Re: halp? [hal support for older kernels]

[Joseph Sacco <jsacco gnome org>]

Carson,

FWIW, "we" are beating them... 

I think that implementing a formal workaround should be straight
forward. [I have not (yet) had the free cycles or machine access to
experiment.]

The build fails primarily because of missing "switch events" defined in 

    <linux/input.h>

        /*
         * Switch events
         */
        
        #define SW_LID      0x00  /* set = lid shut */
        #define SW_TABLET_MODE    0x01  /* set = tablet mode */
        #define SW_HEADPHONE_INSERT 0x02  /* set = inserted */
        #define SW_MAX      0x0f

To my thinking... If the existing kernel does not support the
functionality associated with these switches, HAL should nod politely
and roll on.

A simple grep exposes the "culprits"

    [ignore line wrap]

        [garnome plantain hal-0.5.9]$ grep -r SW_ * | grep -v "*.o"
        hald/linux/device.c:    long bitmask[NBITS(SW_MAX)];
        hald/linux/device.c:            if (test_bit (SW_LID, bitmask)) {
        hald/linux/device.c:            } else if (test_bit (SW_TABLET_MODE, bitmask)) {
        hald/linux/device.c:            } else if (test_bit (SW_HEADPHONE_INSERT, bitmask)) {
        hald/linux/probing/probe-input.c:       long bitmask[NBITS(SW_MAX)];
        hald/linux/probing/probe-input.c:               sw = SW_LID;
        hald/linux/probing/probe-input.c:               sw = SW_TABLET_MODE;
        hald/linux/probing/probe-input.c:               sw = SW_HEADPHONE_INSERT;
        hald/linux/addons/addon-keyboard.c:                     case SW_LID:
        hald/linux/addons/addon-keyboard.c:                     case SW_TABLET_MODE:
        hald/linux/addons/addon-keyboard.c:                     case SW_HEADPHONE_INSERT:
        hald/linux/addons/addon-keyboard.c:                             long bitmask[NBITS(SW_MAX)];


Thoughts
---------
* Can  #ifndef blocks be used to locally specify the missing macros?

* It should not be too difficult to short-circuit the errant code.

* It would be better to have configure probe for the SW_ macros and set
flags that could be used within #ifdef blocks.


-Joseph


==================================================================================
On Tue, 2007-06-19 at 15:01 -0700, Carson Gaspar wrote:
> Joseph Sacco wrote:
> > Dan,
> > 
> > Can you fake it? If by "it" you mean the absence of kernel header files,
> > the answer is "no".  You will need to install the kernel header files
> > for your distro.  
> > 
> > I don't believe that the 2.6.16.1 kernel is new enough to support the
> > recent changes in HAL. 
> > 
> > What to do...
> > 
> > There are choices:
> > * update your kernel
> > [I believe 2.6.17.1 will work]
> > 
> > * retreat to the previous versions of hal and dbus.
> 
> I vote for choice 3:
> 
> * Beat the dbus/hal maintainers about the head and shoulders until they 
> understand that gratuitously breaking older kernels is a Very Bad Idea.
> 
> *sigh* And people wonder why gnome has a bad rep...
> 
> -- 
> Carson
-- 
jsacco [at] gnome [dot] org