1) Assertion gets hit on:
Assertion PdoNameInfo->Name.Length failed at io/deviface.c:606 Entered debugger on embedded INT3 at 0x0008:0x80005813.
2) Cmd.exe outputs garbage.
3) DHCP crashes during boot.
Best regards, Alex Ionescu
Alex Ionescu wrote:
- Assertion gets hit on:
Assertion PdoNameInfo->Name.Length failed at io/deviface.c:606 Entered debugger on embedded INT3 at 0x0008:0x80005813.
Cmd.exe outputs garbage.
DHCP crashes during boot.
Best regards, Alex Ionescu _______________________________________________
4) 14893 breaks deleting a file.
Alex Ionescu wrote:
Alex Ionescu wrote:
- Assertion gets hit on:
Assertion PdoNameInfo->Name.Length failed at io/deviface.c:606 Entered debugger on embedded INT3 at 0x0008:0x80005813.
Cmd.exe outputs garbage.
DHCP crashes during boot.
Best regards, Alex Ionescu _______________________________________________
- 14893 breaks deleting a file.
5) install not possible, because usetup does open files with not enough rights (same as DeleteFile 4)
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Hartmut Birr schrieb:
Alex Ionescu wrote:
Alex Ionescu wrote:
- Assertion gets hit on:
Assertion PdoNameInfo->Name.Length failed at io/deviface.c:606 Entered debugger on embedded INT3 at 0x0008:0x80005813.
Cmd.exe outputs garbage.
DHCP crashes during boot.
Best regards, Alex Ionescu _______________________________________________
- 14893 breaks deleting a file.
- install not possible, because usetup does open files with not enough
rights (same as DeleteFile 4)
6) Freeloader segfaults always. See attatchment.
Greetings, Jan Schiefer!
From: Jan Schiefer
- Freeloader segfaults always. See attatchment.
Can you give some more information: contents of your freeldr.ini file and layout (partition table) of your harddisk(s)? How did you boot, boot cd, live cd, boot floppy or plain harddisk boot? If harddisk boot, is freeldr installed in the boot sector or are you loading freeldr from another boot loader (chain loading or multiboot)?
Thanks, Gé van Geldorp.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Ge van Geldorp schrieb:
From: Jan Schiefer
- Freeloader segfaults always. See attatchment.
Can you give some more information: contents of your freeldr.ini file and layout (partition table) of your harddisk(s)? How did you boot, boot cd, live cd, boot floppy or plain harddisk boot? If harddisk boot, is freeldr installed in the boot sector or are you loading freeldr from another boot loader (chain loading or multiboot)?
Thanks, Gé van Geldorp.
I just build the current SVN code and tried to start from cd. Both, bootcd and livecd fail. My hd is empty.
Greetings, Jan Schiefer!
From: Jan Schiefer
I just build the current SVN code and tried to start from cd. Both, bootcd and livecd fail. My hd is empty.
A few more questions: looking at your screenshot, I guess you're using qemu on Linux? What qemu version? What's the command line used to start qemu? And just to be sure, what's the output of "svn info" if you run it in reactos/boot/freeldr/freeldr?
Thanks, Gé van Geldorp.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Ge van Geldorp schrieb:
From: Jan Schiefer
I just build the current SVN code and tried to start from cd. Both, bootcd and livecd fail. My hd is empty.
A few more questions: looking at your screenshot, I guess you're using qemu on Linux? What qemu version? What's the command line used to start qemu? And just to be sure, what's the output of "svn info" if you run it in reactos/boot/freeldr/freeldr?
Thanks, Gé van Geldorp.
Yes, taht's Linux qemu. Same behavior for 0.7.0 and 0.6.1. Commandline is just "qemu -hda ros-disk.img -cdrom ReactOS.iso ( or roslive.iso ) - -boot d".
"svn info" on reactos/boot/freeldr/freeldr:
Path: . URL: svn://svn.reactos.com/trunk/reactos/boot/freeldr/freeldr Repository UUID: e2f519d9-20a2-bc46-8705-e3bf7245ce19 Revision: 14951 Node Kind: directory Schedule: normal Last Changed Author: ion Last Changed Rev: 14926 Last Changed Date: 2005-05-01 22:38:29 +0200 (Sun, 01 May 2005) Properties Last Updated: 2005-04-26 15:58:11 +0200 (Tue, 26 Apr 2005)
Greetings, Jan Schiefer!
