Andrew Randrianasulu | 19 Jun 2009 16:47
Picon
Favicon

build.sh release failure for cross-compilation (i386->sgimips)


Looks like Makefile in /src/src/external/mit/xorg/server/xorg-server/hw/xfree86/xorgos has wrong
lines about including  bsdResource.c , file was deleted in xorg master branch back in august, 2008:

http://lists.freedesktop.org/pipermail/xorg-commit/2008-August/017651.html

Y. Takizawa | 22 Dec 2004 09:19

Re: build.sh release failure on alpha (-current)

> While cross-building -current alpha on i386, i noticed that `build.sh
> release' fails :
> 
> /local/src/NetBSD/tool/alpha/bin/alpha--netbsd-ld -Ttext 0x20000000 -N -e start -o
bootxx_ffs.sym  start.o bootxx.o booted_dev.o blkdev.o prom.o prom_disp.o putstr.o panic_putstr.o
/local/src/NetBSD/src/sys/arch/alpha/stand/bootxx_ffs/obj/lib/sa/libsa.a
/local/src/NetBSD/src/sys/arch/alpha/stand/bootxx_ffs/obj/lib/kern/libkern.a -Map bootxx_ffs.map
>    text    data     bss     dec     hex filename
>    7507     336     348    8191    1fff bootxx_ffs.sym
> creating bootxx_ffs from bootxx_ffs.sym...
> -rw-r--r--  1 njoly  users  7848 Dec 20 16:54 bootxx_ffs
> checking sizes for bootxx_ffs/bootxx_ffs.sym... MAXIMUM LOAD SIZE EXCEEDED (7848 > 7680)

I caught the same failure on i386 and macppc cross-building.

I think that bootxx_ffs needs the size less than 7680 bytes
to write to 1.2M boot floppy first track, referring to
sys/arch/i386/stand/bootxx/Makefile.bootxx for example.

NetBSD/alpha has no 1.2M boot floppy now, maybe you can
_void_ the failure to increase checking size (like the following
patch to fit 1.4M floppy first track). I don't know it is correct
to fix for boot disks and release building.

Additionaly, the size of bootxx_cd9660 seems to be less than
7680 bytes on -current like this:

creating bootxx_cd9660 from bootxx_cd9660.sym...
-rw-r--r--  1 root  wheel  7176 Dec 22 16:33 bootxx_cd9660
checking sizes for bootxx_cd9660/bootxx_cd9660.sym... OK
(Continue reading)

Y. Takizawa | 22 Dec 2004 15:37

Re: build.sh release failure on alpha (-current)

In message ``Re: build.sh release failure on alpha (-current)''
	Wed Dec 22 17:19:23 JST 2004,
	you <at> nightbreak.org (Y. Takizawa) writes:

> > While cross-building -current alpha on i386, i noticed that `build.sh
> > release' fails :
...
> I caught the same failure on i386 and macppc cross-building.
> 
> I think that bootxx_ffs needs the size less than 7680 bytes
> to write to 1.2M boot floppy first track, referring to
> sys/arch/i386/stand/bootxx/Makefile.bootxx for example.
> 
> NetBSD/alpha has no 1.2M boot floppy now, maybe you can
> _void_ the failure to increase checking size (like the following
> patch to fit 1.4M floppy first track). I don't know it is correct
> to fix for boot disks and release building.

I confirmed that build.sh release is finished using the patch
mentioned above, but could not boot from my AlphaPC 164LX
SCSI disk installed the bootxx_ffs like this:

>>>boot dka0
(boot dka0.0.0.9.0 -flags a)
block 0 of dka0.0.0.9.0 is a valid boot block
reading 16 blocks from dka0.0.0.9.0
...
jumping to bootstrap code

halted CPU 0
(Continue reading)

Bill Studenmund | 24 Dec 2004 08:54
Picon

Re: build.sh release failure on alpha (-current)

On Wed, Dec 22, 2004 at 11:37:26PM +0900, Y. Takizawa wrote:
> In message ``Re: build.sh release failure on alpha (-current)''
> 	Wed Dec 22 17:19:23 JST 2004,
> 	you <at> nightbreak.org (Y. Takizawa) writes:
> 
> > > While cross-building -current alpha on i386, i noticed that `build.sh
> > > release' fails :
> ...
> > I caught the same failure on i386 and macppc cross-building.
> > 
> > I think that bootxx_ffs needs the size less than 7680 bytes
> > to write to 1.2M boot floppy first track, referring to
> > sys/arch/i386/stand/bootxx/Makefile.bootxx for example.

Note that 7680 is 8192 - 512. That's 8k minus one block.

[snip]

> Maybe SRM firmware also read just 16 sectors (8192 bytes)
> at boot. Now I have no idea to fix this problem.

ffs's superblock is at 8k in. So the boot code HAS to stop before it.

Take care,

Bill
Y. Takizawa | 25 Dec 2004 21:53

Re: build.sh release failure on alpha (-current)

In message ``Re: build.sh release failure on alpha (-current)''
	Fri Dec 24 16:54:41 JST 2004,
	wrstuden <at> NetBSD.org (Bill Studenmund) writes:

> > Maybe SRM firmware also read just 16 sectors (8192 bytes)
> > at boot. Now I have no idea to fix this problem.
> 
> ffs's superblock is at 8k in. So the boot code HAS to stop before it.

Now I found the restriction at installboot(8)...

By the way, I used 2.0 boot floppies to write the primary
bootstrap program for booting the secondary bootstrap and
kernel program of current, it seems to be OK. Please tell
me suggestions if it is wrong.

--
Y. Takizawa
you <at> nightbreak.org, takizawa <at> computer.org

James Chacon | 22 Dec 2004 17:33
Picon

Re: build.sh release failure on alpha (-current)

On Wed, Dec 22, 2004 at 11:37:26PM +0900, Y. Takizawa wrote:
> In message ``Re: build.sh release failure on alpha (-current)''
> 	Wed Dec 22 17:19:23 JST 2004,
> 	you <at> nightbreak.org (Y. Takizawa) writes:
> 
> > > While cross-building -current alpha on i386, i noticed that `build.sh
> > > release' fails :
> ...
> > I caught the same failure on i386 and macppc cross-building.
> > 
> > I think that bootxx_ffs needs the size less than 7680 bytes
> > to write to 1.2M boot floppy first track, referring to
> > sys/arch/i386/stand/bootxx/Makefile.bootxx for example.
> > 
> > NetBSD/alpha has no 1.2M boot floppy now, maybe you can
> > _void_ the failure to increase checking size (like the following
> > patch to fit 1.4M floppy first track). I don't know it is correct
> > to fix for boot disks and release building.
> 
> 
> I confirmed that build.sh release is finished using the patch
> mentioned above, but could not boot from my AlphaPC 164LX
> SCSI disk installed the bootxx_ffs like this:

It's a problem w. binutils adding extra sections to all alpha binaries
(and only alpha binaries) which makes this grow too large.

James

(Continue reading)


Gmane