[PATCH v2 07/22] Revert "vga_switcheroo: Add helper function to get the active client"

Lukas Wunner lukas at wunner.de
Wed Aug 12 10:34:32 PDT 2015


Hi Daniel,

thanks for taking a look at the patch set.

On Wed, Aug 12, 2015 at 04:25:52PM +0200, Daniel Vetter wrote:
> On Tue, Apr 21, 2015 at 10:39:45AM +0200, Lukas Wunner wrote:
> > This reverts commit 26814ce68904c9faf977c90edac798156311981f.
> > 
> > The helper function is no longer needed after Dave Airlie's rewrite
> > of vga_switcheroo_switch_ddc(), the commit introducing it was only
> > included because 31f23c3d488e ("drm/edid: Switch DDC when reading
> > the EDID") does not compile without it.
[...]
> > --- a/drivers/gpu/vga/vga_switcheroo.c
> > +++ b/drivers/gpu/vga/vga_switcheroo.c
> > @@ -214,20 +214,6 @@ find_active_client(struct list_head *head)
> >  	return NULL;
> >  }
> >  
> > -struct pci_dev *vga_switcheroo_get_active_client(void)
> > -{
> > -	struct vga_switcheroo_client *client;
> > -	struct pci_dev *pdev = NULL;
> > -
> > -	mutex_lock(&vgasr_mutex);
> > -	client = find_active_client(&vgasr_priv.clients);
> > -	if (client)
> > -		pdev = client->pdev;
> > -	mutex_unlock(&vgasr_mutex);
> > -	return pdev;
> > -}
> > -EXPORT_SYMBOL(vga_switcheroo_get_active_client);
> 
> you just added this earlier in this very series. Please reorder/squash
> patches so that this isn't required.

I would have to squash patches 2, 4 (by Seth Forshee), 5 (by Dave Airlie),
6 and 7 (mine). The work of two of these authors would only be acknowledged
in the commit message and the history how the code evolved over the course
of 3 years would not be reflected in the git repo.

Are you sure? (y/n)

I deliberately didn't squash to preserve authorship and history but if
you're forcing me at point blank I'll do it. ;-)

Context: Seth Forshee of Canonical came up with patches in 2012 which Dave
Airlie didn't like. He rewrote them and left them as a WIP in his git repo
where I picked them up. Matthew Garrett posted patches of his own last year
but since they were based on Seth Forshee's code, they didn't get merged
either.

The first MacBooks with dual GPUs were introduced 2008, it's 2015 now and
we're still missing support in the mainline kernel. The issue is not so
much that GPU switching doesn't work (the screen just turns black) but
energy consumption because the discrete GPU is used by default and the
integrated GPU isn't turned off.

So, machines with huge marketshare + shoddy dual GPU support for years
= problem.

We need to fix this, hence the patch set.

Thanks,

Lukas


More information about the dri-devel mailing list