Re: [Intel-gfx] [PATCH 0/2] Low hanging fruit to reduce the driver size

2014-07-12 Thread Daniel Vetter
On Fri, Jul 11, 2014 at 06:34:12PM +0100, Damien Lespiau wrote: Being able to target a single platform to reduce the driver size has been voiced a few times. These patches provide a Kconfig option to provide the opportunity. Let's start small, and, along side the generic Multi-platform

Re: [Intel-gfx] [PATCH 0/2] Low hanging fruit to reduce the driver size

2014-07-12 Thread Damien Lespiau
On Sat, Jul 12, 2014 at 01:01:42PM +0200, Daniel Vetter wrote: And I'm really not terribly convinced that kb shaving is worth the effort overall, especially if it comes at a fairly steep maintainance burden of piles of new config options and #defines. I wanted to try, I tried, probably not

[Intel-gfx] [PATCH 0/2] Low hanging fruit to reduce the driver size

2014-07-11 Thread Damien Lespiau
Being able to target a single platform to reduce the driver size has been voiced a few times. These patches provide a Kconfig option to provide the opportunity. Let's start small, and, along side the generic Multi-platform option, only present Haswell and Broadwell in the list of platforms to

Re: [Intel-gfx] [PATCH 0/2] Low hanging fruit to reduce the driver size

2014-07-11 Thread Damien Lespiau
On Fri, Jul 11, 2014 at 06:34:12PM +0100, Damien Lespiau wrote: If the approach convinces, I'll do more. That sounds presumptuous, I more likely need to do more to convince. I meant, I still need to have a look at the harder problem, providing a static device_info structure and hardcoding