[Prev][Next][Index][Thread]

Re: Linux ext2 filesystem

On Tue, 16 May 2000, Leigh Stoller wrote:

> > From: Yves Martin <ymartin@ensisun.imag.fr>
> > Subject: Linux ext2 filesystem 
> > 
> > > >     I would like to report some problems
> > > >   with Linux FS library and Linux devices.
> 
> Well, appended below is a patch relative to snapshot oskit-20000505.  It
> fixes the second problem (with chdir("..")). You should be able to apply it
> in the top level of your oskit tree, if you are running 2000505. Make sure
> you clean and rebuild at least the linux/fs and linux/dev directories.

	Thanks.

> As for the first problem. I was not able to reproduce that one. I commented
> out the call to pic_set_irqmask(), and it ran just fine. Note that I was
> talking to a Freebsd partition using the linux UFS filesystem module
> (instead of ext2). But, I kinda doubt that the problem would be local to
> the ext2 code.

  Sorry, it's confusal : I send my example with the correct
    'pic_set_irqmask( 0x0 ) ;'
  but in fact, I would like to start my environment with
    'pic_set_irqmask( 0xFFFF ) ;'
  and let drivers unmask only used IRQs.

	Of course, commenting pic_set_irqmask() make IDE detection
  effective because default mask (to my mind) is 0x0 but

  if you mask IRQs with 0xFFFF value :
{
  pic_set_irqmask( 0xFFFF ) ;
[...]
  oskit_linux_init_net() ;
  oskit_dev_probe() ;
}
	works correctly
  AND
{
  pic_set_irqmask( 0xFFFF ) ;
[...]
  oskit_linux_init_ide() ;
  oskit_dev_probe() ;
}
	remains stuck.

   Is network device probing mecanism different
   from IDE device probing one ?
   To my mind, ethernet code unmask IRQ before probing
   but IDE code doesn't.

> I'm not sure why you redefine base_multiboot_init_mem(). I sure wouldn't,
> although it did run that way. Is there something special about the hardware
> you are running on?

	I'm running my example code on a DELL GX110.
  With just a IDE disk and IDE cdrom, no idea of
  motherboard model (DELL of course - with 3c905C on board,
  Pentium III 500 MHz)
  To my mind, there is nothing special here.
  Except I had to replace OSkit 3c59x.c by last Linux
  driver for 3c905C detection.


-- Yves Martin --- Elève-ingénieur ENSIMAG - 3e année ---
  Adr : R.H.B. Ch. 765     Yves.Martin@ensimag.imag.fr   
  2 av. J. O.  GRENOBLE      Tél: 04 76 44 20 01         
---------------------------- ICQ: 53379907 --------------
References: