Cox [Sun, 11 Feb 2018 21:31:22 +0000 (21:31 +0000)]
libc: const pedanticism
Alan Cox [Sun, 11 Feb 2018 21:31:00 +0000 (21:31 +0000)]
6502: optimise the top byte check to avoid a shift
Alan Cox [Sun, 11 Feb 2018 17:23:28 +0000 (17:23 +0000)]
libs: update another TODO
Alan Cox [Sun, 11 Feb 2018 17:19:01 +0000 (17:19 +0000)]
libs: update a TODO file
Alan Cox [Sun, 11 Feb 2018 17:18:43 +0000 (17:18 +0000)]
libclean: add another rule for SDCC 3.7
Alan Cox [Sun, 11 Feb 2018 17:17:42 +0000 (17:17 +0000)]
start: we fire up sysio directly so must set u_done
Alan Cox [Sun, 11 Feb 2018 17:17:30 +0000 (17:17 +0000)]
INSTALL: note sdcc requirements
Alan Cox [Sat, 10 Feb 2018 19:58:42 +0000 (19:58 +0000)]
n_open: rework so we operate directly from user memory
In addition adjust the logic so we build a correctly parsed last name component
in lastname so that it can be used by the caller to manipulate directories.
This gets rid of filename() which saves us copying and parsing the path twice
as well as getting rid of that code
It gets rid of most of the 30 byte name arrays (rename still has to have one)
It allows us to get rid of ugets()
Our PATH_MAX can now be arbitrarily set (for now still 512)
ugets/_ugets/__ugets all go away
pathbuf for LEVEL2 goes away, LEVEL0 and 1 and 2 now all look the same
The end result on Z80 at least is that we run faster amd we save a whopping 400
bytes.
Alan Cox [Sat, 10 Feb 2018 19:52:47 +0000 (19:52 +0000)]
audio; fix dumb bug
Not that we have any working audio layer yet
Alan Cox [Sat, 10 Feb 2018 00:09:13 +0000 (00:09 +0000)]
writei: workaround for sdcc
sdcc at least isn't bright enough to turn
static uint32 x
if (x >> 25)
into a byte load of x + 3 followed by an &0xFE
Instead it generates a 25 repeat bitshift of a 32bit value across four
registers in a fairly critical code path.
Allow defines to help it out with its dumbness.
Alan Cox [Fri, 9 Feb 2018 23:49:50 +0000 (23:49 +0000)]
kernel: restructure readi/writei to save memory
This saves us 256 bytes on Z80 mostly by removing the amount of 32bit maths
SDCC has the opportunity to **** up.
We introduce a helper (umove) which adjusts all the pointers and offsets
for a given file and use that where we can. We also restructure a bit so we
can use the udata.u_done value in readi/writei
This has some minor consequences:
Callers need to check u_done for completion not u_count
Char and Socket handlers don't affect the offset unless they choose to (which
is saner). Devsys has been modified accordingly.
Handle with care.
Alan Cox [Fri, 9 Feb 2018 23:41:22 +0000 (23:41 +0000)]
ubee: floppy driver write fix