Page 2 of 3
Re:blueillusion v0.036 - request for tests
Posted: Wed Sep 15, 2004 7:03 am
by distantvoices
@Pype, Curufir, Brendan: could you please be so kind and roll a test on your machines (or qemu) again? I'd be grateful for your patience. With bochs it still doesn't work.
I have hopefully resolved this weird problem with that OS working on one machine and refusing completely to work on the other:
I've simply added zeroing out of pages and pagetables - et voila, even my most stubborn and denying toshiba laptop accepts it and boots up fine. Well, the GUI subsystem will still not work properly with it, as I need to implement some ATI register handlings. If that's not a pace forward, what is it then? ;D
thank you
Re:blueillusion v0.036 - request for tests
Posted: Wed Sep 15, 2004 9:25 am
by Pype.Clicker
i'll test it, but don't expect the test to happen soon: i'm in 'mv flat/* new-home/' mode, so laptop is packed in box with other stuffs
Re:blueillusion v0.036 - request for tests
Posted: Thu Sep 16, 2004 12:02 am
by BI lazy
Hey, Good Luck with the moving and with your new home, man!
Re:blueillusion v0.036 - request for tests
Posted: Fri Sep 17, 2004 6:33 am
by Brendan
Hi,
I did some more testing - less than last time because half of them stopped with "16 MBSTOP! memory requirement: at least 32 Mb! System halted"...
Code: Select all
COMPUTER N
AMD-K6(tm) 3D processor 300 Mhz
64 Mb memory
32x CD-rom
1335Mb hard disk
S3 video card
ethernet card
WORK MACHINE
Pentium IV 1.66 Ghz
256 Mb RAM
40 Gb hard drive
8 Gb hard drive
NVIDEA RIVA TNT2 Model64 Pro video card
NETGEAR FA311/FA312 PCI Ethernet card
Both these computers crashed during boot (tried it 5 times with 2 different floppies), with a heap of exceptions. The exception reports where almost the same, except for EDI, proz and pgdir. I'm not sure how many there where (they scrolled past too fast, but I think there was 6 or 7 of them). The last exception report was:
FSSRV: READ mode and no block number read...
exception ...:0 in proz: 6A, pgdir EE7067
k->gs: 23, k->fs: 23, k->es:23, k->ds: 23
k->eax: B000FF20,k->ecx: B000FFF3, k->edx:0, k->ebx: 1
k->esp: B000FFEF,k->ebp: 0, k->esi:B000FF08, k->edi: C0031324
k->eip: 1B, k->cs: 13282, k->eflags:B000FEEC
k->u_esp: 23, k->u_ss: 0
"proz" increased by one each exception
"pgdir" was different for each exception
EDI increased by 0x1000 each exception
Everything else is the same for each exception.
COMPUTER O
Pentium Pro 200 MHz
128 Mb RAM
Adaptec AHA-2940 SCSI controller
Compaq ST32550W SCSI hard-drive
IBM DCAS-34330W SCSI hard-drive
IBM DGHS09U SCSI hard-drive
Sony CDU-76S SCSI CD-ROM
Ethernet on motherboard
It's not a problem with your code, but GRUB doesn't boot this computer after 1 second like all other computers. After pressing enter it booted ok. Tried "gui" and it displayed "VESA nicht unterstuetzt", loaded something from the floppy and locked up. I tried to boot it about 7 times after that but it didn't respond to the keyboard. I found this odd as it worked at least once. It isn't a hardware problem though because the keyboard works fine until GRUB starts booting your OS.
Code: Select all
COMPUTER P
Dual Pentium III 1 GHz
512 Mb RAM
Adaptec AIC-7899 dual-channel SCSI on motherboard
2 * 9 Gb Quantum ATLAS SCSI hard-drives
IDE CD-ROM (50x generic)
Ethernet on motherboard plus 2 PCI ethernet cards
Booted well and everything mostly worked - see comments from last time...
COMPUTER Q (Compaq Proliant 1600)
Dual Pentium II 400 MHz
384 Mb RAM
Dual-channel SCSI on motherboard
5 * 4.5 Gb SCSI hard-drives
IDE CD-ROM
Ethernet on motherboard
Boot ok. Tried "gui" and it displayed "VESA nicht unterstuetzt", loaded something from the floppy and locked up. I rebooted and tried "calc" (no arguments) and got an "address space violation at A0022001 - pid:100" and locked it up again.
Cheers,
Brendan
Re:blueillusion v0.036 - request for tests
Posted: Wed Nov 03, 2004 7:38 am
by distantvoices
After having spent some time doing kernel grunt work, I've released a test image again: here it is.
http://www.distantvoices.org/download/b ... mg.tar.bz2
There 's been quite some change, so I expect a lot of bugs showing up on real machines. Gonna roll some tests myself too, and ask for additional testing in the meantime. (It works great on vmware).
Thanks a bunch
Re:blueillusion v0.036 - request for tests
Posted: Wed Nov 03, 2004 11:03 am
by Curufir
Mostly works on QEMU (Only the gui part is screwed up).
On real hardware (NFORCE 2, Athlon XP, 512Mb RAM) it craps out just after 'guisrv:init ... done' with what looks to be an error reading from the disk by FSSRV.
