Hi Chris,
On 13.05.2016 03:20, Chris Packham wrote:
On Fri, May 13, 2016 at 10:35 AM, Chris Packham judge.packham@gmail.com wrote:
On Thu, May 12, 2016 at 9:50 PM, Stefan Roese sr@denx.de wrote:
Hi Chris,
On 12.05.2016 04:55, Chris Packham wrote:
<snip> >>> struct mvtwsi_registers *twsi = twsi_get_base(adap); >>> /* ensure controller will be enabled by any twsi*() function */ >>> - twsi_control_flags = MVTWSI_CONTROL_TWSIEN; >>> + if (gd->flags & GD_FLG_RELOC) >>> + twsi_control_flags = MVTWSI_CONTROL_TWSIEN; >>> /* reset controller */ >>> writel(0, &twsi->soft_reset); >>> /* wait 2 ms -- this is what the Marvell LSP does */ >> >> >> I've stumbled over this global data variable also before and would >> very much like to get rid of it. Can't you move this variable into >> a (newly created) private data struct instead? > > I'll take a look. The other deficiency with my solution is that > although it avoids the hang the driver still won't work because the > state that is reflected in twsi_control_flags will either cause a new > hang or not be updated. >
Actually I might need some pointers on this. Where could I keep such a private data struct. About the only thing I can think of is either to add to arch_global_data or pass a flags variable through the call chain.
I've checked and you are correct. Unfortunately we are missing a private pointer in the legacy "struct i2c_adapter". So perhaps the flags variable is the way to go (for the non-DM I2C driver).
Thanks, Stefan