megaraid warnings

megaraid warnings

[David Mosberger]

The 2.6 megaraid driver causes several ugly warnings along the lines of:

drivers/scsi/megaraid.c:3500: warning: cast to pointer from integer of different size

I think those may be harmless (I don't have hardware to test this) but
it would still be better to fix the drivers so there are no warnings
(most likely by casting u32 addresses that are guaranteed to be <4GB
to "unsigned long" before converting them to a pointer).

Can someone look into this?

Thanks,

	--david

Re: megaraid warnings

[Matthew Wilcox]

On Tue, Feb 17, 2004 at 11:45:33AM -0800, David Mosberger wrote:
> The 2.6 megaraid driver causes several ugly warnings along the lines of:
> 
> drivers/scsi/megaraid.c:3500: warning: cast to pointer from integer of different size
> 
> I think those may be harmless (I don't have hardware to test this) but
> it would still be better to fix the drivers so there are no warnings
> (most likely by casting u32 addresses that are guaranteed to be <4GB
> to "unsigned long" before converting them to a pointer).
> 
> Can someone look into this?

This keeps coming up on linux-scsi; basically these are ioctls that
are called by some binary-only (and x86-binary-only at that) management
program.  so if they want to ship an ia64 version of their management
program, they'll have to fix this; otherwise the addresses are guaranteed
to be in the lower 32 bits of the address space.

-- 
"Next the statesmen will invent cheap lies, putting the blame upon 
the nation that is attacked, and every man will be glad of those
conscience-soothing falsities, and will diligently study them, and refuse
to examine any refutations of them; and thus he will by and by convince 
himself that the war is just, and will thank God for the better sleep 
he enjoys after this process of grotesque self-deception." -- Mark Twain

Re: megaraid warnings

[David Mosberger]

>>>>> On Tue, 17 Feb 2004 19:58:47 +0000, Matthew Wilcox <willy@debian.org> said:

  Matthew> This keeps coming up on linux-scsi; basically these are
  Matthew> ioctls that are called by some binary-only (and
  Matthew> x86-binary-only at that) management program.  so if they
  Matthew> want to ship an ia64 version of their management program,
  Matthew> they'll have to fix this; otherwise the addresses are
  Matthew> guaranteed to be in the lower 32 bits of the address space.

Then why not disable the offending code for 64-bit platforms for the
time being?

	--david

Re: megaraid warnings

[Matthew Wilcox]

On Tue, Feb 17, 2004 at 12:08:22PM -0800, David Mosberger wrote:
> >>>>> On Tue, 17 Feb 2004 19:58:47 +0000, Matthew Wilcox <willy@debian.org> said:
> 
>   Matthew> This keeps coming up on linux-scsi; basically these are
>   Matthew> ioctls that are called by some binary-only (and
>   Matthew> x86-binary-only at that) management program.  so if they
>   Matthew> want to ship an ia64 version of their management program,
>   Matthew> they'll have to fix this; otherwise the addresses are
>   Matthew> guaranteed to be in the lower 32 bits of the address space.
> 
> Then why not disable the offending code for 64-bit platforms for the
> time being?

Because it works when the x86 app is run in emulation.

-- 
"Next the statesmen will invent cheap lies, putting the blame upon 
the nation that is attacked, and every man will be glad of those
conscience-soothing falsities, and will diligently study them, and refuse
to examine any refutations of them; and thus he will by and by convince 
himself that the war is just, and will thank God for the better sleep 
he enjoys after this process of grotesque self-deception." -- Mark Twain

Re: megaraid warnings

[David Mosberger]

>>>>> On Tue, 17 Feb 2004 20:09:18 +0000, Matthew Wilcox <willy@debian.org> said:

  Matthew> On Tue, Feb 17, 2004 at 12:08:22PM -0800, David Mosberger
  Matthew> wrote:
  >> >>>>> On Tue, 17 Feb 2004 19:58:47 +0000, Matthew Wilcox
  >> <willy@debian.org> said:

  Matthew> This keeps coming up on linux-scsi; basically these are
  Matthew> ioctls that are called by some binary-only (and
  Matthew> x86-binary-only at that) management program.  so if they
  Matthew> want to ship an ia64 version of their management program,
  Matthew> they'll have to fix this; otherwise the addresses are
  Matthew> guaranteed to be in the lower 32 bits of the address space.

  >>  Then why not disable the offending code for 64-bit platforms for
  >> the time being?

  Matthew> Because it works when the x86 app is run in emulation.

Ah, your mail wasn't clear on this.  In that case, I'd suggest to fix
the code as I suggested in the earlier mail.

	--david