Add support for SLIRP NAT networking
ltdl patches from guile project
Removed legacy libltdl code and force using system ltdl instead.
add back link to github project for Bochs on main page
Updated news.html.
Some more changes after release.
Added new logo for Bochs 2.8.
Use new URL for CHANGES file.
Bochs 2.8 released
Started updates after release 2.8.
Also it seems there are more pixels on the right side of bochs client or inner window then on the left side of the client or inner window lmao maybe get that evened out also.
dark flat ui or bordless inner window for ui testing
Ignore SPP in guest paging-structure entries
This part is not a part of genuine definition. It is part of HLAT feature and Paging Write control. Write for A/D update should be okay only of Paging Write control is enabled. This control is not implemented yet, will be implemented soom/ In meanwhile- this is not a bug.
USABILITY: There should be a small message before the debugger's Bochs prompt that the c key must be pressed
Done
write config/read config can fail
Every such case should be considered an issue. Right now we are clean after fixing few issues like that. Closing. If appears again - please open ticket on Github
When using display_library: sdl2 with fullscreen option it's not possible to switch back to window mode with ALT+ENTER key combination
Fullscreen toggle using ALT+ENTER is now implemented in Bochs on Github. On Linux / X11 there are sometimes issues after toggle that need some more investigation.
When Bochs is killed with SIGTERM, it no longer shows a dialog box. A graceful exit in the SIGKILL case is not yet implemented.
Cirrus: Dword granular color expansion messed up
Bochs Enhanced Debugger useful features
Already implemented in github
Bochs does not support two VGA fonts displayed simultaneously.
Already implemented in github
Thank you for fixing this bug! On my end I can confirm that this bug is fixed.
Incorrect "Blocking by NMI" bit when handling NMI VM-exit
Fix was pushed to github Branch: refs/heads/master Home: https://github.com/bochs-emu/Bochs Commit: 6b48d6e33ff8d5a5f8b1986e01aa34c901faffe3 https://github.com/bochs-emu/Bochs/commit/6b48d6e33ff8d5a5f8b1986e01aa34c901faffe3
I can see only garbage on the screen and in the log there is no line containing "modeext 0x01" here. I'm using latest Bochs code from Github (https://github.com/bochs-emu/Bochs).
Ah, it was in my spam folder. Are you sure? Just to confirm that we're seeing the same thing, the last debug messages to appear should be the following, with the first BLT having the Dword granularity bit set: BLT: src:0x00000000,dst 0x00231c2a,block 234x17,mode 0xac,ROP 0x0d BLT: srcpitch:0x000003e0,dstpitch 0x00000c00,modeext 0x01,writemask 0x00 BLT redraw: x = 58, y = 3357, w = 40, h = 17 BLT, cpu-to-video, transparent (17 times) (a bunch of register reads and writes omitted) BLT: src:0x00321000,dst...
I have received your email with floppy image, tested it and replied with my test results. I don't know whether or not your suggested fix is okay, since there is no bitblt function call with dword granularity bit set. So this floppy image doesn't contain a test case.
I tried to email you a floppy image the other day, did it go through? The error is with the pitch calculation, it should be (w + 31) >> 5 << 2.
I have fixed the PCI MMIO access in the Bochs repository on Github.
Cirrus: MMIO space not decoded when in VGA mode
I do not have a real card to test with, but according to the manual, this bit only enables extended display modes and frame buffer accesses.
Add "-kernel" option to boot an ELF64 linux kernel without a disk image
Sorry for the confusion! #561 is the right one -- please discard #556. Thanks for reviewing this patch! Xinyang On Wed, Dec 27, 2023 at 6:37 AM Volker Ruppert vruppert@users.sourceforge.net wrote: There are now two patches from you for implementing the same feature. Which one should be used? [patches:#561] https://sourceforge.net/p/bochs/patches/561/ Support loading linux kernel directly Status: open Group: to_be_validated Created: Fri May 07, 2021 03:43 PM UTC by Xinyang Ge Last Updated: Fri May...
Volker, Checking the file /iodev/usb/usb_ehci.cc on the master branch of the github repo, I can confirm the line is still there, now at line 2143. I'll need some more time to build a minimal example to test if the bug is still present, but seeing as the line of code is still there, I suspect so. I'm currently running Bochs 2.7 (built from SVN August 1, 2021) on (I think) all my machines. I'll test first against that, and if the bug is still there, I'll build a current version and test against it....
Bochs moved to github and we hope we will get "quite a number of cvs developers" there to raise such an issues again
svn guidelines
assign opcode F1 as magic break point
Many versions of magic breakpoints were added to fight 'bad magic breakpoint encoding' problem. Still definitelly not ICEBP because there are real workloads which want to emulate it properly.
Memory module size determination
Unclear requirement, what is S728P.ROM and why it is interesting ? If still relevant - please come up to discuss on github.
Out of bounds memory access in memory.cc
Was never reproduced and the code was altered significantly. If you see it again - please report issue on github
build issue in mac 10.10
The Bochs development has been moved to Github and the EHCI code has been modified in the meantime. The line you mentioned has been moved or modified. Please verify whether or not the bug still exists.
There are now two patches from you for implementing the same feature. Which one should be used?
