[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20220209182455.GA7074@brightrain.aerifal.cx>
Date: Wed, 9 Feb 2022 13:24:55 -0500
From: Rich Felker <dalias@...c.org>
To: Pinghao Wu <xdavidwuph@...il.com>
Cc: musl@...ts.openwall.com
Subject: Re: [PATCH] fgetwc: fix errno when character span aver buffer
end
On Thu, Feb 10, 2022 at 01:55:57AM +0800, Pinghao Wu wrote:
> If attempt on converting character from buffer failed, errno is set to
> EILSEQ by mbtowc. As a result, if further byte-by-byte conversion
> succeeds, fgetwc will return a valid wchar with a misleading EILSEQ as
> errno. This fixes it by saving errno before the from buffer attempt, and
> restore if it fails.
>
> This also fixes fgetws which find the misleading EILSEQ and fails in
> this case.
errno is only meaningful after failure. A program which is checking it
when the last function called did not fail is incorrect and needs to
be fixed.
> ---
> src/stdio/fgetwc.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/src/stdio/fgetwc.c b/src/stdio/fgetwc.c
> index aa10b818..9dad0505 100644
> --- a/src/stdio/fgetwc.c
> +++ b/src/stdio/fgetwc.c
> @@ -8,6 +8,7 @@ static wint_t __fgetwc_unlocked_internal(FILE *f)
> wchar_t wc;
> int c;
> size_t l;
> + int errno_save = errno;
>
> /* Convert character from buffer if possible */
> if (f->rpos != f->rend) {
> @@ -16,6 +17,7 @@ static wint_t __fgetwc_unlocked_internal(FILE *f)
> f->rpos += l + !l; /* l==0 means 1 byte, null */
> return wc;
> }
> + errno = errno_save;
> }
>
> /* Convert character byte-by-byte */
> --
> 2.35.1
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
