Re: modprobe + request_module() deadlock

[Gerd Knorr]

> > > 	MODULE_DEPENDS_ON("somemodule");
> > 
> > On the other hand I don't depend on request_module() waiting for the
> > modprobe being finished.  So maybe we can solve that with a
> > request_module_async()?
> 
> The problem is fairly simple: we don't let you get at the symbols from a
> module which hasn't finished initializing yet.  This means that a
> request_module() inside a module's init() will fail if the requested
> module depends on this one.  async() will race with init() finishing, so
> won't really help.

Hmm, so module loading is not serialized by a lock?  I assumed that
caused the deadlock in $subject which started this thread ...

> The traditional way to do this has been to have saa7134-empress do its
> own probe, and likewise saa7134-dvb.

They can't actually probe themself.  It's _one_ PCI device (driven by
the saa7134 module) which can handle (among other v4l-related things)
the DMA transfer of mpeg streams.  That can be used in different ways
(or not at all) and the different use cases are handled by the
sub-modules.

So the way it is intended to work is that saa7134 has the pci table and
gets autoloaded by hotplug, it will have a look at the hardware and then
load either saa7134-empress or saa7134-dvb or none of them, so you'll
get everything nicely autoloaded.

> Unfortunately, these days modules are not supposed to fail to load
> simply because there are no devices, so wild module loading has a real
> cost.

Yes, thats exactly the problem.  And this is especially true for the
saa7134-dvb module as this one has a bunch of dependencies to other
modules of the dvb subsystem.  Thats why I try to avoid loading them
if they are not needed.

> Otherwise I'd be tempted to make multiple aliases load *all* of them,
> and solve the problem that way.

Well, that will actually work.  You can simply load both saa7134-empress
and saa7134-dvb, the saa7134 core module will take care that they can
only attach to devices they can actually handle.  Drawback is the memory
footprint ...

  Gerd

-- 
#define printk(args...) fprintf(stderr, ## args)
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/