Re: arm: shmobile_defconfig: ld.lld: error: undefined symbol: lynx_pcs_destroy

Re: arm: shmobile_defconfig: ld.lld: error: undefined symbol: lynx_pcs_destroy

[Naresh Kamboju]

+ netdev



On Tue, 6 Jun 2023 at 14:17, Naresh Kamboju <naresh.kamboju@linaro.org> wrote:
>
> Following build regressions found while building arm shmobile_defconfig on
> Linux next-20230606.
>
> Regressions found on arm:
>
>  - build/clang-16-shmobile_defconfig
>  - build/gcc-8-shmobile_defconfig
>  - build/gcc-12-shmobile_defconfig
>  - build/clang-nightly-shmobile_defconfig

And mips defconfig builds failed.
Regressions found on mips:

  - build/clang-16-defconfig
  - build/gcc-12-defconfig
  - build/gcc-8-defconfig
  - build/clang-nightly-defconfig



>
> ld.lld: error: undefined symbol: lynx_pcs_destroy
> >>> referenced by stmmac_mdio.c
> >>>               drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.o:(stmmac_mdio_unregister) in archive vmlinux.a
> make[2]: *** [scripts/Makefile.vmlinux:35: vmlinux] Error 1
>
>
> Reported-by: Linux Kernel Functional Testing <lkft@linaro.org>
>
> links,
> - https://qa-reports.linaro.org/lkft/linux-next-master/build/next-20230606/testrun/17347517/suite/build/test/gcc-12-shmobile_defconfig/history/
> - https://qa-reports.linaro.org/lkft/linux-next-master/build/next-20230606/testrun/17347562/suite/build/test/clang-16-shmobile_defconfig/log
>
> --
> Linaro LKFT
> https://lkft.linaro.org

Re: arm: shmobile_defconfig: ld.lld: error: undefined symbol: lynx_pcs_destroy

[Geert Uytterhoeven]

Hi Naresh,

On Tue, Jun 6, 2023 at 10:53 AM Naresh Kamboju
<naresh.kamboju@linaro.org> wrote:
> On Tue, 6 Jun 2023 at 14:17, Naresh Kamboju <naresh.kamboju@linaro.org> wrote:
> > Following build regressions found while building arm shmobile_defconfig on
> > Linux next-20230606.
> >
> > Regressions found on arm:
> >
> >  - build/clang-16-shmobile_defconfig
> >  - build/gcc-8-shmobile_defconfig
> >  - build/gcc-12-shmobile_defconfig
> >  - build/clang-nightly-shmobile_defconfig
>
> And mips defconfig builds failed.
> Regressions found on mips:
>
>   - build/clang-16-defconfig
>   - build/gcc-12-defconfig
>   - build/gcc-8-defconfig
>   - build/clang-nightly-defconfig

Please give my fix a try:
https://lore.kernel.org/linux-renesas-soc/7b36ac43778b41831debd5c30b5b37d268512195.1686039915.git.geert+renesas@glider.be

Thanks!

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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

Re: arm: shmobile_defconfig: ld.lld: error: undefined symbol: lynx_pcs_destroy

[Arnd Bergmann]

On Tue, Jun 6, 2023, at 11:01, Geert Uytterhoeven wrote:
> Hi Naresh,
>
> On Tue, Jun 6, 2023 at 10:53 AM Naresh Kamboju
> <naresh.kamboju@linaro.org> wrote:
>> On Tue, 6 Jun 2023 at 14:17, Naresh Kamboju <naresh.kamboju@linaro.org> wrote:
>> > Following build regressions found while building arm shmobile_defconfig on
>> > Linux next-20230606.
>> >
>> > Regressions found on arm:
>> >
>> >  - build/clang-16-shmobile_defconfig
>> >  - build/gcc-8-shmobile_defconfig
>> >  - build/gcc-12-shmobile_defconfig
>> >  - build/clang-nightly-shmobile_defconfig
>>
>> And mips defconfig builds failed.
>> Regressions found on mips:
>>
>>   - build/clang-16-defconfig
>>   - build/gcc-12-defconfig
>>   - build/gcc-8-defconfig
>>   - build/clang-nightly-defconfig
>
> Please give my fix a try:
> https://lore.kernel.org/linux-renesas-soc/7b36ac43778b41831debd5c30b5b37d268512195.1686039915.git.geert+renesas@glider.be
>

This won't work when PCS_LYNX is a loadable module and
STMMAC is built-in. I think we should just select PCS_LYNX
unconditionally from stmmac even if no front-end driver
using it is enabled.

