git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: [PATCH v3 1/1] vreportf(): avoid relying on stdio buffering

From
Junio C Hamano <gitster@pobox.com>
Date
Oct 30, 2019, 02:01 UTC
Message-ID
<xmqqo8xz3sa2.fsf@gitster-ct.c.googlers.com>
In-Reply-To
<fce0894ee46f737322551b51edf5cd2a53413a50.1572379280.git.gitgitgadget@gmail.com>

"Johannes Schindelin via GitGitGadget" <gitgitgadget@gmail.com> writes:

> Subject: Re: [PATCH v3 1/1] vreportf(): avoid relying on stdio buffering
perfect ;-)
Show 6 quoted lines
> Let's avoid this predicament altogether by rendering the entire message,
> including the prefix and the trailing newline, into the buffer we
> already have (and which is still fixed size) and then write it out via
> `write_in_full()`.
>
> We still clip the message to at most 4095 characters.
Good.
> Also please note that we `fflush(stderr)` here to help when running in a
> Git Bash on Windows: in this case, `stderr` is not actually truly
> unbuffered, and needs the extra help.

I do not think you have to single out Windows in this case; if we are writing directly to file descriptor #2, as long as there have been something written by the process to the standard error stream before our caller called us, we must flush it to make sure what we are going to write out will come after, no matter what platform we are on.

    The process may have written to the standard error stream and
    there may be something left in the buffer kept in the stdio
    layer.  Call fflush(stderr) before writing the message we
    prepare in this function.
or something along that line would be sufficient.
Show 19 quoted lines
> diff --git a/usage.c b/usage.c
> index 2fdb20086b..471efb2de9 100644
> --- a/usage.c
> +++ b/usage.c
> @@ -9,14 +9,21 @@
>  void vreportf(const char *prefix, const char *err, va_list params)
>  {
>  	char msg[4096];
> -	char *p;
> +	size_t off = strlcpy(msg, prefix, sizeof(msg));
> +	char *p, *pend = msg + sizeof(msg);
>  
> -	vsnprintf(msg, sizeof(msg), err, params);
> -	for (p = msg; *p; p++) {
> +	p = off < pend - msg ? msg + off : pend - 1;
> +	if (vsnprintf(p, pend - p, err, params) < 0)
> +		*p = '\0'; /* vsnprintf() failed, clip at prefix */
> +
> +	for (; p != pend - 1 && *p; p++) {
It may make the result much simpler to start with something like:
	size_t prefix_len = strlen(prefix);
	if (sizeof(msg) <= prefix_len) {
 		fprintf(stderr, "BUG!!! too long a prefix '%s'\n", prefix);
		abort();
	}
	memcpy(msg, prefix, prefix_len);
	p = msg + prefix_len;

as we agreed that we won't have prefix that will fill or overflow msg[] based on your earlier reading of the callers like BUG-fl. That way, we probably may be able to even lose the pend pointer and arithmetic around it.

>  		if (iscntrl(*p) && *p != '\t' && *p != '\n')
>  			*p = '?';
>  	}

Thanks. With this flow it is crystal clear that the prefix is shown as-is, while the payload is sanitized.

Show 8 quoted lines
> -	fprintf(stderr, "%s%s\n", prefix, msg);
> +
> +	*(p++) = '\n'; /* we no longer need a NUL */
> +	fflush(stderr);
> +	write_in_full(2, msg, p - msg);
>  }
>  
>  static NORETURN void usage_builtin(const char *err, va_list params)
Previous: Johannes SchindelinNext: Johannes Schindelin
Message 30 of 34 in “Fix t5516 flakiness in Visual Studio builds”
  1. 0/1 Fix t5516 flakiness in Visual Studio buildsJohannes Schindelin via GitGitGadget, Oct 28, 2019
  2. 1/1 vreportf(): avoid buffered write in favor of unbuffered oneJohannes Schindelin via GitGitGadget, Oct 28, 2019
  3. Junio C HamanoOct 29, 2019
  4. Johannes SchindelinOct 29, 2019
  5. Jeff KingOct 29, 2019
  6. Johannes SchindelinOct 29, 2019
  7. Jeff KingOct 29, 2019
  8. Johannes SchindelinOct 29, 2019
  9. Junio C HamanoOct 30, 2019
  10. Junio C HamanoOct 29, 2019
  11. SZEDER GáborOct 29, 2019
  12. Johannes SchindelinOct 29, 2019
  13. Jeff KingOct 29, 2019
  14. Johannes SchindelinOct 29, 2019
  15. 0/1 Fix t5516 flakiness in Visual Studio buildsJohannes Schindelin via GitGitGadget, Oct 29, 2019
  16. 1/1 vreportf(): avoid relying on stdio bufferingJohannes Schindelin via GitGitGadget, Oct 29, 2019
  17. Alexandr MiloslavskiyOct 29, 2019
  18. Johannes SchindelinOct 29, 2019
  19. Jeff KingOct 29, 2019
  20. Alexandr MiloslavskiyOct 29, 2019
  21. Alexandr MiloslavskiyOct 29, 2019
  22. 0/1 Fix t5516 flakiness in Visual Studio buildsJohannes Schindelin via GitGitGadget, Oct 29, 2019
  23. 1/1 vreportf(): avoid relying on stdio bufferingJohannes Schindelin via GitGitGadget, Oct 29, 2019
  24. Jeff KingOct 29, 2019
  25. Johannes SchindelinOct 30, 2019
  26. Jeff KingOct 31, 2019
  27. Johannes SchindelinOct 31, 2019
  28. Jeff KingOct 31, 2019
  29. Johannes SchindelinNov 1, 2019
  30. Junio C HamanoOct 30, 2019
  31. Johannes SchindelinOct 30, 2019
  32. 0/1 Fix t5516 flakiness in Visual Studio buildsJohannes Schindelin via GitGitGadget, Oct 30, 2019
  33. 1/1 vreportf(): avoid relying on stdio bufferingJohannes Schindelin via GitGitGadget, Oct 30, 2019
  34. Junio C HamanoNov 2, 2019

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.