Bug#504391: Please enable CONFIG_MMC_UNSAFE_RESUME

[Wouter van Heyst <larstiq@larstiq.dyndns.org>]

On Tue, Nov 10, 2009 at 02:51:52PM +0000, Ben Hutchings wrote:
> On Tue, Nov 10, 2009 at 12:51:27PM +0100, Wouter van Heyst wrote:
> > I second the request for turning CONFIG_MMC_UNSAFE_RESUME on,
> > for me with /home on a SD card fixed firmly in it's slot, having it
> > turned on is the only safe option.
> > 
> >     Off: filesystem corruption after suspend/resume
> >     On: everything works fine
>  
> And for the majority of SD/MMC card users, the choice is:
> 
> off: everything works fine
> on: filesystem corruption after card change while system is off
> 
> I don't think we should change the default to suit the minority of users
> with hardware that abuses SD/MMC for fixed storage devices.  

Ah. I had thought that switching a card while the system is off would be
far more rare.  I see my view is limited to netbooks, quite disjunct from, say,
photographers.

> However this probably should be a module option rather than a compile-time
> option.

Yes, rebuilding kernels myself every time there is a security update has
become tiring.

The impact of CONFIG_MMC_UNSAFE_RESUME seems to be contstrained to let 

    #define mmc_suspend NULL
    #define mmc_resume NULL   

    #define mmc_sd_suspend NULL
    #define mmc_sd_resume NULL

if it's not defined, and with implementations otherwise, filling the
mmc_ops and mmc_sd_ops structs.  After that my kernel fu is too weak.


What needs to be done to get this to be a module option?

Wouter van Heyst