I tried to come up with a way to move the dependency into
the altera specific front-end, but couldn't find an obvious
or simple way to do this.

Having a proper abstraction for PCS drivers instead of
directly calling into exported driver symbols might help
here, but that would add complexity elsewhere.

     Arnd

Re: arm: shmobile_defconfig: ld.lld: error: undefined symbol: lynx_pcs_destroy

[Geert Uytterhoeven]

Hi Arnd,

On Tue, Jun 6, 2023 at 11:16 AM Arnd Bergmann <arnd@arndb.de> wrote:
> On Tue, Jun 6, 2023, at 11:01, Geert Uytterhoeven wrote:
> > On Tue, Jun 6, 2023 at 10:53 AM Naresh Kamboju
> > <naresh.kamboju@linaro.org> wrote:
> >> On Tue, 6 Jun 2023 at 14:17, Naresh Kamboju <naresh.kamboju@linaro.org> wrote:
> >> > Following build regressions found while building arm shmobile_defconfig on
> >> > Linux next-20230606.
> >> >
> >> > Regressions found on arm:
> >> >
> >> >  - build/clang-16-shmobile_defconfig
> >> >  - build/gcc-8-shmobile_defconfig
> >> >  - build/gcc-12-shmobile_defconfig
> >> >  - build/clang-nightly-shmobile_defconfig
> >>
> >> And mips defconfig builds failed.
> >> Regressions found on mips:
> >>
> >>   - build/clang-16-defconfig
> >>   - build/gcc-12-defconfig
> >>   - build/gcc-8-defconfig
> >>   - build/clang-nightly-defconfig
> >
> > Please give my fix a try:
> > https://lore.kernel.org/linux-renesas-soc/7b36ac43778b41831debd5c30b5b37d268512195.1686039915.git.geert+renesas@glider.be
>
> This won't work when PCS_LYNX is a loadable module and
> STMMAC is built-in. I think we should just select PCS_LYNX

Oops, you're right, forgot about that case.
What about using IS_REACHABLE() instead?
No, that won't work either, as DWMAC_SOCFPGA can be modular,
with STMMAC builtin.

> unconditionally from stmmac even if no front-end driver
> using it is enabled.
>
> I tried to come up with a way to move the dependency into
> the altera specific front-end, but couldn't find an obvious
> or simple way to do this.
>
> Having a proper abstraction for PCS drivers instead of
> directly calling into exported driver symbols might help
> here, but that would add complexity elsewhere.

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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

Re: arm: shmobile_defconfig: ld.lld: error: undefined symbol: lynx_pcs_destroy

[Arnd Bergmann]

On Tue, Jun 6, 2023, at 11:28, Geert Uytterhoeven wrote:
> On Tue, Jun 6, 2023 at 11:16 AM Arnd Bergmann <arnd@arndb.de> wrote:
>> On Tue, Jun 6, 2023, at 11:01, Geert Uytterhoeven wrote:
>>
>> This won't work when PCS_LYNX is a loadable module and
>> STMMAC is built-in. I think we should just select PCS_LYNX
>
> Oops, you're right, forgot about that case.
> What about using IS_REACHABLE() instead?
> No, that won't work either, as DWMAC_SOCFPGA can be modular,
> with STMMAC builtin.

It would work because of the 'select PCS_LYNX' below DWMAC_SOCFPGA,
but I think that's too fragile and would easily break when another
dwmac front-end starts using PCS_LYNX without have the same select
statement. I think we should always avoid IS_REACHABLE().

