Hello Peng, On Tue, Mar 12, 2013 at 11:43 AM, Peng Haitao <penght@xxxxxxxxxxxxxx> wrote: > The function localeconv() returns a pointer to a structure which might be > overwritten by subsequent calls to localeconv() or by calls to setlocale(), > so it is not thread-safe. I've applied this patch in my local branch for the ATTRIBUTES work. See also my general comments in the reply to the ecvt.3 patch. Here we have a page that could do with other improvements also, and it is probably worth making such improvements as we go. What I mean is: the page make no other mention that the returned pointer may (okay, in glibc *does*) point to a statically allocated object that is overwritten by subsequent calls. If you notice cases such as this, you might like to add an extra piece to your patch, such as I have doe in the patch below (I added some other details as well that I wouldn't expect you to add), or you could alert me, and maybe I will add a suitable patch. Thanks for the patch. Cheers, Michael diff --git a/man3/localeconv.3 b/man3/localeconv.3 index 80a2a21..b449211 100644 --- a/man3/localeconv.3 +++ b/man3/localeconv.3 @@ -49,6 +49,19 @@ Programs may also use the functions and .BR strfmon (3), which behave according to the actual locale in use. +.SH RETURN VALUE +The +.BR localeconv () +function returns a pointer to a filled in +.IR "struct lconv" . +This structure may be (in glibc, +.IR is ) +statically allocated, and may be overwritten by subsequent calls. +According to POSIX, +the caller should not modify the contents of this structure. +The +.BR localeconv () +function always succeeds. .SH ATTRIBUTES .SS Multithreading (see pthreads(7)) The -- To unsubscribe from this list: send the line "unsubscribe linux-man" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
