Current-Users archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Problems with amd64 -current - bootloader cannot load modules?

On Tue, 8 Oct 2013, Paul Goyette wrote:

On Tue, 8 Oct 2013, Paul Goyette wrote:

On Tue, 8 Oct 2013, Paul Goyette wrote:

Using objdump on the ffs.kmod files from 6.99.23 running system and the 6.99.24 distribution directory, I get the following:

Idx Name             Size .23      Size .24     Change
 0 .text            00024469      00024409       -60
 1 .eh_frame        00003e90      00003e90         0
 2 .rodata.str1.8   00001228      00001228         0
 3 .rodata.str1.1   000006bd      000006bd         0
 4 .rodata          00001158      00001158         0
 5 link_set_modules 00000008      00000008         0
 6 .data            000001f0      000001f0         0
 7 .ident           000006a7      000006a1        -6

So, the newer ffs module is actually _smaller_ than the old one?

Hmmm, another interest data point:

If I boot, and just wait in the boot menu for the 5-second timeout, the boot fails with the "heap full" message.

However, if I interrupt the boot menu and drop to the prompt, and then enter "boot netbsd -s" it boots just fine!

As I recall, there were some recent changes to the boot-loader code for command processing - perhaps something there has not correctly free()d some heap?

One more example, which would tend to confirm that the boot-loader is not correctly free()ing memory?

1. Boot
2. Interrupt the countdown, and go to the prompt
3. Execute the 'ls' command (to see what I called my saved kernel!)
4. boot <old-kernel> -s


But if I omit the 'ls' command (step 3), it boots successfully.

Yet one more data-point....

I was suspecting the recent changes to the boot code from Havard (in July) or Jared (in August), so I re-installed the old bootblocks from another machine. Yet even with the old bootblocks, the bad behavior occurs.

Note that in all failure cases, the "heap full" message occurs after loading all the kernel sections, but before the kernel prints its ID line.

| Paul Goyette     | PGP Key fingerprint:     | E-mail addresses:       |
| Customer Service | FA29 0E3B 35AF E8AE 6651 | paul at    |
| Network Engineer | 0786 F758 55DE 53BA 7731 | pgoyette at |
| Kernel Developer |                          | pgoyette at  |

Home | Main Index | Thread Index | Old Index