Re: [PATCH] x86/mce: Add compat_ioctl assignment to make it compatible with 32-bit system

["Luck, Tony" <tony.luck@intel.com>]

On Thu, Apr 16, 2020 at 04:40:31PM +0800, He Zhe wrote:
> Can this be considered for the moment?
> 
> Thanks,
> Zhe
> 
> On 3/4/20 2:39 PM, zhe.he@windriver.com wrote:
> > From: He Zhe <zhe.he@windriver.com>
> >
> > 32-bit user-space program would get errors like the following from ioctl
> > syscall due to missing compat_ioctl.
> > MCE_GET_RECORD_LEN: Inappropriate ioctl for device
> >
> > compat_ptr_ioctl is provided as a generic implementation of .compat_ioctl
> > file operation to ioctl functions that either ignore the argument or pass
> > a pointer to a compatible data type.

I'm not super-familiar with the compat ioctl bits.  But this looks plausible.

All three of the ioctl's for this driver have a "pointer to integer" for the
"return" value.  And "int" is a compatible type between i386 and x86_64.

I don't have a system setup to build a 32-bit binary to test the theory,
but I assume that you have built something that tests all three:

	MCE_GET_RECORD_LEN
	MCE_GET_LOG_LEN
	MCE_GETCLEAR_FLAGS

So I guess:

Acked-by: Tony Luck <tony.luck@intel.com>

> >
> > Signed-off-by: He Zhe <zhe.he@windriver.com>
> > ---
> >  arch/x86/kernel/cpu/mce/dev-mcelog.c | 1 +
> >  1 file changed, 1 insertion(+)
> >
> > diff --git a/arch/x86/kernel/cpu/mce/dev-mcelog.c b/arch/x86/kernel/cpu/mce/dev-mcelog.c
> > index 7c8958d..6c9b91b7 100644
> > --- a/arch/x86/kernel/cpu/mce/dev-mcelog.c
> > +++ b/arch/x86/kernel/cpu/mce/dev-mcelog.c
> > @@ -328,6 +328,7 @@ static const struct file_operations mce_chrdev_ops = {
> >  	.write			= mce_chrdev_write,
> >  	.poll			= mce_chrdev_poll,
> >  	.unlocked_ioctl		= mce_chrdev_ioctl,
> > +	.compat_ioctl		= compat_ptr_ioctl,
> >  	.llseek			= no_llseek,
> >  };
> >  
>