Jan Schiefer schrieb:
Yes, taht's Linux qemu. Same behavior for 0.7.0 and 0.6.1. Commandline is just "qemu -hda ros-disk.img -cdrom ReactOS.iso ( or roslive.iso )
- -boot d".
I also have problems with the ReactOS (CVS HEAD) in Bochs (CVS HEAD). It always aborts with the message:
interrupt: tss_selector.ti=1
Regards, Mark
From: Mark Junker
I also have problems with the ReactOS (CVS HEAD) in Bochs (CVS HEAD). It always aborts with the message:
interrupt: tss_selector.ti=1
Would it be possible for you to check if this also happens with the last stable Bochs release (i.e. ReactOS HEAD on Bochs 2.1.1). I can't reproduce it in that environment.
Gé van Geldorp.
Ge van Geldorp schrieb:
Would it be possible for you to check if this also happens with the last stable Bochs release (i.e. ReactOS HEAD on Bochs 2.1.1). I can't reproduce it in that environment.
I tried it with 2.1.1 but the problem remains.
Regards, Mark
From: Mark Junker
Ge van Geldorp schrieb:
Would it be possible for you to check if this also happens with the last stable Bochs release (i.e. ReactOS HEAD on Bochs 2.1.1). I can't reproduce it in that environment.
I tried it with 2.1.1 but the problem remains.
Ok, thanks for trying. Is this on Linux or Windows? Can you send me your bochsrc?
Gé van Geldorp.
Ge van Geldorp schrieb:
Would it be possible for you to check if this also happens with the last stable Bochs release (i.e. ReactOS HEAD on Bochs 2.1.1). I can't reproduce it in that environment.
After multiple deletions of all *.d, *.o, *.a, *.exe and re-compilation with DEBUG=1 and KDBG=1, it finally worked (don't know why).
Regards, Mark
From: Jan Schiefer
Yes, taht's Linux qemu. Same behavior for 0.7.0 and 0.6.1. Commandline is just "qemu -hda ros-disk.img -cdrom ReactOS.iso ( or roslive.iso )
- -boot d".
Well, that seems pretty normal. I'm trying to figure out why I can't reproduce the crash. Could you please try with ftp://ftp.geldorp.nl/pub/ReactOS/roslive-14955.iso.gz (to rule out a build tools problem).
Gé van Geldorp.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
This one works. :O
My build-tools: Mingw 3.4.2 Mingw binutils 2.15.94 NASM 0.98.38
Greetings, Jan Schiefer!
From: Jan Schiefer
This one works. :O
So it seems build-related.
My build-tools: Mingw 3.4.2 Mingw binutils 2.15.94 NASM 0.98.38
But then again, these are the exact same tool versions I'm using (the .iso I provided was built on Linux too). I'll have to think a bit about how to move on from here. If anyone else has a brilliant idea (actually, let's drop the brilliant, any idea would be fine) please feel free to jump in.
Gé van Geldorp.
On Tuesday 03 May 2005 12:44, Ge van Geldorp wrote:
From: Jan Schiefer
This one works. :O
So it seems build-related.
My build-tools: Mingw 3.4.2 Mingw binutils 2.15.94 NASM 0.98.38
But then again, these are the exact same tool versions I'm using (the .iso I provided was built on Linux too). I'll have to think a bit about how to move on from here. If anyone else has a brilliant idea (actually, let's drop the brilliant, any idea would be fine) please feel free to jump in.
Gé van Geldorp.
I once had problems building ReactOS on Linux - it crashed in pci.sys iirc. The same version built on Windows worked. I did a diff on the objdump output for both pci.sys files then, and iirc in one was an instruction like "movl %eax, %edx(%edi)" and in the other it was "movl %eax, %edi(%edx)" - I replaced the pci.sys built on Linux by the one built on Windows and it crashed in some other .sys file - I dont know if the difference in this instruction was causing the crash or if it was something else (maybe in the .data segment or PE file format)
Did you already try to do a diff on the binaries?
hth
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Anich Gregor schrieb:
On Tuesday 03 May 2005 12:44, Ge van Geldorp wrote:
From: Jan Schiefer
This one works. :O
So it seems build-related.
