On 08/07/2015 05:42 AM, Tomasz Nowicki wrote:
On 07.08.2015 13:56, Robert Richter wrote:
On 07.08.15 12:52:41, Tomasz Nowicki wrote:
[...]
I would not pollute bgx_probe() with acpi and dt specifics, and instead
keep bgx_init_phy(). The typical design pattern for this is:
static int bgx_init_phy(struct bgx *bgx)
{
#ifdef CONFIG_ACPI
if (!acpi_disabled)
return bgx_init_acpi_phy(bgx);
#endif
return bgx_init_of_phy(bgx);
}
This adds acpi runtime detection (acpi=no), does not call dt code in
case of acpi, and saves the #else for bgx_init_acpi_phy().
I am fine with keeping it in bgx_init_phy(), however we can drop there
#ifdefs since both of bgx_init_{acpi,of}_phy calls have empty stub
for !ACPI
and !OF case. Like that:
static int bgx_init_phy(struct bgx *bgx)
{
if (!acpi_disabled)
return bgx_init_acpi_phy(bgx);
else
return bgx_init_of_phy(bgx);
}
As said, keeping it in #ifdefs makes the empty stub function for !acpi
obsolete, which makes the code smaller and better readable. This style
is common practice in the kernel. Apart from that, the 'else' should
be dropped as it is useless.
I would't say the code is better readable (bgx_init_of_phy has still two
stubs) but this is just cosmetic, my point was to use run time detection
using acpi_disabled.
Thanks Tomasz and Robert for the input. Because of this, it seems that
another version of the patch will be necessary. In the interests of
code clarity and asthetics, we will go with the code without the
#ifdefs, and rely on the compiler to optimize away any dead code.
David Daney
Tomasz
