352009e27e761f9ba3881828ca4f519432fcd0f6 wilkie wilkie05@gmail.com http://github.com/xomboverlord/xomb/commit/112c57e90a22698681eb1388cf993fa9f5cf76e9 112c57e90a22698681eb1388cf993fa9f5cf76e9

2009-01-29T09:27:18-08:00

2009-01-16T18:51:06-08:00

I DON'T KNOW! Signed-off-by: xomboverlord <overlord@xomb.org> 7b72cd028075cd5564930a91d44a0a45a6f911ed xomboverlord overlord@xomb.org f9b17e6bc8e32a73043f07375f9348064979053d wilkie wilkie05@gmail.com http://github.com/xomboverlord/xomb/commit/352009e27e761f9ba3881828ca4f519432fcd0f6 352009e27e761f9ba3881828ca4f519432fcd0f6

2009-01-29T09:26:02-08:00

2009-01-09T13:36:57-08:00

Fixed embarrassing bug. Signed-off-by: xomboverlord <overlord@xomb.org> 5ebc374fac03a43a620db74bb9c3143433dc13b0 xomboverlord overlord@xomb.org 0a11fb38a739e169a5848510673debf004acc10a wilkie wilkie05@gmail.com http://github.com/xomboverlord/xomb/commit/f9b17e6bc8e32a73043f07375f9348064979053d f9b17e6bc8e32a73043f07375f9348064979053d

2009-01-29T09:26:01-08:00

2009-01-09T00:24:39-08:00

Some more instructions allows further support within Xen. It switches the video mode, but considering I don't have a console anymore, I can't tell why it fails! I need to dump this BIOS somehow. I need to see how it works. Signed-off-by: xomboverlord <overlord@xomb.org> 01204c9bd2d342f9d4ac8de1f2891c1868114dd8 xomboverlord overlord@xomb.org 4eb80a77e3adf533f508dc176e3921a89d2b90c5 wilkie wilkie05@gmail.com http://github.com/xomboverlord/xomb/commit/0a11fb38a739e169a5848510673debf004acc10a 0a11fb38a739e169a5848510673debf004acc10a

2009-01-29T09:25:32-08:00

2009-01-08T17:13:16-08:00

Added VESA stuff. VESA uses 16 bit real mode code, so x86 emulation of real mode was necessary. This is added as a package called XOmBemu. The VESA driver (only working in Bochs) will execute this BIOS code in userland and run the necessary routines to poll the device and gather information about supported modes. It will then pick the 'best' mode and switch to it with the capability of a double buffer. The driver then writes to the working buffer with simple blitting functions and then flips the buffer. I've added a couple of static arrays that store raw pixel data for our logo for fun. Signed-off-by: xomboverlord <overlord@xomb.org> ee2c3ab0f6f7160485a2d9b6df8bbcadf6231803 xomboverlord overlord@xomb.org 1b33d4928e310cb256321030b1e93e7b59969012 Lindsey hazardous.narayan@gmail.com http://github.com/xomboverlord/xomb/commit/4eb80a77e3adf533f508dc176e3921a89d2b90c5 4eb80a77e3adf533f508dc176e3921a89d2b90c5

2009-01-29T09:24:51-08:00

2009-01-16T22:23:42-08:00

up to date linked list malloc with printed debug statments all removed. Adding and removing works kinda funny right now as one adds to the end of the list and the other to the beginning but that really doesn't matter. I figure this will all eventually be replaced with a more elegant implementation of malloc in the future. Signed-off-by: xomboverlord <overlord@xomb.org> 0eb77b3145d1b81c07893039e365c0816a93d262 xomboverlord overlord@xomb.org 17c4f05452163aaf533c3250c7e36916c427ca38 Lindsey hazardous.narayan@gmail.com http://github.com/xomboverlord/xomb/commit/1b33d4928e310cb256321030b1e93e7b59969012 1b33d4928e310cb256321030b1e93e7b59969012

2009-01-29T09:24:31-08:00

2009-01-09T20:05:20-08:00

Updated malloc with debug statements in for testing. Signed-off-by: xomboverlord <overlord@xomb.org> 1662a3910e734984b5207acdd360d7870a5dd174 xomboverlord overlord@xomb.org e3b3589c0a4595258256c4ad8a0b003b0197bcf5 Lindsey hazardous.narayan@gmail.com http://github.com/xomboverlord/xomb/commit/17c4f05452163aaf533c3250c7e36916c427ca38 17c4f05452163aaf533c3250c7e36916c427ca38

