[PATCH] newlib: libc: define _ctype for AArch64 Cygwin

[Richard Earnshaw (lists)]

On 17/06/2025 15:59, Corinna Vinschen wrote:
> Hi Radek,
> 
> On Jun 12 07:46, Radek Barton wrote:
>> Hello.
>>
>> This change defines `_ctype_` symbol used in Cygwin for AArch64 the same way as for x86_64.
> 
> Thanks for the patch, looks good, but I'm wondering...
> 
> The #else branch is for compatibility with 32 bit x86, which we don't
> support anymore, so it could have long gone.  Shouldn't we just drop
> the condition and the #else branch entirely?  It would clean up this
> code snippet nicely.
> 

Or perhaps there should just be a __CYGWIN32 def that's tested for code like this.  It certainly feels 'sub-optimal' to be testing specific cpu types.

R.

> What do you think? Do you want to provide the change, perhaps?
> 
> 
> Thanks,
> Corinna
> 
> 
>> Radek
>>
>> ---
>> >From 7532c751266a766cb3e289ebd7afb2d2c3d1ff3d Mon Sep 17 00:00:00 2001
>> From: =?UTF-8?q?Radek=20Barto=C5=88?= <radek.barton@microsoft.com>
>> Date: Fri, 6 Jun 2025 16:17:27 +0200
>> Subject: [PATCH] newlib: libc: define _ctype for AArch64 Cygwin
>>
>> ---
>>  newlib/libc/ctype/ctype_.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/newlib/libc/ctype/ctype_.c b/newlib/libc/ctype/ctype_.c
>> index 32ce4f318..bf403d761 100644
>> --- a/newlib/libc/ctype/ctype_.c
>> +++ b/newlib/libc/ctype/ctype_.c
>> @@ -95,7 +95,7 @@ char _ctype_b[128 + 256] = {
>>  /* For backward compatibility */
>>  char __EXPORT *__ctype_ptr__ = DEFAULT_CTYPE_PTR;
>>  
>> -#    ifdef __x86_64__
>> +#    if defined(__x86_64__) || defined(__aarch64__)
>>  __asm__ ("					\n\
>>          .data					\n\
>>  	.globl  _ctype_				\n\
>> -- 
>> 2.49.0.vfs.0.3
>>
>>
>>           
> 
>