Tried a couple of different floppies and all had the same problem.
Re:blueillusion v0.036 - request for tests
Posted: Wed Nov 03, 2004 1:26 pm
by distantvoices
First, thank you very much!
Does gui stuff in QEMU now. a small hack did the trick. The SVGA Bios reports a linear framebuffer at address 0xf000. That's waaay off. Just query for this address and change it to 0xf0000000. on real hardware or in vmware i always get linear framebuffer located at addresses like 0xf3000000 or 0xe6000000. Hm. Must be some trick with the svga bios the qemu guys are delivering.
Have done some debugging with qemu 0.60. Amazing, how fast this emulation stuff is. Bochs crawls sort of, but qemu runs.
*mmpf* wish I had such a hardware to track this floppy bug down.
Re:blueillusion v0.036 - request for tests
Posted: Wed Nov 03, 2004 3:53 pm
by Curufir
GUI works on qemu now, but I just noticed that the available memory is getting reported wrong. With 32Mb it's correct, but with 128Mb it's reporting 130Mb, and with 256Mb it reports 261Mb.
Odd stuff, something for someone to verify on real hardware ;D
Re:blueillusion v0.036 - request for tests
Posted: Thu Nov 04, 2004 1:21 am
by Candy
Curufir wrote:
GUI works on qemu now, but I just noticed that the available memory is getting reported wrong. With 32Mb it's correct, but with 128Mb it's reporting 130Mb, and with 256Mb it reports 261Mb.
Odd stuff, something for someone to verify on real hardware ;D
Are you not reporting cut off thousands of kilobytes?
32MB -> 32768 KB -> 32448 KB usable -> 32k KB
128MB -> 131072 KB -> 130752 KB usable -> 130k KB
256MB -> 262144 KB -> 261824 KB usable -> 261k KB
HTH, Candy
Re:blueillusion v0.036 - request for tests
Posted: Thu Nov 04, 2004 4:36 am
by Pype.Clicker
testing on
toshiba laptop again
i get "VBE2.0 LFB @ fd00000", then fssrv & guisrv are initialized. I then get "you're welcome for an experiment" message.
It looks like my floppy has troubles (hearing re-seek sounds) while running programs
once a program has been launched on a console, it looks like i cannot run more things on that console (no more character echoes and no return to the shell. I got process 102-107 with NO_NAME in state 9 and priority 5 ...
Re:blueillusion v0.036 - request for tests
Posted: Thu Nov 04, 2004 4:49 am
by Pype.Clicker
hum not easy to type with that german keyboard layout ... maybe binding F12 to a "show keyboard map" code would be great
btw, i have trouble with the "backwards" key aswell. My screen currently shows
BlueIllusion OS v0.036 - pre 0.04 - in 1, out 2
You are welcome for an experiment
> semtest -c
file not found: sesemtest
and the shell still doesn't return (with new floppy)
I type "help" on another console and get help, then "press any key to continue", but whatever key pressed, nothing else occurs (including keys "A", "N" and "Y")
"gui" leads to a black blank screen.
Re:blueillusion v0.036 - request for tests
Posted: Thu Nov 04, 2004 6:31 am
by distantvoices
have found the bug almost immediately. My bad not to indicate who is sending the "wakeup message" on exit. Since the process waiting for pid XXX expects a message from SYSTEM, and I 've ommitted so that naught was indicated, the message passing subsystem didn't unblock the process in order to fetch the desired message. *hmpf*
An updated and corrected version is online. Thanks a bunch for testing.
Regarding this LFB-Address, which the toshiba laptop is reporting: hmmm gunna look. what does pkd's vbe-test program report as lfb address?
Re:blueillusion v0.036 - request for tests
Posted: Thu Nov 04, 2004 6:44 am
by Pype.Clicker
i cannot get pkd's test image: i get a 404 ...
note that the LFB address seems to match with the PCI-detection reported address for the LFB, so everything should be fine ...
The commands seems to complete now (just tested 'help' so far), but "gui" still leads to a black screen ...
Is there a magic key to get out of that GUI back to the console ?
Re:blueillusion v0.036 - request for tests
Posted: Thu Nov 04, 2004 7:27 am
by distantvoices
On the test report page about your laptop, it says LFB(base0) is at 0xfd000000. --> "VBE2.0 LFB @ fd00000"
So, either my vbe-lfb fetching routine is buggy (it does this with bochs and qemu LFB's too) or you've omitted a zero at the end by typo:P.
Gonna give it a key to get back to text mode. But if there are no floppy_scratching_sounds and caps_lock doesn't react anymore, Chances are that it 's crapped out without mercy.
Thanks for testing.
I appreciate it
Re:blueillusion v0.036 - request for tests
Posted: Thu Nov 04, 2004 9:20 am
by Curufir
The output reads like this:
Reported Memory - low 639
Reported Memory - upper 261120
Memory total on system: 261Mb
This is on a system with 256Mb of memory.
The reported memory for upper and low is correct, it's the memory total that is wrong. After trying out a few different numbers it almost looks like you're treating 1Mb = 1000Kb.