2009-01-29T09:24:12-08:00

2009-01-08T17:49:34-08:00

Malloc and free written. Signed-off-by: xomboverlord <overlord@xomb.org> 880691bc6d50780713f6981bc263edb5542b71d5 xomboverlord overlord@xomb.org 34ec4ddb8a9d12cea48ae3a2802e53678cfbf606 wolfwood transmethyl@gmail.com http://github.com/xomboverlord/xomb/commit/e3b3589c0a4595258256c4ad8a0b003b0197bcf5 e3b3589c0a4595258256c4ad8a0b003b0197bcf5

2009-01-29T09:23:08-08:00

2009-01-12T19:28:33-08:00

use the x86_64-pc-xomb flavors of ld and gcc Signed-off-by: xomboverlord <overlord@xomb.org> 7a23bc583146c49c0da2348ad5aa61eff1050e24 xomboverlord overlord@xomb.org 23feba7e015e58c3e8c17f03c268a7a7a1970a83 Steve Klabnik steve@steveklabnik.com http://github.com/xomboverlord/xomb/commit/34ec4ddb8a9d12cea48ae3a2802e53678cfbf606 34ec4ddb8a9d12cea48ae3a2802e53678cfbf606

2009-01-10T21:22:04-08:00

2009-01-09T14:19:24-08:00

getting textile right Signed-off-by: xomboverlord <overlord@xomb.org> 22e0d88b6d0e80c0c500c8cd853e4e2c07da5eaa xomboverlord overlord@xomb.org a7eea801f8960da45cff8a441b532794dfd13b99 Steve Klabnik steve@steveklabnik.com http://github.com/xomboverlord/xomb/commit/23feba7e015e58c3e8c17f03c268a7a7a1970a83 23feba7e015e58c3e8c17f03c268a7a7a1970a83

2009-01-10T21:22:03-08:00

2009-01-09T14:18:16-08:00

READMEs are actually in this time. I need to start reading before I commit. Signed-off-by: xomboverlord <overlord@xomb.org> 2f8ef66cd2ecbfdf8b1905ed4e4cef3a0fa48ef6 xomboverlord overlord@xomb.org cdad3b6d9a34ad8c1ca1c67cba8aee5db72e3104 Steve Klabnik steve@steveklabnik.com http://github.com/xomboverlord/xomb/commit/a7eea801f8960da45cff8a441b532794dfd13b99 a7eea801f8960da45cff8a441b532794dfd13b99

2009-01-10T21:22:01-08:00

2009-01-09T14:07:04-08:00

BSD License added. README added. Textile README added. Sweet. Signed-off-by: xomboverlord <overlord@xomb.org> fb910a1dd6a29ac27accce93a38750d153b739c4 xomboverlord overlord@xomb.org 5404084604655cea7e953ec5066a29cefd5ba777 Steve Klabnik steve@steveklabnik.com http://github.com/xomboverlord/xomb/commit/cdad3b6d9a34ad8c1ca1c67cba8aee5db72e3104 cdad3b6d9a34ad8c1ca1c67cba8aee5db72e3104

2008-12-30T06:12:03-08:00

lol. I always forget to add new files to git. Here's the second app. 07d3ad7a774a0b93f18db0906e148ea4f688c516 Steve Klabnik steve@steveklabnik.com 30028c2a2d6dfb83c5eefa9abf2afc31adf3418d Steve Klabnik steve@steveklabnik.com http://github.com/xomboverlord/xomb/commit/5404084604655cea7e953ec5066a29cefd5ba777 5404084604655cea7e953ec5066a29cefd5ba777

2008-12-29T20:47:56-08:00

New syscall: makeEnvironment(id). This creates a new environment from the grub module located at id, and adds it to the scheduler. beec185ef28703bc46dab4f642769bd6aaf94ab4 Steve Klabnik steve@steveklabnik.com e253348e7cf005a8f7cd676be3c1ff49a48ab303 Steve Klabnik steve@steveklabnik.com http://github.com/xomboverlord/xomb/commit/30028c2a2d6dfb83c5eefa9abf2afc31adf3418d 30028c2a2d6dfb83c5eefa9abf2afc31adf3418d

2008-12-29T19:16:14-08:00

Some fun makefile changes. Also, inttochar is now itoa. That was fun to have, but not fun to type... 2be04ee411fce141491a85eb35752cc8253ba528 Steve Klabnik steve@steveklabnik.com 7ed605ba13eeec879b5189bb7b020dbdcf53eb25 Steve Klabnik steve@steveklabnik.com http://github.com/xomboverlord/xomb/commit/e253348e7cf005a8f7cd676be3c1ff49a48ab303 e253348e7cf005a8f7cd676be3c1ff49a48ab303

