Thread: [ros-kernel] reactos/tools/makefile

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

Casper:

There is missing E at line 59 :

ifeq ($(HOST),mingw32-windows)
rsym$(EX_POSTFIX): rsym.c
	$(HOST_CC) $(CFLAGS) -DDOS_PATHS rsym.c -o rsym$(EXE_POSTFIX)

l=F8r, 2002-07-13 kl. 23:41 skrev David Kredba:
> Casper:
>=20
>=20
> There is missing E at line 59 :
>=20
> ifeq ($(HOST),mingw32-windows)
> rsym$(EX_POSTFIX): rsym.c
> 	$(HOST_CC) $(CFLAGS) -DDOS_PATHS rsym.c -o rsym$(EXE_POSTFIX)

Yes, a fix is in CVS.

Casper

Casper et all,

I can't seem to get your new debug info to work, nor can I get addr2line to
work. I must be doing something wrong, but I don't know what :(

I did a make clean, changed config as you suggested, and then did a make,
then ran install.bat. I'm still crashing on the zip drive, and yes, Eric,
it's an ATAPI as a previous e-mail showed. I noticed in atapi.c that if the
BytesPerSector is 0, it is set to 512. I don't know if that was just put in
or not, but it's still crashing.

Here's my files:

=== bootsym.bat ===
loadros system32\ntoskrnl.exe system32\hal.dll /DEBUGPORT=SCREEN bootsym.lst
=== end of bootsym.bat ===

=== bootsym.lst ===
system32\drivers\scsiport.sys
system32\drivers\atapi.sys
system32\drivers\class2.sys
system32\drivers\disk.sys
system32\drivers\vfatfs.sys
system32\config\system.hiv
system32\ntoskrnl\ntoskrnl.sym
hal\halx86\hal.sym
drivers\storage\scsiport\scsiport.sym
drivers\storage\atapi\atapi.sym
drivers\storage\class2\class2.sym
drivers\storage\disk\disk.sym
drivers\fs\vfat\vfatfs.sym
*
=== end of bootsym.lst ===

=== config ===
#
# Architecture to build for
#
# Specify one of: i386
# Possible values in the future: alpha,i386,m68k,mips,powerpc
ARCH := i386

#
# Whether to compile in the kernel debugger
#
KDBG := 0

#
# Whether to compile for debugging
#
DBG := 1

#
# Whether to compile a multiprocessor or single processor version
#
MP := 0

#
# Whether to compile for ACPI compliant systems
#
ACPI := 0
=== end of config ===

> I can't seem to get your new debug info to work, nor can I get
> addr2line to work. I must be doing something wrong, but I don't know 
> what :(

Loadros is a little bit bugy. It forgot to close the opened files. At
least loadros can load 15 files if the config.sys has a statement
files=20 (or greater). I fix this bug.

- Hartmut

s=F8n, 2002-07-14 kl. 03:00 skrev Royce Mitchell III:
> Casper et all,
>=20
> I can't seem to get your new debug info to work, nor can I get addr2line =
to
> work. I must be doing something wrong, but I don't know what :(

I couldn't get addr2line to work either.

>=20
> I did a make clean, changed config as you suggested, and then did a make,
> then ran install.bat. I'm still crashing on the zip drive, and yes, Eric,
> it's an ATAPI as a previous e-mail showed. I noticed in atapi.c that if t=
he
> BytesPerSector is 0, it is set to 512. I don't know if that was just put =
in
> or not, but it's still crashing.

IIRC BytesPerSector is changed to 512 because the code still assumes a
standard DOS compatible harddisk in this case. Eric can tell you wether
this is still true.

> Here's my files:

Your configuration looks okay.

I don't use install.bat because it is not always up-to-date. IIRC
symbols are not copied by install.bat. I do a "make install" or "make
<module>_install". The needed files are then copied to the
reactos/reactos directory. I then have a DOS .bat file that use xcopy to
copy the whole reactos/reactos directory to the installation directory.

Casper

"Casper Hornstrup" <ch...@us...> wrote:

> > I did a make clean, changed config as you suggested, and then did a
make,
> > then ran install.bat. I'm still crashing on the zip drive, and yes,
Eric,
> > it's an ATAPI as a previous e-mail showed. I noticed in atapi.c that if
the
> > BytesPerSector is 0, it is set to 512. I don't know if that was just put
in
> > or not, but it's still crashing.
>
> IIRC BytesPerSector is changed to 512 because the code still assumes a
> standard DOS compatible harddisk in this case. Eric can tell you wether
> this is still true.

Yes, this is true. The ATA specs say that 0 Bytes per Sector really mean 512
Bytes per Sector (default sector size). IMO it is a brain-dead concept but
most of the harddisks I have seen report 0 Bytes per Sector.

