From: Lukasz Majewski <lukma@denx.de>
To: Paul Eggert <eggert@cs.ucla.edu>
Cc: libc-alpha@sourceware.org, Joseph Myers <joseph@codesourcery.com>
Subject: Re: [PATCH v2 1/2] Y2038: make __mktime_internal compatible with __time64_t
Date: Tue, 19 Mar 2019 14:39:56 +0100 [thread overview]
Message-ID: <20190319143956.52f83a48@jawa> (raw)
In-Reply-To: <bf6ca98d-da93-0677-99d3-86c80e5470a2@cs.ucla.edu>
[-- Attachment #1: Type: text/plain, Size: 3346 bytes --]
Hi Paul,
> On 3/11/19 11:58 PM, Lukasz Majewski wrote:
> > The question is if there is a more suitable place than
> > include/time.h header to have fits_in_time() reused by both Glibc
> > and Gnulib?
>
Thanks for your patch.
> Hmm, well, on further thought, fits_in_time_t can live in
> include/time.h since only glibc will need it. Gnulib won't support
> two kinds of time_t types, so it should't need the function. (The
> attached patch renames it to "in_time_t_range" to avoid *_t
> pollution.)
Is the rewritten code correct?
+/* Check whether T fits in time_t. */
+static inline bool
+in_time_t_range (__time64_t t)
+{
+ time_t s = t;
+ return s != t;
+}
Shouldn't we have: return s == t; ?
>
> That being said, we will have to make a place for private .h code
> shared between glibc and Gnulib, because include/time.h can't easily
> be shared with Gnulib. I suggest using time/mktime-internal.h for this
>
ACK/NAK for using time/mktime-internl.h shall be done by more
experienced glibc developer(s).
> Some other comments on that patch that I noticed while looking into
> this.
>
> The patch added duplicate "#include <errno.h>" lines to time/timegm.c.
Ok.
>
> I still don't get why we need __timelocal64. Glibc code can use
> __mktime64. User code shouldn't see that symbol.
If we don't need it for this conversion, then we should omit it. I will
add it if needed when preparing Y2038 patch set.
In the time/mktime.c there is:
weak_alias (mktime, timelocal), which makes the timelocal calls
aliases to mktime for time_t 32 and 64 bit (for Y2038 the proper
__REDIRECT will be added).
>
> Come to think of it, user code shouldn't see __time64_t either. Yes,
> the name begins with two underscores so user code should avoid it,
> posix/bits/types.hbut putting __time64_t in /usr/include will just
> tempt users. It's easy to keep __time64_t out of /usr/include so
> let's do that.
>
Is that the reason for removing __time64_t definition from
posix/bits/types.h ?
> The body of __mktime64 can be simplified; there's no need for locals
> of type __time64_t, time_t, and struct tm. And it should use
> in_time_t_range instead of doing it by hand.
Yes. Also there is no need to check for EOVERFLOW in the __mktime64()
function (as it operates inherently on 64 bit types).
>
> The bodies of mktime and timegm [__TIMESIZE != 64] should not pass tp
> to __mktime64/__timegm64 directly, since *tp should not be updated if
> the result is in __time64_t range but out of time_t range. And timegm
> should use in_time_t_range instead of doing it by hand.
>
Ok.
> The "#ifdef weak_alias" etc. lines can be removed now, since Gnulib
> does this stuff OK now.
>
> timegm.c should include libc-config.h, not config.h, as this is the
> current Gnulib style for code shared with Gnulib.
>
> Revised proposed patch attached.
>
Just out of curiosity:
In the time/mktime-internal.h you added a comment regarding BeOS users
and posix time_t - do you know any :-) ?
Best regards,
Lukasz Majewski
--
DENX Software Engineering GmbH, Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-59 Fax: (+49)-8142-66989-80 Email: lukma@denx.de
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]
next prev parent reply other threads:[~2019-03-19 13:40 UTC|newest]
Thread overview: 31+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-02-27 11:20 [PATCH v2 1/2] Y2038: make __mktime_internal compatible with __time64_t Lukasz Majewski
2019-02-27 11:20 ` [PATCH v2 2/2] Fix time/mktime.c and time/gmtime.c for gnulib compatibility Lukasz Majewski
2019-03-12 0:12 ` Paul Eggert
2019-03-06 11:31 ` [PATCH v2 1/2] Y2038: make __mktime_internal compatible with __time64_t Lukasz Majewski
2019-03-12 0:11 ` Paul Eggert
2019-03-12 0:36 ` Joseph Myers
2019-03-17 22:48 ` Lukasz Majewski
2019-03-18 16:27 ` Joseph Myers
2019-03-19 10:53 ` Lukasz Majewski
2019-03-12 6:58 ` Lukasz Majewski
2019-03-18 21:23 ` Paul Eggert
2019-03-19 13:39 ` Lukasz Majewski [this message]
2019-03-19 23:12 ` Paul Eggert
2019-03-20 7:03 ` Lukasz Majewski
2019-03-22 21:49 ` Paul Eggert
2019-03-23 21:34 ` Lukasz Majewski
2019-03-24 22:17 ` Lukasz Majewski
2019-03-23 11:59 ` Lukasz Majewski
2019-03-27 20:06 ` Paul Eggert
2019-03-28 8:59 ` Lukasz Majewski
2019-03-28 16:09 ` Paul Eggert
2019-03-29 14:24 ` Lukasz Majewski
2019-03-29 21:10 ` Paul Eggert
2019-03-30 14:39 ` Lukasz Majewski
2019-04-01 20:17 ` Joseph Myers
2019-04-01 20:51 ` Lukasz Majewski
2019-03-28 16:34 ` Joseph Myers
[not found] ` <20190404120715.150a5d44@jawa>
[not found] ` <20190424135748.502c34af@jawa>
2019-04-28 22:45 ` Paul Eggert
2019-04-30 7:59 ` Lukasz Majewski
2019-04-30 16:25 ` Paul Eggert
2019-05-02 7:19 ` Lukasz Majewski
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=20190319143956.52f83a48@jawa \
--to=lukma@denx.de \
--cc=eggert@cs.ucla.edu \
--cc=joseph@codesourcery.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).