Re: [PATCH v2 08/10] arm: socfgpa: iossm: extract initialization of one interface

[Michael Tretter <m.tretter@pengutronix.de>]

Hi Sascha,

On Wed, 15 Apr 2026 08:12:35 +0200, Sascha Hauer wrote:
> This patch causes imtermediate compile breakage.
> 
> On Thu, Apr 09, 2026 at 03:52:48PM +0200, Michael Tretter wrote:
> > +	pr_debug("%s: Start memory initialization BIST on full memory address",
> > +		 __func__);
> > +	mem_exp = 0x40;
> > +
> > +	ret = io96b_mb_req(io96b_csr_addr,
> > +			   mb_ctrl->ip_type[interface],
> > +			   mb_ctrl->ip_instance_id[interface],
> > +			   CMD_TRIG_CONTROLLER_OP, BIST_MEM_INIT_START,
> > +			   FIELD_PREP(GENMASK(5, 0), mem_exp),
> 
> 0x40 doesn't fit into GENMASK(5, 0).
> 
> This is detected here because 0x40 is a constant. In a following patch
> you add some equations to mem_exp, so it's no longer constant for the
> compiler and it compiles fine. Nevertheless the 0x40 is still present
> with this series applied and either that or the mask seem wrong. Can
> you have a look?

The mask for mem_exp is correct. The 0x40 case for initializing the full
memory should explicitly set bit 6 in the parameter instead of being a
special value of mem_exp.

Should I send a patch on top of the series or send a v3 with a fix for
the patch that causes the intermediate compile breakage?

Michael