aic7xxx: make target mode enable a device hint

[hpvb]

Previously it was only possible to enable target mode for these drivers by rebuilding the kernel with AHC_TMODE_ENABLE or AHD_TMODE_ENABLE and a bitmask of which units to statically enable for target mode.

There is no space-savings in the driver by not having AHC_TMODE_ENABLE set, so in addition to the compile time option lets also introduce some tunables:

hint.ahc.<unit>.tmode_enable=0/1
hint.ahd.<unit>.tmode_enable=0/1

For compatibility the old behavior is retained, but it can be overridden with tunables

[hpvb]

Sorry for the many edits and force-pushes. I changed my mind several times. I'm pretty sure this is good now though!

[amotin]

Why did you keep AHC_TARGET_MODE in code, while removed from configs? How it is supposed to be set and what for? I would understand if you kept it as alternative method of enabling target to keep compatibility with existing configs, but this way I just don't understand.

[hpvb]

This is how it has always been. AHC_TARGET_MODE gets unconditionally enabled in aic7xxx_osm.h here:

/* This driver supports target mode */
#define AHC_TARGET_MODE 1

If you want a smaller driver I guess you can set that to 0?

[hpvb]

@amotin Do you still feel this should be done differently? To address your concern I'd have to rip out all of AHC_TARGET_MODE from the driver. This has been in there since 2002 in this form, not possible to be disabled by any config setting.

Alternatively I could introduce a way to switch it in config and allow the driver to be (re) built without target support entirely. But I feel that this should be a separate PR then.

EDIT: also note that what I'm removing is the AHC_TMODE_ENABLE config option. This is wholly separate from AHC_TARGET_MODE, the two flags don't interact.

[amotin]

Thanks for the clarification. I indeed mixed AHC_TARGET_MODE with AHC_TMODE_ENABLE. I do not insist in massive driver modifications. But it seems we could trivially keep AHC_TMODE_ENABLE for compatibility, using it as default value to override with tunables.

[hpvb]

@amotin I think I addressed your concerns.

[amotin]

The direction looks good to me, but I would further cut the printfs. BTW, there is device_printf() to print something for specific device without crafting local.

[hpvb]

The direction looks good to me, but I would further cut the printfs. BTW, there is device_printf() to print something for specific device without crafting local.

I have removed most of the debug prints now, I thought they were kinda useful but perhaps too chatty even for verbose.

I'm aware of device_printf() but the driver uses this throughout, I do kind of want to replace all of the existing handcrafted ones with device_printf() but for consistency it seemed better to just keep using the local convention until I got around to replacing all of them.

[hpvb]

@amotin are you happy with this now? I'd love to see this merged as I have some other changes pending that depend on these changes to the man pages :)

[bsdimp]

I'll add this to today's batch.

[bsdimp]

Dang, was hoping I'd pushed the right thing to have this show up as merged.