Randall Q. Huxley III wrote:
> Hi guys,
>
> I am trying to install Solaris 10 11/06 on a new box from CDs that are known
> to be good. I used this set to install Solaris 3 other times on other boxes.
>
> I am getting an error when GNU Grub version 0.95 (637k lower / 65526k upper
> memory) tries to boot the installer.
>
> module /boot/x86.miniroot
>
> Error 28: Selected item can not fit into memory.
>
This can happen if the miniroot gets corrupted. However in your case
it is not.
So I may be wrong here but I think the load-address of multiboot and the
miniroot are overlapping the reserved/ACPI* memory segments. The install
miniroot is quite big (around 250+MB) and only the RAM segment:
"BIOS-e820: 00000000040ffc00 - 0000000028000000 (usable)" has enough
contiguous usable memory to house it.
From an elfdump of boot/multiboot in the cdrom:
...
Program Header[0]:
p_vaddr: 0x1000000 p_flags: [ PF_X PF_W PF_R ]
p_paddr: 0x1000000 p_type: [ PT_LOAD ]
p_filesz: 0x1416f p_memsz: 0x26a24
p_offset: 0x60 p_align: 0
...
The p_paddr is probably what Grub uses to determine the physical load
address
of the multiboot executable. This lies in the RAM segment:
"BIOS-e820: 0000000000100000 - 00000000040fd800 (usable)" which is around
64MB, which is more than enough for loading multiboot. But the
miniroot is loaded
in a subsequent contiguous location after multiboot and is overlaps
with ACPI data.
You can try to boot some other distro like BeleniX or Schillix etc.
and check whether
the same error occurs. If yes then this analysis should be valid.
Another thing to check:
Boot from the install cd.
In the Grub screen type c to go to command line mode
Execute displaymem which shows what Grub has determined to be the
memory map
and check whether this tallies with what OpenBSD is reporting.
> I searched and found many people reporting this problem but no solutions.
> There is a bug that was reported over a year ago here
> http://bugs.opensolaris.org/bugdatab...bug_id=6387379.
>
> I also got the error 28 on other boxes when I tried to install Solaris and
> was not able to boot it, but that was with earlier versions. After 11/06 came
> out I was able to install and run it fine. But I still get this error on the
> new box.
>
> I also tried SolarisExpress build 52 x86 and I'm getting the same error.
>
If what I think is true then I do not know of an easy solution other
rebuilding
multiboot with a different load address and rebuilding the Install
DVD/CD.
Regards,
Moinak.
> Here are some dmesg from OpenBSD and Slackware machines I have running on the
> box that will not boot Solaris/Solaris Express:
>
> OpenBSD 4.0 (GENERIC) #1107: Sat Sep 16 19:15:58 MDT 2006
> ..
> cpu0: Intel Pentium III ("GenuineIntel" 686-class) 599 MHz
> cpu0:
> FPU,V86,DE,PSE,TSC,MSR,PAE,MCE,CX8,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,MMX,FXSR,SSE
> real mem = 670646272 (654928K)
> avail mem = 603004928 (588872K)
>
> Linux version 2.6.17.13 (root at tree) (gcc version 3.4.6) #1 Sat Sep 9
> 01:11:49 CDT 2006
> BIOS-provided physical RAM map:
> BIOS-e820: 0000000000000000 - 000000000009f800 (usable)
> BIOS-e820: 000000000009f800 - 00000000000a0000 (reserved)
> BIOS-e820: 00000000000e6c00 - 0000000000100000 (reserved)
> BIOS-e820: 0000000000100000 - 00000000040fd800 (usable)
> BIOS-e820: 00000000040fd800 - 00000000040ff800 (ACPI data)
> BIOS-e820: 00000000040ff800 - 00000000040ffc00 (ACPI NVS)
> BIOS-e820: 00000000040ffc00 - 0000000028000000 (usable)
> BIOS-e820: 00000000fff80000 - 0000000100000000 (reserved)
> 0MB HIGHMEM available.
> 640MB LOWMEM available.
> Any ideas?
>
> Thanks,
> Rand
>
>
> This message posted from opensolaris.org
> _______________________________________________
> install-discuss mailing list
> install-discuss at opensolaris.org
> http://opensolaris.org/mailman/listinfo/install-discuss
>