      Arnd

Re: arm: shmobile_defconfig: ld.lld: error: undefined symbol: lynx_pcs_destroy

[Geert Uytterhoeven]

Hi Arnd,

On Tue, Jun 6, 2023 at 12:21 PM Arnd Bergmann <arnd@arndb.de> wrote:
> On Tue, Jun 6, 2023, at 11:28, Geert Uytterhoeven wrote:
> > On Tue, Jun 6, 2023 at 11:16 AM Arnd Bergmann <arnd@arndb.de> wrote:
> >> On Tue, Jun 6, 2023, at 11:01, Geert Uytterhoeven wrote:
> >>
> >> This won't work when PCS_LYNX is a loadable module and
> >> STMMAC is built-in. I think we should just select PCS_LYNX
> >
> > Oops, you're right, forgot about that case.
> > What about using IS_REACHABLE() instead?
> > No, that won't work either, as DWMAC_SOCFPGA can be modular,
> > with STMMAC builtin.
>
> It would work because of the 'select PCS_LYNX' below DWMAC_SOCFPGA,

That was my first thought, but it won't work, as DWMAC_SOCFPGA=m
causes PCS_LYNX=m, while main STMMAC can still be builtin.

> but I think that's too fragile and would easily break when another
> dwmac front-end starts using PCS_LYNX without have the same select
> statement. I think we should always avoid IS_REACHABLE().

;-)

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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

Re: arm: shmobile_defconfig: ld.lld: error: undefined symbol: lynx_pcs_destroy

[Arnd Bergmann]

On Tue, Jun 6, 2023, at 12:31, Geert Uytterhoeven wrote:
> On Tue, Jun 6, 2023 at 12:21 PM Arnd Bergmann <arnd@arndb.de> wrote:
>> On Tue, Jun 6, 2023, at 11:28, Geert Uytterhoeven wrote:
>> > On Tue, Jun 6, 2023 at 11:16 AM Arnd Bergmann <arnd@arndb.de> wrote:
>> >> On Tue, Jun 6, 2023, at 11:01, Geert Uytterhoeven wrote:
>> >>
>> >> This won't work when PCS_LYNX is a loadable module and
>> >> STMMAC is built-in. I think we should just select PCS_LYNX
>> >
>> > Oops, you're right, forgot about that case.
>> > What about using IS_REACHABLE() instead?
>> > No, that won't work either, as DWMAC_SOCFPGA can be modular,
>> > with STMMAC builtin.
>>
>> It would work because of the 'select PCS_LYNX' below DWMAC_SOCFPGA,
>
> That was my first thought, but it won't work, as DWMAC_SOCFPGA=m
> causes PCS_LYNX=m, while main STMMAC can still be builtin.

Right, got it now.

     Arnd

Re: arm: shmobile_defconfig: ld.lld: error: undefined symbol: lynx_pcs_destroy

[Aithal, Srikanth]

On 6/6/2023 2:31 PM, Geert Uytterhoeven wrote:
> Hi Naresh,
> 
> On Tue, Jun 6, 2023 at 10:53 AM Naresh Kamboju
> <naresh.kamboju@linaro.org> wrote:
>> On Tue, 6 Jun 2023 at 14:17, Naresh Kamboju <naresh.kamboju@linaro.org> wrote:
>>> Following build regressions found while building arm shmobile_defconfig on
>>> Linux next-20230606.
>>>
>>> Regressions found on arm:
>>>
>>>   - build/clang-16-shmobile_defconfig
>>>   - build/gcc-8-shmobile_defconfig
>>>   - build/gcc-12-shmobile_defconfig
>>>   - build/clang-nightly-shmobile_defconfig
>>
>> And mips defconfig builds failed.
>> Regressions found on mips:
>>
>>    - build/clang-16-defconfig
>>    - build/gcc-12-defconfig
>>    - build/gcc-8-defconfig
>>    - build/clang-nightly-defconfig
> 
> Please give my fix a try:
> https://lore.kernel.org/linux-renesas-soc/7b36ac43778b41831debd5c30b5b37d268512195.1686039915.git.geert+renesas@glider.be
On x86 as well seeing couple of issues related to same, not on defconfig 
though..

