unofficial mirror of libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Adhemerval Zanella Netto <adhemerval.zanella@linaro.org>
To: DJ Delorie <dj@redhat.com>
Cc: libc-alpha@sourceware.org
Subject: Re: [PATCH v4] Always define __USE_TIME_BITS64 when 64 bit time_t is used
Date: Mon, 18 Mar 2024 08:42:10 -0300	[thread overview]
Message-ID: <b374aaa6-e00c-46da-aafe-a1459576a537@linaro.org> (raw)
In-Reply-To: <xnedcb89ke.fsf@greed.delorie.com>



On 15/03/24 18:33, DJ Delorie wrote:
> 
> Fun stuff first...
> 
> Adhemerval Zanella <adhemerval.zanella@linaro.org> writes:
>> diff --git a/manual/maint.texi b/manual/maint.texi
>>  
>> +When @code{__TIMESIZE == 64} is set, @theglibc{} will also define the
>> +@code{__USE_TIME_BITS64}.  It is used by the Linux kernel ABI to set the
>> +expected @code{time_t} size used on some syscalls.
> 
> This part seems OK.
> 
>> +Also, to define whether
>> +the declarations are expanded to 64-bit @code{time_t} symbols (since not
>> +all ABIs do require such expansion), @theglibc{} will also define
>> +@code{__USE_TIME64_REDIRECTS}.
> 
> This part doesn't match the code, because it's in the __TIMESIZE==64
> section, and the code only sets it when __TIMESIZE==32:
> 
> #if defined __USE_TIME_BITS64 && __TIMESIZE == 32
> # define __USE_TIME64_REDIRECTS 1
> #endif
> 
> Perhaps you were thinking of _TIME_BITS==64, which is defined elsewhere?
> 
> _TIME_BITS is defined in creature.texi as a user-defined feature test macro...

Right, I will move it to creature.texi.  It makes more sense, since it
should be a user-visible macro (albeit in the implementation-defined
namespace).

> 
>> diff --git a/sysdeps/unix/sysv/linux/features-time64.h b/sysdeps/unix/sysv/linux/features-time64.h
> 
>>  # if _TIME_BITS == 64
>>  #  if ! defined (_FILE_OFFSET_BITS) || _FILE_OFFSET_BITS != 64
>>  #   error "_TIME_BITS=64 is allowed only with _FILE_OFFSET_BITS=64"
>> -#  elif __TIMESIZE == 32
>> -#   define __USE_TIME_BITS64	1
>>  #  endif
>> +#  define __USE_TIME_BITS64	1
> 
> If the user asks for 64-bit time, we honor it, whether __TIMESIZE is 32
> or 64.  Ok.
> 
> If the user asks for 32-bit time, we allow it, but need do nothing.  Ok.
> 
>>  # elif _TIME_BITS == 32
>>  #  if __TIMESIZE > 32
>>  #   error "_TIME_BITS=32 is not compatible with __TIMESIZE > 32"
>> @@ -34,4 +33,10 @@
>>  # else
>>  #  error Invalid _TIME_BITS value (can only be 32 or 64-bit)
>>  # endif
> 
> This is for when _TIME_BITS is not set:
> 
>> +#elif __TIMESIZE == 64
>> +# define __USE_TIME_BITS64      1
>> +#endif
> 
> So if the user does not specify _TIME_BITS, we default to 64-bit kabi
> for 64-bit time_t.  OK.
> 
>> +#if defined __USE_TIME_BITS64 && __TIMESIZE == 32
>> +# define __USE_TIME64_REDIRECTS 1
>>  #endif
> 
> If we want 64-bit time_t (for any reason) but the default time_t is 32
> bits, we need the redirected functions et al.  Ok.
> 
>> diff --git a/sysdeps/generic/features-time64.h b/sysdeps/generic/features-time64.h
>>  /* The generic configuration only support _TIME_BITS=32.  */
>> +#define __USE_TIME_BITS64       0
>> +/* The following macro defines whether the 64 time_t redirects will be
>> +   used for 64 bit aware symbols.  */
>> +#define __USE_TIME64_REDIRECTS  0
> 
> So the default for non-linux platforms is to not do any of the linux
> magic.  Ok.
> 
> And now for the mechanical part...
> 
>> diff --git a/io/bits/poll2.h b/io/bits/poll2.h
>> . . .
>> diff --git a/time/time.h b/time/time.h
> 
> They're all good :-)
> 

  reply	other threads:[~2024-03-18 11:42 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-03-15 16:57 [PATCH v4] Always define __USE_TIME_BITS64 when 64 bit time_t is used Adhemerval Zanella
2024-03-15 21:33 ` DJ Delorie
2024-03-18 11:42   ` Adhemerval Zanella Netto [this message]
2024-03-16  9:04 ` Mark Harris
2024-03-18 11:21   ` Adhemerval Zanella Netto

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://www.gnu.org/software/libc/involved.html

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=b374aaa6-e00c-46da-aafe-a1459576a537@linaro.org \
    --to=adhemerval.zanella@linaro.org \
    --cc=dj@redhat.com \
    --cc=libc-alpha@sourceware.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).