My build-tools: Mingw 3.4.2 Mingw binutils 2.15.94 NASM 0.98.38
But then again, these are the exact same tool versions I'm using (the .iso I provided was built on Linux too). I'll have to think a bit about how to move on from here. If anyone else has a brilliant idea (actually, let's drop the brilliant, any idea would be fine) please feel free to jump in.
Gé van Geldorp.
I once had problems building ReactOS on Linux - it crashed in pci.sys iirc. The same version built on Windows worked. I did a diff on the objdump output for both pci.sys files then, and iirc in one was an instruction like "movl %eax, %edx(%edi)" and in the other it was "movl %eax, %edi(%edx)" - I replaced the pci.sys built on Linux by the one built on Windows and it crashed in some other .sys file - I dont know if the difference in this instruction was causing the crash or if it was something else (maybe in the .data segment or PE file format)
Did you already try to do a diff on the binaries?
hth
Mybe I can send you one of my builds to debug? How about that?
Greetings, Jan Schiefer!
From: Jan Schiefer
Mybe I can send you one of my builds to debug? How about that?
Yes, let's try that. My mailbox should be big enough, can you try tarring up reactos/boot/freeldr/freeldr and send it to me? If that doesn't work we'll find some other way to transfer the thing.
Gé van Geldorp.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Ge van Geldorp schrieb:
From: Jan Schiefer
Mybe I can send you one of my builds to debug? How about that?
Yes, let's try that. My mailbox should be big enough, can you try tarring up reactos/boot/freeldr/freeldr and send it to me? If that doesn't work we'll find some other way to transfer the thing.
Gé van Geldorp.
Holy shit. Now it works.
I deleted the reactos-folder ( up-to-date with "svn up" ) and did a new svn checkout ( with "svn co" )
I think it has something to do with the source management. "svn up" doesn't seem to work correctly, I had similiar experiences before. When I did "svn up" instead of deleting the whole ros-folder combined with a new checkout I got sometimes wired compile errors.
Greetings, Jan Schiefer!
Jan Schiefer wrote:
Holy shit. Now it works.
I deleted the reactos-folder ( up-to-date with "svn up" ) and did a new svn checkout ( with "svn co" )
I think it has something to do with the source management. "svn up" doesn't seem to work correctly, I had similiar experiences before. When I did "svn up" instead of deleting the whole ros-folder combined with a new checkout I got sometimes wired compile errors.
Greetings, Jan Schiefer!
Are you sure that you always make clean and delete all *.d files from the root folder? Sometimes that can be a problem too... And check your time/date...SVN up can mess up if those are incorrect.
Best regards, Alex Ionescu
From: Jan Schiefer
Holy shit. Now it works.
I deleted the reactos-folder ( up-to-date with "svn up" ) and did a new svn checkout ( with "svn co" )
I think it has something to do with the source management. "svn up" doesn't seem to work correctly, I had similiar experiences before. When I did "svn up" instead of deleting the whole ros-folder combined with a new checkout I got sometimes wired compile errors.
Maybe it's a dependencies problem. There's no (good) dependency tracking in freeldr, so it's entirely possible that a change in a header file (my guess is include/machine.h) doesn't result in a recompilation of files which really should be recompiled. Freeldr needs to be integrated into our build system better (it still shows that it used to be a separate module), but it doesn't make sense to do that on HEAD right now, better wait for the xmlbuild branch.
Gé van Geldorp.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
However it seems that the last pack of changes to frrloader broke it for me. Maybe we should do regression testing there?
Greetings, Jan Schiefer!
Alex Ionescu a écrit :
- Assertion gets hit on:
Assertion PdoNameInfo->Name.Length failed at io/deviface.c:606 Entered debugger on embedded INT3 at 0x0008:0x80005813.
Fixed in revision 14924 http://reactos.com:8080/pipermail/ros-svn/2005-May/002543.html
Alex Ionescu a écrit :
- Cmd.exe outputs garbage.
Fixed in revision 14939 http://reactos.com:8080/pipermail/ros-svn/2005-May/002558.html