2008-12-29T18:59:24-08:00

Two big changes in this commit: 1) We now load up just xsh instead of all applications. This way, we can use grub modules for applications until we get a filesystem working properly. 2) GRUB was having problems when the .iso was bigger than 3MB. So `man mkisofs`... -boot-load-size load_sectors Specifies the number of "virtual" (512-byte) sectors to load in no-emulation mode. The default is to load the entire boot file. Some BIOSes may have problems if this is not a multiple of 4. Yeah. This is now 8 instead of 4. Voila. bdd0ebd6cf98bc42a9ee6f2082078638e33af131 Steve Klabnik steve@steveklabnik.com 53ab9423315579a1247c2605c7015eda97242f7b wilkie wilkie05@gmail.com http://github.com/xomboverlord/xomb/commit/7ed605ba13eeec879b5189bb7b020dbdcf53eb25 7ed605ba13eeec879b5189bb7b020dbdcf53eb25

2008-12-27T01:38:27-08:00

I couldn't sleep, so I implemented a video driver that runs only in userland. This means no syscall is needed to print to the screen. This also means that the shell won't need to call ANY syscalls. I should get HPET working now, since the program will never voluntarily give up control. So, now you have a userland Console library to link to. I have pointed Steve's print function to this Console and turned 'echo' (the previous syscall to write to the screen) into 'error', which will print out to the screen regardless of the environment that owns the console. Any kernel writes to the screen will happen in a random place independent of the environment. But, this is ok, a kernel panic would be the only reason for a real-world kernel write to the screen, or a call to 'error', which should only be used for debug and then removed. It maps the video buffer directly to a read-write page for the environment. The environment handles its own positioning and color information. This also shows how an environment can link to multiple device resources... the shell uses both this and the keyboard, which are separated. It needs some locks. I stepped on Steve's toes again. I rewrote much of your stuff. :/ Apologies. I D-ified your code a little, I suppose. Sorry! But, hey, I also fixed a buffer overflow bug! I await the malloc and free code. :) -- Wilks 8065b90ab2df16e53dc2edda5c44cd105a3d5999 wilkie wilkie05@gmail.com 96b93882cde2fac8de719c9bc0269c43fc4d4e3e wilkie wilkie05@gmail.com http://github.com/xomboverlord/xomb/commit/53ab9423315579a1247c2605c7015eda97242f7b 53ab9423315579a1247c2605c7015eda97242f7b

2008-12-26T22:47:59-08:00

Alright. I added a few things to allow for device pages and buffers to be allocated by the kernel for the environment. The keyboard syscall initKeyboard is implemented. This syscall will allocate two pages to give the user program a keyboard buffer. The first page is the 'readable' page. This page contains the buffer itself and the write-pointer for the circular buffer. The second is the 'writeable' page, which contains the read-pointer. The Keyboard libOS, implemented in the libOS directory, implements the 'get' portion of the keyboard buffer and the code to do translations from keyboard scancodes to printable characters. The grabch() syscall is now completely gone. All character input from the keyboard is done completely in userland. Some keyboard scancode sets are required by hardware. My laptop, for instance, defaults to the old set 1. I cannot get the keyboard controller to tell me what scancode set it is currently using. Nor can I use the controller to set the scancode set. (It hangs in VMWare, it would sometimes ignore it in RL anyway). So, if it outputs random stuff instead of the keys you are pressing...that is currently an issue. Should be fine in Xen, Bochs, and VMWare which default to the set I am using (set 2) which is the most common code set. The code sets suggest how keys are mapped to values. http://www.computer-engineering.org/ps2keyboard/ 79e274089875f680b6baf36382f29b6a756f83e8 wilkie wilkie05@gmail.com 9e4abf179f23acd61c9ef7653b43a274387e338d wilkie wilkie05@gmail.com http://github.com/xomboverlord/xomb/commit/96b93882cde2fac8de719c9bc0269c43fc4d4e3e 96b93882cde2fac8de719c9bc0269c43fc4d4e3e

2008-12-26T13:51:24-08:00

Bah, something weird with ACPI. It hangs on my laptop. I'll disable it until I figure out why it does this and fall back onto the MP tables. 8017838d07c949608769a7796e32a5ba964e2d63 wilkie wilkie05@gmail.com 6b3553f7e8198c09f768ac78785b185b678cdb47 wilkie wilkie05@gmail.com http://github.com/xomboverlord/xomb/commit/9e4abf179f23acd61c9ef7653b43a274387e338d 9e4abf179f23acd61c9ef7653b43a274387e338d

