[PATCH 1/4] Respect that some compression algos can be enabled separately for SPL

[Stefano Babic]

Hi everybody,

On 27.01.21 14:19, Stefano Babic wrote:
> On 27.01.21 08:57, Frieder Schrempf wrote:
>> On 26.01.21 18:53, Tim Harvey wrote:
>>> On Sat, Jan 23, 2021 at 4:39 AM Stefano Babic <sbabic at denx.de> wrote:
>>>>
>>>> Hi Tim,
>>>>
>>>> there is a weird side effect with this patch, breaking m68k
>>>> architecture. For some reason, image.c is not compiled clean because gd
>>>> is not declared anymore.
>>>>
>>>> In file included from tools/common/image.c:1:
>>>> ./tools/../common/image.c: In function ‘get_table_entry_id’:
>>>> ./tools/../common/image.c:982:41: error: ‘gd’ undeclared (first use in
>>>> this function)
>>>>     982 |   if (t->sname && strcasecmp(t->sname + gd->reloc_off, name)
>>>> == 0)
>>>>         |                                         ^~
>>>> ./tools/../common/image.c:982:41: note: each undeclared identifier is
>>>> reported only once for each functi
>>>>
>>>> Can you take a look ? Thanks !
>>>>
>>>
>>> Stefano,
>>>
>>> I have no idea what could cause this... must be something to do with
>>> including kconfig.h?
>>>
>>> What board/target would I build to reproduce this?
>>>
>>> Frieder, do you have any ideas?
>>
>> I can't remember if I did a full test on all platforms when I wrote this
>> patch. Probably not.
>>
>> And I have no idea what is causing this, nor do I currently have the
>> time to have a closer look, sorry.
>>
>> BTW, this is a very good example for what I said earlier:
>>
>> "I failed to do upstreaming work for U-Boot for quite some time. Mainly
>> because whenever I try to, I end up fixing/cleaning some weird U-Boot
>> code/behavior and spending way more time than I actually have available."
> 
> True, but...we cannot even break other boards, too. I have also get a
> short look at the issue, but I could not find myself the reason, too. I
> plan to get a deeper look in the week end.

It looks like that only for m68k the following statement does not work

#if CONFIG_IS_ENABLED(GZIP) && !defined(USE_HOSTCC)

But if I replace this with the equivalent one:

+#ifndef USE_HOSTCC
+#if CONFIG_IS_ENABLED(GZIP)

This works - someone else should have discovered this before, because in 
common/image.c USE_HOSTCC is never anded with another CONFIG.

I have not found why this happens just for m68k, no issue with 
arm/aarch64 or PowerPC. I send a V2 with the changes like above.

@Tim: why do we need include/kconfig ?

Best regards,
Stefano


-- 
=====================================================================
DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: +49-8142-66989-53 Fax: +49-8142-66989-80 Email: sbabic at denx.de
=====================================================================