From: Corinna Vinschen <vinschen@redhat.com>
To: newlib@sourceware.org
Subject: Re: Broken getchar() and putchar() macros?
Date: Wed, 02 Aug 2017 12:37:00 -0000 [thread overview]
Message-ID: <20170802123722.GI25551@calimero.vinschen.de> (raw)
In-Reply-To: <80e430fc-e931-d9a0-173e-8a66e2a494b6@embedded-brains.de>
[-- Attachment #1: Type: text/plain, Size: 2579 bytes --]
On Aug 2 13:36, Sebastian Huber wrote:
> Hello,
>
> we have via <stdio.h>
>
> #define getchar() getc(stdin)
> #define putchar(x) putc(x, stdout)
>
> which is
>
> #ifndef __CYGWIN__
> #ifndef lint
> #define getc(fp) __sgetc_r(_REENT, fp)
> #define putc(x, fp) __sputc_r(_REENT, x, fp)
> #endif /* lint */
> #endif /* __CYGWIN__ */
> #endif /* __cplusplus */
>
> which is mostly
>
> #define __sgetc_r(__ptr, __p) __sgetc_raw_r(__ptr, __p)
>
> #define __sputc_r(__ptr, __c, __p) __sputc_raw_r(__ptr, __c, __p)
>
> which is
>
> /*
> * The __sfoo macros are here so that we can
> * define function versions in the C library.
> */
> #define __sgetc_raw_r(__ptr, __f) (--(__f)->_r < 0 ? __srget_r(__ptr,
> __f) : (int)(*(__f)->_p++))
>
> _ELIDABLE_INLINE int __sputc_r(struct _reent *_ptr, int _c, FILE *_p) {
> #ifdef __SCLE
> if ((_p->_flags & __SCLE) && _c == '\n')
> __sputc_r (_ptr, '\r', _p);
> #endif
> if (--_p->_w >= 0 || (_p->_w >= _p->_lbfsize && (char)_c != '\n'))
> return (*_p->_p++ = _c);
> else
> return (__swbuf_r(_ptr, _c, _p));
> }
>
> which modifies FILE members without protection of the FILE lock!
> [...]
> What is the purpose of these macros?
If you search for them in newlib, you'll notice that the macros are used
from a couple of places in libc/stdio.
The fact that they are directly called on non-Cygwin platforms is probably
historical, as usual. They are reentrant. As such, and under the
assumption that we're on embedded platforms without preemptive
multi-tasking, the macros work nicely and are faster than the function
calls.
I can think of four ways to solve this:
1. Just add RTEMS to the exceptions, same as Cygwin.
2. Replace
#define getc(fp) __sgetc_r(_REENT, fp)
#define putc(x, fp) __sputc_r(_REENT, x, fp)
with
#define getc_unlocked(fp) __sgetc_r(_REENT, fp)
#define putc_unlocked(x, fp) __sputc_r(_REENT, x, fp)
and let getc/putc simply call the functions.
3. Add a newlib setting to sys/config.h and use that in stdio.h, rather
than excluding targets explicitely. Best is probably an opt-in
config.
4. Move the macros to libc/stdio/local.h and don't use them from
user space at all. This may be a bit harsh on some embedded
targets, but actually I don't know. Most targets have space
constraints and a function would be more prudent then, I guess.
Corinna
--
Corinna Vinschen
Cygwin Maintainer
Red Hat
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]
prev parent reply other threads:[~2017-08-02 12:37 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-08-02 11:36 Sebastian Huber
2017-08-02 12:37 ` Corinna Vinschen [this message]
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
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20170802123722.GI25551@calimero.vinschen.de \
--to=vinschen@redhat.com \
--cc=newlib@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).