ERROR: modpost: "lynx_pcs_destroy" 
[drivers/net/ethernet/stmicro/stmmac/stmmac.ko] undefined!
ERROR: modpost: "lynx_pcs_destroy" 
[drivers/net/ethernet/altera/altera_tse.ko] undefined!
make[1]: *** [scripts/Makefile.modpost:136: Module.symvers] Error 1
make: *** [Makefile:1984: modpost] Error 2

Among above issues stmmac issue would be resolved with above mentioned fix.
> 
> Thanks!
> 
> Gr{oetje,eeting}s,
> 
>                          Geert
> 

Re: arm: shmobile_defconfig: ld.lld: error: undefined symbol: lynx_pcs_destroy

[Russell King (Oracle)]

On Wed, Jun 07, 2023 at 06:57:17PM +0530, Aithal, Srikanth wrote:
> On 6/6/2023 2:31 PM, Geert Uytterhoeven wrote:
> > Hi Naresh,
> > 
> > On Tue, Jun 6, 2023 at 10:53 AM Naresh Kamboju
> > <naresh.kamboju@linaro.org> wrote:
> > > On Tue, 6 Jun 2023 at 14:17, Naresh Kamboju <naresh.kamboju@linaro.org> wrote:
> > > > Following build regressions found while building arm shmobile_defconfig on
> > > > Linux next-20230606.
> > > > 
> > > > Regressions found on arm:
> > > > 
> > > >   - build/clang-16-shmobile_defconfig
> > > >   - build/gcc-8-shmobile_defconfig
> > > >   - build/gcc-12-shmobile_defconfig
> > > >   - build/clang-nightly-shmobile_defconfig
> > > 
> > > And mips defconfig builds failed.
> > > Regressions found on mips:
> > > 
> > >    - build/clang-16-defconfig
> > >    - build/gcc-12-defconfig
> > >    - build/gcc-8-defconfig
> > >    - build/clang-nightly-defconfig
> > 
> > Please give my fix a try:
> > https://lore.kernel.org/linux-renesas-soc/7b36ac43778b41831debd5c30b5b37d268512195.1686039915.git.geert+renesas@glider.be
> On x86 as well seeing couple of issues related to same, not on defconfig
> though..
> 
> ERROR: modpost: "lynx_pcs_destroy"
> [drivers/net/ethernet/stmicro/stmmac/stmmac.ko] undefined!
> ERROR: modpost: "lynx_pcs_destroy"
> [drivers/net/ethernet/altera/altera_tse.ko] undefined!
> make[1]: *** [scripts/Makefile.modpost:136: Module.symvers] Error 1
> make: *** [Makefile:1984: modpost] Error 2
> 
> Among above issues stmmac issue would be resolved with above mentioned fix.

Patches are in progress to resolve this issue (on netdev), please be
patient.

Even better would be to check netdev for the patches, review them, and
give them a test, and provide feedback, which will help get them
merged.

Latest version can be found here:

https://lore.kernel.org/netdev/20230607135941.407054-1-maxime.chevallier@bootlin.com/T/#t

Thanks.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

Re: arm: shmobile_defconfig: ld.lld: error: undefined symbol: lynx_pcs_destroy

[Arnd Bergmann]

