meaning of supports_gnu_unique

[Jan Beulich]

On 02.04.2026 12:49, Alan Modra wrote:
> On Wed, Apr 01, 2026 at 10:42:27AM +0200, Jan Beulich wrote:
>> On 01.04.2026 10:17, Alan Modra wrote:
>>> On Tue, Mar 31, 2026 at 03:22:54PM +0200, Jan Beulich wrote:
>>>> testsuite: supports_gnu_osabi vs Arm
>>>>
>>>> Arm is yet more special than the function presently expresses: Even
>>>> targets with "gnu" in the name (e.g. arm-unknown-linux-gnu) aren't
>>>> ELFOSABI_GNU. Handle Arm first, and drop other Arm special casing in
>>>> exchange.
>>>>
>>>> --- a/binutils/testsuite/lib/binutils-common.exp
>>>> +++ b/binutils/testsuite/lib/binutils-common.exp
>>>> @@ -234,11 +234,17 @@ proc match_target { target } {
>>>>  # testcases will generally need to exclude tic6x or use a -m option.)
>>>>  #
>>>>  proc supports_gnu_osabi {} {
>>>> +    if { [istarget arm*-*-*] } {
>>>> +	if { [istarget arm*-*-*eabi*] } {
>>>> +	    return 1
>>>> +	}
>>>> +	return 0
>>>> +    }
>>>>      if { [istarget *-*-gnu*]
>>>>  	 || [istarget *-*-linux*]
>>>> -	 || ( [istarget *-*-*bsd*] && ![istarget arm*-*-netbsd*] )
>>>> +	 || [istarget *-*-*bsd*]
>>>>  	 || [istarget *-*-lynxos]
>>>> -	 || ( [istarget *-*-nto*] && ![istarget arm*-*-*] )
>>>> +	 || [istarget *-*-nto*]
>>>>  	 || [istarget *-*-irix*]
>>>>  	 || [istarget *-*-*eabi*]
>>>>  	 || [istarget *-*-rtems*] } {
>>>> @@ -250,8 +256,7 @@ proc supports_gnu_osabi {} {
>>>>      if { ![istarget "*-*-elf*"] } {
>>>>  	return 0
>>>>      }
>>>> -    if { [istarget "arm*-*-*"]
>>>> -	 || [istarget "msp430-*-*"]
>>>> +    if { [istarget "msp430-*-*"]
>>>>  	 || [istarget "hppa-unknown-elf"]
>>>>  	 || [istarget "kvx*-*-*"]
>>>>  	 || [istarget "visium-*-*"] } {
>>>
>>> Going by what is in gas/configure.tgt, I think other arm targets
>>> should be included, specifically freebsd and rtems.
>>
>> I was wondering, but I went from gas/config/tc-arm.c:meabi_flags handling.
>> Without a command line option, it would be different from
>> EF_ARM_EABI_UNKNOWN only when te-arm*eabi.h is in use. And
>> bfd/elf32-arm.c:elf32_arm_init_file_header() simply goes from what was put
>> there, using ELFOSABI_ARM only when the eabi version still is
>> EF_ARM_EABI_UNKNOWN.
>>
>> Hmm, then neither mine nor your ...
>>
>>> Perhaps the following?  I'm not an arm maintainer though...
>>>
>>>     remove arm-linux-gnu from supports_gnu_osabi
>>>     
>>>     arm-linux-gnueabi allows OS/ABI to be set to ELFOSABI_GNU, but plain
>>>     arm-linux-gnu like arm-elf is ELFOSABI_ARM.
>>>
>>> --- a/binutils/testsuite/lib/binutils-common.exp
>>> +++ b/binutils/testsuite/lib/binutils-common.exp
>>> @@ -234,27 +234,32 @@ proc match_target { target } {
>>>  # testcases will generally need to exclude tic6x or use a -m option.)
>>>  #
>>>  proc supports_gnu_osabi {} {
>>> -    if { [istarget *-*-gnu*]
>>> +    if { [istarget *-*-*eabi*]
>>> +	 || [istarget *-*-freebsd*]
>>
>> ... logic here looks correct. arm{,eb}-*-freebsd[89].* use te-freebsd.h,
>> only other (presumably newer) arm*-*-freebsd* use te-armfbsdeabi.h
> 
> I did see the other arm freebsd targets but I misread the glob and
> tested an arm-freebsd8 target which (unsurprisingly) behaved the same
> as arm-freebsd regarding test results, and I failed to investigate
> why.  Retesting properly with arm-freebsd8.0 show failures, along with
> messages like "as-new: GNU_RETAIN section is supported only by GNU and
> FreeBSD targets".
> 
>>> +	 || [istarget *-*-rtems*] } {
>>
>> Along with this arm-*-genode* would then also look to need including.
> 
> I missed that one.  Hmm, arm-*-genode* is not supported in BFD..
> 
> I probably should have just kept silent on your patch, which really
> was good enough.

Actually no, I'm glad you didn't remain silent.

>  I'll commit the following and we can tweak it further if necessary.

Thanks much.

Jan