2008-12-26T13:21:48-08:00

Fixed some kernel bugs. The ACPI tables will not be parsed if they are an older specification. The keyboard now initializes properly. It should run on actual hardware. I'm going to test that. I also added the stdarg.d from phobos to the std directory. Import std.stdarg. I have rewrote some of Steve's print function to use it. Hopefully he doesn't mind. Us the va_arg!() template to get at the parameters. I really hate the magic GCC does to implement this (including an empty module imports specific types??). I also hate just about all D runtimes. Bah! fcd0d297d78ebcdc98d8e169644cba09a8b14da9 wilkie wilkie05@gmail.com 85aeb6f4b233718918e0c3baeeb38757aa3672d9 373e0366f29e4e6e1b25718e833f51c1752cb156 Steve Klabnik steve@steveklabnik.com http://github.com/xomboverlord/xomb/commit/6b3553f7e8198c09f768ac78785b185b678cdb47 6b3553f7e8198c09f768ac78785b185b678cdb47

2008-12-22T16:25:04-08:00

Merge branch 'master' of git@github.com:untwisted/xomb f3a84525b341d4a8b85703b219194a972f190afb Steve Klabnik steve@steveklabnik.com 74b99326e8e0fa7e6ac651e5fd6c8900b92b0c9b Steve Klabnik steve@steveklabnik.com http://github.com/xomboverlord/xomb/commit/85aeb6f4b233718918e0c3baeeb38757aa3672d9 85aeb6f4b233718918e0c3baeeb38757aa3672d9

2008-12-22T16:22:41-08:00

Initial work on both malloc and a printf-style io function. 21bd622ebca87bdd329952403e81118d1f0d77b4 Steve Klabnik steve@steveklabnik.com 74b99326e8e0fa7e6ac651e5fd6c8900b92b0c9b wilkie wilkie05@gmail.com http://github.com/xomboverlord/xomb/commit/373e0366f29e4e6e1b25718e833f51c1752cb156 373e0366f29e4e6e1b25718e833f51c1752cb156

2008-12-20T20:24:42-08:00

Let's have that userspace heap allocator syscall working. I think the free works? I will test that. allocPage() -- returns a pointer to an allocated page. Sequential calls should have consecutive addresses. freePage() -- frees the last page in the heap. user/constants.d -- contains constants found at kernel/arch/x86_64/user.d, only one so far: PAGE_SIZE so, you know how much you've allocated. import user.constants; ... void* startHeap = allocPage(); void* endHeap = startHeap + Kernel.PAGE_SIZE; Hopefully that helps malloc creators! 33b925d2e45f14f4ebf58a6178eae54739730ce4 wilkie wilkie05@gmail.com 16aff416be346947d014b16a1e2d957390a041f5 wilkie wilkie05@gmail.com http://github.com/xomboverlord/xomb/commit/74b99326e8e0fa7e6ac651e5fd6c8900b92b0c9b 74b99326e8e0fa7e6ac651e5fd6c8900b92b0c9b

2008-12-20T14:19:19-08:00

I hope this works out. GRUB, contrary to what I have read, does not initialize the static segments, so... I will. The kernel still does not have them, but user applications loaded via GRUB modules will now have zeroed out .bss segments. Have fun. (I dislike ELF greatly) a28a0a8cc1622d6825664ba46a3984b00f3a89e2 wilkie wilkie05@gmail.com b06b928912b38d1834faab0558d0811272d8db89 wilkie wilkie05@gmail.com http://github.com/xomboverlord/xomb/commit/16aff416be346947d014b16a1e2d957390a041f5 16aff416be346947d014b16a1e2d957390a041f5

2008-12-19T20:48:19-08:00

Cloned dstubs to libd. We need to anti-stub this for userland. We need a runtime implementation. Anyone? 46351849969e159d434c9cb45b7abd547f92990e wilkie wilkie05@gmail.com 7b25cd3b7c6c8ab9997f40540770ea568f542c99 wilkie wilkie05@gmail.com http://github.com/xomboverlord/xomb/commit/b06b928912b38d1834faab0558d0811272d8db89 b06b928912b38d1834faab0558d0811272d8db89

