This is an undocumented variable that will hold the build configuration,
e.g. Debug, Release, etc., used to create a package with CPack. It is
only set if you provide an explicit configuration argument to CPack.
Add PCRE2 support as an alternative to the original PCRE. PCRE2 is
actively maintained and has a cleaner API than its predecessor.
- Use a function interface around regular expressions to isolate and
reduce the preprocessor #if/#endif forest. Abide by a single function
declaration/implementation approach, vs. multiple function
implentations within an #if/#endif forest.
- Clean up the code so that work isn't recreated or duplicated, such as
compiling regexps.
Platform notes:
- cmake: PCRE2 is the default RegEx support. To revert to PCRE, add
"-DPREFER_PCRE" to the cmake configuration command line.
- makefile: PCRE remains the default RegEx support out of respect for
tradition (cue Chaim Topol singing "Tradition".) Enable PCRE2 by
defining "USE_PCRE2=<some value>" from the make command line or shell
environment.
- Visual Studio Projects: Untouched. Externally built libraries provide
regular expression support residing in a specific place in the file
system, which is outside the control of SIMH proper.
sim_defs.h:
- Change EXPECT::size from int32 to size_t for increased
cross-platform/64-bit compatibility (i.e., array indices and offsets
should be size_t, where practicable.)
- Add typedefs for EXPECT regex support independence: sim_regex_t,
sim_re_capture_t.
- Add regular expression context to EXPTAB that maintains state when
processing the captured matches, making the function interface clean.
Tracks intermediate state while processing captured matches.
On FreeBSD, one must actually link against libpcap
(or give the path to a shared object at link
time). That is, it is not a header-only library
on that platform.
Account for this by adding a new boolean variable
in `platform-quirks.cmake`: `NEED_PCAP_LIBRARY`,
which defaults to false. Condition pcap library
discovery and link argument modifications on that
variable in `FindPCAP.cmake` and `dep-link.cmake`.
They are personal user settings not appropriate for general
distribution. The toplevel .gitigore has a rule to ignore *.user as
well as most things under the Visual Studio Projects.
- macos-13: Deprecated and gone.
- checkout, upload-artifact: Github action script version changes.
- windows-latest (aka windows-2025): Leave future TODO notes in
cmake-builds.yml. It's pointless to build on windows-2025 since there
is no newer Visual Studio version than VS 2022 and doing so will
produce the exact same package outputs.
I added the missing dependency sdl2_ttf for arch linux in the dependency checker as after using the checker I still needed to manually install that to compile.
GCC 15 uses c23 instead of c17 by default. Apparently c23 is not backwards compatible to older C standards... This prevented me from easily compuling simh so I added a setting to tell cmake to use c99 fixing the compiling errors.
Fixes#490
This problem was a latent bug introduced when summarized array register
output was added back in 2013 or 2018.
This fixes the problem reported in https://github.com/open-simh/simh/issues/499
1. Making sure that only expected values appear on the bus when
registers are read by CPU;
2. Fix order of I/O cancellation in case of disk detachment with
pending I/O;
3. Fix WLOA bug and make protection consistent with documentation;
4. Added more comments.
A hobbyist found an original FDOS for 8" drives (FD360, FD3712) that
has turned up a bug in both the FDC+ and AltairZ80 implementation of
the FD3712.
In the FDC+, the data pointer into the 128 byte write "shift register"
is reset upon receipt of a CLEAR command. The real hardware never
resets the shift register, so software should not adjust the pointer
either. AltairZ80 may need the same adjustment.
These are tests for the classic LINC, including the features new in
1965: 2048 words of memory, a Z register, an overflow flag, and an
interrupt facility.
This emulates the classic LINC. The design was settled in 1965,
increasing memory to 2048 words, and adding a Z register, an overflow
flag, and an interrupt facility.
Accept devices mapped at IOPAGEBASE -- text buffer/character generator
memory on video board starts at 160000.
Clear invalid PSW bits in trap handler -- system acceptance test writes
PSW 113705 to vector 34 (TRAP instruction).
Pervasive misuse of "ETH_MAC *" (a pointer to an ETH_MAC, aka a 6
element unsigned char array) when a simple "ETH_MAC" is correct. The
best example of this was eth_mac_fmt() in sim_ether.c with the following
prototype:
t_stat eth_mac_fmt (ETH_MAC* const mac, char* strmac)
The first parameter is a pointer to an array of 6 unsigned characters,
whereas it really just wants to be a pointer to the first element of the
array:
t_stat eth_mac_scan (const ETH_MAC mac, char* strmac)
The "ETH_MAC *" indirection error also results in subtle memcpy() and
memcmp() issues, e.g.:
void network_func(DEVICE *dev, ETH_MAC *mac)
{
ETH_MAC other_mac;
/* ...code... */
/* memcpy() bug: */
memcpy(other_mac, mac, sizeof(ETH_MAC));
/* or worse: */
memcpy(mac, other_mac, sizeof(ETH_MAC));
}
eth_copy_mac() and eth_mac_cmp() replace calls to memcpy() and memcmp()
that copy or compare Ethernet MAC addresses. These are type-enforcing
functions, i.e., the parameters are ETH_MAC-s, to avoid the subtle
memcpy() and memcmp() bugs.
This fix solves at least one Heisenbug in _eth_close() while free()-ing
write request buffers (and possibly other Heisenbugs.)
- Nuke and reinstall HomeBrew for macOS builds. Gets around the whole
problem of stale links and conflicting packages installed in the GitHub
macOS images.
- macOS 12 and Ubuntu 20.04 officially deprecated. Ubuntu 20.04 will be
removed from GitHub's runner images on 01 APR 2025.
- .travis/deps.sh: Add mingw32 as a dependency target, enable MinGW64
32-bit builds.
- Work around LTO bug on Ubuntu and macOS compilers, where LTO appears
to miscalculate the number of bytes stored when using the default byte
swapping 'for' loop. This is a workaround that uses compiler swapping
intrinsics for well known sizes that appeases LTO. The alternative is
to wait for a compiler fix, which is infeasible.
The vmnet interface created for a simh instance has its own MAC address.
It doesn't seem too useful to for simh itself (you can just use whatever
MAC address and send/receive frames with that), but it does make
referencing it in other places (i.e. SHOW ETHERNET vs. ifconfig) easier.
The generated MAC address is provided at interface creation time; while
we could look it up earlier like with pcap, this seems more efficient.