On Wed, Jun 7, 2023, at 15:27, Aithal, Srikanth wrote:
> On 6/6/2023 2:31 PM, Geert Uytterhoeven wrote:
>> On Tue, Jun 6, 2023 at 10:53 AM Naresh Kamboju
>> <naresh.kamboju@linaro.org> wrote:
>>> On Tue, 6 Jun 2023 at 14:17, Naresh Kamboju <naresh.kamboju@linaro.org> wrote:
>>>> Following build regressions found while building arm shmobile_defconfig on
>>>> Linux next-20230606.
>>>>
>>>> Regressions found on arm:
>>>>
>>>>   - build/clang-16-shmobile_defconfig
>>>>   - build/gcc-8-shmobile_defconfig
>>>>   - build/gcc-12-shmobile_defconfig
>>>>   - build/clang-nightly-shmobile_defconfig
>>>
>>> And mips defconfig builds failed.
>>> Regressions found on mips:
>>>
>>>    - build/clang-16-defconfig
>>>    - build/gcc-12-defconfig
>>>    - build/gcc-8-defconfig
>>>    - build/clang-nightly-defconfig
>> 
>> Please give my fix a try:
>> https://lore.kernel.org/linux-renesas-soc/7b36ac43778b41831debd5c30b5b37d268512195.1686039915.git.geert+renesas@glider.be
> On x86 as well seeing couple of issues related to same, not on defconfig 
> though..
>
> ERROR: modpost: "lynx_pcs_destroy" 
> [drivers/net/ethernet/stmicro/stmmac/stmmac.ko] undefined!
> ERROR: modpost: "lynx_pcs_destroy" 
> [drivers/net/ethernet/altera/altera_tse.ko] undefined!
> make[1]: *** [scripts/Makefile.modpost:136: Module.symvers] Error 1
> make: *** [Makefile:1984: modpost] Error 2
>
> Among above issues stmmac issue would be resolved with above mentioned fix.

I sent out my version of the build fixups for altera and stmmac now.

     Arnd

Re: arm: shmobile_defconfig: ld.lld: error: undefined symbol: lynx_pcs_destroy

[Aithal, Srikanth]

On 6/7/2023 7:29 PM, Arnd Bergmann wrote:
> On Wed, Jun 7, 2023, at 15:27, Aithal, Srikanth wrote:
>> On 6/6/2023 2:31 PM, Geert Uytterhoeven wrote:
>>> On Tue, Jun 6, 2023 at 10:53 AM Naresh Kamboju
>>> <naresh.kamboju@linaro.org> wrote:
>>>> On Tue, 6 Jun 2023 at 14:17, Naresh Kamboju <naresh.kamboju@linaro.org> wrote:
>>>>> Following build regressions found while building arm shmobile_defconfig on
>>>>> Linux next-20230606.
>>>>>
>>>>> Regressions found on arm:
>>>>>
>>>>>    - build/clang-16-shmobile_defconfig
>>>>>    - build/gcc-8-shmobile_defconfig
>>>>>    - build/gcc-12-shmobile_defconfig
>>>>>    - build/clang-nightly-shmobile_defconfig
>>>>
>>>> And mips defconfig builds failed.
>>>> Regressions found on mips:
>>>>
>>>>     - build/clang-16-defconfig
>>>>     - build/gcc-12-defconfig
>>>>     - build/gcc-8-defconfig
>>>>     - build/clang-nightly-defconfig
>>>
>>> Please give my fix a try:
>>> https://lore.kernel.org/linux-renesas-soc/7b36ac43778b41831debd5c30b5b37d268512195.1686039915.git.geert+renesas@glider.be
>> On x86 as well seeing couple of issues related to same, not on defconfig
>> though..
>>
>> ERROR: modpost: "lynx_pcs_destroy"
>> [drivers/net/ethernet/stmicro/stmmac/stmmac.ko] undefined!
>> ERROR: modpost: "lynx_pcs_destroy"
>> [drivers/net/ethernet/altera/altera_tse.ko] undefined!
>> make[1]: *** [scripts/Makefile.modpost:136: Module.symvers] Error 1
>> make: *** [Makefile:1984: modpost] Error 2
>>
>> Among above issues stmmac issue would be resolved with above mentioned fix.
> 
> I sent out my version of the build fixups for altera and stmmac now.
Thanks, tested with next-20230608 and it builds fine.
Tested-by: sraithal@amd.com
> 
>       Arnd