Re: [PATCH v2 1/2] build: clarify configuration without IOVA field in mbuf

[Thomas Monjalon <thomas@monjalon.net>]

06/03/2023 17:39, Bruce Richardson:
> On Mon, Mar 06, 2023 at 05:13:27PM +0100, Thomas Monjalon wrote:
> > -            if dpdk_conf.get('RTE_IOVA_AS_PA') == 0 and not pmd_supports_disable_iova_as_pa and not always_enable.contains(drv_path)
> > +            if not get_option('enable_iova_as_pa') and require_iova_in_mbuf and not always_enable.contains(drv_path)
> 
> I don't particularly like the always_enable check at the end. If a driver
> is set to always-enable and it doesn't support the configured mode, we
> should throw an error immediately IMHO, rather than silently continuing to
> build the driver.

We can think about the "forced enabling" as a developer test.
But the developer could change the meson file to enable it.
So I'm OK to replace it with an error.

> >                  build = false
> > -                reason = 'driver does not support disabling IOVA as PA mode'
> > +                reason = 'requires IOVA in mbuf'
> >              endif
> 
> I think the reason given is more accurate, but for those users not familiar
> with the internals of DPDK builds it could be confusing. I think the
> message needs to reference the 'enable_iova_as_pa' option. How about?
> 
> 	reason = 'requires IOVA in mbuf (set enable_iova_as_pa option)'

OK good