I cannot see this behavior here.. Please check the initial time in the log file. If the difference is constant, you should see it at startup, too. If the difference is variable, you could try the clock option parameter "rtc_sync=1" to make sure the "one second timer" runs at real #time speed. The clock sync setting only affects the system timer, not automatically the RTC.
ne2k ISA doesn't support bootrom, document optromimage1 workaround?
Added info in the docs of Bochs on Github.
Cirrus Graphics don't work correctly with windows 3.1 when i use cirrus graphics drivers
I just tried the driver mentioned above with latest Bochs code from Github. Tried 800 x 600 and 1024 x 768 (both 16 bpp) and everything looks okay.
Thank you. This time i recompiled it with: ./configure --prefix=/dev/shm/bochs --enable-debugger --enable-readline --enable-sb16 --with-x11 --with-sdl2 --with-wx --enable-clgd54xx It also does run. But the RTC clock is still 11 minutes and 52 s off. On my host system running linux i have: 20:25 In bochs in my guest OS running Freedos, i have: C:\>time Current time is 8:36:52.52 PM. Enter a new time: C:\>
You did not set up the vga option, so Bochs uses the VBE extension, but you specify the Cirrus version of SeaVGABios as the vgaromimage. It seems to be unable to boot if there is no Cirrus hardware present. The VGABIOS-lgpl-latest would be the right choice for VBE. If you specify Cirrus as the vga extension, you could also use VGABIOS-lgpl-latest-cirrus.
I tried it, but bochs fails at the VGA BIOS. See the added txt file. I compiled it with the following options: ./configure --prefix=/dev/shm/bochs --enable-debugger --enable-readline --enable-sb16 --with-x11 --with-sdl2 --with-wx make # and make install worked without errors. For the display library i tried both display_library: x and display_library: sdl2
This bug is now fixed after addition i/o mapped register access. See latest Bochs code on Github.
E1000 NDIS 2 DOS driver reports "Driver did not load, NDIS environment invalid"
I understand. But if all device names (clock:, memory:, mouse: etc.) contain a colon, then you would only have to check whether the next argument contains a colon. And if that's not the case, then it's a parameter that belongs to the previous device name. Here it would be enough to add a comma followed by this next argument if the last argument did not have a comma because it was forgotten. So if we have something like this, with a missing comma after "sync=realtime" and another device after clock:...
I understand. But if all device names (clock:, memory:, mouse: etc.) contain a colon, then you would only have to check whether the next argument contains a colon. And if that's not the case, then it's a parameter that belongs to the previous device name. Here it would be enough to add a comma followed by this next argument if the last argument did not have a comma because it was forgotten. So if we have something like this, with a missing comma after "sync=realtime" and another device after clock:...
This would be hard to implement. Currently Bochs doesn't have something like a registry or database of bochsrc options and their parameters. A complete rewrite of the parser code would be required for simply checking whether or not the next argument belongs to the current option.
Thank you.
Thank you very much. I will test it.
I read your changes and commit description: https://github.com/bochs-emu/Bochs/commit/20b61b31f2a890fb250c7d665949d3f06830417a and related issue: https://github.com/bochs-emu/Bochs/issues/160 First thank you for the improvement. Something like this: bochs clock: sync=realtime, time0=local should work now. But i still see an issue with this. What if the user forgets a comma? And why do you need the comma at all, a separation by one or more spaces should be sufficient and is more user friendly. This...
I just tried the driver mentioned above with latest Bochs code from Github. Tried 800 x 600 and 10024 x 768 (both 16 bpp) and everything looks okay.
I read your changes and commit description: https://github.com/bochs-emu/Bochs/commit/20b61b31f2a890fb250c7d665949d3f06830417a First thank you for the improvement. Something like this: bochs clock: sync=realtime, time0=local should work now. But i still see an issue with this. What if the user forgets a comma? And why do you need the comma at all, a separation by one or more spaces should be sufficient and is more user friendly. This (one comma removed) will fail: bochs clock: sync=realtime time...
Ticket moved from /p/bochs/bugs/1451/
Do you have a test case for this feature? Bochs implements it the save way as Qemu does.
Do you the behavior of real hardware? Bochs implements it the same way as Qemu does.
bximage ignores -q quiet option if batch command is incomplete
I have now added an error message before entering interactive mode if function or filename is missing and added example to the manual page. Using -q for quiet mode is still mandatory even if all required parameters are given.
Can't enter ANSI code via ALT GR + Numeric keypad in BOCHS with DOS as guest machine
I have added the BIOS capability to use ALT + numeric keypad on Github (see https://github.com/bochs-emu/Bochs).
Apply standard build flags everywhere
Can you please try with the lastest Bochs code from Github?
Ticket moved from /p/bochs/bugs/1452/
Broken mktime() call in cmos.cc
This bug should be fixed with your patch (on github).
Ticket moved from /p/bochs/bugs/1453/
Can't enter ANSI code via ALT GR + Numeric keypad in BOCHS with DOS as guest machine
magic breakpoint instruction reassign ability
Implemented and turned to master on github
PAUSE instruction hack
Obsolete
Japan and Korea key support & Cirrus support for OS/2
All but Japanese keyboard is already in the master. add scancodes for Japanese A01/AX keyboard and Hanguel keyboard win32 and RFB gui : Japanese A01 (106/109) keyboard support These could be supplied as separate patches based on master in github if still relevant.
Real Time Clock /CMOS fix
Merged into master
Bochs command line parse needs a rewrite to improve usability and make bochs less errorprone