Activity for Anders Gavare
-
Anders Gavare committed [r6502]
Names for RAM/ROM ranges.
-
Also check for X11R7 in the configure script, in addition to X11R6.
-
Copyright update.
-
Updating some URLs to point to where OpenBSD still can be found.
-
Implementing enough of the mb89352 to allow OpenBSD/luna88k to work (both the boot loader and the kernel)!
-
debugmsg, and allow INQUIRY for unimplemented case, with a warning for now.
-
debugmsg:ify.
-
xterm update.
-
debugmsg:ifying LUNA-88K prom.
-
These show up when typing e.g. CTRL. so let them be hidden, for now.
-
Max for LUNA-88K2 was probably 112 MB, but for LUNA-88K it was 64 MB.
-
Note about color framebuffers.
-
Fix crash for short output.
-
Copyright year update.
-
Updating the TODO.
-
Adding __builtin_expect (likely/unlikely) detection to the configure script.
-
Add an example of using setroot to play with booting, even though there is no keyboard support.
-
Fixing disassembly of the MIPS TGEI, TGEIU, TLTI, TLTIU, TEQI, and TNEI instructions (NetBSD's pkgsrc patch had it wrong).
-
Starting to move guest OS installation instructions into individual machine pages, rather than having all of them randomly in the general guest OS page.
-
Only warn about unimplemented c for Valid entries.
-
Minor updates to the TODO.
-
Making sure that when invalidating virtual addresses in the dyntrans cache, take into account that not all CPUs can use all 64 bits for virtual addresses. This fixes a very old bug which affected Glatt MU/MK.
-
Minor updates to the TODO.
-
Include missing iomanip header.
-
Adding a skeleton RISC-V CPU and RISC-V "virt" machine. Refactoring some parts of the new framework to deal better with variable length instructions during disassembly.
-
Use bold font rather than brackets around debug messages, when running from a terminal.
-
Making sure that when writing to mirrored RAM, any dyntrans translations made in either the mirror or mirrored ranges are invalidated.
-
Adding disassembly (but not execution yet) of ARM uxtab, uxtah, ldrex, and strex instructions.
-
Updating with comments about devices and their base addresses.
-
Release 0.6.1.
-
Minor update in preparation for 0.6.1.
-
Comments.
-
Minor fix for NetBSD crmfb_fill_rect: x is shifted 16 bits, not 12 bits.
-
Minor comment update.
-
256 MB of RAM is needed in OpenBSD/sgi in order to self-compile GXemul. 128 MB is not enough.
-
Finally fixed the bug which caused scroll-up to not work when running the O2 PROM.
-
Fixing a bug in the generic X11 framebuffer output, so that scaledown is preserved even when a guest OS switches resultion at runtime.
-
Continuing on the SGI GBE refactoring...
-
Slowly continuing on THUMB.
-
Wording.
-
Beginning to implement very slow THUMB interpretation (not dyntrans). Horrible performance-wise, but better than nothing.
-
Continuing on THUMB stuff.
-
Comment.
-
Making ARM disassembly show THUMB instructions, if the lowest bit of the address is 1. (They can not be executed yet though.)
-
Fixing PS/2 (8242) keyboard controller length for SGI O2 emulation.
-
Enabling SCSI again for 3MIN (but it is not fully working).
-
Hm? How is the "ust" device supposed to work?
-
Fixing (?) offset for fpu branches. But it is still wrong since pc is not always where it is disassembling... TODO.
-
Make the PROM run further.
-
Rewrote/refactored enough of the dec_ioasic code to make the 5000/125 original PROM output its first boot messages again (and NetBSD/pmax and OpenBSD/pmax too).
-
Allow testmips programs to detect whether the X11 framebuffer is in use or not (returns 0x0 as resolution if not).
-
Beginning to implement i960 dyntrans instructions.
-
Implementing basic disassembly of some i960 instructions (only the CTRL group so far).
-
Continuing on interrupt stuff for LUNA88K, including the serial controller...
-
Removing notes about "legacy modes" in the documentation, since it was too ugly and those modes are pretty much the main focus of end-users of GXemul still.
-
Updating NetBSD/pmax install instructions: 6.1.4 -> 7.1.1.
-
Making the code build with reasonably modern GNU C++ (6.4) and clang (3.4.1), although there are lots of warnings.
-
Hacking in some bit-reversal for the framebuffer.
-
Silencing a compiler warning for now.
-
Fake Luna 88K dipswitch settings for OpenBSD/luna88k.
-
Adding enough of a hack to let OpenBSD/luna88k distinguish between LUNA 88K and 88K2, by faking FUSE_ROM string.
-
Reviving the (dummy) Luna 88K machine mode from r5458.
-
Fixing color bug in dev_pvr.cc simpleline, which is used when rendering Gouraud shaded triangles.
-
Enabling ELF file support for 88K experiments.
-
Minor update.
-
Updating the TODO.
-
Include more registers in the set of registered registers (so to speak), including debug dump.
-
Implementing fmov_rm_frn for double-sized register acccess.
-
Adding missing Dreamcast entry.
-
Better reset of graphics during "switch to bootstrap 1" in Dreamcast fake ROM emulation.
-
In SH4 instruction trace, include address in printout in a few more cases than before.
-
When unimplemented SuperH instructions cause an abort failure, GXemul now drops into the debugger instead of calling exit() (if started with -K).
-
Updating the NetBSD/hpcmips guest operating system install instructions from 4.0.1 to 5.0.2.
-
Include HTML link to machine templates (not just other components), from generated trees.
-
Man page reflects September 2010.
-
Overwrite memory with nonsense on free().
-
nbjoerg on #GXemul (FreeNode) noticed an easy way to trigger an unhandled case in memory_sh.cc. Hopefully this is the correct solution.
-
Minor update of the guest OS documentation.
-
Adding flush() calls to stderr and stdout, in an attempt to improve clarity in test output during failures.
-
Upgrading NetBSD/evbmips install instructions from 5.0 to 5.0.2.
-
Improve install path guessing (DESTDIR and MANDIR) based on feedback from Tom Callaway.
-
HTML fix.
-
Preparing for a quick 0.6.0 release.
-
Adding function call trace stuff for MIPS (hopefully it works).
-
Reimplementing the MIPS slt and sltu instructions, various branch instructions, and fixing some bugs; the rectangle drawing demo for testmips runs in the new framework :)
-
Minor HTML updates.
-
Reimplementing the MIPS xor, addu, and subu instructions.
-
href fixes.
-
The m88k_cpu now takes a creation argument called r31, for the initial stack pointer value. (Needed to run the rectangle drawing demo, which assumes a valid r31 on startup.)
-
"Quiet mode" is now turned off, if dropping into the debugger, both by using -V and by triggering unimplemented instructions.
-
Renaming old testmips to oldtestmips, so that the name testmips can be used by the new framework.
-
Updating NetBSD/cobalt guest OS install instructions from 5.0 to 5.0.1.
-
Updating NetBSD/pmax guest OS install instructions, and automated regression/install tests, from 5.0 to 5.0.1.
-
Add link to build.html.
-
Reimplementing a slow version of the M88K st instruction.
-
Remove valgrind tmp core files.
-
Dummy page to get both release and debug builds to pass without memory leaks.
-
Include _GLIBCXX_DEBUG and _GLIBCXX_DEBUG_PEDANTIC defines in debug builds, to potentially catch some STL misuses.
-
Implementing a (so far unused) ScopedTemporaryValue template.
-
Updating the TODO with thoughts/plans for the "plugin" part of the new framework.
1 >