2008-12-16T18:39:09-08:00

Don't need any object files! 148308d6856dc86df1e21a15674c783fe671ba8e wilkie wilkie05@gmail.com b6f75265f534e89bf14fe399e2a6d13e47d8b02f d5bdeeffd0284870ffead59d998b7f3b13eae0d0 wilkie wilkie05@gmail.com http://github.com/xomboverlord/xomb/commit/7b25cd3b7c6c8ab9997f40540770ea568f542c99 7b25cd3b7c6c8ab9997f40540770ea568f542c99

2008-12-16T18:38:15-08:00

It was being a pain. Merge branch 'master' of git@github.com:untwisted/xomb Conflicts: src/Makefile src/application/xsh.d src/application/xsh.o src/kernel/core/syscall.d src/kernel/dev/keyboard.d src/kernel/environment/scheduler.d src/kernel/environment/table.d src/kernel/kmain.d src/menu.lst src/user/syscall.d f879e1e9d02c2c7d93c41ca92901b0f236611573 wilkie wilkie05@gmail.com 75a3ff6760a494d66f38dcb80a18dd50d0afb784 wilkie wilkie05@gmail.com http://github.com/xomboverlord/xomb/commit/b6f75265f534e89bf14fe399e2a6d13e47d8b02f b6f75265f534e89bf14fe399e2a6d13e47d8b02f

2008-12-16T18:35:21-08:00

It won't let me push. cd6a6e0e7bf19ad07441d2e9be071421ab938470 wilkie wilkie05@gmail.com 6357e9d51f98d04a747b8e928da029f6d4142325 wilkie wilkie05@gmail.com http://github.com/xomboverlord/xomb/commit/75a3ff6760a494d66f38dcb80a18dd50d0afb784 75a3ff6760a494d66f38dcb80a18dd50d0afb784

2008-12-16T18:12:25-08:00

Hello, I have merged my changes, Brian's queue implementation, and Steve's shell. Auxilary Processors tell the scheduler of their presence and they are marked in a table as a resource. That is all they do, they are still never used. We may use this to provide user apps with these processors and take the boot processor to either become another cpu resource, or make it an option to turn on a kernel-only core. We want to provide both. It will not be hard to do. A new mapping is in place. Four tiers: Kernel shared, CPU specific, Environment-kernel shared, Environment-kernel cpu specific, and then also the user space regions should use up the rest. Devices and such are still in the same places, but stacks have moved around. Interrupts should work in the kernel and in userspace. Weird methods of doing it, but we use sysret when we can, unlike Linux to my knowledge which uses iretq always. I've tested them pretty well. We should be aware of the usage of the frame pointer, as any misuse or nonuse of it in the kernel code will break the interrupt mechanism and clobber the stack for the kernel code. There may be an issue in the schedule() function not scheduling a new environment when only the yield function seems to do so. The scheduling should be done by the schedule() function which should be called by the yield(). At least that is how I expect it to work. Brian can update this when he reverts to a priority queue. The shell polls for a keyboard press. This can be updated to a keyboard libOS instead of storing the keyboard buffer in the kernel space. That should limit the keyboard reads to take no context penalty. Keyboards can be read without switching due to a syscall. I hope to figure out how this simple kernel-known libOS interface will work soon and have it implemented. But, um, it works as is for now. I need to finish the cpu multi-core stuff... should be easy but it takes 25 minutes to boot the kernel via emulation...so yeah... Should be enough to have concurrent work. Yay us. 7c62cdb60b3d453b97ad9c5f76d707be61d7e056 wilkie wilkie05@gmail.com a67ad43a5156782028116a6a1c305f2dce6d63c3 untwisted untwisted@gmail.com http://github.com/xomboverlord/xomb/commit/d5bdeeffd0284870ffead59d998b7f3b13eae0d0 d5bdeeffd0284870ffead59d998b7f3b13eae0d0

2008-12-12T12:04:56-08:00

Added a templated queue to the kernel for use in the scheduler or other places... Scheduling is now done using a simple queue rather than the priority queue. 43ff03291a0c9f695ea8b03ff5b852d68cd0afa8 untwisted untwisted@gmail.com 3488fbf751b811f02ba70fd7d68e911cfcfc986b Steve Klabnik steve@steveklabnik.com http://github.com/xomboverlord/xomb/commit/a67ad43a5156782028116a6a1c305f2dce6d63c3 a67ad43a5156782028116a6a1c305f2dce6d63c3

2008-12-07T00:25:40-08:00

...hey, adding the xsh file to the repo would be great cf13aa60ed44329e0501cb8d9e9b84c21b2cb9f1 Steve Klabnik steve@steveklabnik.com