Eric

s=F8n, 2002-07-14 kl. 03:00 skrev Royce Mitchell III:
> Casper et all,
>=20
> I can't seem to get your new debug info to work, nor can I get addr2line =
to
> work. I must be doing something wrong, but I don't know what :(

Experiments show that you must add 0x11000 (for most drivers) to your
addresses when passed to addr2line. 0x10000 is probably the image base,
but where does the extra 0x1000 come from? You can get the exact offset
by using objdump -h. The value is the VMA of the .text section.

Casper

On Sun, Jul 14, 2002 at 11:42:48AM +0200, Casper Hornstrup wrote:
> Experiments show that you must add 0x11000 (for most drivers) to your
> addresses when passed to addr2line. 0x10000 is probably the image base,
> but where does the extra 0x1000 come from? You can get the exact offset
> by using objdump -h. The value is the VMA of the .text section.
> 
The kernel always prints code addresses relative to the start of the .text
segment of the module in question.

Yes! Thank You! Finally got a line number!!!!

Here's the dump info:

Page fault at high IRQL was 12
Bug detected code: 0x1D
Divide Error Exception: 0(2)
Processor: 0 CS:EIP 8:c00e3f98 <hal.dll: 3f98>

C:\reactos\hal\halx86>addr2line --exe=hal.nostrip.dll 0x14f98
//C/reactos/hal/halx86/time.c:199

Here's time.c:199:
  Val = HalpQueryCMOS(RTC_REGISTER_B);

Why would that get a div/0?

Here's 0x10000 offset:

C:\reactos\hal\halx86>addr2line --exe=hal.nostrip.dll 0x13f98
//C/reactos/hal/halx86/pci.c:411

Here's pci.c:411:
   return Length - Len;

Not sure why that would div/0 either :(

The .sym files aren't text anymore, so how do I determine what the .text
offset is for hal.dll?

----- Original Message -----
From: "Casper Hornstrup" <ch...@us...>
To: <rea...@li...>
Sent: Sunday, July 14, 2002 4:42 AM
Subject: Re: [ros-kernel] reactos/tools/makefile

søn, 2002-07-14 kl. 03:00 skrev Royce Mitchell III:
> Casper et all,
>
> I can't seem to get your new debug info to work, nor can I get addr2line
to
> work. I must be doing something wrong, but I don't know what :(

Experiments show that you must add 0x11000 (for most drivers) to your
addresses when passed to addr2line. 0x10000 is probably the image base,
but where does the extra 0x1000 come from? You can get the exact offset
by using objdump -h. The value is the VMA of the .text section.

Casper

-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
reactos-kernel mailing list
rea...@li...
https://lists.sourceforge.net/lists/listinfo/reactos-kernel

s=F8n, 2002-07-14 kl. 16:12 skrev Royce Mitchell III:
> Yes! Thank You! Finally got a line number!!!!
>=20
> Here's the dump info:
>=20
> Page fault at high IRQL was 12
> Bug detected code: 0x1D
> Divide Error Exception: 0(2)
> Processor: 0 CS:EIP 8:c00e3f98 <hal.dll: 3f98>
>=20
> C:\reactos\hal\halx86>addr2line --exe=3Dhal.nostrip.dll 0x14f98
> //C/reactos/hal/halx86/time.c:199
>=20
> Here's time.c:199:
>   Val =3D HalpQueryCMOS(RTC_REGISTER_B);
>=20
> Why would that get a div/0?
>=20
>=20
> Here's 0x10000 offset:
>=20
> C:\reactos\hal\halx86>addr2line --exe=3Dhal.nostrip.dll 0x13f98
> //C/reactos/hal/halx86/pci.c:411
>=20
> Here's pci.c:411:
>    return Length - Len;
>=20
> Not sure why that would div/0 either :(
>=20

I think the Divide Error Exception is incorrectly displayed.

> The .sym files aren't text anymore, so how do I determine what the .text
> offset is for hal.dll?

All kernel mode drivers have .text offset 0x11000.

You can get it by using objdump -h or nm -n.

Casper

On Sun, Jul 14, 2002 at 09:12:56AM -0500, Royce Mitchell III wrote:
> Yes! Thank You! Finally got a line number!!!!
> 
> Here's the dump info:
> 
> Page fault at high IRQL was 12
> Bug detected code: 0x1D
> Divide Error Exception: 0(2)
> Processor: 0 CS:EIP 8:c00e3f98 <hal.dll: 3f98>
> 
> C:\reactos\hal\halx86>addr2line --exe=hal.nostrip.dll 0x14f98
> //C/reactos/hal/halx86/time.c:199
> 
> Here's time.c:199:
>   Val = HalpQueryCMOS(RTC_REGISTER_B);
> 
> Why would that get a div/0?
> 
> 
> Here's 0x10000 offset:
> 
> C:\reactos\hal\halx86>addr2line --exe=hal.nostrip.dll 0x13f98
> //C/reactos/hal/halx86/pci.c:411
> 
> Here's pci.c:411:
>    return Length - Len;
> 
> Not sure why that would div/0 either :(
> 
> 
> The .sym files aren't text anymore, so how do I determine what the .text
> offset is for hal.dll?
> 
The trap number was reported incorrectly until a few days ago. I do not
think either of the lines you have got are right, is it possible that
the hal you are using for the line numbers was built with different options
to the one that produced the error?

