ext2fs: different b.p.inode->can't execute binary

ext2fs: different b.p.inode->can't execute binary

Post by Heiko Selb » Thu, 16 Nov 1995 04:00:00



Hi there,

I have the following problem:

when I link an executable file on one ext2fs partition to another
ext2fs partition which was formatted with a different bytes-per-inode
ratio, I get the message "can't execute binary" when I try to execute
it. Non-binaries are not affected whatsoever.

What I did was exactly this:
(I have a 486 with on-board NCR53c810 scsi controller and a Quantum
Maverick 540MB scsi disk and Linux 1.2.8 and 1.2.13, respectively)

I had formatted /dev/sda5 with 2048 bpi and /dev/sda6 with 4096 bpi.
Then I did

'ln -s /susi/bin/this_binary /usr/local/bin/that_binary'
       ^                     ^
       |                     |
    /dev/sda6             /dev/sda5

and just couldn't execute that_binary, although this_binary worked
perfectly fine. I tried some other combinations of formats and ended
up with the conclusion that it's the difference of bytes-per-inode
which is to blame.

Is it a bug or a feature? Please explain.

Thanks, Heiko

--
'Pray that there's intelligent life somewhere up in space
because there's * all down here on Earth.' (Monty Python)
Heiko Selber

http://www.veryComputer.com/~selber
Fritz-Haber-Institut  Berlin, Germany

 
 
 

ext2fs: different b.p.inode->can't execute binary

Post by Rene COUGNE » Fri, 17 Nov 1995 04:00:00


Ce brave Heiko Selber ecrit:

Quote:> Maverick 540MB scsi disk and Linux 1.2.8 and 1.2.13, respectively)
> I had formatted /dev/sda5 with 2048 bpi and /dev/sda6 with 4096 bpi.
> Then I did
> 'ln -s /susi/bin/this_binary /usr/local/bin/that_binary'
>        ^                     ^
>        |                     |
>     /dev/sda6             /dev/sda5
> and just couldn't execute that_binary, although this_binary worked
> perfectly fine. I tried some other combinations of formats and ended
> up with the conclusion that it's the difference of bytes-per-inode
> which is to blame.

Well, it must be something else, because I have exactly the same
type of partitions (same bpi values, same kernel), so I just tried
that : for me it works fine...

--


 
 
 

1. ksh ': >a >b' -vs- '>a >b' re speed

time {
  i=0
  while ((i+=1<10000))
  do : >a >b >c
  done
  }

takes 2.54 seconds real time (all user time) on my machine,
but replacing the 'do' line above with 'do >a >b >c' increases the
real time by a factor of 10 (with two thirds of it system time).
Is a fork, or fork-exec being done in the second case?

Curiously, I couldn't get anywhere the speed of the first one even using
the following perl script (which took just as long as the second case):

while($i++<10000) {
  open(A, ">a"); open(B, ">b"); open(C, ">c");
  }

How does the korn shell do this so fast!

Any help would be appreciated.

Thanks,

John

2. Need for LINES & COLUMNS.

3. inode->i_mapping to existing file's d_inode->i_data

4. DNS over SLIP

5. executing a process as another user, works for ps/top but not for 'who'

6. htm file name with apache?

7. Executing 'ps- elf' and reading output?

8. Matrox MGA-G100 Card

9. >>> Inodes and Vnodes <<<

10. /tmp/ps: cannot find/execute "ps" in ISA subdirectories

11. Strange binary which can't be executed

12. can't execute binary

13. Bad blocks, inodes and .badblocks -> Make root inode 2