Re: [PATCH] kernel: Assembly: need add prefix for h8300 architecture.

[Chen Gang]

On 08/27/2013 02:22 PM, Chen Gang wrote:
> On 08/27/2013 01:12 PM, Rusty Russell wrote:
>> Chen Gang <gang.chen@xxxxxxxxxxx> writes:
>>> If this implementation is correct, the implementation still can be
>>> improved (at least, it is not quite well to hard code the architecture
>>> related code in "kernel/" and "usr/" sub-directory).
>>>
>>> But it seems, I have no right to add or move files, so welcome any
>>> members help do it.
>>>
>>> And welcome any additional suggestions or completions.
>>
>> Hmm, how about something like:
>>
>> #ifdef CONFIG_ARCH_ASM_INCBIN_NEEDS_PREFIX
>> #include <asm/incbin_prefix.h>
>>         ASM_INCBIN_PREFIX
>> #endif /* ARCH_ASM_INCBIN_NEEDS_PREFIX */
>>
>> Then h8300 can define ARCH_ASM_INCBIN_NEEDS_PREFIX in Kconfig and define
>> ASM_INCBIN_PREFIX in arch/h8300/include/asm/incbin_prefix.h?
>>
> 
> Thank you for your reply firstly, I think your way is OK (valuable to
> continue think of the details implementation).
> 

The final result is: we have to use this way (mark 'h8300h' explicitly)
to fix this issue. (build-in.o requires "h8300h", but the default value
of the ".S" files are "h8300").

"-no-warn-mismatch" and "-accept-unknown-input-arch" have no effect with
the same issue under firmware sub-directory (after reading 'ld' code, we
know there are additional stricter checking for it).

After let 'h8300' pass allmodconfig, I will send patch v3 (may need some
days for waiting gcc issues).


Welcome any additional suggestions or completions (especially for
modifying the firmware files, I have no experience for it).


Thanks.


> And excuse me, I continued replying multiple times for this thread (I
> am just continue improving it).
> 
> Current status:
> 
>   after check the binutils-2.22 source code, we know if use "-no-warn-mismatch" for 'ld', it can avoid this issue.
> 
>   and also for another 'ld' issue (which I occurs but not send mail), need use "-accept-unknown-input-arch" for 'ld':
>     "Relocatable linking with relocations from format elf32-h8300 (firmware/keyspan/mpr.fw.gen.o) to format elf32-h8300 (firmware/built-in.o) is not supported"
> 
>   ... maybe find more issues during compiling with 'allmodconfig'.
> 
> 
> My opinion:
> 
>   if h8300 is still one of important architecture, we need try to "let our binary match 'ld'"
>   else just use skipping parameters is OK.
> 
>   In my current feeling, h8300 is not one of important architecture, so I choose 2nd.
> 
> 
> Welcome additional suggestion or completions.
> 
> 
> Thanks.
> 
> 
>> Cheers,
>> Rusty.
>>
>>> On 08/26/2013 06:42 PM, Chen Gang wrote:
>>>> Need add related prefix for h8300, or can not pass compiling by the
>>>> latest linker.
>>>>
>>>> The related error (allmodconfig for h8300):
>>>>
>>>>   /usr/local/bin/h8300-gchen-elf-ld: h8300 architecture of input file `usr/initramfs_data.o' is incompatible with h8300h output
>>>>   /usr/local/bin/h8300-gchen-elf-ld: h8300 architecture of input file `kernel/modsign_certificate.o' is incompatible with h8300h output
>>>>
>>>> The cross-compiler's information:
>>>>
>>>>   [root@dhcp122 linux-next]# /usr/local/bin/h8300-gchen-elf-ld -v
>>>>   GNU ld (GNU Binutils) 2.22
>>>>   [root@dhcp122 linux-next]# /usr/local/bin/h8300-gchen-elf-as -v
>>>>   GNU assembler version 2.22 (h8300-gchen-elf) using BFD version (GNU Binutils) 2.22
>>>>   
>>>>   [root@dhcp122 linux-next]# /usr/local/bin/h8300-gchen-elf-gcc -v
>>>>   Using built-in specs.
>>>>   COLLECT_GCC=/usr/local/bin/h8300-gchen-elf-gcc
>>>>   COLLECT_LTO_WRAPPER=/usr/local/libexec/gcc/h8300-gchen-elf/4.8.0/lto-wrapper
>>>>   Target: h8300-gchen-elf
>>>>   Configured with: ../gcc-4.8.0/configure --target=h8300-gchen-elf --without-header --disable-nls --enable-language=c --disable-threads --disable-shared --enable-werror=no
>>>>   Thread model: single
>>>>   gcc version 4.8.0 (GCC) 
>>>>
>>>>
>>>> Signed-off-by: Chen Gang <gang.chen@xxxxxxxxxxx>
>>>> ---
>>>>  kernel/modsign_certificate.S |    7 +++++++
>>>>  usr/initramfs_data.S         |    7 +++++++
>>>>  2 files changed, 14 insertions(+), 0 deletions(-)
>>>>
>>>> diff --git a/kernel/modsign_certificate.S b/kernel/modsign_certificate.S
>>>> index 4a9a86d..d24acac 100644
>>>> --- a/kernel/modsign_certificate.S
>>>> +++ b/kernel/modsign_certificate.S
>>>> @@ -1,5 +1,12 @@
>>>>  #include <linux/export.h>
>>>>  
>>>> +#if defined(CONFIG_CPU_H8300H)
>>>> +	.h8300h
>>>> +#endif
>>>> +#if defined(CONFIG_CPU_H8S)
>>>> +	.h8300s
>>>> +#endif
>>>> +
>>>>  #define GLOBAL(name)	\
>>>>  	.globl VMLINUX_SYMBOL(name);	\
>>>>  	VMLINUX_SYMBOL(name):
>>>> diff --git a/usr/initramfs_data.S b/usr/initramfs_data.S
>>>> index c14322d..6a6c106 100644
>>>> --- a/usr/initramfs_data.S
>>>> +++ b/usr/initramfs_data.S
>>>> @@ -24,6 +24,13 @@
>>>>  #include <linux/stringify.h>
>>>>  #include <asm-generic/vmlinux.lds.h>
>>>>  
>>>> +#if defined(CONFIG_CPU_H8300H)
>>>> +	.h8300h
>>>> +#endif
>>>> +#if defined(CONFIG_CPU_H8S)
>>>> +	.h8300s
>>>> +#endif
>>>> +
>>>>  .section .init.ramfs,"a"
>>>>  __irf_start:
>>>>  .incbin __stringify(INITRAMFS_IMAGE)
>>>>
>>>
>>>
>>> -- 
>>> Chen Gang
>>
>>
> 
> 


-- 
Chen Gang
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/