From: Adhemerval Zanella Netto <adhemerval.zanella@linaro.org>
To: DJ Delorie <dj@redhat.com>, Andreas Schwab <schwab@suse.de>
Cc: libc-alpha@sourceware.org
Subject: Re: [PATCH] Discard any push-back from a memory stream before finishing it (bug 34020)
Date: Thu, 2 Apr 2026 16:12:58 -0300 [thread overview]
Message-ID: <bb4ee3c2-5acf-43c7-b408-0ce1397d2fff@linaro.org> (raw)
In-Reply-To: <xnse9h87av.fsf@greed.delorie.com>
On 30/03/26 16:07, DJ Delorie wrote:
> Andreas Schwab <schwab@suse.de> writes:
>> This makes sure that the write base is pointing to the main area, not the
>> backup area.
>
> So I just realized, after working in this code for years, that "backup"
> here doesn't mean "in case we run out" it means "going backwards" :-P No
> wonder it seemed weird to me.
>
> One question below about fseek, but otherwise OK.
The new testcase triggers some regression on both redhad and Linaro CI [1],
and it seems a real one.
[1] https://patchwork.sourceware.org/project/glibc/patch/mvm7bqtiivh.fsf@suse.de/
>
>> diff --git a/libio/memstream.c b/libio/memstream.c
>> index 0456adb92f..fcc925df87 100644
>> --- a/libio/memstream.c
>> +++ b/libio/memstream.c
>> @@ -100,6 +100,9 @@ _IO_mem_finish (FILE *fp, int dummy)
>> {
>> struct _IO_FILE_memstream *mp = (struct _IO_FILE_memstream *) fp;
>>
>> + if (_IO_in_backup (fp))
>> + _IO_switch_to_main_get_area (fp);
>> +
>> *mp->bufloc = (char *) realloc (fp->_IO_write_base,
>> fp->_IO_write_ptr - fp->_IO_write_base + 1);
>> if (*mp->bufloc != NULL)
>
> Ok.
>
>> diff --git a/libio/tst-memstream6.c b/libio/tst-memstream6.c
>> +static void
>> +mcheck_abort (enum mcheck_status ev)
>> +{
>> + printf ("mecheck failed with status %d\n", (int) ev);
>> + exit (1);
>> +}
>
> Ok.
>
>> +void
>> +do_test_ungetc_1 (void)
>> +{
>> + CHAR_T *buf;
>> + size_t size;
>> +
>> + FILE *fp = OPEN_MEMSTREAM (&buf, &size);
>> + TEST_VERIFY_EXIT (fp != NULL);
>> +
>> + FPUTC (W('A'), fp);
>> + fflush (fp);
>> +
>> + TEST_COMPARE (FGETC (fp), W('A'));
>
> I would have expected you to need an fseek here. fflush() on a
> memstream synchronizes the buffer and size pointers but says nothing
> about moving the file pointer.
>
>> + TEST_COMPARE (UNGETC (W('B'), fp), W('B'));
>> +
>> + TEST_COMPARE (fclose (fp), 0);
>> +
>> + TEST_COMPARE (buf[0], W('A'));
>> +
>> + free (buf);
>> +}
>
> Ok.
>
>> +void
>> +do_test_ungetc_2 (void)
>> +{
>> + CHAR_T *buf;
>> + size_t size;
>> +
>> + FILE *fp = OPEN_MEMSTREAM (&buf, &size);
>> + TEST_VERIFY_EXIT (fp != NULL);
>> +
>> + TEST_COMPARE (UNGETC (W('A'), fp), W('A'));
>> +
>> + TEST_COMPARE (fclose (fp), 0);
>> +
>> + TEST_COMPARE (buf[0], 0);
>> +
>> + free (buf);
>> +}
>
> Ok.
>
>> +static int
>> +do_test (void)
>> +{
>> + mcheck_pedantic (mcheck_abort);
>> +
>> + do_test_ungetc_1 ();
>> + do_test_ungetc_2 ();
>> +
>> + return 0;
>> +}
>
> Ok.
>
>> diff --git a/libio/tst-wmemstream6.c b/libio/tst-wmemstream6.c
>> +#define TEST_WCHAR
>> +#include <libio/tst-memstream6.c>
>
> Ok.
>
>> diff --git a/libio/wmemstream.c b/libio/wmemstream.c
>> index d0c639be70..a0c61a627f 100644
>> --- a/libio/wmemstream.c
>> +++ b/libio/wmemstream.c
>> @@ -101,6 +101,9 @@ _IO_wmem_finish (FILE *fp, int dummy)
>> {
>> struct _IO_FILE_wmemstream *mp = (struct _IO_FILE_wmemstream *) fp;
>>
>> + if (_IO_in_backup (fp))
>> + _IO_switch_to_main_wget_area (fp);
>> +
>> *mp->bufloc = (wchar_t *) realloc (fp->_wide_data->_IO_write_base,
>> (fp->_wide_data->_IO_write_ptr
>> - fp->_wide_data->_IO_write_base + 1)
>
> Ok.
>
next prev parent reply other threads:[~2026-04-02 19:13 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-03-30 12:47 Andreas Schwab
2026-03-30 19:07 ` DJ Delorie
2026-04-02 19:12 ` Adhemerval Zanella Netto [this message]
2026-04-07 10:26 ` Andreas Schwab
2026-04-07 16:14 ` 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
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=bb4ee3c2-5acf-43c7-b408-0ce1397d2fff@linaro.org \
--to=adhemerval.zanella@linaro.org \
--cc=dj@redhat.com \
--cc=libc-alpha@sourceware.org \
--cc=schwab@suse.de \
/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).