I will double-check it again now.

By the way, I noticed something interesting when I tried to boot ros this
morning...

(class2.c:1173) SectorSize: 512  SectorCount: 0

Could SectorCount of 0 cause a problem?

----- Original Message -----
From: "David Welch" <we...@cw...>
To: <rea...@li...>
Sent: Sunday, July 14, 2002 11:13 AM
Subject: Re: [ros-kernel] reactos/tools/makefile

> On Sun, Jul 14, 2002 at 09:12:56AM -0500, Royce Mitchell III wrote:
> > Yes! Thank You! Finally got a line number!!!!
> >
> > Here's the dump info:
> >
> > Page fault at high IRQL was 12
> > Bug detected code: 0x1D
> > Divide Error Exception: 0(2)
> > Processor: 0 CS:EIP 8:c00e3f98 <hal.dll: 3f98>
> >
> > C:\reactos\hal\halx86>addr2line --exe=hal.nostrip.dll 0x14f98
> > //C/reactos/hal/halx86/time.c:199
> >
> > Here's time.c:199:
> >   Val = HalpQueryCMOS(RTC_REGISTER_B);
> >
> > Why would that get a div/0?
> >
> >
> > Here's 0x10000 offset:
> >
> > C:\reactos\hal\halx86>addr2line --exe=hal.nostrip.dll 0x13f98
> > //C/reactos/hal/halx86/pci.c:411
> >
> > Here's pci.c:411:
> >    return Length - Len;
> >
> > Not sure why that would div/0 either :(
> >
> >
> > The .sym files aren't text anymore, so how do I determine what the .text
> > offset is for hal.dll?
> >
> The trap number was reported incorrectly until a few days ago. I do not
> think either of the lines you have got are right, is it possible that
> the hal you are using for the line numbers was built with different
options
> to the one that produced the error?
>
>
> -------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> reactos-kernel mailing list
> rea...@li...
> https://lists.sourceforge.net/lists/listinfo/reactos-kernel
>

"Royce Mitchell III" <ro...@ev...> wrote:

> By the way, I noticed something interesting when I tried to boot ros this
> morning...
>
> (class2.c:1173) SectorSize: 512  SectorCount: 0
>
> Could SectorCount of 0 cause a problem?

It is okay if it is reported by your empty ZIP drive. Obviously, a harddisk
should never report this.

Eric

Crash line: ntoskrnl.exe <50deb>

C:\reactos\ntoskrnl>objdump -h ntoskrnl.exe

ntoskrnl.exe:     file format pei-i386

Sections:
Idx Name          Size      VMA       LMA       File off  Algn
  0 .text         0007d000  c0001000  c0001000  00001000  2**2
                  CONTENTS, ALLOC, LOAD, READONLY, CODE, DATA
  1 .data         00004000  c007e000  c007e000  0007e000  2**2
                  CONTENTS, ALLOC, LOAD, DATA
  2 .edata        00007000  c0082000  c0082000  00082000  2**2
                  CONTENTS, ALLOC, LOAD, DATA
  3 .idata        00001000  c0089000  c0089000  00089000  2**2
                  CONTENTS, ALLOC, LOAD, DATA
  4 .rsrc         00002000  c008a000  c008a000  0008a000  2**2
                  CONTENTS, ALLOC, LOAD, DATA
  5 .bss          000342f0  c008c000  c008c000  00000000  2**2
                  ALLOC
  6 .reloc        00007000  c00c1000  c00c1000  0008c000  2**2
                  CONTENTS, ALLOC, LOAD, DATA

C:\reactos\ntoskrnl>addr2line --exe=ntoskrnl.nostrip.exe 0xc0051deb
//C/reactos/ntoskrnl/mm/section.c:528

section.c:528 is:
    Entry = MmGetPageEntrySectionSegment(Segment, Offset.u.LowPart);

MAKE INSTALL appears to be broken.

It's not copying atapi.sys, disk.sys, or class2.sys to the system32\drivers
directory.

Thread: [ros-kernel] reactos/tools/makefile

An operating system based on the best Windows NT design principles

reactos-kernel