Re: [PATCH] net: phy: Fix phy_modify() semantic difference fallout

[Geert Uytterhoeven]

Hi Russell,

On Tue, Jan 9, 2018 at 7:31 PM, Russell King - ARM Linux
<linux@xxxxxxxxxxxxxxx> wrote:
> On Tue, Jan 09, 2018 at 07:25:40PM +0100, Geert Uytterhoeven wrote:
>> On Tue, Jan 9, 2018 at 3:22 PM, Russell King - ARM Linux
>> <linux@xxxxxxxxxxxxxxx> wrote:
>> > On Tue, Jan 09, 2018 at 03:10:08PM +0100, Andrew Lunn wrote:
>> >> On Tue, Jan 09, 2018 at 12:11:21PM +0100, Geert Uytterhoeven wrote:
>> >> > In case of success, the return values of (__)phy_write() and
>> >> > (__)phy_modify() are not compatible: (__)phy_write() returns 0, while
>> >> > (__)phy_modify() returns the old PHY register value.
>> >> >
>> >> > Apparently this change was catered for in drivers/net/phy/marvell.c, but
>> >> > not in other source files.
>> >> >
>> >> > Hence genphy_restart_aneg() now returns 4416 instead zero, which is
>> >> > considered an error:
>> >> >
>> >> >     ravb e6800000.ethernet eth0: failed to connect PHY
>> >> >     IP-Config: Failed to open eth0
>> >> >     IP-Config: No network devices available
>> >> >
>> >> > Fix this by converting positive values to zero in all callers of
>> >> > phy_modify().
>> >> >
>> >> > Fixes: fea23fb591cce995 ("net: phy: convert read-modify-write to phy_modify()")
>> >> > Signed-off-by: Geert Uytterhoeven <geert+renesas@xxxxxxxxx>
>> >> > ---
>> >> > Alternatively, __phy_modify() could be changed to follow __phy_write()
>> >> > semantics?
>> >>
>> >> Hi Geert, Russell
>> >>
>> >> I took a quick look at the uses of phy_modify(). I don't see any uses
>> >> of the return code other than as an error indicator. So having it
>> >> return 0 on success seems like a better fix.
>> >
>> > I'd like to avoid that, because I don't want to have yet another
>> > accessor that needs to be used for advertisment modification (where
>> > we need to know if we changed any bits.)
>> >
>> > That's why this accessor returns the old value.
>>
>> But this is documented nowhere!
>>
>> I believe there are no current users of (__)phy_modify() that rely on this
>> behavior.  Except perhaps phy_restore_page(), which I don't understand at all.
>>
>> BTW, I think phy_restore_page() may return a strict positive value as well,
>> thus breaking m88e1318_set_wol(), which is not supposed to return strict
>> positive values.
>
> Correct, and it has to for temperature reading in marvell.c to work.

For phy_restore_page()?
Not for breaking m88e1318_set_wol(), I guess?

>> So changing __phy_modify() to return zero on success seems like the way
>> forward...
>
> So what do we call an accessor that returns the original value?
>
> __phy_modify_return_old_value()

__phy_modify_ret()?

Or __phy_modify(...., u16 *oldval) (where oldval can be NULL)?

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds